feat: Module 11, Conditions (NOUN ON NOUN IS ..., checked per object at use; NEAR as the Expansion Drop)

Adds course/11-conditions: lesson, scenarios, nine test_11 tests in both
tracks, the cumulative reference (Module 10's plus conditions) and the NEAR
drop. The engine learns a condition role, rules carry conditions, and
Level.hasProperty and conversion evaluate content's condition checks against
the board whenever a rule is used. Resolution pairs only rules with the same
conditions. Catalog, harness contract, research notes and the diff-tracks
generator follow.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Lucas WintherandClaude Opus 5.5 committed 2026-11-11 13:00:26 +01:00
1 parent 570e5d6826
commit 34c655c1a4
89 files changed
+11703 -4

No files matched your search

+1 -1
View File
@@ -3,7 +3,7 @@
A free, hands-on course on **systemic game design**. You learn by rebuilding the rules machinery of two games, then extending it:
- **Part I: Hearthstone** (cards, zones, triggers, death phases, auras, text assembly)
- **Part II: Baba Is You** (rules as pushable text, parsing, movement, win/lose)
- **Part II: Baba Is You** (rules as pushable text, parsing, movement, win/lose; Module 11 adds conditions, `BABA ON KEY IS WIN`)
- **Part III: SpellGrid**, a small original game that combines both systems
- **Module 13: Break your own engine**, an optional closing module on testing what the tests miss: mutation testing, invariants, metamorphic relations, undo and replay, and a level solver
+1 -1
View File
@@ -4,7 +4,7 @@ In Module 8, a key press moved things and nothing else happened. This module mak
Part I had this too. In Module 3, the end of the outermost Phase ran a fixed sequence of steps: the Death Creation Step, then the Death Phase, then the check again. In Module 4 the sequence grew: "After the outermost Phase ends, Hearthstone does an Aura Update (Health/Attack), then does the Death Creation Step", and another Aura Update follows ([Advanced rulebook](https://hearthstone.wiki.gg/wiki/Advanced_rulebook)). In both games, the order of the steps is a decision written down once, in one place.
This is the last Baba Is You module. Module 10 joins the two halves of the course.
Module 10 joins the two halves of the course. [Module 11](../11-conditions/lesson.md) comes back to Baba Is You for one more word, `ON`, and asks when a rule applies.
You need your Module 8 workspace with all Module 1–8 tests passing, under git. This module's tests also use the meta-noun `ALL` from the [Module 7 Expansion Drop](../07-rule-resolution/lesson.md#expansion-drop). If you skipped that drop, do it first; it is one line of data.
+23
View File
@@ -0,0 +1,23 @@
# Module 11 Expansion Drop: NEAR
The lesson's [Expansion Drop](../lesson.md#expansion-drop) beat explains the task. This file holds the scenario, the test locations and the check.
- Word: `NEAR`, a condition. From the Baba Is You Wiki's [NEAR](https://babaiswiki.fandom.com/wiki/NEAR) page (recorded in `docs/research/baba-is-you.md` §7): "NEAR is an infix condition that makes a rule apply when an object is on a target or in one of the eight spaces around it". The page doesn't mention float levels; the reference ignores them, as `ON` does (the course's choice).
- Tests: `tests/rust/test_11x.rs` and `tests/ts/test_11x.test.ts`. Copy the one for your track into your workspace's `tests/`.
- Reference answer, one change in data and one in content code:
- `solution/words.json`, one file for both tracks: the Module 11 reference's `src/content/words.json` with one entry added, `{ "name": "NEAR", "role": "condition" }`.
- `solution/rust/src/content/conditions.rs` and `solution/ts/src/content/conditions.ts`: the Module 11 reference's condition checks with one check added, `near`, registered under `NEAR`. It looks at the other objects on the object's tile and the eight around it, and needs a different one for each listed noun, as `on` does on one tile.
- Nothing under `src/engine/` changes, and neither does the harness.
- Check: commit your workspace before you start, then run `node tools/expansion-check.mjs <your-workspace> <that-commit>` from the course repository. It fails if any file under `src/engine/` changed.
## Scenario
### test_11x_near
- Setup: `newBaba` with an empty level 5 wide and 5 high. Place `KEY (0,0) NEAR (1,0) ROCK (2,0) IS (3,0) WIN (4,0)`. Place an object `rock` at (2,3). Place an object `key` at (2,3) (`onIt`, on the rock's tile), one at (3,3) (`beside`), one at (1,2) (`diagonal`) and one at (0,4) (`far`, two columns away).
- Assertions:
- `rules()` is exactly `["KEY NEAR ROCK IS WIN", "TEXT IS PUSH"]`.
- `hasProperty(onIt, "WIN")`, `hasProperty(beside, "WIN")` and `hasProperty(diagonal, "WIN")` are true: the rock is on the key's tile or on one of the eight around it.
- `hasProperty(far, "WIN")` is false.
Before the drop the lexicon doesn't know `NEAR`, so the parser never sees it. The words before `IS` are then just `ROCK`, and the sentence spells `ROCK IS WIN`: the first assertion fails.
@@ -0,0 +1,72 @@
//! Part II: what each condition word means (Module 11). The engine asks these checks
//! every time it is about to use a rule with a condition, for one object, against the
//! board as it is then. A new condition word is one entry in `words.json`, with role
//! "condition", and one check here.
use std::collections::HashSet;
use crate::engine::puzzle::{ConditionCheck, ConditionChecks, Level, ObjectId};
/// The condition checks, by condition word.
pub fn conditions() -> ConditionChecks {
ConditionChecks::from([
("ON".to_string(), on as ConditionCheck),
("NEAR".to_string(), near as ConditionCheck),
])
}
/// Can each noun be given an object of its own among `others`, one the noun names?
/// `ON ROCK AND ROCK` needs two rocks (the wiki: "AND in the noun2 part requires at
/// least as many objects as are mentioned"). Tries every assignment.
fn each_has_one(level: &Level, nouns: &[String], others: &[ObjectId]) -> bool {
fn assign(
level: &Level,
nouns: &[String],
others: &[ObjectId],
taken: &mut HashSet<ObjectId>,
) -> bool {
let Some((noun, rest)) = nouns.split_first() else {
return true;
};
for &other in others {
if taken.contains(&other) || !level.names(noun, other) {
continue;
}
taken.insert(other);
if assign(level, rest, others, taken) {
return true;
}
taken.remove(&other);
}
false
}
assign(level, nouns, others, &mut HashSet::new())
}
/// ON: for each noun, another object on the object's tile that the noun names, a
/// different one for each noun. Float levels don't matter ("ON ignores FLOAT levels").
fn on(level: &Level, id: ObjectId, nouns: &[String]) -> bool {
let o = level.board().get(id);
let others: Vec<ObjectId> = level
.board()
.at(o.x, o.y)
.map(|other| other.id)
.filter(|&other| other != id)
.collect();
each_has_one(level, nouns, &others)
}
/// NEAR: for each noun, another object on the object's tile or on one of the eight
/// around it that the noun names, a different one for each noun ("on a target or in
/// one of the eight spaces around it"). Float levels don't matter (the course's
/// choice: the wiki's NEAR page doesn't mention them).
fn near(level: &Level, id: ObjectId, nouns: &[String]) -> bool {
let o = level.board().get(id);
let others: Vec<ObjectId> = level
.board()
.objects()
.filter(|other| other.id != id && (other.x - o.x).abs() <= 1 && (other.y - o.y).abs() <= 1)
.map(|other| other.id)
.collect();
each_has_one(level, nouns, &others)
}
@@ -0,0 +1,62 @@
// Part II: what each condition word means (Module 11). The engine asks these checks
// every time it is about to use a rule with a condition, for one object, against the
// board as it is then. A new condition word is one entry in `words.json`, with role
// "condition", and one check here.
import type { puzzle } from "../engine";
type Level = puzzle.Level;
type ObjectId = puzzle.ObjectId;
/**
* Can each noun be given an object of its own among `others`, one the noun names?
* `ON ROCK AND ROCK` needs two rocks (the wiki: "AND in the noun2 part requires at
* least as many objects as are mentioned"). Tries every assignment.
*/
function eachHasOne(level: Level, nouns: readonly string[], others: readonly ObjectId[]): boolean {
const taken = new Set<ObjectId>();
const assign = (i: number): boolean => {
if (i === nouns.length) return true;
for (const other of others) {
if (taken.has(other) || !level.names(nouns[i], other)) continue;
taken.add(other);
if (assign(i + 1)) return true;
taken.delete(other);
}
return false;
};
return assign(0);
}
/**
* ON: for each noun, another object on the object's tile that the noun names, a
* different one for each noun. Float levels don't matter ("ON ignores FLOAT levels").
*/
function on(level: Level, id: ObjectId, nouns: readonly string[]): boolean {
const o = level.board.get(id);
const others = level.board
.at(o.x, o.y)
.map((other) => other.id)
.filter((other) => other !== id);
return eachHasOne(level, nouns, others);
}
/**
* NEAR: for each noun, another object on the object's tile or on one of the eight
* around it that the noun names, a different one for each noun ("on a target or in
* one of the eight spaces around it"). Float levels don't matter (the course's choice:
* the wiki's NEAR page doesn't mention them).
*/
function near(level: Level, id: ObjectId, nouns: readonly string[]): boolean {
const o = level.board.get(id);
const others = level.board
.objects()
.filter((other) => other.id !== id && Math.abs(other.x - o.x) <= 1 && Math.abs(other.y - o.y) <= 1)
.map((other) => other.id);
return eachHasOne(level, nouns, others);
}
/** The condition checks, by condition word. */
export function conditions(): puzzle.ConditionChecks {
return { ON: on, NEAR: near };
}
@@ -0,0 +1,37 @@
[
{ "name": "BABA", "role": "noun" },
{ "name": "ROCK", "role": "noun" },
{ "name": "WALL", "role": "noun" },
{ "name": "FLAG", "role": "noun" },
{ "name": "LAVA", "role": "noun" },
{ "name": "KEY", "role": "noun" },
{ "name": "DOOR", "role": "noun" },
{ "name": "SKULL", "role": "noun" },
{ "name": "WATER", "role": "noun" },
{ "name": "ALL", "role": "noun", "selects": "things" },
{ "name": "KEG", "role": "noun" },
{ "name": "BRAMBLE", "role": "noun" },
{ "name": "HEDGE", "role": "noun" },
{ "name": "WARDEN", "role": "noun" },
{ "name": "MOTH", "role": "noun" },
{ "name": "SCRIBE", "role": "noun" },
{ "name": "KNIGHT", "role": "noun" },
{ "name": "CRIER", "role": "noun" },
{ "name": "GOLEM", "role": "noun" },
{ "name": "SHARD", "role": "noun" },
{ "name": "DEATHRATTLE", "role": "noun" },
{ "name": "ON", "role": "condition" },
{ "name": "NEAR", "role": "condition" },
{ "name": "YOU", "role": "property" },
{ "name": "WIN", "role": "property" },
{ "name": "PUSH", "role": "property" },
{ "name": "STOP", "role": "property" },
{ "name": "OPEN", "role": "property" },
{ "name": "SHUT", "role": "property" },
{ "name": "HOT", "role": "property" },
{ "name": "MELT", "role": "property" },
{ "name": "SINK", "role": "property" },
{ "name": "DEFEAT", "role": "property" },
{ "name": "FLOAT", "role": "property" },
{ "name": "DOUBLE", "role": "property" }
]
@@ -0,0 +1,25 @@
// Module 11 Expansion Drop: NEAR. The scenario is in expansion/README.md.
use engine::harness::*;
#[test]
fn test_11x_near() {
let mut b = new_baba(LevelDef {
width: 5,
height: 5,
..Default::default()
});
for (word, x) in [("KEY", 0), ("NEAR", 1), ("ROCK", 2), ("IS", 3), ("WIN", 4)] {
b.place(&format!("text:{word}"), x, 0);
}
b.place("rock", 2, 3);
let on_it = b.place("key", 2, 3);
let beside = b.place("key", 3, 3);
let diagonal = b.place("key", 1, 2);
let far = b.place("key", 0, 4);
assert_eq!(b.rules(), vec!["KEY NEAR ROCK IS WIN", "TEXT IS PUSH"]);
assert!(b.has_property(on_it, "WIN"));
assert!(b.has_property(beside, "WIN"));
assert!(b.has_property(diagonal, "WIN"));
assert!(!b.has_property(far, "WIN"));
}
@@ -0,0 +1,27 @@
// Module 11 Expansion Drop: NEAR. The scenario is in expansion/README.md.
import { expect, test } from "vitest";
import { newBaba } from "../src/harness";
test("test_11x_near", () => {
const b = newBaba({ width: 5, height: 5, objects: [] });
for (const [word, x] of [
["KEY", 0],
["NEAR", 1],
["ROCK", 2],
["IS", 3],
["WIN", 4],
] as const) {
b.place(`text:${word}`, x, 0);
}
b.place("rock", 2, 3);
const onIt = b.place("key", 2, 3);
const beside = b.place("key", 3, 3);
const diagonal = b.place("key", 1, 2);
const far = b.place("key", 0, 4);
expect(b.rules()).toEqual(["KEY NEAR ROCK IS WIN", "TEXT IS PUSH"]);
expect(b.hasProperty(onIt, "WIN")).toBe(true);
expect(b.hasProperty(beside, "WIN")).toBe(true);
expect(b.hasProperty(diagonal, "WIN")).toBe(true);
expect(b.hasProperty(far, "WIN")).toBe(false);
});
+463
View File
@@ -0,0 +1,463 @@
# Module 11: Conditions
Module 7 decided which rules are in force. Module 9 decided when, in a turn, the rules are read. This module adds one word, `ON`, and it unsettles both. `BABA ON KEY IS WIN` is a rule about some Babas: the ones standing on a key, at the moment someone asks. The rule is in force the whole time. Whether it *applies* to a given Baba changes every time that Baba moves, or the key does, or the key is destroyed.
This is the third time the course meets a rule that modifies another rule, and the third answer to "when is the modifier read?". In Module 4, Baron Rivendare was read from a snapshot taken at scheduled moments. In Module 7, `BABA IS NOT YOU` was read once, when the rules were compiled. A condition is read every time someone asks. Lesson 10's closing question said both engines answer "when is the world looked at?" with a schedule. This module is the exception, and it shows what the exception costs.
You need your Module 10 workspace with all Module 1–10 tests passing, under git. This module's tests also use the meta-noun `ALL` from the [Module 7 Expansion Drop](../07-rule-resolution/lesson.md#expansion-drop).
## Observe
All quotes come from sources recorded in the course's research notes. The Baba Is You Wiki is an unofficial community wiki; the course treats it as the best available description of the game, not as Hempuli's own word.
What `ON` does, from the wiki's [ON](https://babaiswiki.fandom.com/wiki/ON) page:
> ON is an infix condition that makes a rule apply when object is overlapping another object.
Its examples: "FLAG ON BABA IS WIN makes flag WIN only when it's on a baba." And with two nouns: "FLAG ON BABA AND BABA IS WIN makes flag WIN only when it's on two babas." The same page says "ON ignores FLOAT levels".
How a condition fits into a sentence, from the wiki's [Conditions](https://babaiswiki.fandom.com/wiki/Category:Conditions) page:
> They are parsed as noun1 condition noun2 verb thing, where noun1 verb thing only if the condition is true.
The same page on `AND`. Before the condition, "AND in the noun1 part works like normal": `BABA AND KEKE ON SKULL IS DEFEAT` is two rules, one for each subject. After it, "AND in the noun2 part requires at least as many objects as are mentioned", so `BABA ON KEKE AND KEKE IS WIN` needs two Kekes. And a limit: "It is not possible to write something like noun1 condition1 noun2 condition2 noun3 verb thing."
Negation works on conditional rules too. The [NOT](https://babaiswiki.fandom.com/wiki/NOT) page: "ROCK ON BABA IS NOT PUSH makes ROCK that is on BABA not PUSHable."
Now the part that matters most for an engine. The [Order of Operations](https://babaiswiki.fandom.com/wiki/Order_of_Operations) page, which Module 9 built its turn from, opens with this:
> Conditions process when the game checks for what objects have a rule, which is usually right before it uses it.
It comes back to conditions twice. In the movement phase, "it completely finishes moving for a particular move stage before checking which units are in the next", "so conditions (e.g. IT ON BABA IS YOU AND MOVE) reparse in between". And for `HAS`: "conditions on these rules are rechecked at the time of destruction, even if multiple destructions occur in the same turn."
Notice what this says and what it doesn't. A condition is not part of the rule list that the game builds when it parses. It is evaluated when a rule is used, "usually right before". Module 9's turn parses at fixed points and every phase reads the rules from the parse before it. A condition doesn't wait for a parse.
Put a board together. `BABA IS YOU`, `BABA AND LAVA IS FLOAT`, `LAVA IS HOT`, `KEY IS SINK` and `BABA ON KEY IS MELT`. A tile holds lava, a key and a rock. Baba steps onto it. When the rules are parsed after movement, Baba is on the key, so `BABA ON KEY IS MELT` applies to it. Then the hazards run, in the `block()` order from Module 9: SINK first. The key is SINK and shares its tile with the rock on the same float level, so both go. "Deletion occurs after each step that removes objects". Then HOT/MELT asks whether Baba is MELT. If the condition is checked right before it is used, Baba is on no key now, and **Baba survives**. No page shows this board. The course infers the outcome from the two quotes above, and this module's test says so.
## Predict
Write your answers down before you open the discussion, for example in `notes/11.md` in your workspace.
1. `KEY ON ROCK IS WIN`. One key lies on a rock, another key lies alone. Which keys are WIN? Is the rock?
2. `BABA IS YOU` and `BABA ON KEY IS NOT YOU`. One Baba stands on a key, another on an empty tile. Which of them moves? What does `activeRules()` list?
3. Add `BABA ON KEY IS YOU` to the board of question 2. Does anything change? What does `activeRules()` list now?
4. `BABA IS YOU`, `LAVA IS HOT` and `BABA ON KEY IS MELT`. One Baba steps onto a tile with a key and lava, another onto a tile with lava only. Before the step, is either Baba MELT? After it, who is left?
5. `KEY ON ROCK AND ROCK IS WIN`, with one key on a single rock. WIN?
6. `BABA ON` then a tile holding two stacked words, `KEY` and `ROCK`, then `IS WIN`. How many rules is that, and what are they?
7. `ROCK ON ALL IS WIN`. Two rocks share a tile, and nothing else is on it. Are they WIN? A third rock shares its tile with a piece of text. Is it?
8. The board at the end of Observe. Does Baba survive? Would it in an engine that decided every condition when it parsed the rules?
9. Baron is read from a snapshot; `NOT` is read when the rules are compiled. In your engine, where would you read `ON`, and when?
<details><summary>Discussion</summary>
1. The key on the rock. The other key isn't, and neither is the rock: the rock is what the condition is about, not a subject. The sentence spells one rule, `KEY ON ROCK IS WIN`. It must not spell `ROCK IS WIN`, which is what your Module 10 parser reads from it.
2. Only the Baba on the empty tile moves. For the Baba on the key, both rules apply, and the negation wins, as in Module 7. For the other, only `BABA IS YOU` applies. `activeRules()` lists both rules (and `TEXT IS PUSH`). Module 7's resolution pass would have cancelled `BABA IS YOU` here, because the subject and the complement match. With conditions it can't: whether the negation wins depends on where each Baba stands, which the rule list doesn't know.
3. Nothing changes on the board. `BABA ON KEY IS NOT YOU` cancels `BABA ON KEY IS YOU` exactly as `BABA IS NOT YOU` cancels `BABA IS YOU`: same subject, same complement, and now also the same condition. So `activeRules()` is the same as before. The resolution pass still does its job, for rules whose conditions are identical; the rest is settled for each object.
4. Neither is MELT before the step. After it, the Baba that stepped onto the key is gone and the other stands on its lava. The rules never changed. Only the board did, and `ON` reads the board.
5. No. One rock can't stand for both nouns: "AND in the noun2 part requires at least as many objects as are mentioned". The course needs a different object for each listed noun.
6. Two rules: `BABA ON KEY IS WIN` and `BABA ON ROCK IS WIN`. A stack is a choice of words, as it has been since Module 6, so each noun in the stack makes its own sentence. It is not `BABA ON KEY AND ROCK IS WIN`, which would need both.
7. Yes, both: each is on the other. `ALL` names the rock itself too, but `ON` asks about *another* object. The third rock isn't WIN: `ALL` doesn't name text.
8. It survives, if conditions are checked when they are used. In an engine that decided `BABA ON KEY` during the parse after movement, Baba would stay MELT until the next parse, and HOT/MELT would destroy it. The difference is visible only because something changed the board between the parse and the use, and in Module 9's turn the hazards are such a thing.
9. There is no single place. Baron's count was read by one routine at the moment a trigger ran. `NOT` was applied by one pass over the rules. A condition has to be checked by everything that asks a question about an object: `hasProperty`, and so movement, every hazard and the win check, and conversion. Keep the answer in one function that all of them call, and the question "where" goes back to having one answer.
</details>
## Spec
From the course repository, copy the module's tests into your workspace:
```sh
node tools/start-module.mjs 11 ~/my-engine
```
### The harness
No new functions. The contract of four old ones grows:
- `rules()` and `activeRules()` write a condition between the subject and `IS`, its nouns in the order they are written: `"KEY ON ROCK IS WIN"`, `"KEY ON ROCK AND FLAG IS WIN"`, `"BABA ON KEY IS NOT YOU"`.
- `hasProperty(id, prop)` uses only the rules that apply to the object: its noun names the object, and every condition of the rule holds for the object on the board as it is now.
- `step(dir)`: conversion uses the same rules, decided from the board as movement left it. The hazards and the win check ask `hasProperty`, so each of them checks conditions when it runs.
What `ON Y1 AND … AND Yk` means: the other objects on the object's tile can supply a different object for each `Yi`, one that `Yi` names. Float levels don't matter.
New word: `ON`, with a new role, `condition`. In the reference's word list: `{ "name": "ON", "role": "condition" }`.
### The tests
The full statements are in `course/11-conditions/scenarios.md`.
- `test_11_condition_gates_property`: Predict question 1.
- `test_11_condition_follows_the_object`: Predict question 4. `activeRules()` is the same before and after the turn.
- `test_11_and_requires_both`: `KEY ON ROCK AND FLAG IS WIN`, with keys on a rock, on a flag, and on both.
- `test_11_and_counts_objects`: Predict question 5, and a key on two rocks.
- `test_11_conditional_not_only_where_it_holds`: Predict questions 2 and 3, then a turn.
- `test_11_conditional_conversion`: `KEY ON ROCK IS FLAG`. A key already on a rock converts, and so does a key that a pushed rock arrived on during the same turn. A key alone doesn't.
- `test_11_condition_and_stacked_text`: Predict question 6, and `ON` stacked on `AND`, which reads both ways.
- `test_11_on_group_noun`: Predict question 7.
- `test_11_condition_rechecked_between_hazards`: Predict question 8. The outcome is the course's inference.
The second test:
```rust
#[test]
fn test_11_condition_follows_the_object() {
let mut b = new_baba(empty_level(6, 5));
write(&mut b, &[("BABA", 0, 0), ("IS", 1, 0), ("YOU", 2, 0)]);
write(&mut b, &[("LAVA", 0, 1), ("IS", 1, 1), ("HOT", 2, 1)]);
write(
&mut b,
&[
("BABA", 0, 2),
("ON", 1, 2),
("KEY", 2, 2),
("IS", 3, 2),
("MELT", 4, 2),
],
);
let to_key = b.place("baba", 0, 4);
b.place("key", 1, 4);
b.place("lava", 1, 4);
let to_lava = b.place("baba", 3, 4);
b.place("lava", 4, 4);
let before = b.active_rules();
assert!(!b.has_property(to_key, "MELT"));
assert!(!b.has_property(to_lava, "MELT"));
b.step(Dir::Right);
assert_eq!(b.active_rules(), before);
assert_eq!(b.pos_of(to_key), None);
assert_eq!(b.kinds_at(1, 4), vec!["key", "lava"]);
assert_eq!(b.pos_of(to_lava), Some((4, 4)));
assert_eq!(b.kinds_at(4, 4), vec!["baba", "lava"]);
}
```
```ts
test("test_11_condition_follows_the_object", () => {
const b = newBaba(emptyLevel(6, 5));
write(b, [
["BABA", 0, 0],
["IS", 1, 0],
["YOU", 2, 0],
]);
write(b, [
["LAVA", 0, 1],
["IS", 1, 1],
["HOT", 2, 1],
]);
write(b, [
["BABA", 0, 2],
["ON", 1, 2],
["KEY", 2, 2],
["IS", 3, 2],
["MELT", 4, 2],
]);
const toKey = b.place("baba", 0, 4);
b.place("key", 1, 4);
b.place("lava", 1, 4);
const toLava = b.place("baba", 3, 4);
b.place("lava", 4, 4);
const before = b.activeRules();
expect(b.hasProperty(toKey, "MELT")).toBe(false);
expect(b.hasProperty(toLava, "MELT")).toBe(false);
b.step("right");
expect(b.activeRules()).toEqual(before);
expect(b.posOf(toKey)).toBeNull();
expect(b.kindsAt(1, 4)).toEqual(["key", "lava"]);
expect(b.posOf(toLava)).toEqual([4, 4]);
expect(b.kindsAt(4, 4)).toEqual(["baba", "lava"]);
});
```
Run `cargo test` or `npm test`. The Module 11 tests fail: your lexer doesn't know `ON`, so the parser skips it and reads `KEY IS MELT` or `ROCK IS WIN` where a condition stands. Modules 1–10 still pass. That is RED.
## Build
Make the nine tests pass, and keep Modules 1–10 green. Where code goes:
- **A new role, not a new word in the engine.** `ON` goes in your word list with the role `condition`. The engine learns that role, and where it may stand. It doesn't learn the word: the Expansion Drop brings a second condition, and you won't be allowed to touch `src/engine/`.
- **The parser finds the condition, and only reads it.** Walking back from `IS`, the parser collects a run of nouns, as it has since Module 6. If a condition word stands just before that run, the run was the condition's nouns, and the subjects are the run before the condition word. Otherwise the run is the subjects, as before. A rule now carries its conditions: the word, and its nouns.
- **Stacks still mean every reading.** Subjects distribute, so a stack of subjects was always just more rules. Condition nouns don't distribute (`AND` there means *all of them*), so a stack among them is a choice: one rule per noun in the stack.
- **What a condition means is content.** A condition check is a function from (the level, one object, the condition's nouns) to yes or no, registered by its word, as Module 8's movement responses and Module 9's hazards are. `ON` is one such function.
- **Check at use, in one place.** Write one function that answers "does this rule apply to this object now?": the subject names the object, and every condition holds. `hasProperty` and conversion both call it. Don't decide conditions when you parse, and don't cache the answer.
- **Resolution pairs like with like.** `X ON Z IS NOT Y` cancels `X ON Z IS Y` in the resolution pass, as before. It doesn't cancel `X IS Y`: that one is settled for each object, among the rules that apply to it. `hasProperty` already settles negation per object, so the change there is only which rules it looks at. The identity lock pairs the same way.
- **Print the condition.** `rules()` writes `KEY ON ROCK AND FLAG IS WIN`.
<details><summary>Hint 1 — nudge</summary>
You will be tempted to make conditions disappear early. One way: when you parse `BABA ON KEY IS WIN`, look at the board and emit `BABA IS WIN` "for the Babas on keys". Another: compute the answer for every object at each parse and store it. Both pass the first tests. Then play Predict question 8 by hand: what does each of them think about Baba after SINK has removed the key?
</details>
<details><summary>Hint 2 — concept</summary>
- **Parsing.** The run of nouns before `IS` ends at a tile that holds no `AND` (or a `NOT`). Look at that tile. If it holds a condition word, the run you collected is the condition's nouns, in reverse order; read the run of subjects from the tile before the condition word. If it doesn't, the run is the subjects and there is no condition. A stack among the condition's nouns: walk every path, one noun per tile, and make one condition per path. A stack of a condition word and an `AND` on one tile can be read both ways; the reference does.
- **The rule.** Add `conditions: [{ word, nouns }]`. Rules spelled without a condition have none. Two rules "have the same conditions" if they have the same words, each with the same nouns.
- **One question, one place.** `appliesTo(rule, id)`: the rule's subject names the object, and for every condition, content's check for its word says yes for this object and these nouns. In `hasProperty`, filter the active rules by complement and by `appliesTo`, then do what Module 7 did: one positive and no negated. In conversion, collect the rules that apply to the object, instead of the rules whose subject names it, and do what Module 9 did.
- **`ON`.** The other objects on the object's tile. Each noun needs one of them that it names, a different one for each noun. With one noun that is just "any". With several, try each assignment: a short backtracking search.
- **Resolution.** In both steps, add "and the same conditions" to the pairing.
</details>
<details><summary>Hint 3 — pseudo-code</summary>
```text
before_is(tiles, at, step): // walking back from IS
readings = []
run = subjects(tiles, at, step) // as in Module 7, also returning the tile it stopped on
if no condition word at run.stop:
readings += (run.nouns, no condition)
for each path in noun_runs(tiles, at, step): // one noun per tile, joined by AND
for each condition word c at path.stop:
subs = subjects(tiles, path.stop + step, step).nouns
readings += (subs, condition(c, reverse(path.nouns)))
return readings
applies_to(rule, o):
names(rule.subject, o) and
every c in rule.conditions: content.checks[c.word](level, o, c.nouns)
has_property(o, prop):
about = [r in active where r.complement == prop and applies_to(r, o)]
return any(r positive in about) and not any(r negated in about)
on(level, o, nouns):
others = objects on o's tile, other than o
return each noun can be given its own object in others that it names
negates(n, r): n.negated and not r.negated and same subject, complement and conditions
```
</details>
When your tests pass, compare with the reference engine in `course/11-conditions/reference/rust/src/` or `course/11-conditions/reference/ts/src/`: `engine/puzzle/parse` (`before_is` / `beforeIs` and `noun_runs` / `nounRuns`), the new `engine/puzzle/condition`, `engine/puzzle/level` (`applies_to` / `appliesTo`), `engine/puzzle/resolve`, and the new `content/conditions`.
Decisions in the reference that the tests don't force:
- **A condition word without a check is an error.** `Level` refuses to start if a word with role `condition` has no check in content (TypeScript throws, Rust panics). Otherwise a forgotten check would quietly make every rule with that word apply never, or always.
- **Only other objects count.** `ROCK ON ROCK IS WIN` needs a second rock. The ON page says "overlapping another object"; the course reads that as excluding the object itself, and applies the same reading to `ALL`.
- **A different object for each noun, found by search.** `ON ALL AND KEY` on a tile with a key and a rock holds: the rock stands for `ALL` and the key for `KEY`, even if the key is checked first. The game claims objects in order and the wiki says that can fail (see the box below). The course tries every assignment, so the answer never depends on object order.
- **Same conditions means the same nouns in any order.** `BABA ON KEY AND ROCK IS NOT YOU` cancels `BABA ON ROCK AND KEY IS YOU`. They are true of exactly the same objects. `rules()` still prints each as written.
- **`ON` stacked on `AND` reads both ways.** `FLAG [ON AND] KEY IS PUSH` is `FLAG ON KEY IS PUSH`, and also `FLAG IS PUSH` and `KEY IS PUSH`. That is Module 6's rule for stacks: every reading the grammar allows. No source covers this board.
- **A condition crosses no seam.** In SpellGrid, `DEATHRATTLE ON KEY IS DOUBLE` gives no aura. A condition is about one object on the grid, and a trigger word names no object. Module 10's `ruleAuras` / `rule_auras` skips rules with conditions.
- **Who moves is decided once per movement phase.** Module 8's movement collects the `YOU` objects when it starts. A conditional `YOU` that becomes true halfway through the moves doesn't add a mover. The responses (`PUSH`, `STOP`, `OPEN`/`SHUT`) ask `hasProperty` as each step is planned, so they see the board as the earlier moves left it. The game also fixes its movers per move stage: "it completely finishes moving for a particular move stage before checking which units are in the next".
> **How the real game differs**
> - **`NOT` before a condition.** "Conditions can be negated by prefixing with NOT" ([Conditions](https://babaiswiki.fandom.com/wiki/Category:Conditions)), and `NOT` before a condition's noun names everything else ([NOT](https://babaiswiki.fandom.com/wiki/NOT): "FLAG ON NOT BABA IS WIN makes flag WIN when it is ON something other than baba"). The course reads neither: `BABA NOT ON KEY IS WIN` and `BABA ON NOT KEY IS WIN` spell nothing, as `NOT BABA IS YOU` spells nothing in Module 7.
> - **Several conditions.** The game joins conditions with `AND` (`BABA ON KEKE AND NEAR ME IS WIN`) and has prefix conditions such as `LONELY` and `POWERED`, "parsed as condition noun verb thing" ([Conditions](https://babaiswiki.fandom.com/wiki/Category:Conditions)). The course has one infix condition per sentence. A rule carries a list of conditions, so more would fit, but the parser reads one.
> - **Chained conditions.** `A ON B ON C IS WIN`: "This will parse as noun2 condition2 noun3 verb thing" (same page). The course does the same, by accident of its grammar: the run of subjects before the second `ON` stops at the first one.
> - **Matching nouns to objects.** For `NOT` among a condition's nouns, "the game first claims objects for all non-NOT cases, then claims objects for the NOT cases in order", and the page says this can wrongly fail (same page). The course has no `NOT` there and searches every assignment. What the game does for `ON ALL AND KEY` is not stated by any page read; the course's answer is its own.
> - **`LEVEL ON ROCK`**, and the level icon's position ([ON](https://babaiswiki.fandom.com/wiki/ON)). The course has no `LEVEL`.
> - **When a condition is checked.** The game checks conditions "usually right before it uses it" ([Order of Operations](https://babaiswiki.fandom.com/wiki/Order_of_Operations)). The course always checks at use. The page doesn't list the exceptions.
> - **The hazard board is inferred.** Baba surviving after SINK removes the key follows from "Conditions process ... right before it uses it" and "Deletion occurs after each step that removes objects". No source shows this board.
> - **Conditional `NOT` against an unconditional rule.** The NOT page says a negation overrides "other instances of the rule it disables", and its example "ROCK ON BABA IS NOT PUSH makes ROCK that is on BABA not PUSHable" names only those rocks. No page read spells out what happens with `ROCK IS PUSH` elsewhere on the board. Settling the negation per object, among the rules that apply, is the course's reading.
> - **Object order.** "Objects are parsed in order of object priority" ([ON](https://babaiswiki.fandom.com/wiki/ON)), which the game sets by placement on the grid ([Order of Operations](https://babaiswiki.fandom.com/wiki/Order_of_Operations)). The course's objects are in age order, and no answer in this module depends on it.
## Name it
The pattern is a **guard**: a condition attached to a rule, checked each time the rule is about to be used, against the state as it is then. The rule itself doesn't change. Whether it applies does. Dijkstra's guarded commands are the classic form: "A guarded command consists of a boolean condition or guard, and a statement "guarded" by it." ([Guarded Command Language, Wikipedia](https://en.wikipedia.org/wiki/Guarded_Command_Language)). Another way to say it: a rule with a condition is a **predicate** over objects, not a fact about a noun. `BABA IS WIN` was a fact about every Baba; `BABA ON KEY IS WIN` is a test you run on one Baba.
The reference's whole change to the question "does this object have this property?":
```rust
pub fn has_property(&self, id: ObjectId, property: &str) -> bool {
if self.board.find(id).is_none() {
return false;
}
let about_it: Vec<&Rule> = self
.active
.iter()
.filter(|r| r.complement == property && self.applies_to(r, id))
.collect();
about_it.iter().any(|r| !r.negated) && !about_it.iter().any(|r| r.negated)
}
/// Does `rule` apply to object `id` now? Its subject names the object, and each
/// of its conditions holds for the object on the board as it is (Module 11).
fn applies_to(&self, rule: &Rule, id: ObjectId) -> bool {
self.names(&rule.subject, id)
&& rule
.conditions
.iter()
.all(|c| (self.conditions[&c.word])(self, id, &c.nouns))
}
```
```ts
hasProperty(id: ObjectId, property: string): boolean {
if (this.board.find(id) === undefined) return false;
const aboutIt = this.active.filter((r) => r.complement === property && this.appliesTo(r, id));
return aboutIt.some((r) => !r.negated) && !aboutIt.some((r) => r.negated);
}
/**
* Does `rule` apply to object `id` now? Its subject names the object, and each of
* its conditions holds for the object on the board as it is (Module 11).
*/
private appliesTo(rule: Rule, id: ObjectId): boolean {
return (
this.names(rule.subject, id) && rule.conditions.every((c) => this.conditions[c.word](this, id, c.nouns))
);
}
```
Module 7's "one question, one place" is what made this small. Every system that asks about an object already went through `hasProperty`, so putting the guard there reaches movement, the hazards, the win check and `lost()` at once. Conversion asks the same `appliesTo`.
Now the three modifiers side by side:
| | Baron Rivendare (Module 4) | `BABA IS NOT YOU` (Module 7) | `BABA ON KEY IS WIN` (this module) |
|---|---|---|---|
| Read | from a snapshot, taken at scheduled moments | once, when the rules are compiled | every time someone asks, for one object |
| Changes | how many times a trigger runs | whether a rule is in force | whether a rule applies to this object, now |
| Between two reads | stale until the next Aura Update | stale until the next parse | never stale |
The last row is what a guard buys, and the costs come with it:
- **No table to look things up in.** With unconditional rules, an engine can compile "which nouns have which properties" once per parse and answer every question from that table. The rules in force still compile once per parse here, but the answer for a conditional rule is not in them. Every `hasProperty` on a guarded rule runs the guard. A hazard that asks about every object runs every guard about each of them. On the course's small boards the cost is small. On a board with thousands of objects, it would be the place to look first.
- **The order inside a turn becomes visible.** Before this module, the only state a hazard could see change during the hazards phase was which objects were left. The rules stood still until the turn's last parse. Now any rule can change its meaning mid-phase, because its guard reads the board. The order of the hazards decides what a later hazard sees. Two orders that could not be told apart before can be now (Stress it question 1).
- **Guards must not change anything.** A condition check that moved an object, or destroyed one, would make the answer to "is Baba MELT?" depend on who asked first. Guarded commands make the same demand: "guards cannot have side effects", which is why "an implementation may evaluate the guards in any sequence" ([Guarded Command Language, Wikipedia](https://en.wikipedia.org/wiki/Guarded_Command_Language)). `on` only reads.
The real game pays the same costs, and the Order of Operations page shows where it decided to. Movers are fixed per move stage, and "conditions (e.g. IT ON BABA IS YOU AND MOVE) reparse in between". A `HAS` rule's conditions "are rechecked at the time of destruction, even if multiple destructions occur in the same turn". Each of those is the game writing down exactly when a guard is read, because the answer would otherwise depend on the order nobody wrote down.
**Run the coupling audit** again: `node tools/coupling-audit.mjs ~/my-engine` should print `coupling-audit: ok`. The engine knows the role `condition`; `ON` is in content. The audit flags nouns only, but the rule is the same for condition words: the engine doesn't name them.
## Expansion Drop
A new word arrives: `NEAR`, a second condition. From the wiki's [NEAR](https://babaiswiki.fandom.com/wiki/NEAR) page: "NEAR is an infix condition that makes a rule apply when an object is on a target or in one of the eight spaces around it". `PILLAR NEAR FLAG IS PINK` makes a pillar pink "if a Flag is ON the Pillar, directly adjacent, or diagonally adjacent".
The constraint, as always: **do not edit anything under `src/engine/`**. The harness doesn't change either. Your word list changes, and so may the rest of your content.
1. Commit your workspace and note the commit: `git add -A && git commit -m "module 11 done"`.
2. Copy the expansion test for your track from `course/11-conditions/expansion/tests/rust/test_11x.rs` or `course/11-conditions/expansion/tests/ts/test_11x.test.ts` into your workspace's `tests/`. Run it and watch it fail.
3. Add `NEAR` to your word list, and make it do what it says.
4. When the tests pass, run the check from the course repository:
```sh
node tools/expansion-check.mjs ~/my-engine HEAD
```
The test, `test_11x_near`, writes `KEY NEAR ROCK IS WIN` with one rock and four keys: one on the rock's tile, one beside it, one diagonal to it and one two tiles away. The first three are WIN. The full scenario is in `course/11-conditions/expansion/README.md`.
<details><summary>Discussion: what this drop measures</summary>
If a condition word's meaning is a check that content registers, `NEAR` is one word in the list and one check, beside `on`:
```rust
/// NEAR: for each noun, another object on the object's tile or on one of the eight
/// around it that the noun names, a different one for each noun ("on a target or in
/// one of the eight spaces around it"). Float levels don't matter (the course's
/// choice: the wiki's NEAR page doesn't mention them).
fn near(level: &Level, id: ObjectId, nouns: &[String]) -> bool {
let o = level.board().get(id);
let others: Vec<ObjectId> = level
.board()
.objects()
.filter(|other| other.id != id && (other.x - o.x).abs() <= 1 && (other.y - o.y).abs() <= 1)
.map(|other| other.id)
.collect();
each_has_one(level, nouns, &others)
}
```
```ts
/**
* NEAR: for each noun, another object on the object's tile or on one of the eight
* around it that the noun names, a different one for each noun ("on a target or in
* one of the eight spaces around it"). Float levels don't matter (the course's choice:
* the wiki's NEAR page doesn't mention them).
*/
function near(level: Level, id: ObjectId, nouns: readonly string[]): boolean {
const o = level.board.get(id);
const others = level.board
.objects()
.filter((other) => other.id !== id && Math.abs(other.x - o.x) <= 1 && Math.abs(other.y - o.y) <= 1)
.map((other) => other.id);
return eachHasOne(level, nouns, others);
}
```
That is the reference answer, in `course/11-conditions/expansion/solution/`: `words.json` with `{ "name": "NEAR", "role": "condition" }` added, and the condition checks with `near` registered under `NEAR`. `near` shares `on`'s helper for "a different object for each noun", so `ROCK NEAR GRASS AND GRASS`, which "requires at least 2 grass" ([NEAR](https://babaiswiki.fandom.com/wiki/NEAR)), works with no extra code.
Shapes that fail this drop:
- **`ON` in the engine.** If the parser looks for the word `ON`, or `hasProperty` says `if word == "ON"`, then `NEAR` is a second branch under `src/engine/`, and the parser doesn't even read `KEY NEAR ROCK` as a condition. Make conditions a role and a registry, as a normal commit, then redo the drop.
- **Conditions decided in the parser.** If the parser looked at the board to expand `KEY ON ROCK IS WIN` into facts about particular keys, `NEAR` needs a second expansion in the parser, which is engine code. It also gets Predict question 8 wrong.
- **Conditions as hard-wired tile checks.** If the engine's `appliesTo` compares positions itself ("same tile"), a condition about neighbours can't be written in content. The engine should know only that a condition is a check that takes the level, the object and the nouns.
> **How the real game differs**
> - `NOT NEAR` reverses the condition, and the page shows that `PILLAR NOT NEAR BABA AND KEKE` differs from `PILLAR NOT NEAR BABA AND NOT NEAR KEKE` ([NEAR](https://babaiswiki.fandom.com/wiki/NEAR)). The course has no `NOT` before a condition.
> - The NEAR page says nothing about float levels. The course ignores them, as for `ON`.
> - The game has a related condition, `NEXTTO`, which the NEAR page compares with `NEAR` in a picture. The course doesn't add it.
If the check fails, fix your engine (a normal commit, outside the drop), then redo the drop from step 1.
</details>
## Stress it
Optional. Each question pushes the design past what this module's tests check. Predict first, then open the discussion.
1. Module 13's table lists mutant-style changes to the engine: drop one of the turn's parses, move the win check before the hazards, run the hazards in reverse, run DEFEAT before HOT/MELT. Make each change in your engine, one at a time, and run all the tests. Which of them does a *Module 11* test catch? Then add two mutants of your own: decide every condition when the rules are parsed (keep the answers in a table until the next parse), and let a negation cancel a rule in the resolution pass whatever its conditions. Which test catches each?
<details><summary>Discussion</summary>
On the TypeScript reference, with every test from Modules 1–11:
| Change | Tests that fail |
|---|---|
| No parse between movement and conversion | `test_09_mid_turn_conversion`, `test_10_level_5_solution` |
| No parse between conversion and hazards | `test_09_conversion_reparses_before_hazards` |
| Hazards before conversion | `test_09_conversion_before_hazards`, `test_09_conversion_reparses_before_hazards`, `test_09_undo_restores_everything` |
| Win check before hazards | `test_09_defeat_beats_win`, `test_10_level_2_solution` |
| No last parse | `test_09_last_parse_matches_board` |
| **Hazards in reverse order** | both `test_09_hazard_order_…` tests, `test_10_level_6_solution`, **`test_11_condition_rechecked_between_hazards`** |
| DEFEAT before HOT/MELT | `test_09_hazard_order_melt_before_defeat`, `test_10_level_6_solution` |
| Conditions decided at parse time | `test_11_condition_rechecked_between_hazards` |
| Negation cancels whatever the conditions | `test_11_conditional_not_only_where_it_holds` |
One old mutant gains a new killer: hazards in reverse order. Reversed, HOT/MELT runs before SINK, and Baba is still on the key when HOT/MELT asks. Module 9's Stress it question 2 said the order of SINK and HOT/MELT can't be observed with one hazard property per object, and that "tomorrow's" content would make it visible. This is tomorrow. In the new test, every object has at most one hazard property, and the order shows anyway, because a condition lets one hazard change what another hazard's victim *is*.
The new mutants each have one killer. If you deleted `test_11_condition_rechecked_between_hazards`, an engine that decided conditions at parse time would pass every test in the course. The parse after movement already sees Baba on the key, so the follow-the-object test can't tell the two apart. Only a board that changes between a parse and a use can, and in this turn only a hazard changes the board there.
And one old mutant is *not* caught by the new conversion test, though it looks as if it should be: dropping the parse between movement and conversion. `test_11_conditional_conversion` pushes a rock onto a key during movement and expects the key to convert. It does, with or without that parse: the rule `KEY ON ROCK IS FLAG` was in force before the move, and its condition is checked when conversion uses it. A parse used to be the only moment the engine looked at the board. With guards, it isn't.
</details>
2. Write down, for your engine, the order in which `hasProperty` is asked about each object during one turn. Then imagine a condition word whose check itself asks `hasProperty`: `WITH`, say, so that `BABA WITH HOT IS MELT` means "on a tile with something HOT". What could go wrong?
<details><summary>Discussion</summary>
The order is long, and you have probably never written it down: movement asks about `YOU` for every object, then about `PUSH`, `STOP`, `OPEN` and `SHUT` for each object on each tile a mover steps to; conversion asks every rule about every object; each hazard asks about its victim property for every object, and about its killer property and `FLOAT` for each one's company; the win check asks about `YOU`, `WIN` and `FLOAT`. Before this module, the answer to each of those questions was fixed between parses, so the order didn't matter. Now it can, and the order is an accident of how each phase happens to loop.
A condition whose check calls `hasProperty` makes it worse. Content could write a word `WITH`, as in `BABA WITH HOT IS MELT`, meaning "on the same tile as something HOT". Its check asks `hasProperty(other, "HOT")`, and if `LAVA WITH MELT IS HOT` is also on the board, that asks `hasProperty(baba, "MELT")`, which runs the first check again. Nothing stops the loop. This is the case Module 7's Stress it question 2 warned about: a rule whose applicability depends on the resolved rule set. The game has words of this kind (the Conditions page lists `FEELING`, which "uses a property in noun2"), so it must break such loops somehow; no page read for the course says how. An engine has three choices: forbid checks from asking about properties, detect the cycle and pick an answer, or evaluate in strata so that a check only reads properties decided before it. The reference takes the first, by convention: `on` reads only kinds and positions. Nothing in the engine enforces it.
</details>
3. A conditional rule in SpellGrid: `DEATHRATTLE ON KEY IS DOUBLE`. The reference gives it no aura. Argue for a different design, in which a Deathrattle triggers twice if *the minion that has it* stands on a key. What would the seam need to know that it doesn't now, and when would it have to ask?
<details><summary>Discussion</summary>
It is a reasonable reading of the sentence: `DEATHRATTLE` names a trigger, and a trigger belongs to a minion, which is a grid object with a tile. To support it, seam 1 could no longer compile the rules into a fixed list of auras after each reparse. It would have to hand Part I something that answers per minion: "for this minion's Deathrattle, how many times?", with the minion's id. Part I's `activations` would then ask the grid during a Sequence.
The question of *when* comes straight back. Part I reads auras from a snapshot taken at the Aura Update, on purpose (Module 4). A dying minion is still on the grid until seam 3 removes it after the Sequence. So "is the dying Warden on a key?" has an answer when the Deathrattle starts, but the answer comes from a different engine with a different schedule. You would have to decide whether a conditional rule is read like an aura (at the snapshot) or like a guard (when the Deathrattle starts), and Module 10's Level 3 is the proof that a snapshot and a live reading can disagree. The reference refuses the rule rather than pick. That is the cheap answer, and it is labelled as the course's choice. A level designer who wanted the other one would be choosing a new seam.
</details>
## Further reading
- The Baba Is You Wiki's [ON](https://babaiswiki.fandom.com/wiki/ON), [NEAR](https://babaiswiki.fandom.com/wiki/NEAR) and [Conditions](https://babaiswiki.fandom.com/wiki/Category:Conditions) pages: every condition word in the game, `NOT` before conditions, several conditions in one sentence, and how the game matches nouns to objects. Then reread [Order of Operations](https://babaiswiki.fandom.com/wiki/Order_of_Operations) and mark every line that says when a condition is checked.
- [Guarded Command Language, Wikipedia](https://en.wikipedia.org/wiki/Guarded_Command_Language): Dijkstra's guards, the original form of a statement that runs only if its condition holds, and why guards must have no side effects.
- [Datalog, Wikipedia](https://en.wikipedia.org/wiki/Datalog), on stratified negation, again. Module 7's resolver could stratify because no rule's applicability depended on the board. A condition that asks about properties (Stress it question 2) is where stratification starts to matter.
@@ -0,0 +1 @@
{ "track": "rust", "engineDirs": ["src/engine"], "contentDirs": ["src/content", "cards"], "module": "11" }
@@ -0,0 +1 @@
target/
+107
View File
@@ -0,0 +1,107 @@
# This file is automatically @generated by Cargo.
# It is not intended for manual editing.
version = 4
[[package]]
name = "engine"
version = "0.1.0"
dependencies = [
"serde",
"serde_json",
]
[[package]]
name = "itoa"
version = "1.0.18"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8f42a60cbdf9a97f5d2305f08a87dc4e09308d1276d28c869c684d7777685682"
[[package]]
name = "memchr"
version = "2.8.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cf8baf1c55e62ffcace7a9f06f4bd9cd3f0c4beb022d3b367256b91b87513d98"
[[package]]
name = "proc-macro2"
version = "1.0.107"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "985e7ec9bb745e6ce6535b544d84d6cd6f7ad8bd711c398938ae983b91a766d9"
dependencies = [
"unicode-ident",
]
[[package]]
name = "quote"
version = "1.0.47"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1fbf4db142a473a8d80c26bbf18454ed458bf8d26c8219c331daecfdbd079001"
dependencies = [
"proc-macro2",
]
[[package]]
name = "serde"
version = "1.0.229"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4148590afebada386688f18773da617792bf2ef03ffc1e4cbd2b1d45b023e0ba"
dependencies = [
"serde_core",
"serde_derive",
]
[[package]]
name = "serde_core"
version = "1.0.229"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "67dca2c9c51e58a4791a4b1ed58308b39c64224d349a935ab5039aa360942a48"
dependencies = [
"serde_derive",
]
[[package]]
name = "serde_derive"
version = "1.0.229"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e7a5d71263a5a7d47b41f6b3f06ba276f10cc18b0931f1799f710578e2309348"
dependencies = [
"proc-macro2",
"quote",
"syn",
]
[[package]]
name = "serde_json"
version = "1.0.154"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e7e9cc8b1b85264074fbcc02a88680c4096b1e47df8f739dceb03bf482f04bd6"
dependencies = [
"itoa",
"memchr",
"serde",
"serde_core",
"zmij",
]
[[package]]
name = "syn"
version = "3.0.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d78c8dee4c7bf0e14673097256fed6142ce9d3b85a408189d07482442145823b"
dependencies = [
"proc-macro2",
"quote",
"unicode-ident",
]
[[package]]
name = "unicode-ident"
version = "1.0.27"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a2c754d6c33795a1c324727428e5a7dedb5b06195f9890bdbcba760d3e246563"
[[package]]
name = "zmij"
version = "1.0.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "29666d0abbfad1e3dc4dcf6144730dd3a3ab225bbbdac83319345b1b44ccfc1b"
@@ -0,0 +1,9 @@
[package]
name = "engine"
version = "0.1.0"
edition = "2021"
[dependencies]
# Module 5: cards are JSON data.
serde = { version = "1", features = ["derive"] }
serde_json = "1"
@@ -0,0 +1,17 @@
# Module 11 reference engine (Rust)
The course's own engine as of the end of Module 11: Module 10's engine (Part I's card engine and Part II's grid engine, joined by SpellGrid), with conditions in Part II. Every test from Modules 1–10 still passes. Module 10's reference was the starting point, so a diff between the two shows what this module adds. `course/10-spellgrid/reference/rust/README.md` describes everything else.
- `cards/` and `levels/` (at the crate root; not in this directory): staged by `tools/verify-references.mjs` as for Module 10 (`course/cards/`, `course/10-spellgrid/cards/` and `course/10-spellgrid/levels/`), so SpellGrid's tests run here too.
- Part II (`src/engine/puzzle/`):
- `word.rs`: a fourth role, `Role::Condition`.
- `parse.rs`: `Condition { word, nouns }` and `Rule::conditions`. `before_is` reads what stands before an `IS`: the subjects alone, or, when a condition word stands before the run of nouns next to `IS`, the subjects before that word with the condition (`noun_runs` reads one noun from each tile of a stack). `Display` writes `BABA ON KEY AND ROCK IS WIN`. `Rule::negates` and the new `Rule::same_conditions` pair rules only with the same conditions.
- `resolve.rs`: negation and the identity lock pair rules only when their conditions are the same.
- `condition.rs` (new): `ConditionCheck`, `fn(&Level, ObjectId, &[String]) -> bool`, and `ConditionChecks`, content's checks by word.
- `level.rs`: `Level::new` takes the condition checks and panics if a condition word has none. `applies_to(rule, id)` (the subject names the object and every condition holds for it now) is what `has_property` and conversion use, so every question about an object evaluates its conditions at that moment. `names(word, id)` is public, for content's checks.
- Part III (`src/engine/spellgrid.rs`): `SpellGrid::new` takes the condition checks; `rule_auras` skips a rule with a condition.
- `src/content/conditions.rs` (new): `on`, registered under `ON`. `words.json` adds `{ "name": "ON", "role": "condition" }`.
- `src/harness.rs`: `new_baba` and `new_spell_grid` pass `content::conditions()`. No new harness function: `rules()` and `active_rules()` print conditions.
- `src/cli.rs` and `src/bin/spellgrid.rs`: unchanged from Module 10. `tools/play-spellgrid.mjs` plays Module 10's; `tools/diff-tracks.mjs` uses this module's by default.
Run it with the module's tests: `node tools/verify-references.mjs 11` from the course repository.
@@ -0,0 +1,21 @@
//! SpellGrid on the command line. See `src/cli.rs`.
//!
//! `cargo run --bin spellgrid -- levels/l1-double-tap.json`
use std::io::{self, BufReader};
use engine::cli::{run, Outcome};
fn main() {
let Some(path) = std::env::args().nth(1) else {
eprintln!("usage: cargo run --bin spellgrid -- <level.json>");
std::process::exit(2);
};
let level = std::fs::read_to_string(&path).unwrap_or_else(|e| {
eprintln!("cannot read {path}: {e}");
std::process::exit(2);
});
let input = Box::new(BufReader::new(io::stdin()));
let outcome = run(&level, input, Box::new(io::stdout()));
std::process::exit(if outcome == Outcome::Won { 0 } else { 1 });
}
@@ -0,0 +1,200 @@
//! SpellGrid on the command line: `cargo run --bin spellgrid -- levels/l1-double-tap.json`.
//!
//! A thin loop over the harness: it draws the level, reads a command, and calls
//! `step`, `cast` or `play`. No rules live here. After a command that ran Part I
//! Sequences, it prints their events under `Log:` as an indented tree.
//!
//! ```text
//! w a s d one turn: up, left, down, right
//! c <card> [x y] cast a spell from your hand (at the minion on (x, y), if given),
//! or play a minion from your hand onto the empty tile (x, y)
//! q quit
//! ```
//!
//! Assemble's choices (Graft) are asked for one at a time: answer with a number or a name.
use std::cell::RefCell;
use std::io::{BufRead, Write};
use std::rc::Rc;
use crate::harness::{new_spell_grid, Dir, LevelDef, SpellGrid};
/// How a game ended.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Outcome {
Won,
Lost,
Quit,
}
/// The player's input and the screen, shared with the chooser.
struct Io {
input: Box<dyn BufRead>,
output: Box<dyn Write>,
}
impl Io {
fn read_line(&mut self) -> Option<String> {
let mut line = String::new();
match self.input.read_line(&mut line) {
Ok(0) | Err(_) => None,
Ok(_) => Some(line.trim_end_matches(['\n', '\r']).to_string()),
}
}
fn write(&mut self, s: &str) {
let _ = self.output.write_all(s.as_bytes());
let _ = self.output.flush();
}
}
/// One cell of the drawing: the newest object on the tile, as 3 letters; TEXT in capitals.
fn cell(g: &SpellGrid, x: i32, y: i32) -> String {
let kinds = g.kinds_at(x, y);
// `kinds_at` is sorted; prefer text, which is what the player reads.
let Some(kind) = kinds
.iter()
.find(|k| k.starts_with("text:"))
.or(kinds.first())
else {
return " . ".to_string();
};
let name = match kind.strip_prefix("text:") {
Some(word) => word.to_uppercase(),
None => kind.to_lowercase(),
};
format!("{:<3}", name.chars().take(3).collect::<String>())
}
/// The level as text: the grid, the rules, the minions and the hand.
pub fn render(g: &SpellGrid, level: &LevelDef) -> String {
let mut lines = Vec::new();
for y in 0..level.height {
let row: Vec<String> = (0..level.width).map(|x| cell(g, x, y)).collect();
lines.push(row.join(" "));
}
lines.push(format!("Rules: {}", g.active_rules().join(", ")));
let mut minions = Vec::new();
for y in 0..level.height {
for x in 0..level.width {
if let Some(id) = g.minion_at(x, y) {
let (name, health) = (g.name_of(id), g.health(id));
minions.push(format!("{name} ({x},{y}) {health} Health"));
}
}
}
let or = |list: Vec<String>, separator: &str, none: &str| {
if list.is_empty() {
none.to_string()
} else {
list.join(separator)
}
};
lines.push(format!("Minions: {}", or(minions, "; ", "none")));
lines.push(format!("Hand: {}", or(g.hand(), ", ", "empty")));
lines.join("\n")
}
/// A coordinate, or a chooser's number: one to nine digits, nothing else (no sign, no point).
fn number(word: &str) -> Option<i32> {
let digits = (1..=9).contains(&word.len()) && word.bytes().all(|b| b.is_ascii_digit());
digits.then(|| word.parse().expect("nine digits fit an i32"))
}
/// A `c` command: `c <card words...> [x y]`.
fn parse_cast(args: &[&str]) -> (String, Option<(i32, i32)>) {
let n = args.len();
if n >= 3 {
if let (Some(x), Some(y)) = (number(args[n - 2]), number(args[n - 1])) {
return (args[..n - 2].join(" "), Some((x, y)));
}
}
(args.join(" "), None)
}
/// Play `level_text` (a level in the course's format), reading commands from `input`
/// and writing to `output`.
pub fn run(level_text: &str, input: Box<dyn BufRead>, output: Box<dyn Write>) -> Outcome {
let level: LevelDef =
serde_json::from_str(level_text).expect("a level in the course's level format");
let mut g = new_spell_grid(level.clone(), vec![]);
let io = Rc::new(RefCell::new(Io { input, output }));
let chooser_io = Rc::clone(&io);
g.set_chooser(move |kind, options| {
let mut io = chooser_io.borrow_mut();
let listed: Vec<String> = options
.iter()
.enumerate()
.map(|(i, o)| format!(" {}) {o}", i + 1))
.collect();
io.write(&format!(
"Choose the {kind} (a number or a name):\n{}\n? ",
listed.join("\n")
));
let answer = io.read_line().unwrap_or_default().trim().to_string();
if options.contains(&answer) {
return answer;
}
let picked = number(&answer)
.and_then(|n| usize::try_from(n - 1).ok())
.and_then(|i| options.get(i));
picked.unwrap_or(&options[0]).clone()
});
play(&mut g, &level, &io)
}
/// Draw, read a command, do it; until the level is won or lost, or the player quits.
fn play(g: &mut SpellGrid, level: &LevelDef, io: &Rc<RefCell<Io>>) -> Outcome {
loop {
io.borrow_mut().write(&format!("{}\n", render(g, level)));
if g.won() {
io.borrow_mut().write("You win!\n");
return Outcome::Won;
}
if g.lost() {
io.borrow_mut().write("Nothing is YOU. You lose.\n");
return Outcome::Lost;
}
io.borrow_mut().write("> ");
let Some(line) = io.borrow_mut().read_line() else {
return Outcome::Quit;
};
let words: Vec<&str> = line.split_whitespace().collect();
// The log holds only what this command does.
g.clear_log();
let dir = match words.first() {
Some(&"q") => return Outcome::Quit,
Some(&"w") => Some(Dir::Up),
Some(&"a") => Some(Dir::Left),
Some(&"s") => Some(Dir::Down),
Some(&"d") => Some(Dir::Right),
_ => None,
};
// The harness panics on a move the rules don't allow (a card not in the hand, a
// full tile). The CLI asks first, and reports the reason instead.
let done = match (dir, words.as_slice()) {
(Some(dir), _) => Ok(g.step(dir)),
(None, ["c", args @ ..]) if !args.is_empty() => match parse_cast(args) {
(card, None) => g.check_cast(&card, None).map(|()| g.cast(&card, None)),
(card, Some((x, y))) => match g.minion_at(x, y) {
Some(target) => g
.check_cast(&card, Some(target))
.map(|()| g.cast(&card, Some(target))),
None => g.check_play(&card, x, y).map(|()| {
g.play(&card, x, y);
}),
},
},
_ => {
Err("Commands: w a s d (move), c <card> [x y] (cast or play), q (quit)".to_string())
}
};
if let Err(message) = done {
io.borrow_mut().write(&format!("{message}\n"));
}
let tree = g.log_tree();
if !tree.is_empty() {
io.borrow_mut().write(&format!("Log:\n{tree}"));
}
}
}
@@ -0,0 +1,54 @@
//! Part II: what each condition word means (Module 11). The engine asks these checks
//! every time it is about to use a rule with a condition, for one object, against the
//! board as it is then. A new condition word is one entry in `words.json`, with role
//! "condition", and one check here.
use std::collections::HashSet;
use crate::engine::puzzle::{ConditionCheck, ConditionChecks, Level, ObjectId};
/// The condition checks, by condition word.
pub fn conditions() -> ConditionChecks {
ConditionChecks::from([("ON".to_string(), on as ConditionCheck)])
}
/// Can each noun be given an object of its own among `others`, one the noun names?
/// `ON ROCK AND ROCK` needs two rocks (the wiki: "AND in the noun2 part requires at
/// least as many objects as are mentioned"). Tries every assignment.
fn each_has_one(level: &Level, nouns: &[String], others: &[ObjectId]) -> bool {
fn assign(
level: &Level,
nouns: &[String],
others: &[ObjectId],
taken: &mut HashSet<ObjectId>,
) -> bool {
let Some((noun, rest)) = nouns.split_first() else {
return true;
};
for &other in others {
if taken.contains(&other) || !level.names(noun, other) {
continue;
}
taken.insert(other);
if assign(level, rest, others, taken) {
return true;
}
taken.remove(&other);
}
false
}
assign(level, nouns, others, &mut HashSet::new())
}
/// ON: for each noun, another object on the object's tile that the noun names, a
/// different one for each noun. Float levels don't matter ("ON ignores FLOAT levels").
fn on(level: &Level, id: ObjectId, nouns: &[String]) -> bool {
let o = level.board().get(id);
let others: Vec<ObjectId> = level
.board()
.at(o.x, o.y)
.map(|other| other.id)
.filter(|&other| other != id)
.collect();
each_has_one(level, nouns, &others)
}
@@ -0,0 +1,54 @@
//! Part II: what destroys what at the end of a turn. The engine runs these
//! hazards in this order (the order of the wiki's `block()`: SINK, then HOT/MELT,
//! then DEFEAT), and destroys each one's victims together before the next runs.
//! A hazard only reaches objects on the same float level: it asks the level what
//! shares a tile (`Level::sharing`). A new hazard property is one more function here.
use crate::engine::puzzle::{Hazard, Level, ObjectId};
/// The hazards, in the order the engine runs them.
pub fn hazards() -> Vec<Hazard> {
vec![sink, hot_melt, defeat]
}
/// The objects that have `property`, oldest first.
fn with(level: &Level, property: &str) -> Vec<ObjectId> {
level
.board()
.objects()
.map(|o| o.id)
.filter(|&id| level.has_property(id, property))
.collect()
}
/// A SINK object that shares its tile with anything destroys all of it, and itself.
fn sink(level: &Level) -> Vec<ObjectId> {
let mut doomed = vec![];
for id in with(level, "SINK") {
let others = level.sharing(id);
if !others.is_empty() {
doomed.push(id);
doomed.extend(others);
}
}
doomed
}
/// A MELT object is destroyed if it is HOT itself, or shares its tile with a HOT one.
fn hot_melt(level: &Level) -> Vec<ObjectId> {
destroyed_by(level, "MELT", "HOT")
}
/// A YOU object is destroyed if it is DEFEAT itself, or shares its tile with a DEFEAT one.
fn defeat(level: &Level) -> Vec<ObjectId> {
destroyed_by(level, "YOU", "DEFEAT")
}
/// The objects with `victim` that have `killer` too, or share a tile with one that does.
fn destroyed_by(level: &Level, victim: &str, killer: &str) -> Vec<ObjectId> {
let deadly = |id: ObjectId| level.has_property(id, killer);
with(level, victim)
.into_iter()
.filter(|&id| deadly(id) || level.sharing(id).into_iter().any(deadly))
.collect()
}
@@ -0,0 +1,63 @@
//! Content: the cards and the Baba Is You words. Only this directory (and the card data in `cards/`) may name
//! specific cards, or specific Baba Is You nouns.
//!
//! From Module 5 every card is a JSON file in `cards/` at the crate root, in the
//! course's card format (`course/schema/card.schema.json`). The functions that
//! were card effects in Modules 2–4 are gone: the engine interprets the data.
//! `start-module` copies the course's card files into `cards/`; add your own there.
use std::path::Path;
use std::sync::OnceLock;
use crate::engine::CardDef;
mod conditions;
mod hazards;
mod movement;
pub use conditions::conditions;
pub use hazards::hazards;
pub use movement::responses as movement;
/// Every card in `cards/`, in file-name order. Read once per test run.
pub fn cards() -> Vec<CardDef> {
static CARDS: OnceLock<Vec<CardDef>> = OnceLock::new();
CARDS
.get_or_init(|| load_dir(&Path::new(env!("CARGO_MANIFEST_DIR")).join("cards")))
.clone()
}
/// Parse every `.json` file in `dir` as a card definition.
pub fn load_dir(dir: &Path) -> Vec<CardDef> {
let entries = std::fs::read_dir(dir)
.unwrap_or_else(|e| panic!("cannot read card directory {}: {e}", dir.display()));
let mut paths: Vec<_> = entries
.map(|entry| entry.expect("card directory entry").path())
.filter(|p| p.extension().and_then(|x| x.to_str()) == Some("json"))
.collect();
paths.sort();
paths
.iter()
.map(|path| {
let text = std::fs::read_to_string(path)
.unwrap_or_else(|e| panic!("cannot read {}: {e}", path.display()));
serde_json::from_str(&text)
.unwrap_or_else(|e| panic!("{} is not a valid card: {e}", path.display()))
})
.collect()
}
/// Part III: the words that cross between the engines (`seams.json`). A noun that
/// names a trigger, and a property that multiplies one: `DEATHRATTLE IS DOUBLE`.
pub fn seams() -> crate::engine::SeamWords {
serde_json::from_str(include_str!("seams.json"))
.expect("src/content/seams.json is not a seam word table")
}
/// Part II: the words of Baba Is You the parser can read, besides the engine's own
/// (`IS`, `AND`, `NOT`, `TEXT`). Nouns and properties are data: a new word is one
/// more entry in `words.json` in this directory. A condition word (Module 11) also
/// needs its check in `conditions.rs`.
pub fn words() -> Vec<crate::engine::puzzle::WordDef> {
serde_json::from_str(include_str!("words.json"))
.expect("src/content/words.json is not a word list")
}
@@ -0,0 +1,44 @@
//! Part II: what each property does when an object moves. The engine plans a move
//! one step at a time and runs these responses for every step, in this order (the
//! order of the wiki's movement checks: OPEN/SHUT, then PUSH, then STOP). A new
//! movement property is one more response here.
use crate::engine::puzzle::{Planner, Response, Step};
/// The movement responses, in the order the engine runs them.
pub fn responses() -> Vec<Response> {
vec![open_shut, push, stop]
}
/// An OPEN object stepping onto a SHUT one (or a SHUT one onto an OPEN one)
/// destroys both, and that object doesn't block the step. One partner per step:
/// the oldest.
fn open_shut(p: &mut Planner, s: Step) -> bool {
for (mine, theirs) in [("OPEN", "SHUT"), ("SHUT", "OPEN")] {
if !p.has(s.mover, mine) {
continue;
}
let partner = p.staying_at(s.to).into_iter().find(|&o| p.has(o, theirs));
if let Some(partner) = partner {
p.destroy(partner);
p.destroy(s.mover);
return true;
}
}
true
}
/// Every PUSH object on the tile ahead must step the same way, or the step fails.
fn push(p: &mut Planner, s: Step) -> bool {
for o in p.staying_at(s.to) {
if p.has(o, "PUSH") && !p.try_move(o) {
return false;
}
}
true
}
/// A STOP object still on the tile ahead blocks the step.
fn stop(p: &mut Planner, s: Step) -> bool {
!p.staying_at(s.to).into_iter().any(|o| p.has(o, "STOP"))
}
@@ -0,0 +1,4 @@
{
"triggerWords": { "DEATHRATTLE": "deathrattle" },
"multiplierWords": { "DOUBLE": { "times": 2, "stacking": "max" } }
}
@@ -0,0 +1,36 @@
[
{ "name": "BABA", "role": "noun" },
{ "name": "ROCK", "role": "noun" },
{ "name": "WALL", "role": "noun" },
{ "name": "FLAG", "role": "noun" },
{ "name": "LAVA", "role": "noun" },
{ "name": "KEY", "role": "noun" },
{ "name": "DOOR", "role": "noun" },
{ "name": "SKULL", "role": "noun" },
{ "name": "WATER", "role": "noun" },
{ "name": "ALL", "role": "noun", "selects": "things" },
{ "name": "KEG", "role": "noun" },
{ "name": "BRAMBLE", "role": "noun" },
{ "name": "HEDGE", "role": "noun" },
{ "name": "WARDEN", "role": "noun" },
{ "name": "MOTH", "role": "noun" },
{ "name": "SCRIBE", "role": "noun" },
{ "name": "KNIGHT", "role": "noun" },
{ "name": "CRIER", "role": "noun" },
{ "name": "GOLEM", "role": "noun" },
{ "name": "SHARD", "role": "noun" },
{ "name": "DEATHRATTLE", "role": "noun" },
{ "name": "ON", "role": "condition" },
{ "name": "YOU", "role": "property" },
{ "name": "WIN", "role": "property" },
{ "name": "PUSH", "role": "property" },
{ "name": "STOP", "role": "property" },
{ "name": "OPEN", "role": "property" },
{ "name": "SHUT", "role": "property" },
{ "name": "HOT", "role": "property" },
{ "name": "MELT", "role": "property" },
{ "name": "SINK", "role": "property" },
{ "name": "DEFEAT", "role": "property" },
{ "name": "FLOAT", "role": "property" },
{ "name": "DOUBLE", "role": "property" }
]
@@ -0,0 +1,174 @@
//! Assemble: "Get a random minion. Change half its text."
//!
//! Card text is a list of clauses, so changing half of it is list surgery: the base
//! card's `text[..split]` (top) or `text[split..]` (bottom) is replaced by the same
//! half of a donor card. Name, cost and stats stay the base's. This works only
//! because a card's behaviour is its data: the new card needs no new code.
//!
//! Hearthstone's wiki doesn't define "half". The course's `split` field is its
//! definition: each card's data says where its top half ends.
use std::fmt;
use super::card::{well_formed, CardDef, PoolFilter};
use super::world::{EntityId, PlayerId, World, Zone};
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Half {
Top,
Bottom,
}
impl Half {
pub const BOTH: [Half; 2] = [Half::Top, Half::Bottom];
pub fn label(self) -> &'static str {
match self {
Half::Top => "top",
Half::Bottom => "bottom",
}
}
}
/// Why an Assemble was refused.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum AssembleFailure {
/// The new text would equal the base's.
Unchanged,
/// The new text would not be well-formed (see `well_formed`).
IllFormed,
}
/// Assembling `base` with `donor`'s `half` would leave the base's text unchanged, or ill-formed.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct AssembleError {
pub base: String,
pub donor: String,
pub half: Half,
pub reason: AssembleFailure,
}
impl fmt::Display for AssembleError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
let result = match self.reason {
AssembleFailure::Unchanged => "leaves its text unchanged",
AssembleFailure::IllFormed => "gives ill-formed text",
};
write!(
f,
"assembling {:?} with the {} half of {:?} {result}",
self.base,
self.half.label(),
self.donor
)
}
}
impl std::error::Error for AssembleError {}
/// The base card with `half` of its text replaced by the same half of `donor`.
/// It keeps the base's type, cost and stats, and is named "<base> (Assembled)".
pub fn assemble(base: &CardDef, donor: &CardDef, half: Half) -> Result<CardDef, AssembleError> {
let (base_top, base_bottom) = base.text.split_at(base.split);
let (donor_top, donor_bottom) = donor.text.split_at(donor.split);
let (top, bottom) = match half {
Half::Top => (donor_top, base_bottom),
Half::Bottom => (base_top, donor_bottom),
};
let text = [top, bottom].concat();
let refuse = |reason| AssembleError {
base: base.name.clone(),
donor: donor.name.clone(),
half,
reason,
};
if text == base.text {
return Err(refuse(AssembleFailure::Unchanged));
}
if !well_formed(&text, base.card_type) {
return Err(refuse(AssembleFailure::IllFormed));
}
Ok(CardDef {
name: format!("{} (Assembled)", base.name),
split: top.len(),
text,
..base.clone()
})
}
/// Register an assembled card and return its name. If a different card already has
/// the name (an earlier Assemble of the same base), a number is added: "(Assembled 2)".
pub fn register_assembled(world: &mut World, mut def: CardDef) -> String {
let stem = def.name.clone();
let mut n = 1;
while world.has_card(&def.name) && world.card(&def.name).def != def {
n += 1;
def.name = format!("{} {n})", stem.trim_end_matches(')'));
}
let name = def.name.clone();
world.register_assembled(def);
name
}
/// The `assemble` effect: the owner's chooser picks a base minion from the pool
/// (filtered), a donor, and the half to replace. Only combinations that change the
/// base's text and leave it well-formed are offered: a donor is offered if one of its
/// halves gives such a text, and a base if it has such a donor. The Assembled minion
/// goes to the owner's hand. Real Hearthstone picks the base at random and offers two
/// donors; here every choice is the chooser's.
pub fn assemble_from_pool(
world: &mut World,
player: PlayerId,
filter: &PoolFilter,
) -> Option<EntityId> {
// Each pool card, and whether the filter lets it be the base.
let pool: Vec<(CardDef, bool)> = world
.pool()
.into_iter()
.map(|c| {
(
c.def.clone(),
filter.trigger.map_or(true, |on| c.has_trigger(on)),
)
})
.collect();
// Does assembling `base` with `donor`'s `half` give a new, well-formed text?
let changes = |base: &CardDef, donor: &CardDef, half: Half| assemble(base, donor, half).is_ok();
let donors_for = |base: &CardDef| -> Vec<String> {
pool.iter()
.map(|(d, _)| d)
.filter(|d| d.name != base.name && Half::BOTH.iter().any(|&h| changes(base, d, h)))
.map(|d| d.name.clone())
.collect()
};
let bases: Vec<String> = pool
.iter()
.filter(|(d, allowed)| *allowed && !donors_for(d).is_empty())
.map(|(d, _)| d.name.clone())
.collect();
if bases.is_empty() {
return None;
}
let base_name = world.choose("base", &bases);
let base = world.card(&base_name).def.clone();
let donor_name = world.choose("donor", &donors_for(&base));
let donor = world.card(&donor_name).def.clone();
let halves: Vec<String> = Half::BOTH
.iter()
.filter(|&&h| changes(&base, &donor, h))
.map(|h| h.label().to_string())
.collect();
let half = world.choose("half", &halves);
let half = if half == "top" {
Half::Top
} else {
Half::Bottom
};
let def = assemble(&base, &donor, half)
.expect("only halves that give a new, well-formed text are offered");
let name = register_assembled(world, def);
Some(world.create(&name, player, Zone::Hand))
}
@@ -0,0 +1,84 @@
//! Auras, the one routine that applies trigger multipliers, and the lookup for
//! damage redirects.
//!
//! An aura is data on a card: a continuous effect its minion has while in play.
//! The engine never asks "is that card on the board?" in the middle of resolving
//! something. It takes a snapshot of the auras in force at defined points (an Aura
//! Update) and every rule reads the snapshot.
//!
//! Hearthstone runs two kinds of Aura Update when the outermost Phase ends: one for
//! Health/Attack auras before the Death Creation Step, and one for every other aura
//! after it. Trigger multipliers are "Other" auras, so a minion that dies in a Death
//! Creation Step no longer multiplies the Deathrattles of that step's Death Phase.
//! Summoning a minion also runs an Aura Update.
//!
//! Module 10: auras can come from outside the cards. SpellGrid compiles rules on its
//! grid into auras (`World::set_rule_auras`); they join the snapshot at the next Aura
//! Update, beside the minions' own, and go through the same routine. A rule's
//! multiplier stacks with a card's by the same `stacking` field. A cap (`IS NOT`)
//! limits the total, whatever multiplies it.
use super::card::{Aura, Stacking, TriggerOn};
use super::world::{ActiveAura, EntityId, PlayerId, World};
/// Aura Update (Other): replace the snapshot with the auras of every minion now in
/// play, and the auras from outside the cards, which apply to both players.
pub fn aura_update_other(world: &mut World) {
let mut active = Vec::new();
for id in world.minions_in_play() {
for &aura in &world.card(world.card_name(id)).auras {
active.push(ActiveAura {
owner: world.owner_of(id),
source: Some(id),
aura,
});
}
}
for &aura in world.rule_auras() {
for owner in 0..2 {
active.push(ActiveAura {
owner,
source: None,
aura,
});
}
}
world.set_active_auras(active);
}
/// The minion that takes damage meant for `player`'s hero, if a redirect is in force:
/// the oldest (order of play) of `player`'s minions with a `DamageRedirect` aura in the
/// snapshot. The snapshot lists minions oldest first, so that is the first one found.
pub fn redirect_for(world: &World, player: PlayerId) -> Option<EntityId> {
world
.active_auras()
.iter()
.find(|a| matches!(a.aura, Aura::DamageRedirect { .. }) && a.owner == player)
.and_then(|a| a.source)
}
/// How many times a trigger listening for `on`, on a minion `owner` controls, activates
/// when it resolves. Reads the snapshot from the last Aura Update. With no multiplier in
/// force the answer is 1. Multipliers with `Stacking::Max` contribute their largest
/// `times`; each `Stacking::Additive` one adds `times - 1` extra activations. A cap
/// (`Aura::TriggerCap`) then limits the total to its `times`.
pub fn activations(world: &World, owner: PlayerId, on: TriggerOn) -> u32 {
let mut max = 1;
let mut extra = 0;
let mut cap = u32::MAX;
for active in world.active_auras().iter().filter(|a| a.owner == owner) {
match active.aura {
Aura::TriggerCap { trigger, times } if trigger == on => cap = cap.min(times),
Aura::TriggerMultiplier {
trigger,
times,
stacking,
} if trigger == on => match stacking {
Stacking::Max => max = max.max(times),
Stacking::Additive => extra += times - 1,
},
_ => {}
}
}
cap.min(max + extra)
}
@@ -0,0 +1,342 @@
//! Cards as data. A `CardDef` is the course's card format (`course/schema/card.schema.json`),
//! read straight from JSON. The engine compiles each definition once, when it is
//! registered, into the lists it reads while resolving: triggers, auras and a
//! spell's effects. The definitions themselves live in `cards/`, loaded by `content`.
use std::collections::HashMap;
use serde::Deserialize;
/// What a triggered ability listens for.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "camelCase")]
pub enum TriggerOn {
/// This minion takes damage.
SelfDamaged,
/// A minion with the same owner (this one included) takes damage.
FriendlyMinionDamaged,
/// Any minion takes damage.
AnyMinionDamaged,
/// Either hero takes damage.
HeroDamaged,
/// This minion dies. It resolves after the minion has left play.
Deathrattle,
/// This minion is played from hand.
Battlecry,
/// The end of its owner's turn. The course has no turns, so it never fires; card
/// data may still name it, for example in a trigger multiplier.
EndOfTurn,
}
impl TriggerOn {
/// The trigger's family name, as reported in `Triggered` events.
pub fn label(self) -> &'static str {
match self {
TriggerOn::SelfDamaged
| TriggerOn::FriendlyMinionDamaged
| TriggerOn::AnyMinionDamaged
| TriggerOn::HeroDamaged => "damaged",
TriggerOn::Deathrattle => "deathrattle",
TriggerOn::Battlecry => "battlecry",
TriggerOn::EndOfTurn => "endOfTurn",
}
}
}
/// How several trigger multipliers for the same trigger combine.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "camelCase")]
pub enum Stacking {
/// The largest multiplier wins: two copies of a doubler still double.
Max,
/// Each copy adds its extra activations.
Additive,
}
/// An aura: a continuous effect a minion has while it is in play. The engine reads
/// auras only at the Aura Update points (see `aura.rs`), never mid-Phase.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(tag = "type", rename_all = "camelCase")]
pub enum Aura {
/// Triggers listening for `trigger`, belonging to minions with the aura owner's
/// owner, activate `times` times each.
TriggerMultiplier {
trigger: TriggerOn,
times: u32,
stacking: Stacking,
},
/// Module 10: triggers listening for `trigger` activate at most `times` times,
/// whatever multiplies them. Not in the card format: SpellGrid compiles it from a
/// negated rule (`DEATHRATTLE IS NOT DOUBLE`).
TriggerCap { trigger: TriggerOn, times: u32 },
/// Damage that would be dealt to the aura owner's hero is dealt to the aura's
/// minion instead (see `DAMAGE_STEPS` in `queue.rs`).
DamageRedirect { from: RedirectFrom, to: RedirectTo },
}
/// Whose damage a redirect takes: the card format's `"friendlyHero"`.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "camelCase")]
pub enum RedirectFrom {
FriendlyHero,
}
/// Who takes redirected damage. `This` is the card format's `"self"`: the aura's minion.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
pub enum RedirectTo {
#[serde(rename = "self")]
This,
}
/// Who an effect touches. `This` is the schema's `"self"`.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "camelCase")]
pub enum Target {
#[serde(rename = "self")]
This,
AllMinions,
AllOtherMinions,
AllOtherCharacters,
EnemyHero,
Chosen,
}
/// Which minions an Assemble may pick as its base. An empty filter allows any.
#[derive(Debug, Clone, Default, PartialEq, Eq, Deserialize)]
pub struct PoolFilter {
/// Only minions with a trigger clause listening for this.
#[serde(default)]
pub trigger: Option<TriggerOn>,
}
/// One effect: a node of a card's ability, interpreted by `effect.rs`.
#[derive(Debug, Clone, PartialEq, Eq, Deserialize)]
#[serde(tag = "type", rename_all = "camelCase")]
pub enum Effect {
DealDamage {
target: Target,
amount: i32,
},
Draw {
count: u32,
},
Summon {
card: String,
count: u32,
},
GainArmor {
amount: i32,
},
GainAttack {
target: Target,
amount: i32,
},
Destroy {
target: Target,
},
/// Part III (SpellGrid) only: puts a word on the grid.
SpawnText {
word: String,
at: String,
},
/// Get a minion from the pool, with half its text replaced (see `assemble.rs`).
Assemble {
filter: PoolFilter,
},
/// Give every minion in the owner's hand +attack/+health.
BuffHand {
attack: i32,
health: i32,
},
}
/// One line of card text. A trigger clause applies to the effect clauses after it,
/// until the next trigger, keyword or aura. Effects before any trigger are what a
/// spell does when cast.
#[derive(Debug, Clone, PartialEq, Eq, Deserialize)]
#[serde(tag = "kind", rename_all = "camelCase")]
pub enum Clause {
Keyword { keyword: String },
Trigger { on: TriggerOn },
Aura { modifier: Aura },
Effect { effect: Effect },
}
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "camelCase")]
pub enum CardType {
Minion,
Spell,
}
/// A card as data, in the course's card format. Fields the engine doesn't use
/// (`fictional`, `game`) are ignored. `noun` is read by SpellGrid (Module 10).
#[derive(Debug, Clone, PartialEq, Eq, Deserialize)]
pub struct CardDef {
pub name: String,
#[serde(rename = "type")]
pub card_type: CardType,
pub cost: i32,
#[serde(default)]
pub attack: i32,
#[serde(default)]
pub health: i32,
pub text: Vec<Clause>,
/// `text[..split]` is the top half, `text[split..]` the bottom half.
pub split: usize,
/// The noun a SpellGrid minion has on the grid (Module 10).
#[serde(default)]
pub noun: Option<String>,
}
/// A triggered ability: "whenever <on>, <effects>".
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Trigger {
pub on: TriggerOn,
pub effects: Vec<Effect>,
}
/// A registered card: its definition, and the definition's text compiled into the
/// lists the engine reads while resolving.
#[derive(Debug, Clone)]
pub struct Card {
pub def: CardDef,
/// Triggered abilities, in printed order.
pub triggers: Vec<Trigger>,
/// Auras this minion has while in play.
pub auras: Vec<Aura>,
/// What the card does when cast (a spell's effects).
pub on_cast: Vec<Effect>,
/// May an Assemble offer this card? Minions with text that were not themselves assembled.
pub in_pool: bool,
}
impl Card {
/// Compile a definition's text. Panics if `split` is past the end of the text.
pub fn compile(def: CardDef, in_pool: bool) -> Card {
assert!(
def.split <= def.text.len(),
"{:?}: split {} is past the end of its text",
def.name,
def.split
);
let mut triggers: Vec<Trigger> = Vec::new();
let mut auras = Vec::new();
let mut on_cast = Vec::new();
// Is the latest clause a trigger whose effects are still being listed?
let mut open_trigger = false;
for clause in &def.text {
match clause {
Clause::Trigger { on } => {
triggers.push(Trigger {
on: *on,
effects: Vec::new(),
});
open_trigger = true;
}
Clause::Effect { effect } if open_trigger => {
triggers.last_mut().unwrap().effects.push(effect.clone());
}
Clause::Effect { effect } => on_cast.push(effect.clone()),
Clause::Aura { modifier } => {
auras.push(*modifier);
open_trigger = false;
}
// The course has no combat, so keywords such as Taunt do nothing here.
Clause::Keyword { .. } => open_trigger = false,
}
}
let in_pool = in_pool && def.card_type == CardType::Minion && !def.text.is_empty();
Card {
def,
triggers,
auras,
on_cast,
in_pool,
}
}
pub fn is_spell(&self) -> bool {
self.def.card_type == CardType::Spell
}
pub fn has_trigger(&self, on: TriggerOn) -> bool {
self.triggers.iter().any(|t| t.on == on)
}
}
/// Does `text` mean what it says? For a minion, every effect clause belongs to a
/// trigger (none comes before the first trigger, or after an aura or keyword), and
/// every trigger has at least one effect before the next trigger, aura, keyword or
/// the end. The rule is for minions (Assemble makes only minions): a spell passes.
pub fn well_formed(text: &[Clause], card_type: CardType) -> bool {
if card_type != CardType::Minion {
return true;
}
// The trigger whose effects are being listed: none, one with no effect yet, or one with some.
#[derive(PartialEq)]
enum Open {
None,
Empty,
Filled,
}
let mut open = Open::None;
for clause in text {
match clause {
Clause::Effect { .. } if open == Open::None => return false,
Clause::Effect { .. } => open = Open::Filled,
_ if open == Open::Empty => return false,
Clause::Trigger { .. } => open = Open::Empty,
Clause::Aura { .. } | Clause::Keyword { .. } => open = Open::None,
}
}
open != Open::Empty
}
/// Every card the game can create, looked up by name.
#[derive(Debug, Default)]
pub struct CardLibrary {
cards: HashMap<String, Card>,
}
impl CardLibrary {
pub fn new(defs: Vec<CardDef>) -> CardLibrary {
let mut library = CardLibrary::default();
for def in defs {
library.register(def);
}
library
}
/// Add a card, or replace the card with the same name. It joins the Assemble pool
/// if it is a minion with text.
pub fn register(&mut self, def: CardDef) {
self.insert(Card::compile(def, true));
}
/// Add a card made by Assemble. It never joins the pool.
pub fn register_assembled(&mut self, def: CardDef) {
self.insert(Card::compile(def, false));
}
fn insert(&mut self, card: Card) {
self.cards.insert(card.def.name.clone(), card);
}
pub fn contains(&self, name: &str) -> bool {
self.cards.contains_key(name)
}
pub fn get(&self, name: &str) -> &Card {
self.cards
.get(name)
.unwrap_or_else(|| panic!("unknown card {name:?}: add its file to cards/"))
}
/// The Assemble pool: every card with `in_pool`, sorted by name.
pub fn pool(&self) -> Vec<&Card> {
let mut pool: Vec<&Card> = self.cards.values().filter(|c| c.in_pool).collect();
pool.sort_by(|a, b| a.def.name.cmp(&b.def.name));
pool
}
}
@@ -0,0 +1,41 @@
//! The Death Creation Step: the one place minions leave play.
//!
//! Damage and destroy effects only mark a minion (0 or less Health, or pending
//! destroy). It stays in play, and keeps triggering, until the outermost Phase
//! ends. Then this step removes every marked minion at once.
use super::event::Event;
use super::world::{EntityId, World, Zone};
/// Remove every mortally wounded or pending-destroy minion from play, simultaneously.
/// Returns their Death Events in order of play; empty if nothing died.
pub fn death_creation_step(world: &mut World) -> Vec<Event> {
// Every neighbour is read before anything moves: removal is simultaneous. Each Death
// Event remembers where its minion stood as a handle, the minion that stays to its
// right, not as a board index: an index goes stale as soon as a minion further left
// leaves play in the same step (Module 1: hold handles, not indexes).
let mut deaths = Vec::new();
for id in world.minions_in_play() {
if world.is_doomed(id) {
let before = survivor_to_the_right(world, id);
deaths.push(Event::Destroyed { id, before });
}
}
for death in &deaths {
if let Event::Destroyed { id, before } = *death {
world.kill(id, before);
}
}
deaths
}
/// The nearest minion to the right of `id` on its board that is not leaving play in this step, if any.
fn survivor_to_the_right(world: &World, id: EntityId) -> Option<EntityId> {
world
.zone(world.owner_of(id), Zone::Board)
.iter()
.copied()
.skip_while(|&other| other != id)
.skip(1)
.find(|&other| !world.is_doomed(other))
}
@@ -0,0 +1,98 @@
//! The interpreter: turns one effect, as data, into engine actions.
//!
//! This is the only place that knows what an effect *means*. A card's text is a
//! small tree (clauses holding effects holding targets); the engine walks it. The
//! effect vocabulary is closed: a card can only do what this match can do, which is
//! what makes a data card safe to load.
use super::card::{Effect, Target};
use super::queue::{Action, Ctx};
use super::world::EntityId;
/// The actions one effect performs, in order, given where and for whom it runs.
pub fn actions(effect: &Effect, ctx: &Ctx) -> Vec<Action> {
let owner = ctx.owner;
match effect {
Effect::DealDamage { target, amount } => vec![Action::Damage {
targets: targets(*target, ctx),
amount: *amount,
}],
Effect::Draw { count } => (0..*count)
.map(|_| Action::Draw { player: owner })
.collect(),
// Several tokens stand left to right in summon order: each goes right of the one before.
Effect::Summon { card, count } => {
let at = summon_position(ctx);
(0..*count)
.map(|i| Action::Summon {
card: card.clone(),
player: owner,
at: at.map(|at| at + i as usize),
})
.collect()
}
Effect::GainArmor { amount } => vec![Action::GainArmor {
player: owner,
amount: *amount,
}],
Effect::GainAttack { target, amount } => targets(*target, ctx)
.into_iter()
.map(|id| Action::GainAttack {
id,
amount: *amount,
})
.collect(),
Effect::Destroy { target } => vec![Action::Destroy {
targets: targets(*target, ctx),
}],
Effect::Assemble { filter } => vec![Action::Assemble {
player: owner,
filter: filter.clone(),
}],
Effect::BuffHand { attack, health } => ctx
.world
.minions_in_hand(owner)
.into_iter()
.map(|id| Action::Buff {
id,
attack: *attack,
health: *health,
})
.collect(),
Effect::SpawnText { word, .. } => vec![Action::SpawnText {
source: ctx.minion(),
word: word.clone(),
}],
}
}
/// The entities a target names, from where the effect runs.
fn targets(target: Target, ctx: &Ctx) -> Vec<EntityId> {
let world = ctx.world;
let others = |ids: Vec<EntityId>| -> Vec<EntityId> {
ids.into_iter()
.filter(|&id| Some(id) != ctx.source)
.collect()
};
match target {
Target::This => vec![ctx.minion()],
Target::AllMinions => world.minions_in_play(),
Target::AllOtherMinions => others(world.minions_in_play()),
Target::AllOtherCharacters => {
let mut ids = vec![world.hero(0), world.hero(1)];
ids.extend(world.minions_in_play());
others(ids)
}
Target::EnemyHero => vec![ctx.enemy_hero()],
Target::Chosen => ctx.target.into_iter().collect(),
}
}
/// Where a summon goes: to the right of a source still on the board; where a dead
/// source stood (left of the minion its Death Event remembers); at the right end for a spell.
fn summon_position(ctx: &Ctx) -> Option<usize> {
match ctx.source.and_then(|s| ctx.world.board_index(s)) {
Some(index) => Some(index + 1),
None => ctx.position,
}
}
@@ -0,0 +1,72 @@
//! What happened, in order. The engine records an event each time it changes the game.
//! The world also keeps the nesting depth each event was recorded at (`World::log_depths`).
use super::card::TriggerOn;
use super::world::{EntityId, PlayerId};
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Event {
/// `id` entered the board. `by` is the minion whose triggered ability put it there (the
/// trigger resolving when it entered), or `None` for a spell, a play or a summon from
/// outside a Sequence. Module 10: SpellGrid puts it on `by`'s grid tile. The course's log
/// doesn't show `by`.
Summoned {
id: EntityId,
by: Option<EntityId>,
},
Damaged {
target: EntityId,
amount: i32,
},
Triggered {
source: EntityId,
on: TriggerOn,
},
Drew {
player: PlayerId,
id: EntityId,
},
Fatigue {
player: PlayerId,
},
ArmorGained {
player: PlayerId,
amount: i32,
},
/// Armor absorbed `amount` of a Damage to `player`'s hero. The `Damaged` that follows
/// shows the full amount.
ArmorLost {
player: PlayerId,
amount: i32,
},
StatsChanged {
id: EntityId,
attack: i32,
health: i32,
},
SpellCast {
card: String,
},
/// A Death Event: the Death Creation Step moved `id` from play to the graveyard.
/// `before` is the nearest minion to its right that did not die in the same step, if any:
/// its Deathrattle summons immediately left of that minion, or at the right end. It is a
/// handle because a board index goes stale when a minion further left dies in the same step.
Destroyed {
id: EntityId,
before: Option<EntityId>,
},
DeathPhase {
round: u32,
},
/// A minion was played from hand (it has just entered the board). Its Battlecry answers this.
Played {
id: EntityId,
},
/// Module 10: `source`'s effect asked for `word` to be written on its grid tile.
/// Part I has no grid: it only records the request. SpellGrid reads it when the
/// Sequence ends.
TextSpawned {
source: EntityId,
word: String,
},
}
@@ -0,0 +1,62 @@
//! Engine core: generic rules machinery. Never name a specific card here.
//!
//! Module 1: every card in the game is an entity addressed by an id handle.
//! The engine knows which zone each entity is in, who owns it, and when it
//! entered play.
//!
//! Module 2: heroes, Health, Attack and Armor; triggered abilities; and the
//! action queue that resolves a Sequence depth-first, recording an event log.
//!
//! Module 3: deferred death. Damage and destroy effects only mark minions; the
//! Death Creation Step removes them when the outermost Phase ends, and Death
//! Phases resolve their Deathrattles until no new deaths occur.
//!
//! Module 4: auras as card data, Aura Updates at defined points, and one
//! routine that decides how many times a queued trigger activates. Playing a
//! minion from hand, and Battlecries. A second interceptor, replacement: the
//! steps that rewrite a Damage before it lands (a redirect, then Armor).
//!
//! Module 5: behaviour as data. Cards are definitions in the course's card format,
//! loaded from JSON; one interpreter turns their effects into actions. Assemble
//! builds a new card by swapping half of one card's text for half of another's, and
//! offers only results that are well-formed. A Sequence is bounded: a trigger never
//! nests inside itself, and more than `DEATH_PHASE_LIMIT` Death Phases is an error.
//!
//! Part II lives in `puzzle`: a second engine, for the rule-writing puzzle game,
//! beside the first.
//! Module 6: the grid, the lexer and the parser.
//! Module 7: `IS NOT`, and the resolution pass that turns parsed rules into rules in force.
//! Module 8: movement, planned in full and then committed; push chains.
//! Module 9: the turn, as a list of phases; hazards; winning.
//! Module 11: conditions (`X ON Z IS P`), checked for each object when a rule is used.
//!
//! Part III lives in `spellgrid`, the only module that knows both engines.
//! Module 10: SpellGrid. Part I gains auras from outside the cards
//! (`World::set_rule_auras`), a cap on activations, the `spawnText` effect (recorded as a
//! `TextSpawned` event) and the end-of-Phase steps as a list (`END_OF_PHASE`); Part II
//! gains a destroy function.
mod assemble;
mod aura;
mod card;
mod death;
mod effect;
mod event;
pub mod puzzle;
mod queue;
mod spellgrid;
mod world;
pub use assemble::{assemble, register_assembled, AssembleError, AssembleFailure, Half};
pub use aura::aura_update_other;
pub use card::{
well_formed, Aura, Card, CardDef, CardLibrary, CardType, Clause, Effect, PoolFilter,
RedirectFrom, RedirectTo, Stacking, Target, Trigger, TriggerOn,
};
pub use event::Event;
pub use queue::{
run_sequence, summon, Action, Ctx, DamageStep, EndOfPhaseStep, DAMAGE_STEPS, DEATH_PHASE_LIMIT,
END_OF_PHASE,
};
pub use spellgrid::{rule_auras, Multiplier, SeamWords, SpellGrid};
pub use world::{Chooser, EntityId, PlayerId, World, Zone, HERO_HEALTH};
@@ -0,0 +1,21 @@
//! Conditions (Module 11): `X ON Z IS P` gives an X the property P only while it is on a Z.
//!
//! The parser reads a condition into the rule (`Rule::conditions`); the engine never
//! decides it there. A condition is evaluated for one object, against the board as
//! it is, every time a rule is about to be used: by `Level::has_property` (and so by
//! movement, the hazards and the win check), and by conversion. What a condition
//! word means is a `ConditionCheck` that content provides, registered by word, as
//! movement responses and hazards are. The engine knows only that a word with role
//! `Condition` may stand between the subjects and `IS`.
use std::collections::HashMap;
use super::grid::ObjectId;
use super::level::Level;
/// Content's meaning for one condition word: does the condition, with these nouns
/// (in the order they are written), hold for object `id` on the board as it is now?
pub type ConditionCheck = fn(&Level, ObjectId, &[String]) -> bool;
/// Content's condition checks, by condition word: `{"ON": on}`.
pub type ConditionChecks = HashMap<String, ConditionCheck>;
@@ -0,0 +1,175 @@
//! The grid: objects on tiles. `x` grows right, `y` grows down, and any number of
//! objects may share a tile.
use std::collections::BTreeMap;
use std::fmt;
use serde::Deserialize;
pub type ObjectId = u32;
/// A level in the course's level format (`course/schema/level.schema.json`).
#[derive(Debug, Clone, Default, Deserialize)]
pub struct LevelDef {
pub width: i32,
pub height: i32,
#[serde(default)]
pub objects: Vec<LevelObject>,
/// Module 10, read only by SpellGrid: minions in play, each on its tile.
#[serde(default)]
pub cards: Vec<LevelCard>,
/// Module 10, read only by SpellGrid: the cards in player 0's hand.
#[serde(default)]
pub hand: Vec<String>,
}
/// A minion card on the grid (Module 10).
#[derive(Debug, Clone, Deserialize)]
pub struct LevelCard {
pub card: String,
pub x: i32,
pub y: i32,
}
#[derive(Debug, Clone, Deserialize)]
pub struct LevelObject {
pub kind: String,
pub x: i32,
pub y: i32,
}
/// What an object is: a thing named by a noun (kind `"<noun>"`, the noun in lower case),
/// or a piece of text carrying one word (kind `"text:<WORD>"`).
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Kind {
Thing { noun: String },
Text { word: String },
}
impl Kind {
const TEXT_PREFIX: &'static str = "text:";
pub fn parse(kind: &str) -> Kind {
match kind.strip_prefix(Self::TEXT_PREFIX) {
Some(word) => Kind::Text {
word: word.to_uppercase(),
},
None => Kind::Thing {
noun: kind.to_uppercase(),
},
}
}
}
/// The kind as it is placed: `"<noun>"` in lower case, or `"text:<WORD>"`.
impl fmt::Display for Kind {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
Kind::Thing { noun } => write!(f, "{}", noun.to_lowercase()),
Kind::Text { word } => write!(f, "{}{word}", Self::TEXT_PREFIX),
}
}
}
#[derive(Debug, Clone)]
pub struct Object {
pub id: ObjectId,
pub kind: Kind,
pub x: i32,
pub y: i32,
}
#[derive(Debug, Clone, Default)]
pub struct Board {
pub width: i32,
pub height: i32,
objects: BTreeMap<ObjectId, Object>,
next_id: ObjectId,
}
impl Board {
pub fn new(width: i32, height: i32) -> Board {
Board {
width,
height,
..Default::default()
}
}
pub fn in_bounds(&self, x: i32, y: i32) -> bool {
(0..self.width).contains(&x) && (0..self.height).contains(&y)
}
/// Add an object; ids start at 1 and are never reused. Panics off the board.
pub fn place(&mut self, kind: Kind, x: i32, y: i32) -> ObjectId {
assert!(
self.in_bounds(x, y),
"({x}, {y}) is outside the {}x{} level",
self.width,
self.height
);
self.next_id += 1;
let id = self.next_id;
self.objects.insert(id, Object { id, kind, x, y });
id
}
/// An id that no object will ever have (Module 10): SpellGrid gives one to each
/// card that isn't on the grid, so cards and objects share one id space.
pub fn reserve_id(&mut self) -> ObjectId {
self.next_id += 1;
self.next_id
}
/// The object `id`. Panics if there is none.
pub fn get(&self, id: ObjectId) -> &Object {
self.objects
.get(&id)
.unwrap_or_else(|| panic!("no object {id}"))
}
/// The object `id`, or `None` once it has been removed. Panics on an id this
/// board never issued.
pub fn find(&self, id: ObjectId) -> Option<&Object> {
assert!((1..=self.next_id).contains(&id), "no object {id}");
self.objects.get(&id)
}
/// Take object `id` off the board for good.
pub fn remove(&mut self, id: ObjectId) {
self.objects.remove(&id);
}
/// Move object `id` by `(dx, dy)`. Panics if it would leave the board.
pub fn shift(&mut self, id: ObjectId, (dx, dy): (i32, i32)) {
let (width, height) = (self.width, self.height);
let o = self
.objects
.get_mut(&id)
.unwrap_or_else(|| panic!("no object {id}"));
let (x, y) = (o.x + dx, o.y + dy);
assert!(
(0..width).contains(&x) && (0..height).contains(&y),
"({x}, {y}) is outside the {width}x{height} level"
);
(o.x, o.y) = (x, y);
}
/// Make object `id` a `kind`, where it stands. Panics if there is no such object.
pub fn set_kind(&mut self, id: ObjectId, kind: Kind) {
self.objects
.get_mut(&id)
.unwrap_or_else(|| panic!("no object {id}"))
.kind = kind;
}
/// The objects on tile (x, y), oldest first.
pub fn at(&self, x: i32, y: i32) -> impl Iterator<Item = &Object> {
self.objects().filter(move |o| (o.x, o.y) == (x, y))
}
/// Every object, oldest first.
pub fn objects(&self) -> impl Iterator<Item = &Object> {
self.objects.values()
}
}
@@ -0,0 +1,382 @@
//! A level in play: the board, the words content provides, and the rules compiled
//! from them. Every change to the board recompiles the rules from scratch: parse,
//! then resolve. Every question about what an object is goes through `has_property`.
//!
//! A move (Module 8) is planned for each `YOU` object, then committed; the rules are
//! reparsed once, after all of them.
//!
//! A turn (Module 9) runs the phases of `TURN` in order: movement, parse,
//! conversion, parse, content's hazards, the win check, parse.
//!
//! Undo (Module 9): before each move or turn the level saves the board (every
//! object's id, kind and position, oldest first, and the id counter) and `won`.
//! The rules are not saved: undo restores the board and reparses.
//!
//! Module 10: what happens to the objects a turn destroys is a destroy function the
//! caller may pass in (`step_with`, `move_you_with`). The default, `remove_all`,
//! removes them. SpellGrid passes one that sends minions to Part I's Death Creation
//! Step instead. SpellGrid doesn't offer undo: Part I has state the level's
//! snapshot doesn't cover.
//!
//! Module 11: a rule may carry conditions (`X ON Z IS P`). They are not
//! decided when the rules are compiled. Every question about an object
//! (`has_property`, conversion) asks content's `ConditionCheck`s against the board
//! as it is at that moment, and uses only the rules whose conditions hold for that
//! object.
use std::collections::BTreeSet;
use super::condition::ConditionChecks;
use super::grid::{Board, Kind, LevelDef, ObjectId};
use super::movement::{Dir, Plan, Planner, Response};
use super::parse::{parse, Rule};
use super::resolve::resolve;
use super::turn::{Hazard, Phase, TURN};
use super::word::{Lexicon, Role, WordDef};
#[derive(Debug, Clone)]
pub struct Level {
board: Board,
lexicon: Lexicon,
responses: Vec<Response>,
hazards: Vec<Hazard>,
conditions: ConditionChecks,
rules: Vec<Rule>,
active: Vec<Rule>,
won: bool,
/// What `undo` puts back: one entry per move or turn, the latest last (Module 9).
history: Vec<Snapshot>,
}
/// The level before a move or a turn: the whole board (its objects, oldest first,
/// and its id counter) and `won`.
#[derive(Debug, Clone)]
struct Snapshot {
board: Board,
won: bool,
}
/// The property that makes an object move when the player moves.
pub const YOU: &str = "YOU";
/// The property that wins the level when a `YOU` object is on it, or is it.
pub const WIN: &str = "WIN";
/// The property that puts an object on the other float level: only objects on
/// the same level share a tile for hazards and for winning.
pub const FLOAT: &str = "FLOAT";
/// What to do with objects a move or a turn destroys (Module 10). It may change the
/// board, and reparse, before it returns. An id may appear more than once.
pub type Destroy<'a> = dyn FnMut(&mut Level, &[ObjectId]) + 'a;
/// The default destroy function: take every object off the board.
pub fn remove_all(level: &mut Level, ids: &[ObjectId]) {
for &id in ids {
level.board.remove(id);
}
}
impl Level {
/// A level from its definition, the words content provides, content's movement
/// responses and hazards, each in the order they are to run, and content's
/// condition checks, by word. Panics if a word with role `Condition` has no check.
pub fn new(
def: &LevelDef,
words: &[WordDef],
responses: Vec<Response>,
hazards: Vec<Hazard>,
conditions: ConditionChecks,
) -> Level {
for w in words {
if w.role == Role::Condition && !conditions.contains_key(&w.name.to_uppercase()) {
panic!("the condition word {} has no condition check", w.name);
}
}
let mut level = Level {
board: Board::new(def.width, def.height),
lexicon: Lexicon::new(words),
responses,
hazards,
conditions,
rules: vec![],
active: vec![],
won: false,
history: vec![],
};
for o in &def.objects {
level.board.place(Kind::parse(&o.kind), o.x, o.y);
}
level.reparse();
level
}
pub fn place(&mut self, kind: &str, x: i32, y: i32) -> ObjectId {
let id = self.board.place(Kind::parse(kind), x, y);
self.history.clear();
self.reparse();
id
}
pub fn board(&self) -> &Board {
&self.board
}
/// The board, to change directly (Module 10: SpellGrid writes text and removes
/// minions). Nothing is reparsed until `reparse` is called.
pub fn board_mut(&mut self) -> &mut Board {
&mut self.board
}
/// The movement responses, in the order they run.
pub fn responses(&self) -> &[Response] {
&self.responses
}
/// Every rule the text spells, as parsed: negated rules and duplicates included.
pub fn rules(&self) -> &[Rule] {
&self.rules
}
/// The rules in force after the resolution pass.
pub fn active_rules(&self) -> &[Rule] {
&self.active
}
/// Does object `id` have `property`? Yes if an active rule that applies to it
/// (`applies_to`) gives it the property and no active negated rule that applies
/// to it takes it away. A destroyed object has no properties.
pub fn has_property(&self, id: ObjectId, property: &str) -> bool {
if self.board.find(id).is_none() {
return false;
}
let about_it: Vec<&Rule> = self
.active
.iter()
.filter(|r| r.complement == property && self.applies_to(r, id))
.collect();
about_it.iter().any(|r| !r.negated) && !about_it.iter().any(|r| r.negated)
}
/// Does the noun `word`, as the subject of a rule, name object `id`?
pub fn names(&self, word: &str, id: ObjectId) -> bool {
self.lexicon.names(word, &self.board.get(id).kind)
}
/// Does `rule` apply to object `id` now? Its subject names the object, and each
/// of its conditions holds for the object on the board as it is (Module 11).
fn applies_to(&self, rule: &Rule, id: ObjectId) -> bool {
self.names(&rule.subject, id)
&& rule
.conditions
.iter()
.all(|c| (self.conditions[&c.word])(self, id, &c.nouns))
}
/// The other objects on object `id`'s tile that are on its float level (both
/// `FLOAT` or both not), oldest first. Hazards and the win check ask this.
pub fn sharing(&self, id: ObjectId) -> Vec<ObjectId> {
let o = self.board.get(id);
let floats = self.has_property(id, FLOAT);
self.board
.at(o.x, o.y)
.filter(|other| other.id != id && self.has_property(other.id, FLOAT) == floats)
.map(|other| other.id)
.collect()
}
/// Has a turn's win check found a `YOU` object on a `WIN` one? Stays true.
pub fn won(&self) -> bool {
self.won
}
/// Is there no `YOU` object left?
pub fn lost(&self) -> bool {
!self.board.objects().any(|o| self.has_property(o.id, YOU))
}
/// Move every `YOU` object one tile in `dir`, then reparse (Module 8). Movement
/// only: no conversion, hazard or win.
pub fn move_you(&mut self, dir: Dir) {
self.move_you_with(dir, &mut remove_all);
}
/// `move_you`, with destroyed objects going to `destroy` (Module 10).
pub fn move_you_with(&mut self, dir: Dir, destroy: &mut Destroy) {
self.save();
self.movement(dir, destroy);
self.reparse();
}
/// Play one turn: the phases of `TURN`, in order.
pub fn step(&mut self, dir: Dir) {
self.step_with(dir, &mut remove_all);
}
/// `step`, with destroyed objects going to `destroy` (Module 10).
pub fn step_with(&mut self, dir: Dir, destroy: &mut Destroy) {
self.save();
for phase in TURN {
match phase {
Phase::Movement => self.movement(dir, destroy),
Phase::Parse => self.reparse(),
Phase::Conversion => self.convert(destroy),
Phase::Hazards => self.run_hazards(destroy),
Phase::Win => self.check_win(),
}
}
}
/// Put the level back as it was before the last move or turn (Module 9), and
/// reparse. False, and no change, if there is none; `place` forgets them all.
pub fn undo(&mut self) -> bool {
let Some(last) = self.history.pop() else {
return false;
};
self.board = last.board;
self.won = last.won;
self.reparse();
true
}
fn save(&mut self) {
let snapshot = Snapshot {
board: self.board.clone(),
won: self.won,
};
self.history.push(snapshot);
}
/// The `YOU` objects move one at a time, the one furthest ahead in `dir` first
/// (oldest first on a tie); each one's move is planned in full, then committed,
/// before the next is planned. Every move uses the rules in force before the
/// first.
///
/// An object steps at most once in a movement phase. A `YOU` object that an
/// earlier move already carried along (pushed, or moved by any other response)
/// doesn't start a move of its own, and no later move can make it step again.
fn movement(&mut self, dir: Dir, destroy: &mut Destroy) {
let (dx, dy) = dir.delta();
let mut you: Vec<_> = self
.board
.objects()
.filter(|o| self.has_property(o.id, YOU))
.map(|o| (-(o.x * dx + o.y * dy), o.id))
.collect();
you.sort();
let mut stepped = BTreeSet::new();
for (_, id) in you {
if self.board.find(id).is_none() {
continue; // destroyed by an earlier move
}
if stepped.contains(&id) {
continue; // already moved, by an earlier move
}
let mut planner = Planner::new(self, dir, &stepped);
if planner.try_move(id) {
let plan = planner.into_plan();
self.commit(&plan, dir, destroy);
stepped.extend(plan.moving.difference(&plan.destroyed));
}
}
}
/// Destroy, then move what is still on the board (a destroy function may remove
/// more than it was given).
fn commit(&mut self, plan: &Plan, dir: Dir, destroy: &mut Destroy) {
if !plan.destroyed.is_empty() {
let ids: Vec<ObjectId> = plan.destroyed.iter().copied().collect();
destroy(self, &ids);
}
for &id in plan.moving.difference(&plan.destroyed) {
if self.board.find(id).is_some() {
self.board.shift(id, dir.delta());
}
}
}
/// Every object named by a conversion `X IS Y` (`Y` a noun) becomes a `Y`
/// where it stands, keeping its id; with several `Y`s it becomes the first,
/// and each other `Y` is a new object on its tile. An `X IS X` naming the
/// object blocks this. An object named by `X IS NOT X` is deleted, and still
/// makes its `Y`s, as new objects: the negation cancelled `X IS X`.
///
/// Negation is settled per object, as in `has_property`: an object doesn't
/// become a `Y` if any active `... IS NOT Y` names it, whatever noun gave it the
/// conversion (`ALL IS Y` with `X IS NOT Y` leaves every X an X).
///
/// Every object is decided from the board as movement left it, before any
/// changes, so `X IS Y` and `Y IS X` swap the Xs and the Ys. Only the rules that
/// apply to the object are used: their conditions are checked on that board too.
fn convert(&mut self, destroy: &mut Destroy) {
let mut changes = vec![];
for o in self.board.objects() {
let about_it: Vec<&Rule> = self
.active
.iter()
.filter(|r| self.applies_to(r, o.id))
.collect();
let deleted = about_it.iter().any(|r| r.negated && r.is_identity());
let locked = about_it.iter().any(|r| !r.negated && r.is_identity());
let mut into: Vec<Kind> = vec![];
if !locked {
for r in about_it.iter().filter(|r| !r.negated && r.is_conversion()) {
let negated = |n: &&Rule| n.negated && n.complement == r.complement;
if about_it.iter().any(negated) {
continue;
}
match self.lexicon.thing_named(&r.complement) {
Some(kind) if kind != o.kind && !into.contains(&kind) => into.push(kind),
_ => {}
}
}
}
if deleted || !into.is_empty() {
changes.push((o.id, (o.x, o.y), deleted, into));
}
}
let mut deleted_ids = vec![];
for (id, (x, y), deleted, into) in changes {
let mut into = into.into_iter();
if deleted {
deleted_ids.push(id);
} else if let Some(first) = into.next() {
self.board.set_kind(id, first);
}
for kind in into {
self.board.place(kind, x, y);
}
}
if !deleted_ids.is_empty() {
destroy(self, &deleted_ids);
}
}
/// Run content's hazards in order. Each one's victims are destroyed together
/// before the next runs; no reparse until the turn's last phase.
fn run_hazards(&mut self, destroy: &mut Destroy) {
for i in 0..self.hazards.len() {
let hazard = self.hazards[i];
let victims = hazard(self);
if !victims.is_empty() {
destroy(self, &victims);
}
}
}
/// The level is won if any `YOU` object is `WIN`, or shares its tile with a
/// `WIN` object on its float level.
fn check_win(&mut self) {
let wins = |id: ObjectId| self.has_property(id, WIN);
let won = self
.board
.objects()
.filter(|o| self.has_property(o.id, YOU))
.any(|o| wins(o.id) || self.sharing(o.id).into_iter().any(wins));
self.won |= won;
}
/// Recompile the rules from the board: parse, then resolve.
pub fn reparse(&mut self) {
self.rules = parse(&self.board, &self.lexicon);
self.active = resolve(&self.rules);
}
}
@@ -0,0 +1,47 @@
//! Part II: the rule-writing puzzle game. Objects have no behaviour of their own; every rule is
//! compiled from the text on the board, and recompiled whenever the board changes.
//!
//! Module 6: the grid, the lexer (text objects to words with grammatical roles)
//! and the parser (words to rules). The engine knows the operators `IS`, `AND`,
//! `NOT`, the meta-noun `TEXT` and the base rule `TEXT IS PUSH`. Every other word,
//! nouns included, is data handed in by content.
//!
//! Module 7: `IS NOT` in the parser, and the resolution pass (negation, then the
//! identity lock) that turns the parsed rules into the rules in force. Objects are
//! asked about through `Level::has_property`, which knows which nouns name them.
//!
//! Module 8: movement. `Level::move_you` plans each `YOU` object's move in full
//! (`Planner`), then commits it, then reparses. What a property does when something
//! moves is a `Response` that content provides; the engine runs them in order.
//!
//! Module 9: the turn. `Level::step` runs the phases of `TURN` in order: movement,
//! parse, conversion, parse, hazards, win, parse. What a hazard property destroys
//! is a `Hazard` that content provides; the engine knows `YOU`, `WIN` and `FLOAT`.
//!
//! Module 10: `Level::step_with` and `Level::move_you_with` take a destroy function,
//! which decides what happens to the objects they destroy; `Board::reserve_id` hands
//! out an id no object has. The puzzle engine still knows nothing of cards: SpellGrid
//! (`../spellgrid.rs`) uses these.
//!
//! Module 11: conditions. A word with role `Condition` may stand between a rule's
//! subjects and `IS` (`X ON Z IS P`); the rule carries it (`Rule::conditions`).
//! What it means is a `ConditionCheck` content registers by word. `Level` asks the
//! checks every time it uses a rule, against the board as it is then.
mod condition;
mod grid;
mod level;
mod movement;
mod parse;
mod resolve;
mod turn;
mod word;
pub use condition::{ConditionCheck, ConditionChecks};
pub use grid::{Board, Kind, LevelCard, LevelDef, LevelObject, Object, ObjectId};
pub use level::{remove_all, Level, FLOAT, WIN, YOU};
pub use movement::{Dir, Plan, Planner, Response, Step};
pub use parse::{parse, Condition, Rule, BASE_RULES};
pub use resolve::resolve;
pub use turn::{Hazard, Phase, TURN};
pub use word::{lex, Lexicon, Role, Selects, Token, WordDef};
@@ -0,0 +1,144 @@
//! Movement: validate, then commit.
//!
//! A move is planned before anything on the board changes. The `Planner` asks
//! whether one object can step one tile; to answer, it runs every movement
//! response content provides for that step, and a response may ask the planner
//! to move other objects too (a push), or to destroy objects. If any response
//! says no, the planner forgets everything that step added to the plan, so a
//! failed push leaves no trace. Only a plan that succeeded is committed.
//!
//! The engine knows the board, its edge, and the order of the steps. What each
//! property does when something moves (`PUSH`, `STOP`, `OPEN`/`SHUT`, ...) is a
//! response in content, run in the order content lists them.
use std::collections::BTreeSet;
use super::grid::ObjectId;
use super::level::Level;
/// One of the four directions a move can take.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Dir {
Up,
Down,
Left,
Right,
}
impl Dir {
/// One tile in this direction, as `(dx, dy)`; `y` grows down.
pub fn delta(self) -> (i32, i32) {
match self {
Dir::Up => (0, -1),
Dir::Down => (0, 1),
Dir::Left => (-1, 0),
Dir::Right => (1, 0),
}
}
}
/// One object stepping one tile, from `from` to `to`, in direction `dir`.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct Step {
pub mover: ObjectId,
pub from: (i32, i32),
pub to: (i32, i32),
pub dir: Dir,
}
/// A movement response: content's rule for what happens when an object steps.
/// It returns `false` to block the step. It may plan other moves or destroy
/// objects through the planner.
pub type Response = fn(&mut Planner<'_>, Step) -> bool;
/// What a move will do: the objects that step one tile, and the objects destroyed.
/// A destroyed object doesn't step, even if it was pushed.
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct Plan {
pub moving: BTreeSet<ObjectId>,
pub destroyed: BTreeSet<ObjectId>,
}
/// Plans one move on a board it only reads. `stepped` holds the objects that have
/// already stepped in this movement phase, in an earlier committed plan: an object
/// steps at most once per phase, so none of them steps again.
pub struct Planner<'a> {
level: &'a Level,
dir: Dir,
stepped: &'a BTreeSet<ObjectId>,
plan: Plan,
}
impl<'a> Planner<'a> {
pub fn new(level: &'a Level, dir: Dir, stepped: &'a BTreeSet<ObjectId>) -> Planner<'a> {
Planner {
level,
dir,
stepped,
plan: Plan::default(),
}
}
/// Does object `id` have `property`, under the rules in force before the move?
pub fn has(&self, id: ObjectId, property: &str) -> bool {
self.level.has_property(id, property)
}
/// The objects that will still be on tile `at` when the plan so far is
/// committed: those there now, less those planned to move away or be destroyed.
/// Oldest first.
pub fn staying_at(&self, (x, y): (i32, i32)) -> Vec<ObjectId> {
self.level
.board()
.at(x, y)
.map(|o| o.id)
.filter(|id| !self.plan.moving.contains(id) && !self.plan.destroyed.contains(id))
.collect()
}
/// Plan object `id`'s destruction.
pub fn destroy(&mut self, id: ObjectId) {
self.plan.destroyed.insert(id);
}
/// Plan object `id` stepping one tile in the move's direction, with whatever
/// the responses add. Returns whether it can. If it can't, the plan is left as
/// it was before the call. An object already planned to move, or to be
/// destroyed, is out of the way: the answer is yes, and nothing is added. An
/// object that has already stepped in this movement phase doesn't step again, and
/// doesn't hold the mover back either: the answer is yes, and nothing is added, so
/// objects that began the phase stacked on one tile end it stacked.
pub fn try_move(&mut self, id: ObjectId) -> bool {
if self.plan.moving.contains(&id) || self.plan.destroyed.contains(&id) {
return true;
}
if self.stepped.contains(&id) {
return true;
}
let level = self.level;
let o = level.board().get(id);
let (dx, dy) = self.dir.delta();
let to = (o.x + dx, o.y + dy);
// The edge of the level blocks every step (the course's choice; see the lesson).
if !level.board().in_bounds(to.0, to.1) {
return false;
}
let before = self.plan.clone();
self.plan.moving.insert(id);
let step = Step {
mover: id,
from: (o.x, o.y),
to,
dir: self.dir,
};
let allowed = level.responses().iter().all(|respond| respond(self, step));
if !allowed {
self.plan = before;
}
allowed
}
pub fn into_plan(self) -> Plan {
self.plan
}
}
@@ -0,0 +1,313 @@
//! The parser: words on the grid become rules.
//!
//! Grammar (Module 7): `NOUN [AND NOUN]* IS [NOT] (NOUN|PROPERTY) [AND [NOT] (NOUN|PROPERTY)]*`,
//! read left to right in a row and top to bottom in a column. The parser starts at
//! every `IS`, collects the subjects before it and the complements after it, and
//! emits one rule per pair: AND distributes over both sides. A tile holding several
//! words contributes each of them, so stacked text yields every possible rule.
//!
//! `NOT` before a complement negates that one complement. `NOT` before a subject is
//! not in the grammar: the run of subjects stops there and that subject is dropped.
//!
//! Module 11: conditions. `NOUN [AND NOUN]* [CONDITION NOUN [AND NOUN]*] IS ...`: a word
//! whose role is `Condition` may stand between the subjects and `IS`, followed by
//! the nouns it is about. The condition belongs to every subject of the sentence,
//! and AND among its nouns means all of them. A run of nouns before `IS` is the
//! condition's when a condition word stands before it, and then it is not a run of
//! subjects. Stacked text still yields every possible rule: a stack of nouns after
//! the condition word gives one rule per noun, as a stack after `IS` does.
//!
//! The parser only reads; which rules win is the resolution pass (`resolve`), and
//! whether a condition holds is decided for each object when someone asks (`Level`).
use std::collections::BTreeMap;
use std::fmt;
use super::grid::{Board, ObjectId};
use super::word::{lex, Lexicon, Role, Token, AND, IS, NOT, TEXT};
/// A condition on a rule (Module 11): the condition word and the nouns after it, in
/// the order they are written. `X ON Y AND Z IS P` has `word: "ON"`,
/// `nouns: ["Y", "Z"]`.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Condition {
pub word: String,
pub nouns: Vec<String>,
}
/// One rule: `subject IS complement`, or `subject IS NOT complement` when `negated`,
/// applying to an object only while each of its `conditions` holds for that object
/// (none for an unconditional rule). `text` is the text objects that spell it
/// (subject, condition word and nouns if any, `IS`, `NOT` if any, complement), or
/// empty for a base rule.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Rule {
pub subject: String,
pub complement: String,
pub complement_role: Role,
pub negated: bool,
pub conditions: Vec<Condition>,
pub text: Vec<ObjectId>,
}
impl Rule {
/// `X IS Y` where `Y` is a noun: every X is to become a Y.
pub fn is_conversion(&self) -> bool {
self.complement_role == Role::Noun
}
/// `X IS X` (or `X IS NOT X`).
pub fn is_identity(&self) -> bool {
self.subject == self.complement
}
/// Same subject, complement and conditions, opposite sign: `X IS NOT Y` negates
/// `X IS Y`, and `X ON Z IS NOT Y` negates `X ON Z IS Y` (Module 11), but not `X IS Y`.
pub fn negates(&self, other: &Rule) -> bool {
self.negated
&& !other.negated
&& self.subject == other.subject
&& self.complement == other.complement
&& self.same_conditions(other)
}
/// The same conditions: the same words, each with the same nouns in any order.
pub fn same_conditions(&self, other: &Rule) -> bool {
let key = |r: &Rule| -> Vec<(String, Vec<String>)> {
r.conditions
.iter()
.map(|c| {
let mut nouns = c.nouns.clone();
nouns.sort();
(c.word.clone(), nouns)
})
.collect()
};
key(self) == key(other)
}
}
/// The normalised form: `"TEXT IS PUSH"`, `"TEXT IS NOT PUSH"`, `"X ON Y AND Z IS P"`.
impl fmt::Display for Rule {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "{}", self.subject)?;
for c in &self.conditions {
write!(f, " {} {}", c.word, c.nouns.join(&format!(" {AND} ")))?;
}
let not = if self.negated { " NOT" } else { "" };
write!(f, " {IS}{not} {}", self.complement)
}
}
/// Rules in force with no text: the game's base rules (`(subject, property)`).
pub const BASE_RULES: &[(&str, &str)] = &[(TEXT, "PUSH")];
type Tiles = BTreeMap<(i32, i32), Vec<Token>>;
/// A word in a run, with the `NOT` in front of it if there is one.
struct Word<'a> {
not: Option<&'a Token>,
token: &'a Token,
}
/// Every rule the board spells, base rules first, negated rules included. A rule
/// spelled by two different sets of text objects appears twice.
pub fn parse(board: &Board, lexicon: &Lexicon) -> Vec<Rule> {
let tiles = lex(board, lexicon);
let mut rules: Vec<Rule> = BASE_RULES
.iter()
.map(|&(subject, property)| Rule {
subject: subject.to_string(),
complement: property.to_string(),
complement_role: Role::Property,
negated: false,
conditions: vec![],
text: vec![],
})
.collect();
for (&(x, y), tokens) in &tiles {
for is in tokens.iter().filter(|t| t.word == IS) {
for (dx, dy) in [(1, 0), (0, 1)] {
let complements = complements(&tiles, (x + dx, y + dy), (dx, dy));
for reading in before_is(&tiles, (x - dx, y - dy), (-dx, -dy)) {
for s in &reading.subjects {
for c in &complements {
let mut text = vec![s.id];
text.extend(&reading.text);
text.push(is.id);
text.extend(c.not.map(|n| n.id));
text.push(c.token.id);
rules.push(Rule {
subject: s.word.clone(),
complement: c.token.word.clone(),
complement_role: c.token.role,
negated: c.not.is_some(),
conditions: reading.conditions.clone(),
text,
});
}
}
}
}
}
}
rules
}
fn words_at(tiles: &Tiles, p: (i32, i32)) -> &[Token] {
tiles.get(&p).map(Vec::as_slice).unwrap_or(&[])
}
fn holds(tiles: &Tiles, p: (i32, i32), word: &str) -> bool {
words_at(tiles, p).iter().any(|t| t.word == word)
}
fn ahead(p: (i32, i32), step: (i32, i32)) -> (i32, i32) {
(p.0 + step.0, p.1 + step.1)
}
/// What stands before an `IS` (Module 11): the subjects, with the condition on them
/// if there is one.
struct Reading<'a> {
subjects: Vec<&'a Token>,
conditions: Vec<Condition>,
/// The condition word and its nouns, in reading order (empty without a condition).
text: Vec<ObjectId>,
}
/// A run of words walking backwards, and the tile the walk stopped on, before the
/// run's first word.
struct Run<'a> {
tokens: Vec<&'a Token>,
stop: (i32, i32),
}
fn has_condition(tiles: &Tiles, p: (i32, i32)) -> bool {
words_at(tiles, p).iter().any(|t| t.role == Role::Condition)
}
/// Every way to read the words before an `IS`, walking backwards from `at`.
///
/// - The run of subjects, without a condition, unless a condition word stands
/// before the run: then those nouns are the condition's, not subjects.
/// - For every run of nouns that has a condition word before it, the run of
/// subjects before that word, with the condition. A stack among the condition's
/// nouns gives one reading per noun (`noun_runs`).
fn before_is(tiles: &Tiles, at: (i32, i32), step: (i32, i32)) -> Vec<Reading<'_>> {
let mut readings = vec![];
let run = subjects(tiles, at, step);
if !has_condition(tiles, run.stop) {
readings.push(Reading {
subjects: run.tokens,
conditions: vec![],
text: vec![],
});
}
for nouns in noun_runs(tiles, at, step) {
let in_order: Vec<&Token> = nouns.tokens.iter().rev().copied().collect();
let words = words_at(tiles, nouns.stop).iter();
for word in words.filter(|t| t.role == Role::Condition) {
let mut text = vec![word.id];
text.extend(in_order.iter().map(|t| t.id));
readings.push(Reading {
subjects: subjects(tiles, ahead(nouns.stop, step), step).tokens,
conditions: vec![Condition {
word: word.word.clone(),
nouns: in_order.iter().map(|t| t.word.clone()).collect(),
}],
text,
});
}
}
readings
}
/// The run of subjects before an `IS`, walking backwards from `at`: the nouns at
/// `at`, then, if the tile before them holds an AND, the run beyond it. A `NOT`
/// before a noun ends the run without it (a negated subject is not in the grammar).
fn subjects(tiles: &Tiles, at: (i32, i32), step: (i32, i32)) -> Run<'_> {
let before = ahead(at, step);
if holds(tiles, before, NOT) {
return Run {
tokens: vec![],
stop: before,
};
}
let mut found: Vec<&Token> = words_at(tiles, at)
.iter()
.filter(|t| t.role == Role::Noun)
.collect();
if !found.is_empty() && holds(tiles, before, AND) {
let rest = subjects(tiles, ahead(before, step), step);
if !rest.tokens.is_empty() {
found.extend(rest.tokens);
return Run {
tokens: found,
stop: rest.stop,
};
}
}
Run {
tokens: found,
stop: before,
}
}
/// Every run of nouns walking backwards from `at`, one noun from each tile, joined
/// by AND; each with the tile before its first noun. Runs that end early, at an
/// AND, are included: a condition word stacked on that AND starts a reading. A
/// `NOT` before a noun ends every run through it.
fn noun_runs(tiles: &Tiles, at: (i32, i32), step: (i32, i32)) -> Vec<Run<'_>> {
let before = ahead(at, step);
if holds(tiles, before, NOT) {
return vec![];
}
let mut runs = vec![];
for noun in words_at(tiles, at).iter().filter(|t| t.role == Role::Noun) {
runs.push(Run {
tokens: vec![noun],
stop: before,
});
if !holds(tiles, before, AND) {
continue;
}
for rest in noun_runs(tiles, ahead(before, step), step) {
let mut tokens = vec![noun];
tokens.extend(rest.tokens);
runs.push(Run {
tokens,
stop: rest.stop,
});
}
}
runs
}
/// The run of complements after an `IS`, walking forwards from `at`: the nouns
/// and properties at `at`, or, if `at` holds a `NOT`, those on the tile after it,
/// negated; then, if the next tile holds an AND, the run beyond it.
fn complements(tiles: &Tiles, at: (i32, i32), step: (i32, i32)) -> Vec<Word<'_>> {
let fits = |t: &&Token| matches!(t.role, Role::Noun | Role::Property);
let mut found = vec![];
// Plain words here, and negated words one tile on: a stack may hold both.
let mut branches: Vec<(Option<&Token>, (i32, i32))> = vec![(None, at)];
for not in words_at(tiles, at).iter().filter(|t| t.word == NOT) {
branches.push((Some(not), ahead(at, step)));
}
for (not, p) in branches {
let words: Vec<Word> = words_at(tiles, p)
.iter()
.filter(fits)
.map(|token| Word { not, token })
.collect();
if words.is_empty() {
continue;
}
found.extend(words);
let and_at = ahead(p, step);
if holds(tiles, and_at, AND) {
found.extend(complements(tiles, ahead(and_at, step), step));
}
}
found
}
@@ -0,0 +1,39 @@
//! The resolution pass: from every rule the text spells to the rules in force.
//!
//! The parser reports what is written; this pass decides what wins, over the whole
//! rule list at once, so it doesn't matter where on the board each rule is written:
//!
//! 1. **Negation.** `X IS NOT Y` cancels every `X IS Y`. The negated rule stays in
//! force: it keeps overriding, and `X IS NOT X` is also the record that every X is
//! to vanish (the turn removes them, Module 9).
//! 2. **Identity.** A surviving `X IS X` cancels every conversion `X IS Y` (`Y` a noun):
//! X can't become anything else. It doesn't stop other nouns becoming X. Because
//! negation runs first, `X IS NOT X` cancels `X IS X` and conversions work again.
//!
//! Base rules are rules like any other, so `TEXT IS NOT PUSH` cancels `TEXT IS PUSH`.
//!
//! Module 11: both steps pair rules only when their conditions are the same.
//! `X ON Z IS NOT P` cancels `X ON Z IS P` here, but not `X IS P`:
//! which of those wins depends on where each X stands, so it is settled per
//! object when someone asks (`Level::has_property`, and conversion), among the
//! rules whose conditions hold for that object.
use super::parse::Rule;
/// The rules in force, in the order they were parsed.
pub fn resolve(rules: &[Rule]) -> Vec<Rule> {
let survives_negation: Vec<&Rule> = rules
.iter()
.filter(|r| !rules.iter().any(|n| n.negates(r)))
.collect();
let locked = |r: &Rule| {
survives_negation.iter().any(|l| {
!l.negated && l.is_identity() && l.subject == r.subject && l.same_conditions(r)
})
};
survives_negation
.iter()
.filter(|r| r.negated || !r.is_conversion() || r.is_identity() || !locked(r))
.map(|&r| r.clone())
.collect()
}
@@ -0,0 +1,44 @@
//! The turn (Module 9): one move of the player, as a fixed list of phases.
//!
//! The course's turn is a labelled simplification of the game's (see the lesson):
//! movement, parse, conversion, parse, hazards, win, and a last parse so that
//! the rules match the board when the turn ends. Each phase reads the rules from
//! the parse before it, so a rule that movement breaks is gone before hazards run.
//!
//! What a hazard property does (`SINK`, `HOT`/`MELT`, `DEFEAT`, ...) is a `Hazard`
//! that content provides, as with movement's responses. The engine runs them in
//! content's order and owns only the phases, `YOU`, `WIN` and float levels.
use super::grid::ObjectId;
use super::level::Level;
/// One phase of the turn.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Phase {
/// Every `YOU` object moves (Module 8), without the reparse.
Movement,
/// The rules are recompiled from the board.
Parse,
/// `X IS Y` conversions, and `X IS NOT X` deletions.
Conversion,
/// Content's hazards, in content's order.
Hazards,
/// A `YOU` object on a `WIN` one (or `WIN` itself) wins the level.
Win,
}
/// The turn, in order.
pub const TURN: [Phase; 7] = [
Phase::Movement,
Phase::Parse,
Phase::Conversion,
Phase::Parse,
Phase::Hazards,
Phase::Win,
Phase::Parse,
];
/// A hazard: content's rule for what destroys what at the end of a turn. It reads
/// the level and returns the objects to destroy; the engine destroys them all at
/// once, before the next hazard runs. An id may appear more than once.
pub type Hazard = fn(&Level) -> Vec<ObjectId>;
@@ -0,0 +1,135 @@
//! The lexer: text objects become words with a grammatical role.
//!
//! The engine owns the grammar words (`IS`, `AND`, `NOT`) and the meta-noun `TEXT`.
//! Every other word comes from content as data: the engine never names a noun.
//! A noun also says which objects it names (`Selects`), so a meta-noun that names
//! a whole group of objects can be data too.
//!
//! Module 11: a fourth role, `Condition`. The engine knows only where a condition
//! word may stand in a sentence (between the subjects and `IS`); what it means is a
//! check content provides (`ConditionCheck`).
use std::collections::{BTreeMap, HashMap};
use serde::Deserialize;
use super::grid::{Board, Kind, ObjectId};
#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserialize)]
#[serde(rename_all = "lowercase")]
pub enum Role {
Noun,
Property,
Operator,
Condition,
}
/// Which objects a noun names when it is the subject of a rule.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default, Deserialize)]
#[serde(rename_all = "lowercase")]
pub enum Selects {
/// The things of its own name: `<NOUN>` names every object of kind `"<noun>"`.
#[default]
Named,
/// Every text object (the meta-noun `TEXT`).
Text,
/// Every object that is not text, whatever its noun.
Things,
}
/// One entry of the word list content provides: `{ "name": "<NOUN>", "role": "noun" }`,
/// optionally with `"selects"` (default `"named"`).
#[derive(Debug, Clone, Deserialize)]
pub struct WordDef {
pub name: String,
pub role: Role,
#[serde(default)]
pub selects: Selects,
}
pub const IS: &str = "IS";
pub const AND: &str = "AND";
pub const NOT: &str = "NOT";
/// The meta-noun for every text object.
pub const TEXT: &str = "TEXT";
/// Every word the parser can read, with its role.
#[derive(Debug, Clone, Default)]
pub struct Lexicon {
roles: HashMap<String, Role>,
selects: HashMap<String, Selects>,
}
impl Lexicon {
/// The engine's own words plus `words`. Content can't redefine an engine word.
pub fn new(words: &[WordDef]) -> Lexicon {
let mut roles: HashMap<String, Role> = words
.iter()
.map(|w| (w.name.to_uppercase(), w.role))
.collect();
let mut selects: HashMap<String, Selects> = words
.iter()
.map(|w| (w.name.to_uppercase(), w.selects))
.collect();
for op in [IS, AND, NOT] {
roles.insert(op.to_string(), Role::Operator);
}
roles.insert(TEXT.to_string(), Role::Noun);
selects.insert(TEXT.to_string(), Selects::Text);
Lexicon { roles, selects }
}
/// `None` for a word the lexicon doesn't know: such text spells nothing.
pub fn role(&self, word: &str) -> Option<Role> {
self.roles.get(word).copied()
}
/// Does the noun `word`, as the subject of a rule, name an object of this kind?
pub fn names(&self, word: &str, kind: &Kind) -> bool {
let selects = self.selects.get(word).copied().unwrap_or_default();
match (selects, kind) {
(Selects::Named, Kind::Thing { noun }) => noun == word,
(Selects::Text, Kind::Text { .. }) => true,
(Selects::Things, Kind::Thing { .. }) => true,
_ => false,
}
}
/// What a conversion into the noun `word` makes (Module 9): a thing of that
/// name, if `word` names the things of its own name. `None` for a noun that
/// names a group (`TEXT`, or `ALL` if content adds it): the course doesn't
/// convert into those.
pub fn thing_named(&self, word: &str) -> Option<Kind> {
let selects = self.selects.get(word).copied().unwrap_or_default();
(selects == Selects::Named).then(|| Kind::Thing {
noun: word.to_string(),
})
}
}
/// A piece of text the parser can read: which object, which word, which role.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Token {
pub id: ObjectId,
pub word: String,
pub role: Role,
}
/// The words on the board, by tile; a tile's words in object order (oldest first).
/// Text whose word the lexicon doesn't know is left out.
pub fn lex(board: &Board, lexicon: &Lexicon) -> BTreeMap<(i32, i32), Vec<Token>> {
let mut tiles: BTreeMap<(i32, i32), Vec<Token>> = BTreeMap::new();
for object in board.objects() {
let Kind::Text { word } = &object.kind else {
continue;
};
if let Some(role) = lexicon.role(word) {
tiles.entry((object.x, object.y)).or_default().push(Token {
id: object.id,
word: word.clone(),
role,
});
}
}
tiles
}
@@ -0,0 +1,400 @@
//! The action queue: how one player action (a Sequence) resolves.
//!
//! Performing an action changes the world and produces events. Every trigger that
//! responds to those events goes into one Queue, oldest listener first, and the Queue
//! is locked. Each queued trigger then resolves completely, including everything
//! its own actions cause, before the next one starts: depth-first.
//!
//! Module 3: nothing leaves play mid-Sequence. When the outermost Phase ends, the
//! Death Creation Step removes the dead, and a Death Phase resolves their
//! Deathrattles. That repeats until a step finds no new deaths.
//!
//! Module 4: a queued trigger may activate more than once. How many times is
//! decided by one routine, `activations` in `aura.rs`, from aura data on cards.
//! Nothing here knows which cards multiply which triggers.
//!
//! Module 4, second interceptor: replacement. Before a Damage changes the world,
//! the steps in `DAMAGE_STEPS` may rewrite it, in that order: a redirect changes
//! who takes it, then Armor takes what it can of damage to a hero.
//!
//! Module 5: a card's effects are data. Each one is interpreted (`effect.rs`) and
//! its actions performed before the next effect of the same ability is read, so
//! "Assemble a minion. Give all minions in your hand +2/+2." buffs the new minion.
//!
//! Module 5, robustness: once a test can write any card as data, one card can make a
//! Sequence endless. A trigger is never queued while that same trigger of that same
//! minion is resolving (the nesting guard in `react`), and a Sequence that needs more
//! than `DEATH_PHASE_LIMIT` Death Phases is an error.
//!
//! Module 10: the steps at the end of the outermost Phase are a list, `END_OF_PHASE`,
//! read from the world, as Part II's turn is a list of phases. SpellGrid's tests run
//! the same Sequence with the list reversed, to show what the order decides.
use super::assemble::assemble_from_pool;
use super::aura::{activations, aura_update_other, redirect_for};
use super::card::{Effect, PoolFilter, TriggerOn};
use super::death::death_creation_step;
use super::effect::actions;
use super::event::Event;
use super::world::{EntityId, PlayerId, World, Zone};
/// Something the engine can do. Card effects return these instead of changing the world.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Action {
/// Deal `amount` to each target at once: all damage first, then the reactions.
Damage {
targets: Vec<EntityId>,
amount: i32,
},
Draw {
player: PlayerId,
},
GainArmor {
player: PlayerId,
amount: i32,
},
GainAttack {
id: EntityId,
amount: i32,
},
Cast {
card: String,
player: PlayerId,
target: Option<EntityId>,
},
/// Mark each target pending destroy. They leave play in the next Death Creation Step.
Destroy {
targets: Vec<EntityId>,
},
/// Summon `card` for `player` at board index `at`, or at the right end if `at` is `None`.
Summon {
card: String,
player: PlayerId,
at: Option<usize>,
},
/// Play a minion from its owner's hand: it enters the right end of the board, then its
/// Battlecry resolves.
Play {
id: EntityId,
},
/// Put an Assembled minion into `player`'s hand (see `assemble.rs`).
Assemble {
player: PlayerId,
filter: PoolFilter,
},
/// Give one entity +attack/+health.
Buff {
id: EntityId,
attack: i32,
health: i32,
},
/// Ask for `word` to be written on `source`'s grid tile (Module 10). Part I only records it.
SpawnText {
source: EntityId,
word: String,
},
}
/// A step that runs when the outermost Phase ends.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum EndOfPhaseStep {
/// The Death Creation Step (Module 3).
DeathCreation,
/// An Aura Update (Other) (Module 4).
AuraUpdate,
}
/// The end of the outermost Phase, in Hearthstone's order: the dead leave play first,
/// then the auras are read again, so a minion that just died multiplies nothing in
/// the Death Phase that follows.
pub const END_OF_PHASE: [EndOfPhaseStep; 2] =
[EndOfPhaseStep::DeathCreation, EndOfPhaseStep::AuraUpdate];
/// A step that may rewrite one Damage to one target before the world changes.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum DamageStep {
/// Damage to a hero whose owner has a `DamageRedirect` aura in force goes to that
/// aura's minion instead (the oldest one's, if several).
Redirect,
/// Damage to a hero lowers its owner's Armor first, then Health.
Armor,
}
/// The order the steps run in. Each runs once per Damage, so a rewritten Damage is
/// never offered to a redirect again. A redirect comes first: Armor protects only
/// the hero that actually takes the damage.
pub const DAMAGE_STEPS: [DamageStep; 2] = [DamageStep::Redirect, DamageStep::Armor];
/// What an effect may look at when it runs.
pub struct Ctx<'a> {
pub world: &'a World,
/// The player the effect works for.
pub owner: PlayerId,
/// The minion whose trigger is resolving; `None` for a spell.
pub source: Option<EntityId>,
/// The chosen target, if any.
pub target: Option<EntityId>,
/// The source's board position. For a Deathrattle, the source has left play, and
/// this is where it stood: immediately left of the minion its Death Event remembers.
pub position: Option<usize>,
}
impl Ctx<'_> {
/// The minion whose trigger is resolving.
pub fn minion(&self) -> EntityId {
self.source.expect("only a triggered ability has a source")
}
pub fn enemy_hero(&self) -> EntityId {
self.world.hero(1 - self.owner)
}
}
/// The most Death Phases one Sequence may have. One more is an error, not a cut-off:
/// a Sequence that stopped early would resolve differently without saying so.
pub const DEATH_PHASE_LIMIT: u32 = 64;
/// Run one Sequence: a player action and everything it causes.
pub fn run_sequence(world: &mut World, action: Action) {
perform(world, action);
// The outermost Phase has ended. Each Death Phase is itself an outermost
// Phase, so the end-of-Phase steps run again after it, until nothing dies.
let mut deaths = end_of_outermost_phase(world);
let mut round = 0;
while !deaths.is_empty() {
round += 1;
// Between Death Phases no trigger is resolving, so the panic leaves no block open.
assert!(
round <= DEATH_PHASE_LIMIT,
"Death Phase limit: more than {DEATH_PHASE_LIMIT} Death Phases in one Sequence"
);
world.record(Event::DeathPhase { round });
react(world, &deaths);
deaths = end_of_outermost_phase(world);
}
}
/// The steps that run only when the outermost Phase ends, in the world's order
/// (`END_OF_PHASE` unless changed); returns the Death Events. Hearthstone runs an
/// Aura Update (Health/Attack) first. The course has no Health/Attack auras, so that
/// step has nothing to do and is left out.
fn end_of_outermost_phase(world: &mut World) -> Vec<Event> {
let mut deaths = Vec::new();
for step in world.end_of_phase().to_vec() {
match step {
EndOfPhaseStep::DeathCreation => deaths = death_creation_step(world),
EndOfPhaseStep::AuraUpdate => aura_update_other(world),
}
}
deaths
}
/// Summon `card` for `player` at board index `at` (or the right end). Summoning runs an
/// Aura Update.
pub fn summon(world: &mut World, card: &str, player: PlayerId, at: Option<usize>) -> EntityId {
let id = world.create_at(card, player, Zone::Board, at);
aura_update_other(world);
id
}
fn perform(world: &mut World, action: Action) {
match action {
Action::Damage {
mut targets,
amount,
} => {
// Area damage: every Damage Event is created, in order of play, before any reaction.
targets.sort_by_key(|&id| world.order_of_play(id));
let mut events = Vec::new();
for target in targets {
events.push(apply_damage(world, target, amount));
}
react(world, &events);
}
Action::Draw { player } => world.draw(player),
Action::GainArmor { player, amount } => world.gain_armor(player, amount),
Action::GainAttack { id, amount } => world.gain_attack(id, amount),
Action::Destroy { targets } => {
for target in targets {
world.mark_destroyed(target);
}
}
Action::Summon { card, player, at } => {
summon(world, &card, player, at);
}
Action::Play { id } => {
assert!(
world.zone_of(id) == Zone::Hand,
"play: entity {id} is not in a hand"
);
// Playing is not summoning: no Aura Update here. The played minion's own
// auras join at the end of the outermost Phase.
world.move_to(id, Zone::Board);
let played = Event::Played { id };
world.record(played.clone());
react(world, &[played]);
}
Action::Cast {
card,
player,
target,
} => {
let spell = world.card(&card);
assert!(spell.is_spell(), "{card:?} is not a spell");
let effects = spell.on_cast.clone();
world.record(Event::SpellCast { card });
run_effects(world, &effects, player, None, target, None);
}
Action::Assemble { player, filter } => {
assemble_from_pool(world, player, &filter);
}
Action::Buff { id, attack, health } => world.buff(id, attack, health),
Action::SpawnText { source, word } => world.record(Event::TextSpawned { source, word }),
}
}
/// Interpret and perform an ability's effects in printed order. Each effect sees the
/// world as the effects before it left it. Depth-first: each action, and every
/// reaction it causes, finishes before the next.
fn run_effects(
world: &mut World,
effects: &[Effect],
owner: PlayerId,
source: Option<EntityId>,
target: Option<EntityId>,
position: Option<usize>,
) {
for effect in effects {
let ctx = Ctx {
world,
owner,
source,
target,
position,
};
for action in actions(effect, &ctx) {
perform(world, action);
}
}
}
/// Deal one Damage to one target, after the steps of `DAMAGE_STEPS` have rewritten it.
/// Returns its Damage Event, which names whoever took it.
fn apply_damage(world: &mut World, target: EntityId, amount: i32) -> Event {
let mut taker = target;
let mut absorbed = 0;
for step in DAMAGE_STEPS {
match step {
DamageStep::Redirect => {
if world.is_hero(taker) {
taker = redirect_for(world, world.owner_of(taker)).unwrap_or(taker);
}
}
DamageStep::Armor => {
if world.is_hero(taker) {
let player = world.owner_of(taker);
absorbed = world.armor(player).min(amount);
if absorbed > 0 {
world.lose_armor(player, absorbed);
}
}
}
}
}
world.damage(taker, amount, absorbed);
Event::Damaged {
target: taker,
amount,
}
}
/// A Phase: queue every trigger that responds to `events`, then resolve them one by one.
fn react(world: &mut World, events: &[Event]) {
// Fill the Queue: for each event (already in order of play), each listener in order of play.
// The nesting guard: a trigger that is resolving right now is not queued again, so
// a trigger is never nested inside itself.
let mut queue = Vec::new();
for event in events {
for listener in listeners(world, event) {
for (index, trigger) in world
.card(world.card_name(listener))
.triggers
.iter()
.enumerate()
{
if responds(world, trigger.on, listener, event)
&& !world.is_resolving(listener, index)
{
queue.push((listener, index, trigger.clone(), event));
}
}
}
}
// The Queue is now locked: nothing that happens below adds to it.
for (source, index, trigger, event) in queue {
// The interceptor: the entry was queued once; the auras in force when it
// starts decide how many times it activates, back to back.
let times = activations(world, world.owner_of(source), trigger.on);
for _ in 0..times {
world.record(Event::Triggered {
source,
on: trigger.on,
});
let position = match *event {
Event::Destroyed { id, before } if id == source => {
Some(position_left_of(world, world.owner_of(source), before))
}
_ => world.board_index(source),
};
// The trigger's consequences are one level deeper in the log tree.
world.open_block(source, index);
let owner = world.owner_of(source);
run_effects(world, &trigger.effects, owner, Some(source), None, position);
world.close_block();
}
}
}
/// The board index, right now, immediately left of `before` on `owner`'s board; the
/// right end if `before` is `None` or no longer there. A Death Event remembers the
/// minion, not an index, so this still holds after other minions have left or arrived.
fn position_left_of(world: &World, owner: PlayerId, before: Option<EntityId>) -> usize {
let board = world.zone(owner, Zone::Board);
before
.and_then(|before| board.iter().position(|&id| id == before))
.unwrap_or(board.len())
}
/// Who may respond to `event`, in order of play: every minion in play and, for a
/// Death Event, the minion that died. It has left play, but its Deathrattle still answers.
fn listeners(world: &World, event: &Event) -> Vec<EntityId> {
let mut ids = world.minions_in_play();
if let Event::Destroyed { id, .. } = *event {
ids.push(id);
ids.sort_by_key(|&id| world.order_of_play(id));
}
ids
}
/// Does a trigger listening for `on`, belonging to `listener`, respond to `event`?
fn responds(world: &World, on: TriggerOn, listener: EntityId, event: &Event) -> bool {
if let Event::Destroyed { id, .. } = *event {
return on == TriggerOn::Deathrattle && id == listener;
}
if let Event::Played { id } = *event {
return on == TriggerOn::Battlecry && id == listener;
}
let Event::Damaged { target, .. } = *event else {
return false;
};
let is_minion = !world.is_hero(target);
match on {
TriggerOn::SelfDamaged => target == listener,
TriggerOn::FriendlyMinionDamaged => {
is_minion && world.owner_of(target) == world.owner_of(listener)
}
TriggerOn::AnyMinionDamaged => is_minion,
TriggerOn::HeroDamaged => !is_minion,
TriggerOn::Deathrattle | TriggerOn::Battlecry | TriggerOn::EndOfTurn => false,
}
}
@@ -0,0 +1,418 @@
//! Part III: SpellGrid, the two engines in one level.
//!
//! Part I (the card engine) and Part II (the grid engine, `puzzle`) don't use each
//! other. This file is the only place that knows both, and they meet at four seams:
//!
//! 1. Rules become modifiers. After every reparse, the rules in force that join a
//! trigger word to a multiplier word (`DEATHRATTLE IS DOUBLE`) are compiled into
//! auras, and handed to Part I with `World::set_rule_auras`. A negated one
//! (`DEATHRATTLE IS NOT DOUBLE`) becomes a cap. Which words are which is content's
//! data (`SeamWords`), not code here.
//! 2. Hazards mark, the death step removes. The grid engine hands the objects a turn
//! destroys to `CardSide::destroy`. A minion among them is marked pending destroy,
//! in one Sequence, and Part I's Death Creation Step removes it.
//! 3. Grid changes reparse. While a Sequence runs, the grid doesn't change. When it
//! ends, the text its effects asked for (`TextSpawned` events) is written, the
//! minions that died (`Destroyed` events) leave the grid, and the rules are parsed
//! again, before anything else happens.
//! 4. Summoned minions enter the grid. In the same place, before the dead leave: a
//! minion that a minion's ability summoned (`Summoned` events with a `by`), whose card
//! has a noun, becomes an object of that noun on its summoner's tile. One summoned by
//! a spell, or with no noun, is in play but not on the grid.
//!
//! One id space: a minion's id is its grid object's id. Every other card (a hero, a
//! card in hand) gets an id from the same counter that no object will have.
use std::collections::{HashMap, HashSet};
use serde::Deserialize;
use super::aura::aura_update_other;
use super::card::{Aura, CardLibrary, CardType, Effect, Stacking, Target, TriggerOn};
use super::event::Event;
use super::puzzle::{
ConditionChecks, Dir, Hazard, Kind, Level, LevelDef, ObjectId, Response, Rule, WordDef,
};
use super::queue::{run_sequence, Action};
use super::world::{EntityId, World, Zone};
/// A multiplier word's effect: `"DOUBLE": { "times": 2, "stacking": "max" }`.
#[derive(Debug, Clone, Copy, Deserialize)]
pub struct Multiplier {
pub times: u32,
pub stacking: Stacking,
}
/// The words that cross the seam, from content: nouns that name a trigger
/// (`"DEATHRATTLE": "deathrattle"`) and properties that multiply one.
#[derive(Debug, Clone, Default, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct SeamWords {
pub trigger_words: HashMap<String, TriggerOn>,
pub multiplier_words: HashMap<String, Multiplier>,
}
/// Seam 1: the auras the rules in force give. `T IS M` (T a trigger word, M a
/// multiplier word) multiplies T's triggers; `T IS NOT M` caps them at one activation,
/// so it cancels a card's multiplier too, as `X IS NOT Y` cancels every `X IS Y`.
///
/// Module 11: a rule with a condition gives no aura. A condition holds or not for one
/// object on the grid, and a trigger word names no object (the course's choice).
pub fn rule_auras(rules: &[Rule], words: &SeamWords) -> Vec<Aura> {
let mut auras = Vec::new();
for rule in rules {
if !rule.conditions.is_empty() {
continue;
}
let (Some(&trigger), Some(&m)) = (
words.trigger_words.get(&rule.subject),
words.multiplier_words.get(&rule.complement),
) else {
continue;
};
auras.push(if rule.negated {
Aura::TriggerCap { trigger, times: 1 }
} else {
Aura::TriggerMultiplier {
trigger,
times: m.times,
stacking: m.stacking,
}
});
}
auras
}
/// A SpellGrid level in play: a grid and a game, kept in step. Ids are grid ids (see above).
pub struct SpellGrid {
level: Level,
side: CardSide,
}
/// Part I's half: the game, and which entity is which id. Kept apart from the level so
/// that the level can lend itself to the destroy function while this half changes.
struct CardSide {
world: World,
/// Every entity's id.
ids: HashMap<EntityId, ObjectId>,
/// The entity behind each id (a card's old id still finds it after it is played).
entities: HashMap<ObjectId, EntityId>,
seam_words: SeamWords,
}
impl SpellGrid {
/// The level's objects; its `cards` as minions in play for player 0, each a grid
/// object named by the card's noun; its `hand` in player 0's hand.
pub fn new(
def: &LevelDef,
library: CardLibrary,
words: &[WordDef],
responses: Vec<Response>,
hazards: Vec<Hazard>,
conditions: ConditionChecks,
seam_words: SeamWords,
) -> SpellGrid {
let mut level = Level::new(def, words, responses, hazards, conditions);
let mut side = CardSide {
world: World::new(library),
ids: HashMap::new(),
entities: HashMap::new(),
seam_words,
};
for c in &def.cards {
let kind = side.noun_of(&c.card);
let id = level.board_mut().place(kind, c.x, c.y);
let entity = side.world.create(&c.card, 0, Zone::Board);
side.link(entity, id);
}
// Every minion is in play: one Aura Update, as for a summon.
aura_update_other(&mut side.world);
for card in &def.hand {
side.world.create(card, 0, Zone::Hand);
}
side.link_the_rest(&mut level);
side.grid_changed(&mut level);
SpellGrid { level, side }
}
pub fn level(&self) -> &Level {
&self.level
}
pub fn world(&self) -> &World {
&self.side.world
}
pub fn world_mut(&mut self) -> &mut World {
&mut self.side.world
}
// ---- Ids ----
/// The id of entity `entity`.
pub fn id_of(&self, entity: EntityId) -> ObjectId {
self.side.ids[&entity]
}
/// The entity behind `id`. Panics if `id` is not a card's.
pub fn entity(&self, id: ObjectId) -> EntityId {
self.side.entity(id)
}
/// The id of the minion on tile (x, y), if there is one.
pub fn minion_at(&self, x: i32, y: i32) -> Option<ObjectId> {
self.level
.board()
.at(x, y)
.map(|o| o.id)
.find(|id| self.side.entities.contains_key(id))
}
// ---- What the player does ----
/// Add an object or a piece of text, then reparse.
pub fn place(&mut self, kind: &str, x: i32, y: i32) -> ObjectId {
let id = self.level.place(kind, x, y);
self.side.grid_changed(&mut self.level);
id
}
/// One whole turn (Module 9); what it destroys goes through seam 2.
pub fn step(&mut self, dir: Dir) {
let side = &mut self.side;
self.level
.step_with(dir, &mut |level, ids| side.destroy(level, ids));
self.side.grid_changed(&mut self.level);
}
/// Movement only (Module 8), then a reparse.
pub fn move_dir(&mut self, dir: Dir) {
let side = &mut self.side;
self.level
.move_you_with(dir, &mut |level, ids| side.destroy(level, ids));
self.side.grid_changed(&mut self.level);
}
/// Why `cast(card, target)` can't be done now, or `Ok` if it can: the card must be a
/// spell in player 0's hand, and a spell that hits a chosen minion needs a target
/// minion on the grid. Nothing changes either way.
pub fn check_cast(&self, card: &str, target: Option<ObjectId>) -> Result<(), String> {
self.side.check_hand(card, CardType::Spell)?;
let chooses = self.side.world.card(card).on_cast.iter().any(|e| {
matches!(
e,
Effect::DealDamage {
target: Target::Chosen,
..
} | Effect::GainAttack {
target: Target::Chosen,
..
} | Effect::Destroy {
target: Target::Chosen
}
)
});
match target {
None if chooses => Err(format!("\"{card}\" needs a target minion")),
None => Ok(()),
Some(id) => {
// Every id in `entities` was issued by the board, so `find` may ask about it.
let card = self.side.entities.contains_key(&id);
if card && self.level.board().find(id).is_some() {
Ok(())
} else {
Err(format!("{id} is not a minion on the grid"))
}
}
}
}
/// Why `play(card, x, y)` can't be done now, or `Ok` if it can.
pub fn check_play(&self, card: &str, x: i32, y: i32) -> Result<(), String> {
self.side.check_hand(card, CardType::Minion)?;
if self.side.world.card(card).def.noun.is_none() {
return Err(format!(
"\"{card}\" has no noun: it can't stand on the grid"
));
}
if !self.level.board().in_bounds(x, y) {
return Err(format!("({x}, {y}) is off the grid"));
}
if self.level.board().at(x, y).next().is_some() {
return Err(format!("({x}, {y}) is not empty"));
}
Ok(())
}
/// Cast a spell from player 0's hand, as one Sequence. `target` is a minion's id.
/// Panics if `check_cast` says no.
pub fn cast(&mut self, card: &str, target: Option<ObjectId>) {
if let Err(problem) = self.check_cast(card, target) {
panic!("{problem}");
}
let in_hand = self.side.in_hand(card).expect("checked");
self.side.world.move_to(in_hand, Zone::Graveyard);
let target = target.map(|id| self.side.entity(id));
let action = Action::Cast {
card: card.to_string(),
player: 0,
target,
};
self.side.sequence(&mut self.level, action);
}
/// Play a minion from player 0's hand onto the empty tile (x, y), as one Sequence.
/// Returns its new id. Panics if `check_play` says no.
pub fn play(&mut self, card: &str, x: i32, y: i32) -> ObjectId {
if let Err(problem) = self.check_play(card, x, y) {
panic!("{problem}");
}
let in_hand = self.side.in_hand(card).expect("checked");
let kind = self.side.noun_of(card);
let id = self.level.board_mut().place(kind, x, y);
self.side.link(in_hand, id);
self.side
.sequence(&mut self.level, Action::Play { id: in_hand });
id
}
}
impl CardSide {
fn entity(&self, id: ObjectId) -> EntityId {
*self
.entities
.get(&id)
.unwrap_or_else(|| panic!("{id} is not a card"))
}
fn link(&mut self, entity: EntityId, id: ObjectId) {
self.ids.insert(entity, id);
self.entities.insert(id, entity);
}
/// Give every entity that has no id one that no object will have.
fn link_the_rest(&mut self, level: &mut Level) {
for entity in 0..self.world.entity_count() {
if !self.ids.contains_key(&entity) {
let id = level.board_mut().reserve_id();
self.link(entity, id);
}
}
}
fn noun_of(&self, card: &str) -> Kind {
let def = &self.world.card(card).def;
match (&def.noun, def.card_type) {
(Some(noun), CardType::Minion) => Kind::parse(&noun.to_lowercase()),
_ => panic!("{card:?} is not a minion with a noun"),
}
}
fn in_hand(&self, card: &str) -> Option<EntityId> {
self.world
.zone(0, Zone::Hand)
.iter()
.copied()
.find(|&id| self.world.card_name(id) == card)
}
fn check_hand(&self, card: &str, card_type: CardType) -> Result<(), String> {
if self.in_hand(card).is_none() {
return Err(format!("\"{card}\" is not in the hand"));
}
if self.world.card(card).def.card_type != card_type {
let kind = match card_type {
CardType::Minion => "minion",
CardType::Spell => "spell",
};
return Err(format!("\"{card}\" is not a {kind}"));
}
Ok(())
}
/// Seam 2. Objects that aren't cards leave the board. Minions are marked pending
/// destroy, in one Sequence, under the rules as they stand now.
fn destroy(&mut self, level: &mut Level, ids: &[ObjectId]) {
let mut seen = HashSet::new();
let mut minions = Vec::new();
for &id in ids {
if !seen.insert(id) || level.board().find(id).is_none() {
continue;
}
match self.entities.get(&id) {
Some(&entity) => minions.push(entity),
None => level.board_mut().remove(id),
}
}
if minions.is_empty() {
return;
}
self.sync_rules(level);
self.sequence(level, Action::Destroy { targets: minions });
}
/// Run one Sequence in Part I, then seams 3 and 4: apply its grid changes and reparse.
fn sequence(&mut self, level: &mut Level, action: Action) {
let from = self.world.log().len();
run_sequence(&mut self.world, action);
let events = self.world.log()[from..].to_vec();
// Text first, while every minion that wrote it is still on its tile. A minion
// with no id (summoned by this Sequence) is not on the grid yet: nowhere to write.
for event in &events {
if let Event::TextSpawned { source, word } = event {
let tile = self
.ids
.get(source)
.and_then(|&id| level.board().find(id))
.map(|o| (o.x, o.y));
if let Some((x, y)) = tile {
let text = Kind::Text { word: word.clone() };
level.board_mut().place(text, x, y);
}
}
}
// Seam 4: summoned minions, in summon order, on their summoner's tile. The dead are
// still on the grid here, so a Deathrattle's summoner is found where it died; a
// minion summoned earlier in the list is already placed, so it can be a summoner too.
for event in &events {
let Event::Summoned { id, by: Some(by) } = *event else {
continue;
};
let card = self.world.card_name(id).to_string();
if self.ids.contains_key(&id) || self.world.card(&card).def.noun.is_none() {
continue;
}
let tile = self
.ids
.get(&by)
.and_then(|&object| level.board().find(object))
.map(|o| (o.x, o.y));
if let Some((x, y)) = tile {
let object = level.board_mut().place(self.noun_of(&card), x, y);
self.link(id, object);
}
}
for event in &events {
if let Event::Destroyed { id, .. } = event {
if let Some(&object) = self.ids.get(id) {
level.board_mut().remove(object);
}
}
}
self.link_the_rest(level);
self.grid_changed(level);
}
/// The grid changed: reparse it, then seam 1.
fn grid_changed(&mut self, level: &mut Level) {
level.reparse();
self.sync_rules(level);
}
/// Seam 1: compile the rules in force into auras for Part I, then run an Aura Update.
fn sync_rules(&mut self, level: &Level) {
let auras = rule_auras(level.active_rules(), &self.seam_words);
self.world.set_rule_auras(auras);
aura_update_other(&mut self.world);
}
}
@@ -0,0 +1,456 @@
//! Entities, zones, order of play, stats and the event log: the game's state.
//! The rules for resolving actions live in `queue.rs`.
use super::card::{Aura, Card, CardDef, CardLibrary};
use super::event::Event;
use super::queue::{EndOfPhaseStep, END_OF_PHASE};
/// A handle to an entity. Ids are never reused and stay valid when the entity changes zone.
pub type EntityId = u32;
pub type PlayerId = u8;
/// Starting Health of each hero.
pub const HERO_HEALTH: i32 = 30;
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Zone {
Deck,
Hand,
Board,
Graveyard,
}
#[derive(Debug)]
struct Entity {
card: String,
owner: PlayerId,
zone: Zone,
/// Stamped each time the entity enters the board; `None` until it first does.
order_of_play: Option<u32>,
attack: i32,
/// Current Health. At 0 or below the entity is mortally wounded: still in play until the
/// Death Creation Step.
health: i32,
/// Hit by a destroy effect: still in play until the Death Creation Step.
pending_destroy: bool,
}
/// An aura in force: which minion has it, and that minion's owner. An aura from outside
/// the cards (Module 10) has no minion (`source` is `None`), and is in force once for
/// each player.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct ActiveAura {
pub owner: PlayerId,
pub source: Option<EntityId>,
pub aura: Aura,
}
/// One player's zones. Each is an ordered list of ids (the board is left to right).
#[derive(Debug, Default)]
struct Side {
deck: Vec<EntityId>,
hand: Vec<EntityId>,
board: Vec<EntityId>,
graveyard: Vec<EntityId>,
hero: EntityId,
armor: i32,
}
impl Side {
fn zone(&self, zone: Zone) -> &Vec<EntityId> {
match zone {
Zone::Deck => &self.deck,
Zone::Hand => &self.hand,
Zone::Board => &self.board,
Zone::Graveyard => &self.graveyard,
}
}
fn zone_mut(&mut self, zone: Zone) -> &mut Vec<EntityId> {
match zone {
Zone::Deck => &mut self.deck,
Zone::Hand => &mut self.hand,
Zone::Board => &mut self.board,
Zone::Graveyard => &mut self.graveyard,
}
}
}
/// Makes a player's choices: called with what is being chosen and the options, it
/// returns one of the options. Hearthstone would ask the player or pick at random.
pub type Chooser = Box<dyn FnMut(&str, &[String]) -> String>;
pub struct World {
library: CardLibrary,
chooser: Chooser,
/// Indexed by `EntityId`.
entities: Vec<Entity>,
sides: [Side; 2],
next_order_of_play: u32,
log: Vec<Event>,
/// The nesting depth at which each event in `log` was recorded, index for index.
depths: Vec<u32>,
/// The triggers resolving right now, one inside another, outermost first: each is a
/// minion and the index of the trigger in its card's `triggers`. How many there are
/// is the nesting depth of the log tree.
resolving: Vec<(EntityId, usize)>,
/// The auras in force as of the last Aura Update. Changes only there (see `aura.rs`).
auras: Vec<ActiveAura>,
/// Module 10: auras that no minion carries, for every player's minions. SpellGrid
/// compiles them from the rules on its grid. The next Aura Update brings them in.
rule_auras: Vec<Aura>,
/// Module 10: the steps at the end of the outermost Phase, in order.
end_of_phase: Vec<EndOfPhaseStep>,
}
impl World {
/// A new game: both heroes are in play from the start, so they are the oldest entities.
pub fn new(library: CardLibrary) -> World {
let mut world = World {
library,
// Until a chooser is set, take the first option.
chooser: Box::new(|_, options| options[0].clone()),
entities: Vec::new(),
sides: Default::default(),
next_order_of_play: 0,
log: Vec::new(),
depths: Vec::new(),
resolving: Vec::new(),
auras: Vec::new(),
rule_auras: Vec::new(),
end_of_phase: END_OF_PHASE.to_vec(),
};
for owner in 0..2 {
let id = world.entities.len() as EntityId;
world.next_order_of_play += 1;
world.entities.push(Entity {
card: "Hero".to_string(),
owner,
zone: Zone::Board,
order_of_play: Some(world.next_order_of_play),
attack: 0,
health: HERO_HEALTH,
pending_destroy: false,
});
// In play, but not on the board list: the board holds minions only.
world.sides[owner as usize].hero = id;
}
world
}
/// Create an entity of `card` for `owner` at the end of `zone` (the right end for the board).
pub fn create(&mut self, card: &str, owner: PlayerId, zone: Zone) -> EntityId {
self.create_at(card, owner, zone, None)
}
/// Create an entity of `card` for `owner` in `zone` at index `at` (clamped to the
/// zone's length), or at the end if `at` is `None`.
pub fn create_at(
&mut self,
card: &str,
owner: PlayerId,
zone: Zone,
at: Option<usize>,
) -> EntityId {
assert!(owner < 2, "no player {owner}");
let def = &self.library.get(card).def;
let entity = Entity {
card: def.name.clone(),
owner,
zone,
order_of_play: None,
attack: def.attack,
health: def.health,
pending_destroy: false,
};
let id = self.entities.len() as EntityId;
self.entities.push(entity);
self.enter(id, zone, at);
id
}
/// Move an entity to the end of `zone` on its owner's side, keeping its id.
pub fn move_to(&mut self, id: EntityId, zone: Zone) {
let (owner, from) = (self.entity(id).owner, self.entity(id).zone);
self.sides[owner as usize]
.zone_mut(from)
.retain(|&e| e != id);
self.enter(id, zone, None);
}
/// The single place an entity arrives in a zone. Entering the board starts a new order of play.
fn enter(&mut self, id: EntityId, zone: Zone, at: Option<usize>) {
let owner = self.entity(id).owner;
let list = self.sides[owner as usize].zone_mut(zone);
list.insert(at.map_or(list.len(), |i| i.min(list.len())), id);
let entity = &mut self.entities[id as usize];
entity.zone = zone;
if zone == Zone::Board {
self.next_order_of_play += 1;
entity.order_of_play = Some(self.next_order_of_play);
// The summoner is the minion whose trigger is resolving (the innermost one), if any.
let by = self.resolving.last().map(|&(source, _)| source);
self.record(Event::Summoned { id, by });
}
}
fn entity(&self, id: EntityId) -> &Entity {
self.entities
.get(id as usize)
.unwrap_or_else(|| panic!("unknown entity id {id}"))
}
fn entity_mut(&mut self, id: EntityId) -> &mut Entity {
self.entities
.get_mut(id as usize)
.unwrap_or_else(|| panic!("unknown entity id {id}"))
}
// ---- Queries -------------------------------------------------------
pub fn card(&self, name: &str) -> &Card {
self.library.get(name)
}
pub fn has_card(&self, name: &str) -> bool {
self.library.contains(name)
}
/// The cards an Assemble may offer, sorted by name.
pub fn pool(&self) -> Vec<&Card> {
self.library.pool()
}
pub fn zone_of(&self, id: EntityId) -> Zone {
self.entity(id).zone
}
pub fn card_name(&self, id: EntityId) -> &str {
&self.entity(id).card
}
pub fn owner_of(&self, id: EntityId) -> PlayerId {
self.entity(id).owner
}
/// When the entity last entered the board (larger is later); `None` if it never has.
pub fn order_of_play(&self, id: EntityId) -> Option<u32> {
self.entity(id).order_of_play
}
/// The ids in one player's zone, in zone order (left to right for the board).
pub fn zone(&self, owner: PlayerId, zone: Zone) -> &[EntityId] {
assert!(owner < 2, "no player {owner}");
self.sides[owner as usize].zone(zone)
}
pub fn hero(&self, owner: PlayerId) -> EntityId {
assert!(owner < 2, "no player {owner}");
self.sides[owner as usize].hero
}
pub fn is_hero(&self, id: EntityId) -> bool {
self.sides.iter().any(|s| s.hero == id)
}
/// Every minion on either board, oldest first (order of play).
pub fn minions_in_play(&self) -> Vec<EntityId> {
let mut ids: Vec<EntityId> = self.sides.iter().flat_map(|s| s.board.clone()).collect();
ids.sort_by_key(|&id| self.order_of_play(id));
ids
}
pub fn attack(&self, id: EntityId) -> i32 {
self.entity(id).attack
}
pub fn health(&self, id: EntityId) -> i32 {
self.entity(id).health
}
/// The cost printed on the entity's card.
pub fn cost(&self, id: EntityId) -> i32 {
self.card(self.card_name(id)).def.cost
}
/// The minions in a player's hand, in hand order.
pub fn minions_in_hand(&self, owner: PlayerId) -> Vec<EntityId> {
self.zone(owner, Zone::Hand)
.iter()
.copied()
.filter(|&id| !self.card(self.card_name(id)).is_spell())
.collect()
}
/// Mortally wounded (0 or less Health) or pending destroy: it leaves play at the next
/// Death Creation Step.
pub fn is_doomed(&self, id: EntityId) -> bool {
let entity = self.entity(id);
entity.health <= 0 || entity.pending_destroy
}
/// Index of a minion on its owner's board, left to right; `None` if it is not on the board.
pub fn board_index(&self, id: EntityId) -> Option<usize> {
self.zone(self.owner_of(id), Zone::Board)
.iter()
.position(|&e| e == id)
}
pub fn armor(&self, owner: PlayerId) -> i32 {
self.sides[owner as usize].armor
}
pub fn active_auras(&self) -> &[ActiveAura] {
&self.auras
}
/// Auras from outside the cards (Module 10), as last set.
pub fn rule_auras(&self) -> &[Aura] {
&self.rule_auras
}
/// The steps at the end of the outermost Phase, in order.
pub fn end_of_phase(&self) -> &[EndOfPhaseStep] {
&self.end_of_phase
}
/// How many entities exist: ids run from 0 to this, exclusive.
pub fn entity_count(&self) -> u32 {
self.entities.len() as u32
}
pub fn log(&self) -> &[Event] {
&self.log
}
/// The nesting depth of each event in `log()`, index for index: 0 for the outermost
/// action's own events.
pub fn log_depths(&self) -> &[u32] {
&self.depths
}
// ---- Changes: each one records its event ----------------------------
pub fn record(&mut self, event: Event) {
self.log.push(event);
self.depths.push(self.resolving.len() as u32);
}
/// Trigger number `trigger` of `source` starts resolving: the events it causes are
/// recorded one level deeper.
pub fn open_block(&mut self, source: EntityId, trigger: usize) {
self.resolving.push((source, trigger));
}
/// The trigger that last opened a block has finished resolving.
pub fn close_block(&mut self) {
self.resolving.pop();
}
/// Is trigger number `trigger` of `source` resolving right now (at any depth)?
pub fn is_resolving(&self, source: EntityId, trigger: usize) -> bool {
self.resolving.contains(&(source, trigger))
}
/// Add a card to the library, or replace the card with the same name.
pub fn register(&mut self, def: CardDef) {
self.library.register(def);
}
/// Add a card made by Assemble (see `assemble.rs`).
pub fn register_assembled(&mut self, def: CardDef) {
self.library.register_assembled(def);
}
pub fn set_chooser(&mut self, chooser: Chooser) {
self.chooser = chooser;
}
/// Ask the chooser to pick one of `options` (not empty). Panics if it answers
/// something that isn't an option.
pub fn choose(&mut self, kind: &str, options: &[String]) -> String {
let answer = (self.chooser)(kind, options);
assert!(
options.contains(&answer),
"chooser answered {answer:?} for {kind:?}; the options were {options:?}"
);
answer
}
/// Replace the auras from outside the cards. They take effect at the next Aura Update.
pub fn set_rule_auras(&mut self, auras: Vec<Aura>) {
self.rule_auras = auras;
}
/// Change the order of the steps at the end of the outermost Phase.
pub fn set_end_of_phase(&mut self, steps: Vec<EndOfPhaseStep>) {
self.end_of_phase = steps;
}
/// Only an Aura Update calls this.
pub fn set_active_auras(&mut self, auras: Vec<ActiveAura>) {
self.auras = auras;
}
pub fn clear_log(&mut self) {
self.log.clear();
self.depths.clear();
}
/// `target` takes `amount` damage, of which `absorbed` was already taken by Armor
/// (see `lose_armor`): Health drops by the rest. The event shows the full amount.
pub fn damage(&mut self, target: EntityId, amount: i32, absorbed: i32) {
self.entity_mut(target).health -= amount - absorbed;
self.record(Event::Damaged { target, amount });
}
/// `player`'s hero's Armor absorbs `amount` damage.
pub fn lose_armor(&mut self, player: PlayerId, amount: i32) {
self.sides[player as usize].armor -= amount;
self.record(Event::ArmorLost { player, amount });
}
/// Draw the top card of `player`'s deck (the first in the list). An empty deck
/// records `Fatigue` and changes nothing else.
pub fn draw(&mut self, player: PlayerId) {
match self.zone(player, Zone::Deck).first() {
Some(&id) => {
self.move_to(id, Zone::Hand);
self.record(Event::Drew { player, id });
}
None => self.record(Event::Fatigue { player }),
}
}
/// Mark for destruction. Nothing leaves play until the Death Creation Step.
pub fn mark_destroyed(&mut self, id: EntityId) {
self.entity_mut(id).pending_destroy = true;
}
/// Move a minion from play to its owner's graveyard, recording its Death Event.
pub fn kill(&mut self, id: EntityId, before: Option<EntityId>) {
self.entity_mut(id).pending_destroy = false;
self.move_to(id, Zone::Graveyard);
self.record(Event::Destroyed { id, before });
}
pub fn gain_armor(&mut self, player: PlayerId, amount: i32) {
self.sides[player as usize].armor += amount;
self.record(Event::ArmorGained { player, amount });
}
pub fn buff(&mut self, id: EntityId, attack: i32, health: i32) {
let entity = self.entity_mut(id);
entity.attack += attack;
entity.health += health;
let (attack, health) = (entity.attack, entity.health);
self.record(Event::StatsChanged { id, attack, health });
}
pub fn gain_attack(&mut self, id: EntityId, amount: i32) {
let entity = self.entity_mut(id);
entity.attack += amount;
let (attack, health) = (entity.attack, entity.health);
self.record(Event::StatsChanged { id, attack, health });
}
}
@@ -0,0 +1,531 @@
//! Test harness: the thin adapter the course tests call. It translates each
//! call into engine operations and holds no rules of its own.
use crate::content;
use crate::engine::{
self, puzzle, run_sequence, summon, Action, CardLibrary, Event, SpellGrid as Grid, World,
END_OF_PHASE,
};
pub use crate::engine::{AssembleError, CardDef, Half, Zone};
pub type Id = u32;
pub type Player = u8;
#[derive(Debug, Default, Clone)]
pub struct GameOpts {
pub decks: Option<[Vec<String>; 2]>,
}
/// The events the tests read from `log()` (see course/harness.md).
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum GameEvent {
Summoned { id: Id },
Damaged { target: Id, amount: i32 },
Triggered { source: Id, trigger: String },
Destroyed { id: Id },
DeathPhase { round: u32 },
Drew { player: Player, id: Id },
Fatigue { player: Player },
ArmorGained { player: Player, amount: i32 },
ArmorLost { player: Player, amount: i32 },
StatsChanged { id: Id, attack: i32, health: i32 },
SpellCast { card: String },
}
impl GameEvent {
/// The course's view of an engine event; `None` for engine events the course log
/// doesn't show.
fn from_engine(event: &Event) -> Option<GameEvent> {
Some(match event.clone() {
Event::Summoned { id, .. } => GameEvent::Summoned { id },
Event::Damaged { target, amount } => GameEvent::Damaged { target, amount },
Event::Triggered { source, on } => GameEvent::Triggered {
source,
trigger: on.label().to_string(),
},
Event::Drew { player, id } => GameEvent::Drew { player, id },
Event::Fatigue { player } => GameEvent::Fatigue { player },
Event::ArmorGained { player, amount } => GameEvent::ArmorGained { player, amount },
Event::ArmorLost { player, amount } => GameEvent::ArmorLost { player, amount },
Event::StatsChanged { id, attack, health } => {
GameEvent::StatsChanged { id, attack, health }
}
Event::SpellCast { card } => GameEvent::SpellCast { card },
Event::Destroyed { id, .. } => GameEvent::Destroyed { id },
Event::DeathPhase { round } => GameEvent::DeathPhase { round },
// A played minion's `Summoned` already shows it entering the board.
Event::Played { .. } => return None,
// SpellGrid writes the text on the grid; the course's log doesn't show the request.
Event::TextSpawned { .. } => return None,
})
}
}
/// One event as `scenarios.md` writes it, such as `Triggered{source: 3, trigger: "deathrattle"}`.
impl std::fmt::Display for GameEvent {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
GameEvent::Summoned { id } => write!(f, "Summoned{{id: {id}}}"),
GameEvent::Damaged { target, amount } => {
write!(f, "Damaged{{target: {target}, amount: {amount}}}")
}
GameEvent::Triggered { source, trigger } => {
write!(f, "Triggered{{source: {source}, trigger: {trigger:?}}}")
}
GameEvent::Destroyed { id } => write!(f, "Destroyed{{id: {id}}}"),
GameEvent::DeathPhase { round } => write!(f, "DeathPhase{{round: {round}}}"),
GameEvent::Drew { player, id } => write!(f, "Drew{{player: {player}, id: {id}}}"),
GameEvent::Fatigue { player } => write!(f, "Fatigue{{player: {player}}}"),
GameEvent::ArmorGained { player, amount } => {
write!(f, "ArmorGained{{player: {player}, amount: {amount}}}")
}
GameEvent::ArmorLost { player, amount } => {
write!(f, "ArmorLost{{player: {player}, amount: {amount}}}")
}
GameEvent::StatsChanged { id, attack, health } => {
write!(
f,
"StatsChanged{{id: {id}, attack: {attack}, health: {health}}}"
)
}
GameEvent::SpellCast { card } => write!(f, "SpellCast{{card: {card:?}}}"),
}
}
}
/// The log as a tree: one event per line, two spaces of indent per level of nesting.
fn log_tree(world: &World, view: impl Fn(&Event) -> Option<GameEvent>) -> String {
world
.log()
.iter()
.zip(world.log_depths())
.filter_map(|(event, &depth)| {
let shown = view(event)?;
Some(format!("{}{shown}\n", " ".repeat(depth as usize)))
})
.collect()
}
pub struct Game {
world: World,
}
pub fn new_game(opts: GameOpts) -> Game {
let mut world = World::new(CardLibrary::new(content::cards()));
if let Some(decks) = opts.decks {
for (player, deck) in decks.iter().enumerate() {
for card in deck {
world.create(card, player as Player, Zone::Deck);
}
}
}
Game { world }
}
impl Game {
/// Put `card` at the right end of `player`'s board.
pub fn summon(&mut self, card: &str, player: Player) -> Id {
summon(&mut self.world, card, player, None)
}
pub fn add_to_hand(&mut self, card: &str, player: Player) -> Id {
self.world.create(card, player, Zone::Hand)
}
pub fn add_to_deck(&mut self, card: &str, player: Player) -> Id {
self.world.create(card, player, Zone::Deck)
}
/// Play a card from hand, as one Sequence: it enters the right end of the board, then
/// its Battlecry resolves. (The Module 1 Expansion Drop; required from Module 4.)
pub fn play(&mut self, hand_id: Id) {
run_sequence(&mut self.world, Action::Play { id: hand_id });
}
pub fn zone_of(&self, id: Id) -> Zone {
self.world.zone_of(id)
}
/// Board ids, left to right.
pub fn board(&self, player: Player) -> Vec<Id> {
self.world.zone(player, Zone::Board).to_vec()
}
pub fn hand(&self, player: Player) -> Vec<Id> {
self.world.zone(player, Zone::Hand).to_vec()
}
pub fn order_of_play(&self, id: Id) -> u32 {
self.world
.order_of_play(id)
.unwrap_or_else(|| panic!("entity {id} has not entered play"))
}
pub fn name_of(&self, id: Id) -> String {
self.world.card_name(id).to_string()
}
// ---- Module 2 ----
pub fn hero(&self, player: Player) -> Id {
self.world.hero(player)
}
/// Deal damage to one character, as one Sequence.
pub fn deal_damage(&mut self, target: Id, amount: i32) {
let targets = vec![target];
run_sequence(&mut self.world, Action::Damage { targets, amount });
}
pub fn cast(&mut self, card: &str, player: Player, target: Option<Id>) {
let card = card.to_string();
run_sequence(
&mut self.world,
Action::Cast {
card,
player,
target,
},
);
}
pub fn health(&self, id: Id) -> i32 {
self.world.health(id)
}
pub fn attack(&self, id: Id) -> i32 {
self.world.attack(id)
}
pub fn armor(&self, player: Player) -> i32 {
self.world.armor(player)
}
pub fn log(&self) -> Vec<GameEvent> {
self.world
.log()
.iter()
.filter_map(GameEvent::from_engine)
.collect()
}
pub fn clear_log(&mut self) {
self.world.clear_log();
}
/// The log as an indented tree (see course/harness.md).
pub fn log_tree(&self) -> String {
log_tree(&self.world, GameEvent::from_engine)
}
// ---- Module 3 ----
/// Mark `target` pending destroy, as one Sequence. It leaves play in the death step.
pub fn destroy(&mut self, target: Id) {
let targets = vec![target];
run_sequence(&mut self.world, Action::Destroy { targets });
}
// ---- Module 5 ----
/// Add card definitions to this game, replacing any with the same name.
pub fn register_cards(&mut self, defs: Vec<CardDef>) {
for def in defs {
self.world.register(def);
}
}
/// Register `base` with `half` of its text replaced by the same half of `donor`,
/// and return the new card's name, `"<base> (Assembled)"`.
pub fn assemble(
&mut self,
base: &str,
donor: &str,
half: Half,
) -> Result<String, AssembleError> {
let base = &self.world.card(base).def;
let donor = &self.world.card(donor).def;
let def = engine::assemble(base, donor, half)?;
Ok(engine::register_assembled(&mut self.world, def))
}
pub fn cost(&self, id: Id) -> i32 {
self.world.cost(id)
}
/// Who makes choices (Assemble's base, donor and half): called with the kind of
/// choice (`"base"`, `"donor"`, `"half"`) and the options; returns one of them.
pub fn set_chooser(&mut self, chooser: impl FnMut(&str, &[String]) -> String + 'static) {
self.world.set_chooser(Box::new(chooser));
}
}
// ======== Part II: Baba Is You ========
pub use crate::engine::puzzle::{Dir, LevelCard, LevelDef, LevelObject};
/// A Baba Is You level in play (Part II).
pub struct Baba {
level: puzzle::Level,
}
/// A level from a level definition; its text is parsed straight away.
pub fn new_baba(level: LevelDef) -> Baba {
Baba {
level: puzzle::Level::new(
&level,
&content::words(),
content::movement(),
content::hazards(),
content::conditions(),
),
}
}
impl Baba {
// ---- Module 6 ----
/// Add an object (`"baba"`) or a piece of text (`"text:BABA"`) at (x, y).
pub fn place(&mut self, kind: &str, x: i32, y: i32) -> Id {
self.level.place(kind, x, y)
}
/// Every distinct rule the text spells, before resolution, normalised
/// (`"BABA IS YOU"`, `"BABA IS NOT YOU"`, `"BABA ON KEY IS WIN"`) and sorted.
pub fn rules(&self) -> Vec<String> {
normalised(self.level.rules())
}
// ---- Module 7 ----
/// Every distinct rule in force after resolution, normalised and sorted.
pub fn active_rules(&self) -> Vec<String> {
normalised(self.level.active_rules())
}
/// Does object `id` have the property `prop` under the rules in force?
pub fn has_property(&self, id: Id, prop: &str) -> bool {
self.level.has_property(id, prop)
}
// ---- Module 8 ----
/// Move every `YOU` object one tile in `dir`, pushing what it pushes, then
/// reparse the rules. Movement only: no conversion, hazard or win (Module 9).
pub fn move_dir(&mut self, dir: Dir) {
self.level.move_you(dir);
}
/// Where object `id` is, or `None` once it has been destroyed.
pub fn pos_of(&self, id: Id) -> Option<(i32, i32)> {
self.level.board().find(id).map(|o| (o.x, o.y))
}
/// The kinds of the objects on tile (x, y), as placed (`"baba"`, `"text:ROCK"`), sorted.
pub fn kinds_at(&self, x: i32, y: i32) -> Vec<String> {
let mut kinds: Vec<String> = self
.level
.board()
.at(x, y)
.map(|o| o.kind.to_string())
.collect();
kinds.sort();
kinds
}
/// The kinds on tile (x, y), as `kinds_at`, but oldest object first.
pub fn stack_at(&self, x: i32, y: i32) -> Vec<String> {
let board = self.level.board();
board.at(x, y).map(|o| o.kind.to_string()).collect()
}
// ---- Module 9 ----
/// Play one whole turn: movement, parse, conversion, parse, hazards, win
/// (and a last parse, so the rules match the board).
pub fn step(&mut self, dir: Dir) {
self.level.step(dir);
}
/// Has a turn found a `YOU` object on a `WIN` one (or `WIN` itself)? Stays true.
pub fn won(&self) -> bool {
self.level.won()
}
/// Is no object `YOU`?
pub fn lost(&self) -> bool {
self.level.lost()
}
/// Put the level back as it was before the last `step` or `move_dir`. False, and
/// no change, if there is none; repeated calls walk further back.
pub fn undo(&mut self) -> bool {
self.level.undo()
}
}
fn normalised(rules: &[puzzle::Rule]) -> Vec<String> {
let mut rules: Vec<String> = rules.iter().map(|r| r.to_string()).collect();
rules.sort();
rules.dedup();
rules
}
// ======== Part III: SpellGrid ========
/// A SpellGrid level in play (Part III): a Baba Is You level whose minions are cards.
/// Every id is a grid id: a minion's is its grid object's (see `engine/spellgrid.rs`).
pub struct SpellGrid {
grid: Grid,
}
/// A SpellGrid level; `extra` cards are added to the course's cards, replacing any
/// with the same name.
pub fn new_spell_grid(level: LevelDef, extra: Vec<CardDef>) -> SpellGrid {
let mut library = CardLibrary::new(content::cards());
for def in extra {
library.register(def);
}
SpellGrid {
grid: Grid::new(
&level,
library,
&content::words(),
content::movement(),
content::hazards(),
content::conditions(),
content::seams(),
),
}
}
impl SpellGrid {
// ---- Part II, as on `Baba` ----
pub fn place(&mut self, kind: &str, x: i32, y: i32) -> Id {
self.grid.place(kind, x, y)
}
pub fn rules(&self) -> Vec<String> {
normalised(self.grid.level().rules())
}
pub fn active_rules(&self) -> Vec<String> {
normalised(self.grid.level().active_rules())
}
pub fn has_property(&self, id: Id, prop: &str) -> bool {
self.grid.level().has_property(id, prop)
}
pub fn move_dir(&mut self, dir: Dir) {
self.grid.move_dir(dir);
}
pub fn pos_of(&self, id: Id) -> Option<(i32, i32)> {
self.grid.level().board().find(id).map(|o| (o.x, o.y))
}
pub fn kinds_at(&self, x: i32, y: i32) -> Vec<String> {
let mut kinds: Vec<String> = self
.grid
.level()
.board()
.at(x, y)
.map(|o| o.kind.to_string())
.collect();
kinds.sort();
kinds
}
pub fn stack_at(&self, x: i32, y: i32) -> Vec<String> {
let board = self.grid.level().board();
board.at(x, y).map(|o| o.kind.to_string()).collect()
}
pub fn step(&mut self, dir: Dir) {
self.grid.step(dir);
}
pub fn won(&self) -> bool {
self.grid.level().won()
}
pub fn lost(&self) -> bool {
self.grid.level().lost()
}
/// Not supported: Part I has state (health, hands, the log) that the level's
/// snapshot doesn't cover. Panics.
pub fn undo(&mut self) -> bool {
panic!("undo is not supported in SpellGrid");
}
// ---- Module 10 ----
/// Cast a spell from player 0's hand, as one Sequence. `target` is a minion's id.
pub fn cast(&mut self, card: &str, target: Option<Id>) {
self.grid.cast(card, target);
}
/// Play a minion from player 0's hand onto the empty tile (x, y); returns its id.
pub fn play(&mut self, card: &str, x: i32, y: i32) -> Id {
self.grid.play(card, x, y)
}
pub fn health(&self, id: Id) -> i32 {
self.grid.world().health(self.grid.entity(id))
}
pub fn hero(&self, player: Player) -> Id {
self.grid.id_of(self.grid.world().hero(player))
}
pub fn name_of(&self, id: Id) -> String {
self.grid
.world()
.card_name(self.grid.entity(id))
.to_string()
}
/// The minion on tile (x, y), if any.
pub fn minion_at(&self, x: i32, y: i32) -> Option<Id> {
self.grid.minion_at(x, y)
}
/// The card names in player 0's hand, in order.
pub fn hand(&self) -> Vec<String> {
let world = self.grid.world();
world
.zone(0, Zone::Hand)
.iter()
.map(|&id| world.card_name(id).to_string())
.collect()
}
/// The course's view of an engine event, with every entity id as a grid id.
fn grid_event(&self, event: &Event) -> Option<GameEvent> {
let id = |entity: Id| self.grid.id_of(entity);
Some(match GameEvent::from_engine(event)? {
GameEvent::Summoned { id: e } => GameEvent::Summoned { id: id(e) },
GameEvent::Destroyed { id: e } => GameEvent::Destroyed { id: id(e) },
GameEvent::Drew { player, id: e } => GameEvent::Drew { player, id: id(e) },
GameEvent::StatsChanged {
id: e,
attack,
health,
} => GameEvent::StatsChanged {
id: id(e),
attack,
health,
},
GameEvent::Damaged { target, amount } => GameEvent::Damaged {
target: id(target),
amount,
},
GameEvent::Triggered { source, trigger } => GameEvent::Triggered {
source: id(source),
trigger,
},
other => other,
})
}
/// Part I's log, with grid ids.
pub fn log(&self) -> Vec<GameEvent> {
let log = self.grid.world().log();
log.iter().filter_map(|e| self.grid_event(e)).collect()
}
pub fn clear_log(&mut self) {
self.grid.world_mut().clear_log();
}
/// Part I's log as an indented tree, with grid ids (see course/harness.md).
pub fn log_tree(&self) -> String {
log_tree(self.grid.world(), |e| self.grid_event(e))
}
/// Who makes Assemble's choices (`"base"`, `"donor"`, `"half"`).
pub fn set_chooser(&mut self, chooser: impl FnMut(&str, &[String]) -> String + 'static) {
self.grid.world_mut().set_chooser(Box::new(chooser));
}
/// For the command line: why `cast` would panic, or `Ok` if it wouldn't.
pub fn check_cast(&self, card: &str, target: Option<Id>) -> Result<(), String> {
self.grid.check_cast(card, target)
}
/// For the command line: why `play` would panic, or `Ok` if it wouldn't.
pub fn check_play(&self, card: &str, x: i32, y: i32) -> Result<(), String> {
self.grid.check_play(card, x, y)
}
/// Test hook: the naive engine reads the auras before the dead leave play, the
/// same end-of-Phase steps reversed.
pub fn with_naive_aura_timing(&mut self) {
let mut steps = END_OF_PHASE.to_vec();
steps.reverse();
self.grid.world_mut().set_end_of_phase(steps);
}
}
@@ -0,0 +1,4 @@
pub mod cli;
pub mod content;
pub mod engine;
pub mod harness;
@@ -0,0 +1,113 @@
// The SpellGrid command line, played from a script (the reference's own check; not a course test).
use std::cell::RefCell;
use std::io::{Cursor, Write};
use std::rc::Rc;
use engine::cli::{run, Outcome};
/// Collects what the CLI prints.
#[derive(Clone, Default)]
struct Screen(Rc<RefCell<Vec<u8>>>);
impl Write for Screen {
fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
self.0.borrow_mut().extend_from_slice(buf);
Ok(buf.len())
}
fn flush(&mut self) -> std::io::Result<()> {
Ok(())
}
}
fn play(level: &str, input: &[&str]) -> (Outcome, String) {
let screen = Screen::default();
let input = Box::new(Cursor::new(input.join("\n") + "\n"));
let outcome = run(level, input, Box::new(screen.clone()));
let out = String::from_utf8(screen.0.borrow().clone()).unwrap();
(outcome, out)
}
#[test]
fn cli_wins_level_1() {
let input = ["w", "w", "c Whirlwind", "s", "s", "a", "s", "s", "d", "d"];
let (outcome, out) = play(include_str!("../levels/l1-double-tap.json"), &input);
assert_eq!(outcome, Outcome::Won);
assert!(out.contains("DEA IS DOU"));
assert!(out.contains("You win!"));
}
#[test]
fn cli_wins_level_2_with_choices_and_play() {
let input = [
"c Graft",
"Ember Moth",
"Rune Scribe",
"bottom",
"c Ember Moth (Assembled) 4 3",
"a",
"a",
];
let (outcome, out) = play(include_str!("../levels/l2-spliced-corpse.json"), &input);
assert_eq!(outcome, Outcome::Won);
assert!(out.contains("Choose the base"));
assert!(out.contains("Hand: Ember Moth (Assembled)"));
}
#[test]
fn cli_prints_the_log_tree_of_each_command() {
let level = include_str!("../levels/l1-double-tap.json");
let (_, out) = play(level, &["w", "w", "c Whirlwind", "q"]);
// The moves run no Sequence, so only the cast prints a log: its tree, Deathrattles nested.
assert_eq!(out.matches("Log:\n").count(), 1, "{out}");
let tree = [
"Log:",
"SpellCast{card: \"Whirlwind\"}",
"Damaged{target: 32, amount: 1}",
"Damaged{target: 33, amount: 1}",
"Destroyed{id: 32}",
"DeathPhase{round: 1}",
"Triggered{source: 32, trigger: \"deathrattle\"}",
" Damaged{target: 33, amount: 1}",
"Triggered{source: 32, trigger: \"deathrattle\"}",
" Damaged{target: 33, amount: 1}",
"Destroyed{id: 33}",
"DeathPhase{round: 2}",
"",
];
assert!(out.contains(&tree.join("\n")), "{out}");
}
#[test]
fn cli_reports_errors_and_quits() {
let input = ["c Spark", "x", "q"];
let (outcome, out) = play(
include_str!("../levels/l3-simultaneous-paradox.json"),
&input,
);
assert_eq!(outcome, Outcome::Quit);
assert!(out.contains("\"Spark\" is not in the hand"));
assert!(out.contains("Commands:"));
}
// The two reference CLIs print the same text for the same input (tools/diff-tracks.mjs checks it).
#[test]
fn cli_prints_card_names_as_typed() {
let level = include_str!("../levels/l3-simultaneous-paradox.json");
let (_, out) = play(level, &["c a\"b", "q"]);
assert!(out.contains("\"a\"b\" is not in the hand"), "{out}");
}
#[test]
fn cli_numbers_are_plain_digits() {
// "-1" is not a coordinate, so it is part of the card's name.
let level = include_str!("../levels/l1-double-tap.json");
let (_, out) = play(level, &["c Whirlwind -1 0", "q"]);
assert!(
out.contains("\"Whirlwind -1 0\" is not in the hand"),
"{out}"
);
// "+2" is not a number, so the chooser takes the first option.
let level = include_str!("../levels/l2-spliced-corpse.json");
let (_, out) = play(level, &["c Graft", "+2", "1", "1", "q"]);
assert!(out.contains("Hand: Clay Golem (Assembled)"), "{out}");
}
@@ -0,0 +1 @@
{ "track": "ts", "engineDirs": ["src/engine"], "contentDirs": ["src/content", "cards"], "module": "11" }
@@ -0,0 +1 @@
node_modules/
@@ -0,0 +1,17 @@
# Module 11 reference engine (TypeScript)
The course's own engine as of the end of Module 11: Module 10's engine (Part I's card engine and Part II's grid engine, joined by SpellGrid), with conditions in Part II. Every test from Modules 1–10 still passes. Module 10's reference was the starting point, so a diff between the two shows what this module adds. `course/10-spellgrid/reference/ts/README.md` describes everything else.
- `cards/` and `levels/` (at the package root; not in this directory): staged by `tools/verify-references.mjs` as for Module 10 (`course/cards/`, `course/10-spellgrid/cards/` and `course/10-spellgrid/levels/`), so SpellGrid's tests run here too.
- Part II (`src/engine/puzzle/`):
- `word.ts`: a fourth role, `"condition"`.
- `parse.ts`: `Condition { word, nouns }` and `Rule.conditions`. `beforeIs` reads what stands before an `IS`: the subjects alone, or, when a condition word stands before the run of nouns next to `IS`, the subjects before that word with the condition (`nounRuns` reads one noun from each tile of a stack). `ruleText` writes `BABA ON KEY AND ROCK IS WIN`. `negates` and the new `sameConditions` pair rules only with the same conditions.
- `resolve.ts`: negation and the identity lock pair rules only when their conditions are the same.
- `condition.ts` (new): `ConditionCheck`, `(level, id, nouns) => boolean`, and `ConditionChecks`, content's checks by word.
- `level.ts`: the constructor takes the condition checks and throws if a condition word has none. `appliesTo(rule, id)` (the subject names the object and every condition holds for it now) is what `hasProperty` and conversion use, so every question about an object evaluates its conditions at that moment. `names(word, id)` is public, for content's checks.
- Part III (`src/engine/spellgrid.ts`): `SpellGrid`'s constructor takes the condition checks; `ruleAuras` skips a rule with a condition.
- `src/content/conditions.ts` (new): `on`, registered under `ON`. `words.json` adds `{ "name": "ON", "role": "condition" }`.
- `src/harness.ts`: `newBaba` and `newSpellGrid` pass `conditions()`. No new harness function: `rules()` and `activeRules()` print conditions.
- `src/cli.ts`: unchanged from Module 10. `tools/play-spellgrid.mjs` plays Module 10's; `tools/diff-tracks.mjs` uses this module's by default.
Run it with the module's tests: `node tools/verify-references.mjs 11` from the course repository.
File diff suppressed because it is too large. Load diff
@@ -0,0 +1,16 @@
{
"name": "systemic-design-workspace",
"version": "0.1.0",
"private": true,
"type": "module",
"scripts": {
"test": "vitest run",
"typecheck": "tsc --noEmit"
},
"devDependencies": {
"@types/node": "^22.20.5",
"tsx": "^4.23.15",
"typescript": "^5.6.0",
"vitest": "^2.1.0"
}
}
@@ -0,0 +1,138 @@
// SpellGrid on the command line: `npx tsx src/cli.ts levels/l1-double-tap.json`.
//
// A thin loop over the harness: it draws the level, reads a command, and calls
// `step`, `cast` or `play`. No rules live here. After a command that ran Part I
// Sequences, it prints their events under `Log:` as an indented tree.
//
// w a s d one turn: up, left, down, right
// c <card> [x y] cast a spell from your hand (at the minion on (x, y), if given),
// or play a minion from your hand onto the empty tile (x, y)
// q quit
//
// Assemble's choices (Graft) are asked for one at a time: answer with a number or a name.
import { readFileSync, readSync } from "node:fs";
import { pathToFileURL } from "node:url";
import { newSpellGrid, type LevelDef, type SpellGrid } from "./harness";
const DIRS = { w: "up", a: "left", s: "down", d: "right" } as const;
/** One cell of the drawing: the newest object on the tile, as 3 letters; TEXT in capitals. */
function cell(g: SpellGrid, x: number, y: number): string {
const kinds = g.kindsAt(x, y);
if (kinds.length === 0) return " . ";
// kindsAt is sorted; prefer text, which is what the player reads.
const kind = kinds.find((k) => k.startsWith("text:")) ?? kinds[0];
const name = kind.startsWith("text:") ? kind.slice(5).toUpperCase() : kind.toLowerCase();
return name.slice(0, 3).padEnd(3);
}
/** The level as text: the grid, the rules, the minions and the hand. */
export function render(g: SpellGrid, level: LevelDef): string {
const lines: string[] = [];
for (let y = 0; y < level.height; y++) {
const row: string[] = [];
for (let x = 0; x < level.width; x++) row.push(cell(g, x, y));
lines.push(row.join(" "));
}
lines.push(`Rules: ${g.activeRules().join(", ")}`);
const minions: string[] = [];
for (let y = 0; y < level.height; y++) {
for (let x = 0; x < level.width; x++) {
const id = g.minionAt(x, y);
if (id !== null) minions.push(`${g.nameOf(id)} (${x},${y}) ${g.health(id)} Health`);
}
}
lines.push(`Minions: ${minions.join("; ") || "none"}`);
lines.push(`Hand: ${g.hand().join(", ") || "empty"}`);
return lines.join("\n");
}
/** A coordinate, or a chooser's number: one to nine digits, nothing else (no sign, no point). */
const NUMBER = /^\d{1,9}$/;
/** A `c` command: `c <card words...> [x y]`. */
function parseCast(args: string[]): { card: string; at?: [number, number] } {
const n = args.length;
if (n >= 3 && NUMBER.test(args[n - 2]) && NUMBER.test(args[n - 1])) {
return { card: args.slice(0, n - 2).join(" "), at: [Number(args[n - 2]), Number(args[n - 1])] };
}
return { card: args.join(" ") };
}
/**
* Play `levelText` (a level in the course's format), reading commands from
* `readLine` (undefined at the end of input) and writing to `write`. Returns
* `"won"`, `"lost"` or `"quit"`.
*/
export function run(levelText: string, readLine: () => string | undefined, write: (s: string) => void): string {
const level = JSON.parse(levelText) as LevelDef;
const g = newSpellGrid(level, []);
g.setChooser((kind, options) => {
write(`Choose the ${kind} (a number or a name):\n${options.map((o, i) => ` ${i + 1}) ${o}`).join("\n")}\n? `);
const answer = readLine()?.trim() ?? "";
if (options.includes(answer)) return answer;
return (NUMBER.test(answer) ? options[Number(answer) - 1] : undefined) ?? options[0];
});
for (;;) {
write(`${render(g, level)}\n`);
if (g.won()) {
write("You win!\n");
return "won";
}
if (g.lost()) {
write("Nothing is YOU. You lose.\n");
return "lost";
}
write("> ");
const line = readLine();
if (line === undefined) return "quit";
const [command, ...args] = line.trim().split(/\s+/);
// The log holds only what this command does.
g.clearLog();
try {
if (command === "q") return "quit";
if (command in DIRS) g.step(DIRS[command as keyof typeof DIRS]);
else if (command === "c" && args.length > 0) {
const { card, at } = parseCast(args);
const target = at === undefined ? null : g.minionAt(...at);
if (at === undefined) g.cast(card);
else if (target !== null) g.cast(card, target);
else g.play(card, ...at);
} else write("Commands: w a s d (move), c <card> [x y] (cast or play), q (quit)\n");
} catch (e) {
write(`${e instanceof Error ? e.message : String(e)}\n`);
}
const tree = g.logTree();
if (tree !== "") write(`Log:\n${tree}`);
}
}
/** Lines from standard input, read synchronously, so the chooser can ask mid-Sequence. */
function stdinLines(): () => string | undefined {
const buffer = Buffer.alloc(4096);
let pending = "";
let ended = false;
return () => {
while (!pending.includes("\n") && !ended) {
const n = readSync(0, buffer, 0, buffer.length, null);
if (n === 0) ended = true;
else pending += buffer.toString("utf8", 0, n);
}
if (pending === "" && ended) return undefined;
const end = pending.indexOf("\n");
const line = end < 0 ? pending : pending.slice(0, end);
pending = end < 0 ? "" : pending.slice(end + 1);
return line;
};
}
if (process.argv[1] !== undefined && import.meta.url === pathToFileURL(process.argv[1]).href) {
const path = process.argv[2];
if (path === undefined) {
process.stderr.write("usage: npx tsx src/cli.ts <level.json>\n");
process.exit(2);
}
const outcome = run(readFileSync(path, "utf8"), stdinLines(), (s) => process.stdout.write(s));
process.exit(outcome === "won" ? 0 : 1);
}
@@ -0,0 +1,47 @@
// Part II: what each condition word means (Module 11). The engine asks these checks
// every time it is about to use a rule with a condition, for one object, against the
// board as it is then. A new condition word is one entry in `words.json`, with role
// "condition", and one check here.
import type { puzzle } from "../engine";
type Level = puzzle.Level;
type ObjectId = puzzle.ObjectId;
/**
* Can each noun be given an object of its own among `others`, one the noun names?
* `ON ROCK AND ROCK` needs two rocks (the wiki: "AND in the noun2 part requires at
* least as many objects as are mentioned"). Tries every assignment.
*/
function eachHasOne(level: Level, nouns: readonly string[], others: readonly ObjectId[]): boolean {
const taken = new Set<ObjectId>();
const assign = (i: number): boolean => {
if (i === nouns.length) return true;
for (const other of others) {
if (taken.has(other) || !level.names(nouns[i], other)) continue;
taken.add(other);
if (assign(i + 1)) return true;
taken.delete(other);
}
return false;
};
return assign(0);
}
/**
* ON: for each noun, another object on the object's tile that the noun names, a
* different one for each noun. Float levels don't matter ("ON ignores FLOAT levels").
*/
function on(level: Level, id: ObjectId, nouns: readonly string[]): boolean {
const o = level.board.get(id);
const others = level.board
.at(o.x, o.y)
.map((other) => other.id)
.filter((other) => other !== id);
return eachHasOne(level, nouns, others);
}
/** The condition checks, by condition word. */
export function conditions(): puzzle.ConditionChecks {
return { ON: on };
}
@@ -0,0 +1,48 @@
// Part II: what destroys what at the end of a turn. The engine runs these
// hazards in this order (the order of the wiki's `block()`: SINK, then HOT/MELT,
// then DEFEAT), and destroys each one's victims together before the next runs.
// A hazard only reaches objects on the same float level: it asks the level what
// shares a tile (`Level.sharing`). A new hazard property is one more function here.
import type { puzzle } from "../engine";
type Level = puzzle.Level;
type ObjectId = puzzle.ObjectId;
/** The objects that have `property`, oldest first. */
const withProperty = (level: Level, property: string): ObjectId[] =>
level.board
.objects()
.map((o) => o.id)
.filter((id) => level.hasProperty(id, property));
/** A SINK object that shares its tile with anything destroys all of it, and itself. */
function sink(level: Level): ObjectId[] {
const doomed: ObjectId[] = [];
for (const id of withProperty(level, "SINK")) {
const others = level.sharing(id);
if (others.length > 0) doomed.push(id, ...others);
}
return doomed;
}
/** A MELT object is destroyed if it is HOT itself, or shares its tile with a HOT one. */
function hotMelt(level: Level): ObjectId[] {
return destroyedBy(level, "MELT", "HOT");
}
/** A YOU object is destroyed if it is DEFEAT itself, or shares its tile with a DEFEAT one. */
function defeat(level: Level): ObjectId[] {
return destroyedBy(level, "YOU", "DEFEAT");
}
/** The objects with `victim` that have `killer` too, or share a tile with one that does. */
function destroyedBy(level: Level, victim: string, killer: string): ObjectId[] {
const deadly = (id: ObjectId) => level.hasProperty(id, killer);
return withProperty(level, victim).filter((id) => deadly(id) || level.sharing(id).some(deadly));
}
/** The hazards, in the order the engine runs them. */
export function hazards(): puzzle.Hazard[] {
return [sink, hotMelt, defeat];
}
@@ -0,0 +1,50 @@
// Content: the cards and the Baba Is You words. Only this directory (and the card
// data in `cards/`) may name specific cards, or specific Baba Is You nouns.
//
// From Module 5 every card is a JSON file in `cards/` at the package root, in the
// course's card format (`course/schema/card.schema.json`). The functions that were
// card effects in Modules 2–4 are gone: the engine interprets the data.
// `start-module` copies the course's card files into `cards/`; add your own there.
//
// Module 10: the cards are read with Node's file system, not Vite's `import.meta.glob`,
// so the SpellGrid command line (`npx tsx src/cli.ts`), which runs without Vite, reads
// the same cards as the tests.
import { readdirSync, readFileSync } from "node:fs";
import type { CardDef, puzzle, SeamWords } from "../engine";
import seamWords from "./seams.json";
import wordList from "./words.json";
export { conditions } from "./conditions";
export { hazards } from "./hazards";
export { responses as movement } from "./movement";
const CARDS = new URL("../../cards/", import.meta.url);
let loaded: CardDef[] | undefined;
/** Every card in `cards/`, in file-name order. Read once. */
export function cards(): CardDef[] {
loaded ??= readdirSync(CARDS)
.filter((file) => file.endsWith(".json"))
.sort()
.map((file) => JSON.parse(readFileSync(new URL(file, CARDS), "utf8")) as CardDef);
return loaded;
}
/**
* Part II: the words of Baba Is You the parser can read, besides the engine's own
* (`IS`, `AND`, `NOT`, `TEXT`). Nouns and properties are data: a new word is one
* more entry in `words.json` in this directory. A condition word (Module 11) also
* needs its check in `conditions.ts`.
*/
export function words(): puzzle.WordDef[] {
return wordList as puzzle.WordDef[];
}
/**
* Part III: the words that cross between the engines (`seams.json`). A noun that
* names a trigger, and a property that multiplies one: `DEATHRATTLE IS DOUBLE`.
*/
export function seams(): SeamWords {
return seamWords as SeamWords;
}
@@ -0,0 +1,45 @@
// Part II: what each property does when an object moves. The engine plans a move
// one step at a time and runs these responses for every step, in this order (the
// order of the wiki's movement checks: OPEN/SHUT, then PUSH, then STOP). A new
// movement property is one more response here.
import type { puzzle } from "../engine";
type Planner = puzzle.Planner;
type Step = puzzle.Step;
/**
* An OPEN object stepping onto a SHUT one (or a SHUT one onto an OPEN one)
* destroys both, and that object doesn't block the step. One partner per step:
* the oldest.
*/
function openShut(p: Planner, s: Step): boolean {
for (const [mine, theirs] of [
["OPEN", "SHUT"],
["SHUT", "OPEN"],
]) {
if (!p.has(s.mover, mine)) continue;
const partner = p.stayingAt(s.to).find((o) => p.has(o, theirs));
if (partner !== undefined) {
p.destroy(partner);
p.destroy(s.mover);
return true;
}
}
return true;
}
/** Every PUSH object on the tile ahead must step the same way, or the step fails. */
function push(p: Planner, s: Step): boolean {
return p.stayingAt(s.to).every((o) => !p.has(o, "PUSH") || p.tryMove(o));
}
/** A STOP object still on the tile ahead blocks the step. */
function stop(p: Planner, s: Step): boolean {
return !p.stayingAt(s.to).some((o) => p.has(o, "STOP"));
}
/** The movement responses, in the order the engine runs them. */
export function responses(): puzzle.Response[] {
return [openShut, push, stop];
}
@@ -0,0 +1,4 @@
{
"triggerWords": { "DEATHRATTLE": "deathrattle" },
"multiplierWords": { "DOUBLE": { "times": 2, "stacking": "max" } }
}
@@ -0,0 +1,36 @@
[
{ "name": "BABA", "role": "noun" },
{ "name": "ROCK", "role": "noun" },
{ "name": "WALL", "role": "noun" },
{ "name": "FLAG", "role": "noun" },
{ "name": "LAVA", "role": "noun" },
{ "name": "KEY", "role": "noun" },
{ "name": "DOOR", "role": "noun" },
{ "name": "SKULL", "role": "noun" },
{ "name": "WATER", "role": "noun" },
{ "name": "ALL", "role": "noun", "selects": "things" },
{ "name": "KEG", "role": "noun" },
{ "name": "BRAMBLE", "role": "noun" },
{ "name": "HEDGE", "role": "noun" },
{ "name": "WARDEN", "role": "noun" },
{ "name": "MOTH", "role": "noun" },
{ "name": "SCRIBE", "role": "noun" },
{ "name": "KNIGHT", "role": "noun" },
{ "name": "CRIER", "role": "noun" },
{ "name": "GOLEM", "role": "noun" },
{ "name": "SHARD", "role": "noun" },
{ "name": "DEATHRATTLE", "role": "noun" },
{ "name": "ON", "role": "condition" },
{ "name": "YOU", "role": "property" },
{ "name": "WIN", "role": "property" },
{ "name": "PUSH", "role": "property" },
{ "name": "STOP", "role": "property" },
{ "name": "OPEN", "role": "property" },
{ "name": "SHUT", "role": "property" },
{ "name": "HOT", "role": "property" },
{ "name": "MELT", "role": "property" },
{ "name": "SINK", "role": "property" },
{ "name": "DEFEAT", "role": "property" },
{ "name": "FLOAT", "role": "property" },
{ "name": "DOUBLE", "role": "property" }
]
@@ -0,0 +1,117 @@
// Assemble: "Get a random minion. Change half its text."
//
// Card text is a list of clauses, so changing half of it is list surgery: the base
// card's `text.slice(0, split)` (top) or `text.slice(split)` (bottom) is replaced by
// the same half of a donor card. Name, cost and stats stay the base's. This works
// only because a card's behaviour is its data: the new card needs no new code.
//
// Hearthstone's wiki doesn't define "half". The course's `split` field is its
// definition: each card's data says where its top half ends.
import { hasTrigger, wellFormed, type CardDef, type PoolFilter } from "./card";
import type { EntityId, PlayerId, World } from "./world";
export type Half = "top" | "bottom";
const HALVES: Half[] = ["top", "bottom"];
/** Why an Assemble was refused: the text would not change, or would not be well-formed (see `wellFormed`). */
export type AssembleFailure = "unchanged" | "illFormed";
/** Assembling `base` with `donor`'s `half` would leave the base's text unchanged, or ill-formed. */
export class AssembleError extends Error {
constructor(
readonly base: string,
readonly donor: string,
readonly half: Half,
readonly reason: AssembleFailure,
) {
const result = reason === "unchanged" ? "leaves its text unchanged" : "gives ill-formed text";
super(`assembling "${base}" with the ${half} half of "${donor}" ${result}`);
this.name = "AssembleError";
}
}
/** JSON with object keys sorted, so two clauses written with their keys in a different order compare equal. */
const canonical = (value: unknown): string =>
JSON.stringify(value, (_key, v) =>
v && typeof v === "object" && !Array.isArray(v)
? Object.fromEntries(Object.entries(v).sort(([a], [b]) => (a < b ? -1 : 1)))
: v,
);
const sameText = (a: CardDef["text"], b: CardDef["text"]): boolean => canonical(a) === canonical(b);
/** Two definitions the engine can't tell apart (it ignores `fictional` and `game`). */
const sameCard = (a: CardDef, b: CardDef): boolean => {
const read = (d: CardDef) =>
canonical([d.name, d.type, d.cost, d.attack ?? 0, d.health ?? 0, d.noun ?? null, d.text, d.split]);
return read(a) === read(b);
};
/**
* The base card with `half` of its text replaced by the same half of `donor`.
* It keeps the base's type, cost and stats, and is named "<base> (Assembled)".
*/
export function assemble(base: CardDef, donor: CardDef, half: Half): CardDef {
const topOf = (c: CardDef) => c.text.slice(0, c.split);
const bottomOf = (c: CardDef) => c.text.slice(c.split);
const top = half === "top" ? topOf(donor) : topOf(base);
const bottom = half === "bottom" ? bottomOf(donor) : bottomOf(base);
const text = [...top, ...bottom];
if (sameText(text, base.text)) throw new AssembleError(base.name, donor.name, half, "unchanged");
if (!wellFormed(text, base.type)) throw new AssembleError(base.name, donor.name, half, "illFormed");
return { ...base, name: `${base.name} (Assembled)`, text, split: top.length };
}
/** Does assembling `base` with `donor`'s `half` give a new, well-formed text? */
const changes = (base: CardDef, donor: CardDef, half: Half): boolean => {
try {
assemble(base, donor, half);
return true;
} catch (e) {
if (e instanceof AssembleError) return false;
throw e;
}
};
/**
* Register an assembled card and return its name. If a different card already has
* the name (an earlier Assemble of the same base), a number is added: "(Assembled 2)".
*/
export function registerAssembled(world: World, def: CardDef): string {
const stem = def.name.replace(/\)$/, "");
let name = def.name;
// A different card is any difference the engine reads: the text, the split, or a base re-registered since.
for (let n = 2; world.hasCard(name) && !sameCard(world.card(name).def, { ...def, name }); n++) {
name = `${stem} ${n})`;
}
world.registerAssembled({ ...def, name });
return name;
}
/**
* The `assemble` effect: the owner's chooser picks a base minion from the pool
* (filtered), a donor, and the half to replace. Only combinations that change the
* base's text and leave it well-formed are offered: a donor is offered if one of its
* halves gives such a text, and a base if it has such a donor. The Assembled minion
* goes to the owner's hand. Real Hearthstone picks the base at random and offers two
* donors; here every choice is the chooser's.
*/
export function assembleFromPool(world: World, player: PlayerId, filter: PoolFilter): EntityId | undefined {
const pool = world.pool();
const donorsFor = (base: CardDef): string[] =>
pool
.map((c) => c.def)
.filter((d) => d.name !== base.name && HALVES.some((h) => changes(base, d, h)))
.map((d) => d.name);
const bases = pool
.filter((c) => (filter.trigger === undefined || hasTrigger(c, filter.trigger)) && donorsFor(c.def).length > 0)
.map((c) => c.def.name);
if (bases.length === 0) return undefined;
const base = world.card(world.choose("base", bases)).def;
const donor = world.card(world.choose("donor", donorsFor(base))).def;
const half = world.choose(
"half",
HALVES.filter((h) => changes(base, donor, h)),
) as Half;
const name = registerAssembled(world, assemble(base, donor, half));
return world.create(name, player, "hand");
}
@@ -0,0 +1,65 @@
// Auras, the one routine that applies trigger multipliers, and the lookup for
// damage redirects.
//
// An aura is data on a card: a continuous effect its minion has while in play.
// The engine never asks "is that card on the board?" in the middle of resolving
// something. It takes a snapshot of the auras in force at defined points (an Aura
// Update) and every rule reads the snapshot.
//
// Hearthstone runs two kinds of Aura Update when the outermost Phase ends: one for
// Health/Attack auras before the Death Creation Step, and one for every other aura
// after it. Trigger multipliers are "Other" auras, so a minion that dies in a Death
// Creation Step no longer multiplies the Deathrattles of that step's Death Phase.
// Summoning a minion also runs an Aura Update.
//
// Module 10: auras can come from outside the cards. SpellGrid compiles rules on its
// grid into auras (`World.setRuleAuras`); they join the snapshot at the next Aura
// Update, beside the minions' own, and go through the same routine. A rule's
// multiplier stacks with a card's by the same `stacking` field. A cap (`IS NOT`)
// limits the total, whatever multiplies it.
import type { TriggerOn } from "./card";
import type { EntityId, PlayerId, World } from "./world";
/**
* Aura Update (Other): replace the snapshot with the auras of every minion now in
* play, and the auras from outside the cards, which apply to both players.
*/
export function auraUpdateOther(world: World): void {
const players: PlayerId[] = [0, 1];
world.setActiveAuras([
...world.minionsInPlay().flatMap((id) =>
world.card(world.cardName(id)).auras.map((aura) => ({ owner: world.ownerOf(id), source: id, aura })),
),
...world.ruleAuras().flatMap((aura) => players.map((owner) => ({ owner, aura }))),
]);
}
/**
* The minion that takes damage meant for `player`'s hero, if a redirect is in force:
* the oldest (order of play) of `player`'s minions with a `damageRedirect` aura in the
* snapshot. The snapshot lists minions oldest first, so that is the first one found.
*/
export function redirectFor(world: World, player: PlayerId): EntityId | undefined {
return world.activeAuras().find((a) => a.aura.type === "damageRedirect" && a.owner === player)?.source;
}
/**
* How many times a trigger listening for `on`, on a minion `owner` controls, activates
* when it resolves. Reads the snapshot from the last Aura Update. With no multiplier in
* force the answer is 1. Multipliers with `stacking: "max"` contribute their largest
* `times`; each `"additive"` one adds `times - 1` extra activations. A cap
* (`triggerCap`) then limits the total to its `times`.
*/
export function activations(world: World, owner: PlayerId, on: TriggerOn): number {
let max = 1;
let extra = 0;
let cap = Infinity;
for (const { owner: auraOwner, aura } of world.activeAuras()) {
if (aura.type === "damageRedirect" || auraOwner !== owner || aura.trigger !== on) continue;
if (aura.type === "triggerCap") cap = Math.min(cap, aura.times);
else if (aura.stacking === "max") max = Math.max(max, aura.times);
else extra += aura.times - 1;
}
return Math.min(cap, max + extra);
}
@@ -0,0 +1,209 @@
// Cards as data. A `CardDef` is the course's card format (`course/schema/card.schema.json`),
// read straight from JSON. The engine compiles each definition once, when it is
// registered, into the lists it reads while resolving: triggers, auras and a
// spell's effects. The definitions themselves live in `cards/`, loaded by `content`.
/** What a triggered ability listens for. */
export type TriggerOn =
| "selfDamaged" // this minion takes damage
| "friendlyMinionDamaged" // a minion with the same owner (this one included) takes damage
| "anyMinionDamaged" // any minion takes damage
| "heroDamaged" // either hero takes damage
| "deathrattle" // this minion dies (it resolves after the minion has left play)
| "battlecry" // this minion is played from hand
| "endOfTurn"; // the end of its owner's turn: the course has no turns, so it never fires
/** The trigger's family name, as reported in `Triggered` events. */
export function triggerLabel(on: TriggerOn): string {
switch (on) {
case "selfDamaged":
case "friendlyMinionDamaged":
case "anyMinionDamaged":
case "heroDamaged":
return "damaged";
case "deathrattle":
case "battlecry":
case "endOfTurn":
return on;
}
}
/**
* An aura: a continuous effect a minion has while it is in play. The engine reads
* auras only at the Aura Update points (see `aura.ts`), never mid-Phase.
*
* `triggerMultiplier`: triggers listening for `trigger`, belonging to minions with
* the aura owner's owner, activate `times` times each. `stacking` says how several
* multipliers for the same trigger combine: `"max"` takes the largest (two copies of
* a doubler still double); `"additive"` adds each copy's extra activations.
*/
export type Aura =
| { type: "triggerMultiplier"; trigger: TriggerOn; times: number; stacking: "max" | "additive" }
/**
* Damage that would be dealt to the aura owner's hero is dealt to the aura's minion
* instead (see `DAMAGE_STEPS` in `queue.ts`).
*/
| { type: "damageRedirect"; from: "friendlyHero"; to: "self" }
/**
* Module 10: triggers listening for `trigger` activate at most `times` times, whatever
* multiplies them. Not in the card format: SpellGrid compiles it from a negated rule
* (`DEATHRATTLE IS NOT DOUBLE`).
*/
| { type: "triggerCap"; trigger: TriggerOn; times: number };
/** Who an effect touches. */
export type Target = "self" | "allMinions" | "allOtherMinions" | "allOtherCharacters" | "enemyHero" | "chosen";
/** Which minions an Assemble may pick as its base. An empty filter allows any. */
export interface PoolFilter {
/** Only minions with a trigger clause listening for this. */
trigger?: TriggerOn;
}
/** One effect: a node of a card's ability, interpreted by `effect.ts`. */
export type Effect =
| { type: "dealDamage"; target: Target; amount: number }
| { type: "draw"; count: number }
| { type: "summon"; card: string; count: number }
| { type: "gainArmor"; amount: number }
| { type: "gainAttack"; target: Target; amount: number }
| { type: "destroy"; target: Target }
/** Part III (SpellGrid) only: puts a word on the grid. */
| { type: "spawnText"; word: string; at: "self" }
/** Get a minion from the pool, with half its text replaced (see `assemble.ts`). */
| { type: "assemble"; filter: PoolFilter }
/** Give every minion in the owner's hand +attack/+health. */
| { type: "buffHand"; attack: number; health: number };
/**
* One line of card text. A trigger clause applies to the effect clauses after it,
* until the next trigger, keyword or aura. Effects before any trigger are what a
* spell does when cast.
*/
export type Clause =
| { kind: "keyword"; keyword: string }
| { kind: "trigger"; on: TriggerOn }
| { kind: "aura"; modifier: Aura }
| { kind: "effect"; effect: Effect };
/** A card as data, in the course's card format. */
export interface CardDef {
name: string;
type: "minion" | "spell";
cost: number;
attack?: number;
health?: number;
noun?: string;
text: Clause[];
/** `text.slice(0, split)` is the top half, `text.slice(split)` the bottom half. */
split: number;
fictional?: boolean;
game?: string;
}
/** A triggered ability: "whenever <on>, <effects>". */
export interface Trigger {
on: TriggerOn;
effects: Effect[];
}
/** A registered card: its definition, and the definition's text compiled into the lists the engine reads. */
export interface Card {
def: CardDef;
/** Triggered abilities, in printed order. */
triggers: Trigger[];
/** Auras this minion has while in play. */
auras: Aura[];
/** What the card does when cast (a spell's effects). */
onCast: Effect[];
/** May an Assemble offer this card? Minions with text that were not themselves assembled. */
inPool: boolean;
}
/** Compile a definition's text. Throws if `split` is past the end of the text. */
export function compile(def: CardDef, inPool: boolean): Card {
if (def.split > def.text.length) throw new Error(`"${def.name}": split ${def.split} is past the end of its text`);
const card: Card = { def, triggers: [], auras: [], onCast: [], inPool: false };
// The trigger whose effects are still being listed, if any.
let open: Trigger | undefined;
for (const clause of def.text) {
switch (clause.kind) {
case "trigger":
open = { on: clause.on, effects: [] };
card.triggers.push(open);
break;
case "effect":
(open ? open.effects : card.onCast).push(clause.effect);
break;
case "aura":
card.auras.push(clause.modifier);
open = undefined;
break;
case "keyword": // The course has no combat, so keywords such as Taunt do nothing here.
open = undefined;
break;
}
}
card.inPool = inPool && def.type === "minion" && def.text.length > 0;
return card;
}
export const hasTrigger = (card: Card, on: TriggerOn): boolean => card.triggers.some((t) => t.on === on);
/**
* Does `text` mean what it says? For a minion, every effect clause belongs to a
* trigger (none comes before the first trigger, or after an aura or keyword), and
* every trigger has at least one effect before the next trigger, aura, keyword or
* the end. The rule is for minions (Assemble makes only minions): a spell passes.
*/
export function wellFormed(text: readonly Clause[], type: CardDef["type"]): boolean {
if (type !== "minion") return true;
// The trigger whose effects are being listed: none, one with no effect yet, or one with some.
let open: "none" | "empty" | "filled" = "none";
for (const clause of text) {
if (clause.kind === "effect") {
if (open === "none") return false;
open = "filled";
} else {
if (open === "empty") return false;
open = clause.kind === "trigger" ? "empty" : "none";
}
}
return open !== "empty";
}
/** Every card the game can create, looked up by name. */
export class CardLibrary {
private readonly cards = new Map<string, Card>();
constructor(defs: CardDef[]) {
for (const def of defs) this.register(def);
}
/** Add a card, or replace the card with the same name. It joins the Assemble pool if it is a minion with text. */
register(def: CardDef): void {
this.cards.set(def.name, compile(def, true));
}
/** Add a card made by Assemble. It never joins the pool. */
registerAssembled(def: CardDef): void {
this.cards.set(def.name, compile(def, false));
}
has(name: string): boolean {
return this.cards.has(name);
}
get(name: string): Card {
const card = this.cards.get(name);
if (!card) throw new Error(`unknown card "${name}": add its file to cards/`);
return card;
}
/** The Assemble pool: every card with `inPool`, sorted by name. */
pool(): Card[] {
return [...this.cards.values()]
.filter((c) => c.inPool)
.sort((a, b) => (a.def.name < b.def.name ? -1 : a.def.name > b.def.name ? 1 : 0));
}
}
@@ -0,0 +1,29 @@
// The Death Creation Step: the one place minions leave play.
//
// Damage and destroy effects only mark a minion (0 or less Health, or pending
// destroy). It stays in play, and keeps triggering, until the outermost Phase
// ends. Then this step removes every marked minion at once.
import type { Event } from "./event";
import type { EntityId, World } from "./world";
/**
* Remove every mortally wounded or pending-destroy minion from play, simultaneously.
* Returns their Death Events in order of play; empty if nothing died.
*/
export function deathCreationStep(world: World): Event[] {
const dying = world.minionsInPlay().filter((id) => world.isDoomed(id));
// Every neighbour is read before anything moves: removal is simultaneous. Each Death
// Event remembers where its minion stood as a handle, the minion that stays to its
// right, not as a board index: an index goes stale as soon as a minion further left
// leaves play in the same step (Module 1: hold handles, not indexes).
const deaths = dying.map((id) => ({ type: "Destroyed" as const, id, before: survivorToTheRight(world, id) }));
for (const { id, before } of deaths) world.kill(id, before);
return deaths;
}
/** The nearest minion to the right of `id` on its board that is not leaving play in this step, if any. */
function survivorToTheRight(world: World, id: EntityId): EntityId | undefined {
const board = world.zone(world.ownerOf(id), "board");
return board.slice(board.indexOf(id) + 1).find((other) => !world.isDoomed(other));
}
@@ -0,0 +1,76 @@
// The interpreter: turns one effect, as data, into engine actions.
//
// This is the only place that knows what an effect *means*. A card's text is a
// small tree (clauses holding effects holding targets); the engine walks it. The
// effect vocabulary is closed: a card can only do what this switch can do, which is
// what makes a data card safe to load.
import type { Effect, Target } from "./card";
import type { Action, Ctx } from "./queue";
import type { EntityId } from "./world";
const times = <T>(count: number, make: () => T): T[] => Array.from({ length: count }, make);
/** The actions one effect performs, in order, given where and for whom it runs. */
export function actions(effect: Effect, ctx: Ctx): Action[] {
const player = ctx.owner;
switch (effect.type) {
case "dealDamage":
return [{ type: "damage", targets: targets(effect.target, ctx), amount: effect.amount }];
case "draw":
return times(effect.count, () => ({ type: "draw", player }));
case "summon": {
// Several tokens stand left to right in summon order: each goes right of the one before.
const at = summonPosition(ctx);
return Array.from({ length: effect.count }, (_, i) => ({
type: "summon" as const,
card: effect.card,
player,
at: at === undefined ? undefined : at + i,
}));
}
case "gainArmor":
return [{ type: "gainArmor", player, amount: effect.amount }];
case "gainAttack":
return targets(effect.target, ctx).map((id) => ({ type: "gainAttack", id, amount: effect.amount }));
case "destroy":
return [{ type: "destroy", targets: targets(effect.target, ctx) }];
case "assemble":
return [{ type: "assemble", player, filter: effect.filter }];
case "buffHand":
return ctx.world
.minionsInHand(player)
.map((id) => ({ type: "buff", id, attack: effect.attack, health: effect.health }));
case "spawnText":
return [{ type: "spawnText", source: ctx.minion(), word: effect.word }];
}
}
/** The entities a target names, from where the effect runs. */
function targets(target: Target, ctx: Ctx): EntityId[] {
const { world } = ctx;
const others = (ids: EntityId[]) => ids.filter((id) => id !== ctx.source);
switch (target) {
case "self":
return [ctx.minion()];
case "allMinions":
return world.minionsInPlay();
case "allOtherMinions":
return others(world.minionsInPlay());
case "allOtherCharacters":
return others([world.hero(0), world.hero(1), ...world.minionsInPlay()]);
case "enemyHero":
return [ctx.enemyHero()];
case "chosen":
return ctx.target === undefined ? [] : [ctx.target];
}
}
/**
* Where a summon goes: to the right of a source still on the board; where a dead
* source stood (left of the minion its Death Event remembers); at the right end for a spell.
*/
function summonPosition(ctx: Ctx): number | undefined {
const index = ctx.source === undefined ? undefined : ctx.world.boardIndex(ctx.source);
return index === undefined ? ctx.position : index + 1;
}
@@ -0,0 +1,38 @@
// What happened, in order. The engine records an event each time it changes the game.
// The world also keeps the nesting depth each event was recorded at (`World.logDepths`).
import type { TriggerOn } from "./card";
import type { EntityId, PlayerId } from "./world";
export type Event =
/**
* `id` entered the board. `by` is the minion whose triggered ability put it there (the
* trigger resolving when it entered), or undefined for a spell, a play or a summon from
* outside a Sequence. Module 10: SpellGrid puts it on `by`'s grid tile. The course's log
* doesn't show `by`.
*/
| { type: "Summoned"; id: EntityId; by: EntityId | undefined }
| { type: "Damaged"; target: EntityId; amount: number }
| { type: "Triggered"; source: EntityId; on: TriggerOn }
| { type: "Drew"; player: PlayerId; id: EntityId }
| { type: "Fatigue"; player: PlayerId }
| { type: "ArmorGained"; player: PlayerId; amount: number }
/** Armor absorbed `amount` of a Damage to `player`'s hero. The `Damaged` that follows shows the full amount. */
| { type: "ArmorLost"; player: PlayerId; amount: number }
| { type: "StatsChanged"; id: EntityId; attack: number; health: number }
| { type: "SpellCast"; card: string }
/**
* A Death Event: the Death Creation Step moved `id` from play to the graveyard.
* `before` is the nearest minion to its right that did not die in the same step, if any:
* its Deathrattle summons immediately left of that minion, or at the right end. It is a
* handle because a board index goes stale when a minion further left dies in the same step.
*/
| { type: "Destroyed"; id: EntityId; before: EntityId | undefined }
| { type: "DeathPhase"; round: number }
/** A minion was played from hand (it has just entered the board). Its Battlecry answers this. */
| { type: "Played"; id: EntityId }
/**
* Module 10: `source`'s effect asked for `word` to be written on its grid tile. Part I
* has no grid: it only records the request. SpellGrid reads it when the Sequence ends.
*/
| { type: "TextSpawned"; source: EntityId; word: string };
@@ -0,0 +1,68 @@
// Engine core: generic rules machinery. Never name a specific card here.
//
// Module 1: every card in the game is an entity addressed by an id handle.
// The engine knows which zone each entity is in, who owns it, and when it
// entered play.
//
// Module 2: heroes, Health, Attack and Armor; triggered abilities; and the
// action queue that resolves a Sequence depth-first, recording an event log.
//
// Module 3: deferred death. Damage and destroy effects only mark minions; the
// Death Creation Step removes them when the outermost Phase ends, and Death
// Phases resolve their Deathrattles until no new deaths occur.
//
// Module 4: auras as card data, Aura Updates at defined points, and one
// routine that decides how many times a queued trigger activates. Playing a
// minion from hand, and Battlecries. A second interceptor, replacement: the
// steps that rewrite a Damage before it lands (a redirect, then Armor).
//
// Module 5: behaviour as data. Cards are definitions in the course's card format,
// loaded from JSON; one interpreter turns their effects into actions. Assemble
// builds a new card by swapping half of one card's text for half of another's, and
// offers only results that are well-formed. A Sequence is bounded: a trigger never
// nests inside itself, and more than `DEATH_PHASE_LIMIT` Death Phases is an error.
//
// Part II lives in `puzzle/`: a second engine, for the rule-writing puzzle game,
// beside the first.
// Module 6: the grid, the lexer and the parser.
// Module 7: `IS NOT`, and the resolution pass that turns parsed rules into rules in force.
// Module 8: movement, planned in full and then committed; push chains.
// Module 9: the turn, as a list of phases; hazards; winning.
// Module 11: conditions (`X ON Z IS P`), checked for each object when a rule is used.
//
// Part III lives in `spellgrid.ts`, the only file that knows both engines.
// Module 10: SpellGrid. Part I gains auras from outside the cards (`World.setRuleAuras`),
// a cap on activations, the `spawnText` effect (recorded as a `TextSpawned` event) and
// the end-of-Phase steps as a list (`END_OF_PHASE`); Part II gains a `Destroy` hook.
export * as puzzle from "./puzzle";
export { AssembleError, assemble, registerAssembled, type AssembleFailure, type Half } from "./assemble";
export {
CardLibrary,
triggerLabel,
type Aura,
type Card,
type CardDef,
type Clause,
type Effect,
type PoolFilter,
type Target,
type Trigger,
type TriggerOn,
wellFormed,
} from "./card";
export type { Event } from "./event";
export { auraUpdateOther } from "./aura";
export {
Ctx,
DAMAGE_STEPS,
DEATH_PHASE_LIMIT,
END_OF_PHASE,
runSequence,
summon,
type Action,
type DamageStep,
type EndOfPhaseStep,
} from "./queue";
export { ruleAuras, SpellGrid, type SeamWords } from "./spellgrid";
export { HERO_HEALTH, World, type Chooser, type EntityId, type PlayerId, type Zone } from "./world";
@@ -0,0 +1,21 @@
// Conditions (Module 11): `X ON Z IS P` gives an X the property P only while it is on a Z.
//
// The parser reads a condition into the rule (`Rule.conditions`); the engine never
// decides it there. A condition is evaluated for one object, against the board as
// it is, every time a rule is about to be used: by `Level.hasProperty` (and so by
// movement, the hazards and the win check), and by conversion. What a condition
// word means is a `ConditionCheck` that content provides, registered by word, as
// movement responses and hazards are. The engine knows only that a word with role
// `"condition"` may stand between the subjects and `IS`.
import type { ObjectId } from "./grid";
import type { Level } from "./level";
/**
* Content's meaning for one condition word: does the condition, with these nouns
* (in the order they are written), hold for object `id` on the board as it is now?
*/
export type ConditionCheck = (level: Level, id: ObjectId, nouns: readonly string[]) => boolean;
/** Content's condition checks, by condition word: `{ ON: on }`. */
export type ConditionChecks = Readonly<Record<string, ConditionCheck>>;
@@ -0,0 +1,145 @@
// The grid: objects on tiles. `x` grows right, `y` grows down, and any number of
// objects may share a tile.
export type ObjectId = number;
/** A level in the course's level format (`course/schema/level.schema.json`). */
export interface LevelDef {
width: number;
height: number;
objects: LevelObject[];
/** Module 10, read only by SpellGrid: minions in play, each on its tile. */
cards?: LevelCard[];
/** Module 10, read only by SpellGrid: the cards in player 0's hand. */
hand?: string[];
}
/** A minion card on the grid (Module 10). */
export interface LevelCard {
card: string;
x: number;
y: number;
}
export interface LevelObject {
kind: string;
x: number;
y: number;
}
/**
* What an object is: a thing named by a noun (kind `"<noun>"`, the noun in lower case),
* or a piece of text carrying one word (kind `"text:<WORD>"`).
*/
export type Kind = { type: "thing"; noun: string } | { type: "text"; word: string };
const TEXT_PREFIX = "text:";
export const Kind = {
parse(kind: string): Kind {
return kind.startsWith(TEXT_PREFIX)
? { type: "text", word: kind.slice(TEXT_PREFIX.length).toUpperCase() }
: { type: "thing", noun: kind.toUpperCase() };
},
/** The kind as it is placed: `"<noun>"` in lower case, or `"text:<WORD>"`. */
format(kind: Kind): string {
return kind.type === "text" ? TEXT_PREFIX + kind.word : kind.noun.toLowerCase();
},
};
export interface Obj {
id: ObjectId;
kind: Kind;
x: number;
y: number;
}
/** What `Board.restore` puts back (Module 9, undo): every object, oldest first, and the id counter. */
export interface BoardState {
objects: Obj[];
nextId: ObjectId;
}
export class Board {
private readonly objectsById = new Map<ObjectId, Obj>();
private nextId = 0;
constructor(
readonly width: number,
readonly height: number,
) {}
inBounds(x: number, y: number): boolean {
return x >= 0 && x < this.width && y >= 0 && y < this.height;
}
/** Add an object; ids start at 1 and are never reused. Throws off the board. */
place(kind: Kind, x: number, y: number): ObjectId {
if (!this.inBounds(x, y)) throw new Error(`(${x}, ${y}) is outside the ${this.width}x${this.height} level`);
const id = ++this.nextId;
this.objectsById.set(id, { id, kind, x, y });
return id;
}
/**
* An id that no object will ever have (Module 10): SpellGrid gives one to each
* card that isn't on the grid, so cards and objects share one id space.
*/
reserveId(): ObjectId {
return ++this.nextId;
}
/** The object `id`. Throws if there is none. */
get(id: ObjectId): Obj {
const o = this.objectsById.get(id);
if (o === undefined) throw new Error(`no object ${id}`);
return o;
}
/** The object `id`, or `undefined` once it has been removed. Throws on an id this board never issued. */
find(id: ObjectId): Obj | undefined {
if (!(id >= 1 && id <= this.nextId)) throw new Error(`no object ${id}`);
return this.objectsById.get(id);
}
/** Take object `id` off the board for good. */
remove(id: ObjectId): void {
this.objectsById.delete(id);
}
/** Move object `id` by `[dx, dy]`. Throws if it would leave the board. */
shift(id: ObjectId, [dx, dy]: readonly [number, number]): void {
const o = this.get(id);
const [x, y] = [o.x + dx, o.y + dy];
if (!this.inBounds(x, y)) throw new Error(`(${x}, ${y}) is outside the ${this.width}x${this.height} level`);
o.x = x;
o.y = y;
}
/** Make object `id` a `kind`, where it stands. Throws if there is no such object. */
setKind(id: ObjectId, kind: Kind): void {
this.get(id).kind = kind;
}
/** The objects on tile (x, y), oldest first. */
at(x: number, y: number): Obj[] {
return this.objects().filter((o) => o.x === x && o.y === y);
}
/** Every object, oldest first. */
objects(): Obj[] {
return [...this.objectsById.values()];
}
/** A copy of everything the board holds (Module 9, undo). */
snapshot(): BoardState {
return { objects: this.objects().map((o) => ({ ...o })), nextId: this.nextId };
}
/** Put the board back as `state` recorded it: the same objects, ids, ages and id counter. */
restore(state: BoardState): void {
this.objectsById.clear();
for (const o of state.objects) this.objectsById.set(o.id, { ...o });
this.nextId = state.nextId;
}
}
@@ -0,0 +1,47 @@
// Part II: the rule-writing puzzle game. Objects have no behaviour of their own; every rule is
// compiled from the text on the board, and recompiled whenever the board changes.
//
// Module 6: the grid, the lexer (text objects to words with grammatical roles)
// and the parser (words to rules). The engine knows the operators `IS`, `AND`,
// `NOT`, the meta-noun `TEXT` and the base rule `TEXT IS PUSH`. Every other word,
// nouns included, is data handed in by content.
//
// Module 7: `IS NOT` in the parser, and the resolution pass (negation, then the
// identity lock) that turns the parsed rules into the rules in force. Objects are
// asked about through `Level.hasProperty`, which knows which nouns name them.
//
// Module 8: movement. `Level.moveYou` plans each `YOU` object's move in full
// (`Planner`), then commits it, then reparses. What a property does when something
// moves is a `Response` that content provides; the engine runs them in order.
//
// Module 9: the turn. `Level.step` runs the phases of `TURN` in order: movement,
// parse, conversion, parse, hazards, win, parse. What a hazard property destroys
// is a `Hazard` that content provides; the engine knows `YOU`, `WIN` and `FLOAT`.
//
// Module 10: `Level.step` and `Level.moveYou` take a `Destroy`, which decides what
// happens to the objects they destroy; `Board.reserveId` hands out an id no object has.
// The puzzle engine still knows nothing of cards: SpellGrid (`../spellgrid.ts`) uses these.
//
// Module 11: conditions. A word with role `"condition"` may stand between a rule's
// subjects and `IS` (`X ON Z IS P`); the rule carries it (`Rule.conditions`).
// What it means is a `ConditionCheck` content registers by word. `Level` asks the
// checks every time it uses a rule, against the board as it is then.
export type { ConditionCheck, ConditionChecks } from "./condition";
export { Board, Kind, type LevelCard, type LevelDef, type LevelObject, type Obj, type ObjectId } from "./grid";
export { FLOAT, Level, removeAll, WIN, YOU, type Destroy } from "./level";
export { delta, Planner, type Dir, type MoveContext, type Plan, type Pos, type Response, type Step } from "./movement";
export {
BASE_RULES,
isConversion,
isIdentity,
negates,
parse,
ruleText,
sameConditions,
type Condition,
type Rule,
} from "./parse";
export { resolve } from "./resolve";
export { TURN, type Hazard, type Phase } from "./turn";
export { lex, Lexicon, type Role, type Selects, type Token, type WordDef } from "./word";
@@ -0,0 +1,316 @@
// A level in play: the board, the words content provides, and the rules compiled
// from them. Every change to the board recompiles the rules from scratch: parse,
// then resolve. Every question about what an object is goes through `hasProperty`.
//
// A move (Module 8) is planned for each `YOU` object, then committed; the rules are
// reparsed once, after all of them.
//
// A turn (Module 9) runs the phases of `TURN` in order: movement, parse,
// conversion, parse, content's hazards, the win check, parse.
//
// Undo (Module 9): before each move or turn the level saves the board (every
// object's id, kind and position, oldest first, and the id counter) and `won`.
// The rules are not saved: undo restores the board and reparses.
//
// Module 10: what happens to the objects a turn destroys is a `Destroy` the caller
// may pass in. The default removes them. SpellGrid passes one that sends minions to
// Part I's Death Creation Step instead. SpellGrid doesn't offer undo: Part I has
// state the level's snapshot doesn't cover.
//
// Module 11: a rule may carry conditions (`X ON Z IS P`). They are not
// decided when the rules are compiled. Every question about an object (`hasProperty`,
// conversion) asks content's `ConditionCheck`s against the board as it is at that
// moment, and uses only the rules whose conditions hold for that object.
import type { ConditionChecks } from "./condition";
import { Board, Kind, type BoardState, type LevelDef, type ObjectId } from "./grid";
import { delta, Planner, type Dir, type Plan, type Response } from "./movement";
import { isConversion, isIdentity, parse, type Rule } from "./parse";
import { resolve } from "./resolve";
import { TURN, type Hazard } from "./turn";
import { Lexicon, type WordDef } from "./word";
/** The property that makes an object move when the player moves. */
export const YOU = "YOU";
/** The property that wins the level when a `YOU` object is on it, or is it. */
export const WIN = "WIN";
/**
* The property that puts an object on the other float level: only objects on
* the same level share a tile for hazards and for winning.
*/
export const FLOAT = "FLOAT";
/**
* What to do with objects a move or a turn destroys (Module 10). It may change the
* board, and reparse, before it returns. An id may appear more than once.
*/
export type Destroy = (level: Level, ids: ObjectId[]) => void;
/** The default `Destroy`: take every object off the board. */
export const removeAll: Destroy = (level, ids) => {
for (const id of ids) level.board.remove(id);
};
export class Level {
readonly board: Board;
private readonly lexicon: Lexicon;
private compiled: Rule[] = [];
private active: Rule[] = [];
private hasWon = false;
/** What `undo` puts back: one entry per move or turn, the latest last (Module 9). */
private history: { board: BoardState; won: boolean }[] = [];
/**
* A level from its definition, the words content provides, content's movement
* responses and hazards, each in the order they are to run, and content's
* condition checks, by word. Throws if a word with role `"condition"` has no check.
*/
constructor(
def: LevelDef,
words: readonly WordDef[],
private readonly moveResponses: readonly Response[],
private readonly hazards: readonly Hazard[],
private readonly conditions: ConditionChecks,
) {
for (const w of words) {
if (w.role === "condition" && !Object.hasOwn(conditions, w.name.toUpperCase())) {
throw new Error(`the condition word ${w.name} has no condition check`);
}
}
this.board = new Board(def.width, def.height);
this.lexicon = new Lexicon(words);
for (const o of def.objects) this.board.place(Kind.parse(o.kind), o.x, o.y);
this.reparse();
}
place(kind: string, x: number, y: number): ObjectId {
const id = this.board.place(Kind.parse(kind), x, y);
this.history = [];
this.reparse();
return id;
}
/** The movement responses, in the order they run. */
responses(): readonly Response[] {
return this.moveResponses;
}
/** Every rule the text spells, as parsed: negated rules and duplicates included. */
rules(): readonly Rule[] {
return this.compiled;
}
/** The rules in force after the resolution pass. */
activeRules(): readonly Rule[] {
return this.active;
}
/**
* Does object `id` have `property`? Yes if an active rule that applies to it
* (`appliesTo`) gives it the property and no active negated rule that applies to
* it takes it away. A destroyed object has no properties.
*/
hasProperty(id: ObjectId, property: string): boolean {
if (this.board.find(id) === undefined) return false;
const aboutIt = this.active.filter((r) => r.complement === property && this.appliesTo(r, id));
return aboutIt.some((r) => !r.negated) && !aboutIt.some((r) => r.negated);
}
/** Does the noun `word`, as the subject of a rule, name object `id`? */
names(word: string, id: ObjectId): boolean {
return this.lexicon.names(word, this.board.get(id).kind);
}
/**
* Does `rule` apply to object `id` now? Its subject names the object, and each of
* its conditions holds for the object on the board as it is (Module 11).
*/
private appliesTo(rule: Rule, id: ObjectId): boolean {
return (
this.names(rule.subject, id) && rule.conditions.every((c) => this.conditions[c.word](this, id, c.nouns))
);
}
/**
* The other objects on object `id`'s tile that are on its float level (both
* `FLOAT` or both not), oldest first. Hazards and the win check ask this.
*/
sharing(id: ObjectId): ObjectId[] {
const o = this.board.get(id);
const floats = this.hasProperty(id, FLOAT);
return this.board
.at(o.x, o.y)
.filter((other) => other.id !== id && this.hasProperty(other.id, FLOAT) === floats)
.map((other) => other.id);
}
/** Has a turn's win check found a `YOU` object on a `WIN` one? Stays true. */
won(): boolean {
return this.hasWon;
}
/** Is there no `YOU` object left? */
lost(): boolean {
return !this.board.objects().some((o) => this.hasProperty(o.id, YOU));
}
/**
* Move every `YOU` object one tile in `dir`, then reparse (Module 8). Movement
* only: no conversion, hazard or win.
*/
moveYou(dir: Dir, destroy: Destroy = removeAll): void {
this.save();
this.movement(dir, destroy);
this.reparse();
}
/** Play one turn: the phases of `TURN`, in order. Destroyed objects go to `destroy`. */
step(dir: Dir, destroy: Destroy = removeAll): void {
this.save();
for (const phase of TURN) {
switch (phase) {
case "movement":
this.movement(dir, destroy);
break;
case "parse":
this.reparse();
break;
case "conversion":
this.convert(destroy);
break;
case "hazards":
this.runHazards(destroy);
break;
case "win":
this.checkWin();
break;
}
}
}
/**
* Put the level back as it was before the last `moveYou` or `step` (Module 9),
* and reparse. False, and no change, if there is none; `place` forgets them all.
*/
undo(): boolean {
const last = this.history.pop();
if (last === undefined) return false;
this.board.restore(last.board);
this.hasWon = last.won;
this.reparse();
return true;
}
private save(): void {
this.history.push({ board: this.board.snapshot(), won: this.hasWon });
}
/**
* The `YOU` objects move one at a time, the one furthest ahead in `dir` first
* (oldest first on a tie); each one's move is planned in full, then committed,
* before the next is planned. Every move uses the rules in force before the first.
*
* An object steps at most once in a movement phase. A `YOU` object that an
* earlier move already carried along (pushed, or moved by any other response)
* doesn't start a move of its own, and no later move can make it step again.
*/
private movement(dir: Dir, destroy: Destroy): void {
const [dx, dy] = delta(dir);
const ahead = (o: { x: number; y: number }) => o.x * dx + o.y * dy;
const you = this.board
.objects()
.filter((o) => this.hasProperty(o.id, YOU))
.sort((a, b) => ahead(b) - ahead(a) || a.id - b.id)
.map((o) => o.id);
const stepped = new Set<ObjectId>();
for (const id of you) {
if (this.board.find(id) === undefined) continue; // destroyed by an earlier move
if (stepped.has(id)) continue; // already moved, by an earlier move
const planner = new Planner(this, dir, stepped);
if (!planner.tryMove(id)) continue;
const plan = planner.result();
this.commit(plan, dir, destroy);
for (const moved of plan.moving) if (!plan.destroyed.has(moved)) stepped.add(moved);
}
}
/** Destroy, then move what is still on the board (a `Destroy` may remove more than it was given). */
private commit(plan: Plan, dir: Dir, destroy: Destroy): void {
if (plan.destroyed.size > 0) destroy(this, [...plan.destroyed]);
for (const id of plan.moving) {
if (!plan.destroyed.has(id) && this.board.find(id) !== undefined) this.board.shift(id, delta(dir));
}
}
/**
* Every object named by a conversion `X IS Y` (`Y` a noun) becomes a `Y` where
* it stands, keeping its id; with several `Y`s it becomes the first, and each
* other `Y` is a new object on its tile. An `X IS X` naming the object blocks
* this. An object named by `X IS NOT X` is deleted, and still makes its `Y`s,
* as new objects: the negation cancelled `X IS X`.
*
* Negation is settled per object, as in `hasProperty`: an object doesn't become
* a `Y` if any active `... IS NOT Y` names it, whatever noun gave it the
* conversion (`ALL IS Y` with `X IS NOT Y` leaves every X an X).
*
* Every object is decided from the board as movement left it, before any
* changes, so `X IS Y` and `Y IS X` swap the Xs and the Ys. Only the rules that
* apply to the object are used: their conditions are checked on that board too.
*/
private convert(destroy: Destroy): void {
const changes: { id: ObjectId; x: number; y: number; deleted: boolean; into: Kind[] }[] = [];
for (const o of this.board.objects()) {
const aboutIt = this.active.filter((r) => this.appliesTo(r, o.id));
const deleted = aboutIt.some((r) => r.negated && isIdentity(r));
const locked = aboutIt.some((r) => !r.negated && isIdentity(r));
const into: Kind[] = [];
if (!locked) {
for (const r of aboutIt) {
if (r.negated || !isConversion(r)) continue;
if (aboutIt.some((n) => n.negated && n.complement === r.complement)) continue;
const kind = this.lexicon.thingNamed(r.complement);
if (kind === undefined) continue;
const same = (k: Kind) => Kind.format(k) === Kind.format(kind);
if (!same(o.kind) && !into.some(same)) into.push(kind);
}
}
if (deleted || into.length > 0) changes.push({ id: o.id, x: o.x, y: o.y, deleted, into });
}
for (const { id, x, y, deleted, into } of changes) {
if (!deleted) this.board.setKind(id, into[0]);
for (const kind of deleted ? into : into.slice(1)) this.board.place(kind, x, y);
}
const deleted = changes.filter((c) => c.deleted).map((c) => c.id);
if (deleted.length > 0) destroy(this, deleted);
}
/**
* Run content's hazards in order. Each one's victims are destroyed together
* before the next runs; no reparse until the turn's last phase.
*/
private runHazards(destroy: Destroy): void {
for (const hazard of this.hazards) {
const victims = hazard(this);
if (victims.length > 0) destroy(this, victims);
}
}
/**
* The level is won if any `YOU` object is `WIN`, or shares its tile with a
* `WIN` object on its float level.
*/
private checkWin(): void {
const wins = (id: ObjectId) => this.hasProperty(id, WIN);
const won = this.board
.objects()
.filter((o) => this.hasProperty(o.id, YOU))
.some((o) => wins(o.id) || this.sharing(o.id).some(wins));
this.hasWon ||= won;
}
/** Recompile the rules from the board: parse, then resolve. */
reparse(): void {
this.compiled = parse(this.board, this.lexicon);
this.active = resolve(this.compiled);
}
}
@@ -0,0 +1,130 @@
// Movement: validate, then commit.
//
// A move is planned before anything on the board changes. The `Planner` asks
// whether one object can step one tile; to answer, it runs every movement
// response content provides for that step, and a response may ask the planner
// to move other objects too (a push), or to destroy objects. If any response
// says no, the planner forgets everything that step added to the plan, so a
// failed push leaves no trace. Only a plan that succeeded is committed.
//
// The engine knows the board, its edge, and the order of the steps. What each
// property does when something moves (`PUSH`, `STOP`, `OPEN`/`SHUT`, ...) is a
// response in content, run in the order content lists them.
import type { Board, ObjectId } from "./grid";
/** One of the four directions a move can take. */
export type Dir = "up" | "down" | "left" | "right";
export type Pos = readonly [number, number];
/** One tile in direction `dir`, as `[dx, dy]`; `y` grows down. */
export function delta(dir: Dir): Pos {
switch (dir) {
case "up":
return [0, -1];
case "down":
return [0, 1];
case "left":
return [-1, 0];
case "right":
return [1, 0];
}
}
/** One object stepping one tile, from `from` to `to`, in direction `dir`. */
export interface Step {
mover: ObjectId;
from: Pos;
to: Pos;
dir: Dir;
}
/**
* A movement response: content's rule for what happens when an object steps.
* It returns `false` to block the step. It may plan other moves or destroy
* objects through the planner.
*/
export type Response = (p: Planner, s: Step) => boolean;
/**
* What a move will do: the objects that step one tile, and the objects destroyed.
* A destroyed object doesn't step, even if it was pushed.
*/
export interface Plan {
moving: Set<ObjectId>;
destroyed: Set<ObjectId>;
}
/** What the planner reads: the board, the properties in force, the responses. */
export interface MoveContext {
readonly board: Board;
hasProperty(id: ObjectId, property: string): boolean;
responses(): readonly Response[];
}
/**
* Plans one move on a board it only reads. `stepped` holds the objects that have
* already stepped in this movement phase, in an earlier committed plan: an object
* steps at most once per phase, so none of them steps again.
*/
export class Planner {
private plan: Plan = { moving: new Set(), destroyed: new Set() };
constructor(
private readonly level: MoveContext,
private readonly dir: Dir,
private readonly stepped: ReadonlySet<ObjectId>,
) {}
/** Does object `id` have `property`, under the rules in force before the move? */
has(id: ObjectId, property: string): boolean {
return this.level.hasProperty(id, property);
}
/**
* The objects that will still be on tile `at` when the plan so far is
* committed: those there now, less those planned to move away or be destroyed.
* Oldest first.
*/
stayingAt([x, y]: Pos): ObjectId[] {
return this.level.board
.at(x, y)
.map((o) => o.id)
.filter((id) => !this.plan.moving.has(id) && !this.plan.destroyed.has(id));
}
/** Plan object `id`'s destruction. */
destroy(id: ObjectId): void {
this.plan.destroyed.add(id);
}
/**
* Plan object `id` stepping one tile in the move's direction, with whatever
* the responses add. Returns whether it can. If it can't, the plan is left as
* it was before the call. An object already planned to move, or to be
* destroyed, is out of the way: the answer is yes, and nothing is added. An
* object that has already stepped in this movement phase doesn't step again, and
* doesn't hold the mover back either: the answer is yes, and nothing is added, so
* objects that began the phase stacked on one tile end it stacked.
*/
tryMove(id: ObjectId): boolean {
if (this.plan.moving.has(id) || this.plan.destroyed.has(id)) return true;
if (this.stepped.has(id)) return true;
const o = this.level.board.get(id);
const [dx, dy] = delta(this.dir);
const to: Pos = [o.x + dx, o.y + dy];
// The edge of the level blocks every step (the course's choice; see the lesson).
if (!this.level.board.inBounds(...to)) return false;
const before: Plan = { moving: new Set(this.plan.moving), destroyed: new Set(this.plan.destroyed) };
this.plan.moving.add(id);
const step: Step = { mover: id, from: [o.x, o.y], to, dir: this.dir };
const allowed = this.level.responses().every((respond) => respond(this, step));
if (!allowed) this.plan = before;
return allowed;
}
result(): Plan {
return this.plan;
}
}
@@ -0,0 +1,227 @@
// The parser: words on the grid become rules.
//
// Grammar (Module 7): `NOUN [AND NOUN]* IS [NOT] (NOUN|PROPERTY) [AND [NOT] (NOUN|PROPERTY)]*`,
// read left to right in a row and top to bottom in a column. The parser starts at
// every `IS`, collects the subjects before it and the complements after it, and
// emits one rule per pair: AND distributes over both sides. A tile holding several
// words contributes each of them, so stacked text yields every possible rule.
//
// `NOT` before a complement negates that one complement. `NOT` before a subject is
// not in the grammar: the run of subjects stops there and that subject is dropped.
//
// Module 11: conditions. `NOUN [AND NOUN]* [CONDITION NOUN [AND NOUN]*] IS ...`: a word
// whose role is `"condition"` may stand between the subjects and `IS`, followed by
// the nouns it is about. The condition belongs to every subject of the sentence,
// and AND among its nouns means all of them. A run of nouns before `IS` is the
// condition's when a condition word stands before it, and then it is not a run of
// subjects. Stacked text still yields every possible rule: a stack of nouns after
// the condition word gives one rule per noun, as a stack after `IS` does.
//
// The parser only reads; which rules win is the resolution pass (`resolve`), and
// whether a condition holds is decided for each object when someone asks (`Level`).
import type { Board, ObjectId } from "./grid";
import { AND, IS, lex, NOT, TEXT, tileKey, type Lexicon, type Role, type Token } from "./word";
/**
* A condition on a rule (Module 11): the condition word and the nouns after it, in
* the order they are written. `X ON Y AND Z IS P` has `{ word: "ON",
* nouns: ["Y", "Z"] }`.
*/
export interface Condition {
word: string;
nouns: string[];
}
/**
* One rule: `subject IS complement`, or `subject IS NOT complement` when `negated`,
* applying to an object only while each of its `conditions` holds for that object
* (none for an unconditional rule). `text` is the text objects that spell it
* (subject, condition word and nouns if any, `IS`, `NOT` if any, complement), or
* empty for a base rule.
*/
export interface Rule {
subject: string;
complement: string;
complementRole: Role;
negated: boolean;
conditions: Condition[];
text: ObjectId[];
}
/** The normalised form: `"TEXT IS PUSH"`, `"TEXT IS NOT PUSH"`, `"X ON Y AND Z IS P"`. */
export const ruleText = (r: Rule): string => {
const conditions = r.conditions.map((c) => ` ${c.word} ${c.nouns.join(` ${AND} `)}`).join("");
return `${r.subject}${conditions} ${IS}${r.negated ? ` ${NOT}` : ""} ${r.complement}`;
};
/** The same conditions: the same words, each with the same nouns in any order. */
export const sameConditions = (a: Rule, b: Rule): boolean => {
const key = (r: Rule) => JSON.stringify(r.conditions.map((c) => [c.word, [...c.nouns].sort()]));
return key(a) === key(b);
};
/** `X IS Y` where `Y` is a noun: every X is to become a Y. */
export const isConversion = (r: Rule): boolean => r.complementRole === "noun";
/** `X IS X` (or `X IS NOT X`). */
export const isIdentity = (r: Rule): boolean => r.subject === r.complement;
/**
* Same subject, complement and conditions, opposite sign: `X IS NOT Y` negates
* `X IS Y`, and `X ON Z IS NOT Y` negates `X ON Z IS Y` (Module 11), but not `X IS Y`.
*/
export const negates = (n: Rule, r: Rule): boolean =>
n.negated && !r.negated && n.subject === r.subject && n.complement === r.complement && sameConditions(n, r);
/** Rules in force with no text: the game's base rules (`[subject, property]`). */
export const BASE_RULES: readonly (readonly [string, string])[] = [[TEXT, "PUSH"]];
type Tiles = Map<string, Token[]>;
type Pos = readonly [number, number];
/** A word in a run, with the `NOT` in front of it if there is one. */
interface Word {
not?: Token;
token: Token;
}
/**
* Every rule the board spells, base rules first, negated rules included. A rule
* spelled by two different sets of text objects appears twice.
*/
export function parse(board: Board, lexicon: Lexicon): Rule[] {
const tiles = lex(board, lexicon);
const rules: Rule[] = BASE_RULES.map(([subject, property]) => ({
subject,
complement: property,
complementRole: "property",
negated: false,
conditions: [],
text: [],
}));
for (const o of board.objects()) {
if (o.kind.type !== "text" || o.kind.word !== IS) continue;
for (const [dx, dy] of [
[1, 0],
[0, 1],
] as const) {
const cs = complements(tiles, [o.x + dx, o.y + dy], [dx, dy]);
for (const { subjects, conditions, text } of beforeIs(tiles, [o.x - dx, o.y - dy], [-dx, -dy])) {
for (const s of subjects) {
for (const c of cs) {
rules.push({
subject: s.word,
complement: c.token.word,
complementRole: c.token.role,
negated: c.not !== undefined,
conditions,
text: [s.id, ...text, o.id, ...(c.not ? [c.not.id] : []), c.token.id],
});
}
}
}
}
}
return rules;
}
const wordsAt = (tiles: Tiles, [x, y]: Pos): Token[] => tiles.get(tileKey(x, y)) ?? [];
const holds = (tiles: Tiles, p: Pos, word: string): boolean => wordsAt(tiles, p).some((t) => t.word === word);
const ahead = ([x, y]: Pos, [dx, dy]: Pos): Pos => [x + dx, y + dy];
/** What stands before an `IS` (Module 11): the subjects, with the condition on them if there is one. */
interface Reading {
subjects: Token[];
conditions: Condition[];
/** The condition word and its nouns, in reading order (empty without a condition). */
text: ObjectId[];
}
const isCondition = (t: Token) => t.role === "condition";
/**
* Every way to read the words before an `IS`, walking backwards from `at`.
*
* - The run of subjects, without a condition, unless a condition word stands
* before the run: then those nouns are the condition's, not subjects.
* - For every run of nouns that has a condition word before it, the run of
* subjects before that word, with the condition. A stack among the condition's
* nouns gives one reading per noun (`nounRuns`).
*/
function beforeIs(tiles: Tiles, at: Pos, step: Pos): Reading[] {
const readings: Reading[] = [];
const run = subjects(tiles, at, step);
if (!wordsAt(tiles, run.stop).some(isCondition)) readings.push({ subjects: run.tokens, conditions: [], text: [] });
for (const nouns of nounRuns(tiles, at, step)) {
const inOrder = [...nouns.tokens].reverse();
for (const word of wordsAt(tiles, nouns.stop).filter(isCondition)) {
readings.push({
subjects: subjects(tiles, ahead(nouns.stop, step), step).tokens,
conditions: [{ word: word.word, nouns: inOrder.map((t) => t.word) }],
text: [word.id, ...inOrder.map((t) => t.id)],
});
}
}
return readings;
}
/**
* The run of subjects before an `IS`, walking backwards from `at`: the nouns at
* `at`, then, if the tile before them holds an AND, the run beyond it. A `NOT`
* before a noun ends the run without it (a negated subject is not in the grammar).
* `stop` is the tile the walk stopped on, before the run's first word.
*/
function subjects(tiles: Tiles, at: Pos, step: Pos): { tokens: Token[]; stop: Pos } {
const before = ahead(at, step);
if (holds(tiles, before, NOT)) return { tokens: [], stop: before };
const found = wordsAt(tiles, at).filter((t) => t.role === "noun");
if (found.length > 0 && holds(tiles, before, AND)) {
const rest = subjects(tiles, ahead(before, step), step);
if (rest.tokens.length > 0) return { tokens: [...found, ...rest.tokens], stop: rest.stop };
}
return { tokens: found, stop: before };
}
/**
* Every run of nouns walking backwards from `at`, one noun from each tile, joined
* by AND; each with the tile before its first noun (`stop`). Runs that end early,
* at an AND, are included: a condition word stacked on that AND starts a reading.
* A `NOT` before a noun ends every run through it.
*/
function nounRuns(tiles: Tiles, at: Pos, step: Pos): { tokens: Token[]; stop: Pos }[] {
const before = ahead(at, step);
if (holds(tiles, before, NOT)) return [];
const runs: { tokens: Token[]; stop: Pos }[] = [];
for (const noun of wordsAt(tiles, at).filter((t) => t.role === "noun")) {
runs.push({ tokens: [noun], stop: before });
if (!holds(tiles, before, AND)) continue;
for (const rest of nounRuns(tiles, ahead(before, step), step)) {
runs.push({ tokens: [noun, ...rest.tokens], stop: rest.stop });
}
}
return runs;
}
/**
* The run of complements after an `IS`, walking forwards from `at`: the nouns
* and properties at `at`, or, if `at` holds a `NOT`, those on the tile after it,
* negated; then, if the next tile holds an AND, the run beyond it.
*/
function complements(tiles: Tiles, at: Pos, step: Pos): Word[] {
const fits = (t: Token) => t.role === "noun" || t.role === "property";
const found: Word[] = [];
// Plain words here, and negated words one tile on: a stack may hold both.
const branches: [Token | undefined, Pos][] = [[undefined, at]];
for (const not of wordsAt(tiles, at).filter((t) => t.word === NOT)) branches.push([not, ahead(at, step)]);
for (const [not, p] of branches) {
const words = wordsAt(tiles, p)
.filter(fits)
.map((token) => ({ not, token }));
if (words.length === 0) continue;
found.push(...words);
const andAt = ahead(p, step);
if (holds(tiles, andAt, AND)) found.push(...complements(tiles, ahead(andAt, step), step));
}
return found;
}
@@ -0,0 +1,29 @@
// The resolution pass: from every rule the text spells to the rules in force.
//
// The parser reports what is written; this pass decides what wins, over the whole
// rule list at once, so it doesn't matter where on the board each rule is written:
//
// 1. Negation. `X IS NOT Y` cancels every `X IS Y`. The negated rule stays in
// force: it keeps overriding, and `X IS NOT X` is also the record that every X is
// to vanish (the turn removes them, Module 9).
// 2. Identity. A surviving `X IS X` cancels every conversion `X IS Y` (`Y` a noun):
// X can't become anything else. It doesn't stop other nouns becoming X. Because
// negation runs first, `X IS NOT X` cancels `X IS X` and conversions work again.
//
// Base rules are rules like any other, so `TEXT IS NOT PUSH` cancels `TEXT IS PUSH`.
//
// Module 11: both steps pair rules only when their conditions are the same.
// `X ON Z IS NOT P` cancels `X ON Z IS P` here, but not `X IS P`:
// which of those wins depends on where each X stands, so it is settled per
// object when someone asks (`Level.hasProperty`, and conversion), among the rules
// whose conditions hold for that object.
import { isConversion, isIdentity, negates, sameConditions, type Rule } from "./parse";
/** The rules in force, in the order they were parsed. */
export function resolve(rules: readonly Rule[]): Rule[] {
const survivesNegation = rules.filter((r) => !rules.some((n) => negates(n, r)));
const locked = (r: Rule) =>
survivesNegation.some((l) => !l.negated && isIdentity(l) && l.subject === r.subject && sameConditions(l, r));
return survivesNegation.filter((r) => r.negated || !isConversion(r) || isIdentity(r) || !locked(r));
}
@@ -0,0 +1,33 @@
// The turn (Module 9): one move of the player, as a fixed list of phases.
//
// The course's turn is a labelled simplification of the game's (see the lesson):
// movement, parse, conversion, parse, hazards, win, and a last parse so that
// the rules match the board when the turn ends. Each phase reads the rules from
// the parse before it, so a rule that movement breaks is gone before hazards run.
//
// What a hazard property does (`SINK`, `HOT`/`MELT`, `DEFEAT`, ...) is a `Hazard`
// that content provides, as with movement's responses. The engine runs them in
// content's order and owns only the phases, `YOU`, `WIN` and float levels.
import type { ObjectId } from "./grid";
import type { Level } from "./level";
/**
* One phase of the turn:
* - `"movement"`: every `YOU` object moves (Module 8), without the reparse;
* - `"parse"`: the rules are recompiled from the board;
* - `"conversion"`: `X IS Y` conversions, and `X IS NOT X` deletions;
* - `"hazards"`: content's hazards, in content's order;
* - `"win"`: a `YOU` object on a `WIN` one (or `WIN` itself) wins the level.
*/
export type Phase = "movement" | "parse" | "conversion" | "hazards" | "win";
/** The turn, in order. */
export const TURN: readonly Phase[] = ["movement", "parse", "conversion", "parse", "hazards", "win", "parse"];
/**
* A hazard: content's rule for what destroys what at the end of a turn. It reads
* the level and returns the objects to destroy; the engine destroys them all at
* once, before the next hazard runs. An id may appear more than once.
*/
export type Hazard = (level: Level) => ObjectId[];
@@ -0,0 +1,110 @@
// The lexer: text objects become words with a grammatical role.
//
// The engine owns the grammar words (`IS`, `AND`, `NOT`) and the meta-noun `TEXT`.
// Every other word comes from content as data: the engine never names a noun.
// A noun also says which objects it names (`Selects`), so a meta-noun that names
// a whole group of objects can be data too.
//
// Module 11: a fourth role, `"condition"`. The engine knows only where a condition
// word may stand in a sentence (between the subjects and `IS`); what it means is a
// check content provides (`ConditionCheck`).
import type { Board, Kind, ObjectId } from "./grid";
export type Role = "noun" | "property" | "operator" | "condition";
/**
* Which objects a noun names when it is the subject of a rule:
* - `"named"` (the default): the things of its own name, `<NOUN>` names every object of kind `"<noun>"`;
* - `"text"`: every text object (the meta-noun `TEXT`);
* - `"things"`: every object that is not text, whatever its noun.
*/
export type Selects = "named" | "text" | "things";
/**
* One entry of the word list content provides: `{ "name": "<NOUN>", "role": "noun" }`,
* optionally with `"selects"` (default `"named"`).
*/
export interface WordDef {
name: string;
role: Role;
selects?: Selects;
}
export const IS = "IS";
export const AND = "AND";
export const NOT = "NOT";
/** The meta-noun for every text object. */
export const TEXT = "TEXT";
/** Every word the parser can read, with its role. */
export class Lexicon {
private readonly roles = new Map<string, Role>();
private readonly selects = new Map<string, Selects>();
/** The engine's own words plus `words`. Content can't redefine an engine word. */
constructor(words: readonly WordDef[]) {
for (const w of words) {
this.roles.set(w.name.toUpperCase(), w.role);
this.selects.set(w.name.toUpperCase(), w.selects ?? "named");
}
for (const op of [IS, AND, NOT]) this.roles.set(op, "operator");
this.roles.set(TEXT, "noun");
this.selects.set(TEXT, "text");
}
/** `undefined` for a word the lexicon doesn't know: such text spells nothing. */
role(word: string): Role | undefined {
return this.roles.get(word);
}
/** Does the noun `word`, as the subject of a rule, name an object of this kind? */
names(word: string, kind: Kind): boolean {
switch (this.selects.get(word) ?? "named") {
case "named":
return kind.type === "thing" && kind.noun === word;
case "text":
return kind.type === "text";
case "things":
return kind.type === "thing";
}
}
/**
* What a conversion into the noun `word` makes (Module 9): a thing of that
* name, if `word` names the things of its own name. `undefined` for a noun that
* names a group (`TEXT`, or `ALL` if content adds it): the course doesn't
* convert into those.
*/
thingNamed(word: string): Kind | undefined {
return (this.selects.get(word) ?? "named") === "named" ? { type: "thing", noun: word } : undefined;
}
}
/** A piece of text the parser can read: which object, which word, which role. */
export interface Token {
id: ObjectId;
word: string;
role: Role;
}
/** Tile id for maps: `"x,y"`. */
export const tileKey = (x: number, y: number): string => `${x},${y}`;
/**
* The words on the board, by tile; a tile's words in object order (oldest first).
* Text whose word the lexicon doesn't know is left out.
*/
export function lex(board: Board, lexicon: Lexicon): Map<string, Token[]> {
const tiles = new Map<string, Token[]>();
for (const o of board.objects()) {
if (o.kind.type !== "text") continue;
const role = lexicon.role(o.kind.word);
if (role === undefined) continue;
const tile = tileKey(o.x, o.y);
const tokens = tiles.get(tile) ?? [];
tokens.push({ id: o.id, word: o.kind.word, role });
tiles.set(tile, tokens);
}
return tiles;
}
@@ -0,0 +1,335 @@
// The action queue: how one player action (a Sequence) resolves.
//
// Performing an action changes the world and produces events. Every trigger that
// responds to those events goes into one Queue, oldest listener first, and the Queue
// is locked. Each queued trigger then resolves completely, including everything
// its own actions cause, before the next one starts: depth-first.
//
// Module 3: nothing leaves play mid-Sequence. When the outermost Phase ends, the
// Death Creation Step removes the dead, and a Death Phase resolves their
// Deathrattles. That repeats until a step finds no new deaths.
//
// Module 4: a queued trigger may activate more than once. How many times is
// decided by one routine, `activations` in `aura.ts`, from aura data on cards.
// Nothing here knows which cards multiply which triggers.
//
// Module 4, second interceptor: replacement. Before a Damage changes the world,
// the steps in `DAMAGE_STEPS` may rewrite it, in that order: a redirect changes
// who takes it, then Armor takes what it can of damage to a hero.
//
// Module 5: a card's effects are data. Each one is interpreted (`effect.ts`) and
// its actions performed before the next effect of the same ability is read, so
// "Assemble a minion. Give all minions in your hand +2/+2." buffs the new minion.
//
// Module 5, robustness: once a test can write any card as data, one card can make a
// Sequence endless. A trigger is never queued while that same trigger of that same
// minion is resolving (the nesting guard in `react`), and a Sequence that needs more
// than `DEATH_PHASE_LIMIT` Death Phases is an error.
//
// Module 10: the steps at the end of the outermost Phase are a list, `END_OF_PHASE`,
// read from the world, as Part II's turn is a list of phases. SpellGrid's tests run
// the same Sequence with the list reversed, to show what the order decides.
import { assembleFromPool } from "./assemble";
import { activations, auraUpdateOther, redirectFor } from "./aura";
import type { Effect, PoolFilter, Trigger, TriggerOn } from "./card";
import { deathCreationStep } from "./death";
import { actions } from "./effect";
import type { Event } from "./event";
import type { EntityId, PlayerId, World } from "./world";
/** Something the engine can do. Card effects return these instead of changing the world. */
export type Action =
/** Deal `amount` to each target at once: all damage first, then the reactions. */
| { type: "damage"; targets: EntityId[]; amount: number }
| { type: "draw"; player: PlayerId }
| { type: "gainArmor"; player: PlayerId; amount: number }
| { type: "gainAttack"; id: EntityId; amount: number }
| { type: "cast"; card: string; player: PlayerId; target?: EntityId }
/** Mark each target pending destroy. They leave play in the next Death Creation Step. */
| { type: "destroy"; targets: EntityId[] }
/** Summon `card` for `player` at board index `at`, or at the right end if `at` is undefined. */
| { type: "summon"; card: string; player: PlayerId; at?: number }
/** Play a minion from its owner's hand: it enters the right end of the board, then its Battlecry resolves. */
| { type: "play"; id: EntityId }
/** Put an Assembled minion into `player`'s hand (see `assemble.ts`). */
| { type: "assemble"; player: PlayerId; filter: PoolFilter }
/** Give one entity +attack/+health. */
| { type: "buff"; id: EntityId; attack: number; health: number }
/** Ask for `word` to be written on `source`'s grid tile (Module 10). Part I only records it. */
| { type: "spawnText"; source: EntityId; word: string };
/**
* A step that runs when the outermost Phase ends:
* - `"deathCreation"`: the Death Creation Step (Module 3);
* - `"auraUpdate"`: an Aura Update (Other) (Module 4).
*/
export type EndOfPhaseStep = "deathCreation" | "auraUpdate";
/**
* The end of the outermost Phase, in Hearthstone's order: the dead leave play first,
* then the auras are read again, so a minion that just died multiplies nothing in
* the Death Phase that follows.
*/
export const END_OF_PHASE: readonly EndOfPhaseStep[] = ["deathCreation", "auraUpdate"];
/**
* A step that may rewrite one Damage to one target before the world changes:
* - `"redirect"`: damage to a hero whose owner has a `damageRedirect` aura in force
* goes to that aura's minion instead (the oldest one's, if several);
* - `"armor"`: damage to a hero lowers its owner's Armor first, then Health.
*/
export type DamageStep = "redirect" | "armor";
/**
* The order the steps run in. Each runs once per Damage, so a rewritten Damage is
* never offered to a redirect again. A redirect comes first: Armor protects only
* the hero that actually takes the damage.
*/
export const DAMAGE_STEPS: readonly DamageStep[] = ["redirect", "armor"];
/** What an effect may look at when it runs. */
export class Ctx {
constructor(
readonly world: World,
/** The player the effect works for. */
readonly owner: PlayerId,
/** The minion whose trigger is resolving; undefined for a spell. */
readonly source?: EntityId,
/** The chosen target, if any. */
readonly target?: EntityId,
/**
* The source's board position. For a Deathrattle, the source has left play, and
* this is where it stood: immediately left of the minion its Death Event remembers.
*/
readonly position?: number,
) {}
/** The minion whose trigger is resolving. */
minion(): EntityId {
if (this.source === undefined) throw new Error("only a triggered ability has a source");
return this.source;
}
enemyHero(): EntityId {
return this.world.hero(this.owner === 0 ? 1 : 0);
}
}
/**
* The most Death Phases one Sequence may have. One more is an error, not a cut-off:
* a Sequence that stopped early would resolve differently without saying so.
*/
export const DEATH_PHASE_LIMIT = 64;
/** Run one Sequence: a player action and everything it causes. */
export function runSequence(world: World, action: Action): void {
perform(world, action);
// The outermost Phase has ended. Each Death Phase is itself an outermost
// Phase, so the end-of-Phase steps run again after it, until nothing dies.
let deaths = endOfOutermostPhase(world);
for (let round = 1; deaths.length > 0; round++) {
// Between Death Phases no trigger is resolving, so the error leaves no block open.
if (round > DEATH_PHASE_LIMIT) {
throw new Error(`Death Phase limit: more than ${DEATH_PHASE_LIMIT} Death Phases in one Sequence`);
}
world.record({ type: "DeathPhase", round });
react(world, deaths);
deaths = endOfOutermostPhase(world);
}
}
/**
* The steps that run only when the outermost Phase ends, in the world's order
* (`END_OF_PHASE` unless changed); returns the Death Events. Hearthstone runs an
* Aura Update (Health/Attack) first. The course has no Health/Attack auras, so that
* step has nothing to do and is left out.
*/
function endOfOutermostPhase(world: World): Event[] {
let deaths: Event[] = [];
for (const step of world.endOfPhase()) {
switch (step) {
case "deathCreation":
deaths = deathCreationStep(world);
break;
case "auraUpdate":
auraUpdateOther(world);
break;
}
}
return deaths;
}
/** Summon `card` for `player` at board index `at` (or the right end). Summoning runs an Aura Update. */
export function summon(world: World, card: string, player: PlayerId, at?: number): EntityId {
const id = world.create(card, player, "board", at);
auraUpdateOther(world);
return id;
}
function perform(world: World, action: Action): void {
switch (action.type) {
case "damage": {
// Area damage: every Damage Event is created, in order of play, before any reaction.
const targets = [...action.targets].sort((a, b) => world.orderOfPlay(a)! - world.orderOfPlay(b)!);
const events: Event[] = [];
for (const target of targets) events.push(applyDamage(world, target, action.amount));
react(world, events);
break;
}
case "draw":
world.draw(action.player);
break;
case "gainArmor":
world.gainArmor(action.player, action.amount);
break;
case "gainAttack":
world.gainAttack(action.id, action.amount);
break;
case "destroy":
for (const target of action.targets) world.markDestroyed(target);
break;
case "summon":
summon(world, action.card, action.player, action.at);
break;
case "play": {
if (world.zoneOf(action.id) !== "hand") throw new Error(`play: entity ${action.id} is not in a hand`);
// Playing is not summoning: no Aura Update here. The played minion's own
// auras join at the end of the outermost Phase.
world.moveTo(action.id, "board");
const played: Event = { type: "Played", id: action.id };
world.record(played);
react(world, [played]);
break;
}
case "cast": {
const spell = world.card(action.card);
if (spell.def.type !== "spell") throw new Error(`"${action.card}" is not a spell`);
world.record({ type: "SpellCast", card: action.card });
runEffects(world, spell.onCast, new Ctx(world, action.player, undefined, action.target));
break;
}
case "assemble":
assembleFromPool(world, action.player, action.filter);
break;
case "buff":
world.buff(action.id, action.attack, action.health);
break;
case "spawnText":
world.record({ type: "TextSpawned", source: action.source, word: action.word });
break;
}
}
/**
* Interpret and perform an ability's effects in printed order. Each effect sees the
* world as the effects before it left it. Depth-first: each action, and every
* reaction it causes, finishes before the next.
*/
function runEffects(world: World, effects: Effect[], ctx: Ctx): void {
for (const effect of effects) {
for (const action of actions(effect, ctx)) perform(world, action);
}
}
/**
* Deal one Damage to one target, after the steps of `DAMAGE_STEPS` have rewritten it.
* Returns its Damage Event, which names whoever took it.
*/
function applyDamage(world: World, target: EntityId, amount: number): Event {
let taker = target;
let absorbed = 0;
for (const step of DAMAGE_STEPS) {
switch (step) {
case "redirect":
if (world.isHero(taker)) taker = redirectFor(world, world.ownerOf(taker)) ?? taker;
break;
case "armor":
if (world.isHero(taker)) {
const player = world.ownerOf(taker);
absorbed = Math.min(world.armor(player), amount);
if (absorbed > 0) world.loseArmor(player, absorbed);
}
break;
}
}
world.damage(taker, amount, absorbed);
return { type: "Damaged", target: taker, amount };
}
/** A Phase: queue every trigger that responds to `events`, then resolve them one by one. */
function react(world: World, events: Event[]): void {
// Fill the Queue: for each event (already in order of play), each listener in order of play.
// The nesting guard: a trigger that is resolving right now is not queued again, so
// a trigger is never nested inside itself.
const queue: [EntityId, number, Trigger, Event][] = [];
for (const event of events) {
for (const listener of listeners(world, event)) {
world.card(world.cardName(listener)).triggers.forEach((trigger, index) => {
if (responds(world, trigger.on, listener, event) && !world.isResolving(listener, index)) {
queue.push([listener, index, trigger, event]);
}
});
}
}
// The Queue is now locked: nothing that happens below adds to it.
for (const [source, index, trigger, event] of queue) {
// The interceptor: the entry was queued once; the auras in force when it
// starts decide how many times it activates, back to back.
const times = activations(world, world.ownerOf(source), trigger.on);
for (let i = 0; i < times; i++) {
world.record({ type: "Triggered", source, on: trigger.on });
const position =
event.type === "Destroyed" && event.id === source
? positionLeftOf(world, world.ownerOf(source), event.before)
: world.boardIndex(source);
// The trigger's consequences are one level deeper in the log tree.
world.openBlock(source, index);
runEffects(world, trigger.effects, new Ctx(world, world.ownerOf(source), source, undefined, position));
world.closeBlock();
}
}
}
/**
* The board index, right now, immediately left of `before` on `owner`'s board; the
* right end if `before` is undefined or no longer there. A Death Event remembers the
* minion, not an index, so this still holds after other minions have left or arrived.
*/
function positionLeftOf(world: World, owner: PlayerId, before: EntityId | undefined): number {
const board = world.zone(owner, "board");
const index = before === undefined ? -1 : board.indexOf(before);
return index < 0 ? board.length : index;
}
/**
* Who may respond to `event`, in order of play: every minion in play and, for a
* Death Event, the minion that died. It has left play, but its Deathrattle still answers.
*/
function listeners(world: World, event: Event): EntityId[] {
const inPlay = world.minionsInPlay();
if (event.type !== "Destroyed") return inPlay;
return [...inPlay, event.id].sort((a, b) => world.orderOfPlay(a)! - world.orderOfPlay(b)!);
}
/** Does a trigger listening for `on`, belonging to `listener`, respond to `event`? */
function responds(world: World, on: TriggerOn, listener: EntityId, event: Event): boolean {
if (event.type === "Destroyed") return on === "deathrattle" && event.id === listener;
if (event.type === "Played") return on === "battlecry" && event.id === listener;
if (event.type !== "Damaged") return false;
const isMinion = !world.isHero(event.target);
switch (on) {
case "selfDamaged":
return event.target === listener;
case "friendlyMinionDamaged":
return isMinion && world.ownerOf(event.target) === world.ownerOf(listener);
case "anyMinionDamaged":
return isMinion;
case "heroDamaged":
return !isMinion;
case "deathrattle":
case "battlecry":
case "endOfTurn":
return false;
}
}
@@ -0,0 +1,282 @@
// Part III: SpellGrid, the two engines in one level.
//
// Part I (the card engine) and Part II (the grid engine, `puzzle/`) don't import each
// other. This file is the only place that knows both, and they meet at four seams:
//
// 1. Rules become modifiers. After every reparse, the rules in force that join a
// trigger word to a multiplier word (`DEATHRATTLE IS DOUBLE`) are compiled into
// auras, and handed to Part I with `World.setRuleAuras`. A negated one
// (`DEATHRATTLE IS NOT DOUBLE`) becomes a cap. Which words are which is content's
// data (`SeamWords`), not code here.
// 2. Hazards mark, the death step removes. The grid engine hands the objects a turn
// destroys to `destroy` below. A minion among them is marked pending destroy, in
// one Sequence, and Part I's Death Creation Step removes it.
// 3. Grid changes reparse. While a Sequence runs, the grid doesn't change. When it
// ends, the text its effects asked for (`TextSpawned` events) is written, the
// minions that died (`Destroyed` events) leave the grid, and the rules are parsed
// again, before anything else happens.
// 4. Summoned minions enter the grid. In the same place, before the dead leave: a
// minion that a minion's ability summoned (`Summoned` events with a `by`), whose card
// has a noun, becomes an object of that noun on its summoner's tile. One summoned by
// a spell, or with no noun, is in play but not on the grid.
//
// One id space: a minion's id is its grid object's id. Every other card (a hero, a
// card in hand) gets an id from the same counter that no object will have.
import { auraUpdateOther } from "./aura";
import type { Aura, CardLibrary, TriggerOn } from "./card";
import type { Event } from "./event";
import * as puzzle from "./puzzle";
import { runSequence, type Action } from "./queue";
import { World, type EntityId } from "./world";
/**
* The words that cross the seam, from content: nouns that name a trigger
* (`"DEATHRATTLE": "deathrattle"`) and properties that multiply one
* (`"DOUBLE": { "times": 2, "stacking": "max" }`).
*/
export interface SeamWords {
triggerWords: Record<string, TriggerOn>;
multiplierWords: Record<string, { times: number; stacking: "max" | "additive" }>;
}
/**
* Seam 1: the auras the rules in force give. `T IS M` (T a trigger word, M a multiplier
* word) multiplies T's triggers; `T IS NOT M` caps them at one activation, so it
* cancels a card's multiplier too, as `X IS NOT Y` cancels every `X IS Y`.
*
* Module 11: a rule with a condition gives no aura. A condition holds or not for one
* object on the grid, and a trigger word names no object (the course's choice).
*/
export function ruleAuras(rules: readonly puzzle.Rule[], words: SeamWords): Aura[] {
const auras: Aura[] = [];
for (const rule of rules) {
if (rule.conditions.length > 0) continue;
const trigger = words.triggerWords[rule.subject];
const multiplier = words.multiplierWords[rule.complement];
if (trigger === undefined || multiplier === undefined) continue;
auras.push(
rule.negated
? { type: "triggerCap", trigger, times: 1 }
: { type: "triggerMultiplier", trigger, ...multiplier },
);
}
return auras;
}
/** A SpellGrid level in play: a grid and a game, kept in step. Ids are grid ids (see above). */
export class SpellGrid {
readonly world: World;
readonly level: puzzle.Level;
/** Every entity's id. */
private readonly ids = new Map<EntityId, puzzle.ObjectId>();
/** The entity behind each id (a card's old id still finds it after it is played). */
private readonly entities = new Map<puzzle.ObjectId, EntityId>();
/**
* The level's objects; its `cards` as minions in play for player 0, each a grid
* object named by the card's noun; its `hand` in player 0's hand.
*/
constructor(
def: puzzle.LevelDef,
library: CardLibrary,
words: readonly puzzle.WordDef[],
responses: readonly puzzle.Response[],
hazards: readonly puzzle.Hazard[],
conditions: puzzle.ConditionChecks,
private readonly seamWords: SeamWords,
) {
this.world = new World(library);
this.level = new puzzle.Level(def, words, responses, hazards, conditions);
for (const { card, x, y } of def.cards ?? []) {
const id = this.level.board.place(this.nounOf(card), x, y);
this.link(this.world.create(card, 0, "board"), id);
}
// Every minion is in play: one Aura Update, as for a summon.
auraUpdateOther(this.world);
for (const card of def.hand ?? []) this.world.create(card, 0, "hand");
this.linkTheRest();
this.gridChanged();
}
// ---- Ids ----
/** The id of entity `entity`. */
idOf(entity: EntityId): puzzle.ObjectId {
const id = this.ids.get(entity);
if (id === undefined) throw new Error(`entity ${entity} has no id`);
return id;
}
/** The entity behind `id`. Throws if `id` is not a card's. */
entity(id: puzzle.ObjectId): EntityId {
const entity = this.entities.get(id);
if (entity === undefined) throw new Error(`${id} is not a card`);
return entity;
}
/** The id of the minion on tile (x, y), if there is one. */
minionAt(x: number, y: number): puzzle.ObjectId | undefined {
return this.level.board.at(x, y).find((o) => this.entities.has(o.id))?.id;
}
private link(entity: EntityId, id: puzzle.ObjectId): void {
this.ids.set(entity, id);
this.entities.set(id, entity);
}
/** Give every entity that has no id one that no object will have. */
private linkTheRest(): void {
for (let entity = 0; entity < this.world.entityCount(); entity++) {
if (!this.ids.has(entity)) this.link(entity, this.level.board.reserveId());
}
}
private nounOf(card: string): puzzle.Kind {
const { def } = this.world.card(card);
if (def.type !== "minion" || def.noun === undefined) throw new Error(`"${card}" is not a minion with a noun`);
return puzzle.Kind.parse(def.noun.toLowerCase());
}
// ---- What the player does ----
/** Add an object or a piece of text, then reparse. */
place(kind: string, x: number, y: number): puzzle.ObjectId {
const id = this.level.place(kind, x, y);
this.gridChanged();
return id;
}
/** One whole turn (Module 9); what it destroys goes through seam 2. */
step(dir: puzzle.Dir): void {
this.level.step(dir, this.destroy);
this.gridChanged();
}
/** Movement only (Module 8), then a reparse. */
move(dir: puzzle.Dir): void {
this.level.moveYou(dir, this.destroy);
this.gridChanged();
}
/**
* Why `cast(card, target)` can't be done now, or `undefined` if it can: the card must
* be a spell in player 0's hand, and a spell that hits a chosen minion needs a target
* minion on the grid. Nothing changes when the answer is no.
*/
checkCast(card: string, target?: puzzle.ObjectId): string | undefined {
const problem = this.checkHand(card, "spell");
if (problem !== undefined) return problem;
const chooses = this.world.card(card).onCast.some((e) => "target" in e && e.target === "chosen");
if (target === undefined) return chooses ? `"${card}" needs a target minion` : undefined;
const object = this.level.board.find(target);
if (object === undefined || !this.entities.has(target)) return `${target} is not a minion on the grid`;
return undefined;
}
/** Why `play(card, x, y)` can't be done now, or `undefined` if it can. */
checkPlay(card: string, x: number, y: number): string | undefined {
const problem = this.checkHand(card, "minion");
if (problem !== undefined) return problem;
if (this.world.card(card).def.noun === undefined) return `"${card}" has no noun: it can't stand on the grid`;
if (!this.level.board.inBounds(x, y)) return `(${x}, ${y}) is off the grid`;
if (this.level.board.at(x, y).length > 0) return `(${x}, ${y}) is not empty`;
return undefined;
}
/** Cast a spell from player 0's hand, as one Sequence. `target` is a minion's id. Throws if `checkCast` says no. */
cast(card: string, target?: puzzle.ObjectId): void {
const problem = this.checkCast(card, target);
if (problem !== undefined) throw new Error(problem);
this.world.moveTo(this.inHand(card)!, "graveyard");
const chosen = target === undefined ? undefined : this.entity(target);
this.sequence({ type: "cast", card, player: 0, target: chosen });
}
/**
* Play a minion from player 0's hand onto the empty tile (x, y), as one Sequence.
* Returns its new id. Throws if `checkPlay` says no.
*/
play(card: string, x: number, y: number): puzzle.ObjectId {
const problem = this.checkPlay(card, x, y);
if (problem !== undefined) throw new Error(problem);
const inHand = this.inHand(card)!;
const id = this.level.board.place(this.nounOf(card), x, y);
this.link(inHand, id);
this.sequence({ type: "play", id: inHand });
return id;
}
private inHand(card: string): EntityId | undefined {
return this.world.zone(0, "hand").find((id) => this.world.cardName(id) === card);
}
private checkHand(card: string, type: "minion" | "spell"): string | undefined {
if (this.inHand(card) === undefined) return `"${card}" is not in the hand`;
if (this.world.card(card).def.type !== type) return `"${card}" is not a ${type}`;
return undefined;
}
// ---- The seams ----
/**
* Seam 2. Objects that aren't cards leave the board. Minions are marked pending
* destroy, in one Sequence, under the rules as they stand now.
*/
private readonly destroy: puzzle.Destroy = (level, ids) => {
const present = [...new Set(ids)].filter((id) => level.board.find(id) !== undefined);
const minions = present.filter((id) => this.entities.has(id));
for (const id of present) if (!this.entities.has(id)) level.board.remove(id);
if (minions.length === 0) return;
this.syncRules();
this.sequence({ type: "destroy", targets: minions.map((id) => this.entity(id)) });
};
/** Run one Sequence in Part I, then seams 3 and 4: apply its grid changes and reparse. */
private sequence(action: Action): void {
const from = this.world.log().length;
runSequence(this.world, action);
const events = this.world.log().slice(from);
const board = this.level.board;
// Text first, while every minion that wrote it is still on its tile.
for (const event of events) {
if (event.type !== "TextSpawned") continue;
// A minion with no id (summoned by this Sequence) is not on the grid yet: nowhere to write.
const id = this.ids.get(event.source);
const source = id === undefined ? undefined : board.find(id);
if (source !== undefined) board.place({ type: "text", word: event.word }, source.x, source.y);
}
// Seam 4: summoned minions, in summon order, on their summoner's tile. The dead are still
// on the grid here, so a Deathrattle's summoner is found where it died; a minion summoned
// earlier in the list is already placed, so it can be a summoner too.
for (const event of events) {
if (event.type !== "Summoned" || event.by === undefined) continue;
const card = this.world.cardName(event.id);
if (this.ids.has(event.id) || this.world.card(card).def.noun === undefined) continue;
const by = this.ids.get(event.by);
const summoner = by === undefined ? undefined : board.find(by);
if (summoner !== undefined) this.link(event.id, board.place(this.nounOf(card), summoner.x, summoner.y));
}
for (const event of deaths(events)) {
const id = this.ids.get(event.id);
if (id !== undefined) board.remove(id);
}
this.linkTheRest();
this.gridChanged();
}
/** The grid changed: reparse it, then seam 1. */
private gridChanged(): void {
this.level.reparse();
this.syncRules();
}
/** Seam 1: compile the rules in force into auras for Part I, then run an Aura Update. */
private syncRules(): void {
this.world.setRuleAuras(ruleAuras(this.level.activeRules(), this.seamWords));
auraUpdateOther(this.world);
}
}
const deaths = (events: readonly Event[]) =>
events.filter((e): e is Extract<Event, { type: "Destroyed" }> => e.type === "Destroyed");
@@ -0,0 +1,382 @@
// Entities, zones, order of play, stats and the event log: the game's state.
// The rules for resolving actions live in `queue.ts`.
import type { Aura, Card, CardDef, CardLibrary } from "./card";
import type { Event } from "./event";
import { END_OF_PHASE, type EndOfPhaseStep } from "./queue";
/** A handle to an entity. Ids are never reused and stay valid when the entity changes zone. */
export type EntityId = number;
export type PlayerId = 0 | 1;
export type Zone = "deck" | "hand" | "board" | "graveyard";
/** Starting Health of each hero. */
export const HERO_HEALTH = 30;
interface Entity {
card: string;
owner: PlayerId;
zone: Zone;
/** Stamped each time the entity enters the board; undefined until it first does. */
orderOfPlay?: number;
attack: number;
/** Current Health. At 0 or below the entity is mortally wounded: still in play until the Death Creation Step. */
health: number;
/** Hit by a destroy effect: still in play until the Death Creation Step. */
pendingDestroy: boolean;
}
/**
* An aura in force: which minion has it, and that minion's owner. An aura from outside
* the cards (Module 10) has no minion, and is in force once for each player.
*/
export interface ActiveAura {
owner: PlayerId;
source?: EntityId;
aura: Aura;
}
/**
* Makes a player's choices: called with what is being chosen and the options, it
* returns one of the options. Hearthstone would ask the player or pick at random.
*/
export type Chooser = (kind: string, options: readonly string[]) => string;
/** One player's zones (each an ordered list of ids; the board is left to right), hero and Armor. */
type Side = Record<Zone, EntityId[]> & { hero: EntityId; armor: number };
export class World {
/** Indexed by EntityId. */
private readonly entities: Entity[] = [];
private readonly sides: [Side, Side];
private nextOrderOfPlay = 0;
private events: Event[] = [];
/** The nesting depth at which each event in `events` was recorded, index for index. */
private depths: number[] = [];
/**
* The triggers resolving right now, one inside another, outermost first: each is a
* minion and the index of the trigger in its card's `triggers`. How many there are
* is the nesting depth of the log tree.
*/
private resolving: [EntityId, number][] = [];
/** The auras in force as of the last Aura Update. Changes only there (see `aura.ts`). */
private auras: ActiveAura[] = [];
/** Until a chooser is set, take the first option. */
private chooser: Chooser = (_kind, options) => options[0];
/**
* Module 10: auras that no minion carries, for every player's minions. SpellGrid
* compiles them from the rules on its grid. The next Aura Update brings them in.
*/
private rules: Aura[] = [];
/** Module 10: the steps at the end of the outermost Phase, in order. */
private endSteps: readonly EndOfPhaseStep[] = END_OF_PHASE;
/** A new game: both heroes are in play from the start, so they are the oldest entities. */
constructor(private readonly library: CardLibrary) {
const side = (owner: PlayerId): Side => {
const hero = this.entities.length;
this.entities.push({
card: "Hero",
owner,
zone: "board",
orderOfPlay: ++this.nextOrderOfPlay,
attack: 0,
health: HERO_HEALTH,
pendingDestroy: false,
});
// In play, but not on the board list: the board holds minions only.
return { deck: [], hand: [], board: [], graveyard: [], hero, armor: 0 };
};
this.sides = [side(0), side(1)];
}
/**
* Create an entity of `card` for `owner` in `zone`: at the end (the right end for the board),
* or at index `at` when given (clamped to the zone's length).
*/
create(card: string, owner: PlayerId, zone: Zone, at?: number): EntityId {
if (owner !== 0 && owner !== 1) throw new Error(`no player ${owner}`);
const { def } = this.library.get(card);
const id = this.entities.length;
this.entities.push({
card: def.name,
owner,
zone,
attack: def.attack ?? 0,
health: def.health ?? 0,
pendingDestroy: false,
});
this.enter(id, zone, at);
return id;
}
/** Move an entity to the end of `zone` on its owner's side, keeping its id. */
moveTo(id: EntityId, zone: Zone): void {
const { owner, zone: from } = this.entity(id);
const list = this.sides[owner][from];
list.splice(list.indexOf(id), 1);
this.enter(id, zone);
}
/** The single place an entity arrives in a zone. Entering the board starts a new order of play. */
private enter(id: EntityId, zone: Zone, at?: number): void {
const entity = this.entity(id);
const list = this.sides[entity.owner][zone];
list.splice(at === undefined ? list.length : Math.min(at, list.length), 0, id);
entity.zone = zone;
if (zone === "board") {
entity.orderOfPlay = ++this.nextOrderOfPlay;
// The summoner is the minion whose trigger is resolving (the innermost one), if any.
this.record({ type: "Summoned", id, by: this.resolving.at(-1)?.[0] });
}
}
private entity(id: EntityId): Entity {
const entity = this.entities[id];
if (!entity) throw new Error(`unknown entity id ${id}`);
return entity;
}
// ---- Queries ----
card(name: string): Card {
return this.library.get(name);
}
hasCard(name: string): boolean {
return this.library.has(name);
}
/** The cards an Assemble may offer, sorted by name. */
pool(): Card[] {
return this.library.pool();
}
zoneOf(id: EntityId): Zone {
return this.entity(id).zone;
}
cardName(id: EntityId): string {
return this.entity(id).card;
}
ownerOf(id: EntityId): PlayerId {
return this.entity(id).owner;
}
/** When the entity last entered the board (larger is later); undefined if it never has. */
orderOfPlay(id: EntityId): number | undefined {
return this.entity(id).orderOfPlay;
}
/** The ids in one player's zone, in zone order (left to right for the board). */
zone(owner: PlayerId, zone: Zone): readonly EntityId[] {
if (owner !== 0 && owner !== 1) throw new Error(`no player ${owner}`);
return this.sides[owner][zone];
}
hero(owner: PlayerId): EntityId {
if (owner !== 0 && owner !== 1) throw new Error(`no player ${owner}`);
return this.sides[owner].hero;
}
isHero(id: EntityId): boolean {
return this.sides.some((s) => s.hero === id);
}
/** Every minion on either board, oldest first (order of play). */
minionsInPlay(): EntityId[] {
return [...this.sides[0].board, ...this.sides[1].board].sort(
(a, b) => this.orderOfPlay(a)! - this.orderOfPlay(b)!,
);
}
attack(id: EntityId): number {
return this.entity(id).attack;
}
health(id: EntityId): number {
return this.entity(id).health;
}
/** The cost printed on the entity's card. */
cost(id: EntityId): number {
return this.card(this.cardName(id)).def.cost;
}
/** The minions in a player's hand, in hand order. */
minionsInHand(owner: PlayerId): EntityId[] {
return this.zone(owner, "hand").filter((id) => this.card(this.cardName(id)).def.type === "minion");
}
/** Mortally wounded (0 or less Health) or pending destroy: it leaves play at the next Death Creation Step. */
isDoomed(id: EntityId): boolean {
const entity = this.entity(id);
return entity.health <= 0 || entity.pendingDestroy;
}
/** Index of a minion on its owner's board, left to right; undefined if it is not on the board. */
boardIndex(id: EntityId): number | undefined {
const index = this.sides[this.ownerOf(id)].board.indexOf(id);
return index < 0 ? undefined : index;
}
armor(owner: PlayerId): number {
return this.sides[owner].armor;
}
activeAuras(): readonly ActiveAura[] {
return this.auras;
}
/** Auras from outside the cards (Module 10), as last set. */
ruleAuras(): readonly Aura[] {
return this.rules;
}
/** The steps at the end of the outermost Phase, in order. */
endOfPhase(): readonly EndOfPhaseStep[] {
return this.endSteps;
}
/** How many entities exist: ids run from 0 to this, exclusive. */
entityCount(): number {
return this.entities.length;
}
log(): readonly Event[] {
return this.events;
}
/** The nesting depth of each event in `log()`, index for index: 0 for the outermost action's own events. */
logDepths(): readonly number[] {
return this.depths;
}
// ---- Changes: each one records its event ----
record(event: Event): void {
this.events.push(event);
this.depths.push(this.resolving.length);
}
/**
* Trigger number `trigger` of `source` starts resolving: the events it causes are
* recorded one level deeper.
*/
openBlock(source: EntityId, trigger: number): void {
this.resolving.push([source, trigger]);
}
/** The trigger that last opened a block has finished resolving. */
closeBlock(): void {
this.resolving.pop();
}
/** Is trigger number `trigger` of `source` resolving right now (at any depth)? */
isResolving(source: EntityId, trigger: number): boolean {
return this.resolving.some(([s, t]) => s === source && t === trigger);
}
/** Add a card to the library, or replace the card with the same name. */
register(def: CardDef): void {
this.library.register(def);
}
/** Add a card made by Assemble (see `assemble.ts`). */
registerAssembled(def: CardDef): void {
this.library.registerAssembled(def);
}
setChooser(chooser: Chooser): void {
this.chooser = chooser;
}
/** Ask the chooser to pick one of `options` (not empty). Throws if it answers something that isn't an option. */
choose(kind: string, options: readonly string[]): string {
const answer = this.chooser(kind, options);
if (!options.includes(answer)) {
throw new Error(`chooser answered "${answer}" for "${kind}"; the options were ${JSON.stringify(options)}`);
}
return answer;
}
/** Replace the auras from outside the cards. They take effect at the next Aura Update. */
setRuleAuras(auras: Aura[]): void {
this.rules = auras;
}
/** Change the order of the steps at the end of the outermost Phase. */
setEndOfPhase(steps: readonly EndOfPhaseStep[]): void {
this.endSteps = steps;
}
/** Only an Aura Update calls this. */
setActiveAuras(auras: ActiveAura[]): void {
this.auras = auras;
}
clearLog(): void {
this.events = [];
this.depths = [];
}
/**
* `target` takes `amount` damage, of which `absorbed` was already taken by Armor
* (see `loseArmor`): Health drops by the rest. The event shows the full amount.
*/
damage(target: EntityId, amount: number, absorbed = 0): void {
this.entity(target).health -= amount - absorbed;
this.record({ type: "Damaged", target, amount });
}
/** `player`'s hero's Armor absorbs `amount` damage. */
loseArmor(player: PlayerId, amount: number): void {
this.sides[player].armor -= amount;
this.record({ type: "ArmorLost", player, amount });
}
/**
* Draw the top card of `player`'s deck (the first in the list). An empty deck
* records `Fatigue` and changes nothing else.
*/
draw(player: PlayerId): void {
const id = this.zone(player, "deck")[0];
if (id === undefined) {
this.record({ type: "Fatigue", player });
return;
}
this.moveTo(id, "hand");
this.record({ type: "Drew", player, id });
}
/** Mark for destruction. Nothing leaves play until the Death Creation Step. */
markDestroyed(id: EntityId): void {
this.entity(id).pendingDestroy = true;
}
/** Move a minion from play to its owner's graveyard, recording its Death Event. */
kill(id: EntityId, before: EntityId | undefined): void {
this.entity(id).pendingDestroy = false;
this.moveTo(id, "graveyard");
this.record({ type: "Destroyed", id, before });
}
gainArmor(player: PlayerId, amount: number): void {
this.sides[player].armor += amount;
this.record({ type: "ArmorGained", player, amount });
}
buff(id: EntityId, attack: number, health: number): void {
const entity = this.entity(id);
entity.attack += attack;
entity.health += health;
this.record({ type: "StatsChanged", id, attack: entity.attack, health: entity.health });
}
gainAttack(id: EntityId, amount: number): void {
const entity = this.entity(id);
entity.attack += amount;
this.record({ type: "StatsChanged", id, attack: entity.attack, health: entity.health });
}
}
@@ -0,0 +1,301 @@
// Test harness: the thin adapter the course tests call. It translates each
// call into engine operations and holds no rules of its own.
import { cards, conditions, hazards, movement, seams, words } from "./content";
import {
assemble,
CardLibrary,
END_OF_PHASE,
SpellGrid as Grid,
puzzle,
registerAssembled,
runSequence,
summon,
triggerLabel,
World,
type CardDef,
type Chooser,
type Event,
type Half,
type Zone,
} from "./engine";
export { AssembleError } from "./engine";
export type { CardDef, Chooser, Half, Zone };
export type Id = number;
export type Player = 0 | 1;
/** The events the tests read from `log()` (see course/harness.md). */
export type GameEvent =
| { type: "Summoned"; id: Id }
| { type: "Damaged"; target: Id; amount: number }
| { type: "Triggered"; source: Id; trigger: string }
| { type: "Destroyed"; id: Id }
| { type: "DeathPhase"; round: number }
| { type: "Drew"; player: Player; id: Id }
| { type: "Fatigue"; player: Player }
| { type: "ArmorGained"; player: Player; amount: number }
| { type: "ArmorLost"; player: Player; amount: number }
| { type: "StatsChanged"; id: Id; attack: number; health: number }
| { type: "SpellCast"; card: string };
/** The course's view of an engine event; `undefined` for engine events the course log doesn't show. */
function toGameEvent(event: Event): GameEvent | undefined {
if (event.type === "Triggered") return { type: "Triggered", source: event.source, trigger: triggerLabel(event.on) };
if (event.type === "Destroyed") return { type: "Destroyed", id: event.id };
if (event.type === "Summoned") return { type: "Summoned", id: event.id };
// A played minion's `Summoned` already shows it entering the board.
if (event.type === "Played") return undefined;
// SpellGrid writes the text on the grid; the course's log doesn't show the request.
if (event.type === "TextSpawned") return undefined;
return { ...event };
}
/** One event as `scenarios.md` writes it, such as `Triggered{source: 3, trigger: "deathrattle"}`. */
function eventText(event: GameEvent): string {
const { type, ...fields } = event;
const shown = Object.entries(fields).map(
([key, value]) => `${key}: ${typeof value === "string" ? JSON.stringify(value) : value}`,
);
return `${type}{${shown.join(", ")}}`;
}
/** The log as a tree: one event per line, two spaces of indent per level of nesting. */
function logTree(world: World, view: (event: Event) => GameEvent | undefined): string {
const depths = world.logDepths();
return world
.log()
.map((event, i) => {
const shown = view(event);
return shown === undefined ? "" : `${" ".repeat(depths[i])}${eventText(shown)}\n`;
})
.join("");
}
export interface Game {
summon(card: string, player?: Player): Id; // to right end of board
addToHand(card: string, player?: Player): Id;
addToDeck(card: string, player?: Player): Id;
play(handId: Id): void; // Module 1 Expansion Drop; required from Module 4 (Battlecry)
zoneOf(id: Id): Zone;
board(player: Player): Id[]; // left to right
hand(player: Player): Id[];
orderOfPlay(id: Id): number;
nameOf(id: Id): string;
// Module 2
hero(player: Player): Id;
dealDamage(target: Id, amount: number): void; // one Sequence
cast(card: string, player?: Player, target?: Id): void;
health(id: Id): number;
attack(id: Id): number;
armor(player: Player): number;
log(): GameEvent[];
clearLog(): void;
logTree(): string; // the log as an indented tree
// Module 3
destroy(target: Id): void; // marks pending destroy, one Sequence
// Module 5
registerCards(defs: CardDef[]): void; // add card definitions, replacing any with the same name
assemble(base: string, donor: string, half: Half): string; // registers "<base> (Assembled)"; throws AssembleError
cost(id: Id): number;
setChooser(choose: Chooser): void; // makes Assemble's choices: kind "base" | "donor" | "half"
}
export function newGame(opts?: { decks?: [string[], string[]] }): Game {
const world = new World(new CardLibrary(cards()));
opts?.decks?.forEach((deck, player) => {
for (const card of deck) world.create(card, player as Player, "deck");
});
return {
summon: (card, player = 0) => summon(world, card, player),
addToHand: (card, player = 0) => world.create(card, player, "hand"),
addToDeck: (card, player = 0) => world.create(card, player, "deck"),
// Play a card from hand, as one Sequence: it enters the right end of the board, then its Battlecry resolves.
play: (handId) => runSequence(world, { type: "play", id: handId }),
zoneOf: (id) => world.zoneOf(id),
board: (player) => [...world.zone(player, "board")],
hand: (player) => [...world.zone(player, "hand")],
orderOfPlay: (id) => {
const order = world.orderOfPlay(id);
if (order === undefined) throw new Error(`entity ${id} has not entered play`);
return order;
},
nameOf: (id) => world.cardName(id),
hero: (player) => world.hero(player),
dealDamage: (target, amount) => runSequence(world, { type: "damage", targets: [target], amount }),
cast: (card, player = 0, target) => runSequence(world, { type: "cast", card, player, target }),
health: (id) => world.health(id),
attack: (id) => world.attack(id),
armor: (player) => world.armor(player),
log: () => world.log().flatMap((event) => toGameEvent(event) ?? []),
clearLog: () => world.clearLog(),
logTree: () => logTree(world, toGameEvent),
destroy: (target) => runSequence(world, { type: "destroy", targets: [target] }),
registerCards: (defs) => defs.forEach((def) => world.register(def)),
assemble: (base, donor, half) =>
registerAssembled(world, assemble(world.card(base).def, world.card(donor).def, half)),
cost: (id) => world.cost(id),
setChooser: (choose) => world.setChooser(choose),
};
}
// ======== Part II: Baba Is You ========
export type { Dir, LevelCard, LevelDef, LevelObject } from "./engine/puzzle";
/** A Baba Is You level in play (Part II). */
export interface Baba {
// Module 6
place(kind: string, x: number, y: number): Id; // "baba" or "text:BABA"
rules(): string[]; // distinct, normalised ("BABA IS YOU", "BABA IS NOT YOU", "BABA ON KEY IS WIN"), sorted; before resolution
// Module 7
activeRules(): string[]; // the same, after resolution (NOT, identity)
hasProperty(id: Id, prop: string): boolean;
// Module 8
move(dir: puzzle.Dir): void; // every YOU object one tile, with what it pushes; then reparse. Movement only.
posOf(id: Id): [number, number] | null; // null once destroyed
kindsAt(x: number, y: number): string[]; // kinds as placed ("baba", "text:ROCK"), sorted
stackAt(x: number, y: number): string[]; // the same tile, oldest object first
// Module 9
step(dir: puzzle.Dir): void; // one whole turn: movement, parse, conversion, parse, hazards, win (then parse)
won(): boolean; // a turn found a YOU object on a WIN one (or WIN itself); stays true
lost(): boolean; // no object is YOU
undo(): boolean; // back to before the last step or move; false if there is none
}
/** Distinct rules as normalised strings, sorted. */
const normalised = (rules: readonly puzzle.Rule[]): string[] => [...new Set(rules.map(puzzle.ruleText))].sort();
/** A level from a level definition; its text is parsed straight away. */
export function newBaba(level: puzzle.LevelDef): Baba {
const lvl = new puzzle.Level(level, words(), movement(), hazards(), conditions());
return {
place: (kind, x, y) => lvl.place(kind, x, y),
rules: () => normalised(lvl.rules()),
activeRules: () => normalised(lvl.activeRules()),
hasProperty: (id, prop) => lvl.hasProperty(id, prop),
move: (dir) => lvl.moveYou(dir),
posOf: (id) => {
const o = lvl.board.find(id);
return o === undefined ? null : [o.x, o.y];
},
kindsAt: (x, y) =>
lvl.board
.at(x, y)
.map((o) => puzzle.Kind.format(o.kind))
.sort(),
stackAt: (x, y) =>
lvl.board
.at(x, y)
.map((o) => puzzle.Kind.format(o.kind)),
step: (dir) => lvl.step(dir),
won: () => lvl.won(),
lost: () => lvl.lost(),
undo: () => lvl.undo(),
};
}
// ======== Part III: SpellGrid ========
/**
* A SpellGrid level in play (Part III): a Baba Is You level whose minions are cards.
* Every id is a grid id: a minion's is its grid object's (see `engine/spellgrid.ts`).
*/
export interface SpellGrid extends Baba {
// Module 10
cast(card: string, target?: Id): void; // a spell from player 0's hand, as one Sequence; target: a minion
play(card: string, x: number, y: number): Id; // a minion from player 0's hand onto an empty tile
health(id: Id): number;
hero(player: Player): Id;
nameOf(id: Id): string;
minionAt(x: number, y: number): Id | null;
hand(): string[]; // card names in player 0's hand, in order
log(): GameEvent[]; // Part I's log, with grid ids
clearLog(): void;
logTree(): string; // Part I's log as an indented tree, with grid ids
setChooser(choose: Chooser): void; // makes Assemble's choices
withNaiveAuraTiming(): void; // test hook: Aura Update before the Death Creation Step
// For the command line: why `cast`/`play` would fail, or null if they wouldn't
checkCast(card: string, target?: Id): string | null;
checkPlay(card: string, x: number, y: number): string | null;
}
/** A SpellGrid level; `extra` cards are added to the course's cards, replacing any with the same name. */
export function newSpellGrid(level: puzzle.LevelDef, extra: CardDef[]): SpellGrid {
const library = new CardLibrary(cards());
for (const def of extra) library.register(def);
const g = new Grid(level, library, words(), movement(), hazards(), conditions(), seams());
const lvl = g.level;
/** An engine event with every entity id as a grid id. */
const withIds = (event: GameEvent): GameEvent => {
switch (event.type) {
case "Summoned":
case "Destroyed":
case "Drew":
case "StatsChanged":
return { ...event, id: g.idOf(event.id) };
case "Damaged":
return { ...event, target: g.idOf(event.target) };
case "Triggered":
return { ...event, source: g.idOf(event.source) };
default:
return event;
}
};
return {
place: (kind, x, y) => g.place(kind, x, y),
rules: () => normalised(lvl.rules()),
activeRules: () => normalised(lvl.activeRules()),
hasProperty: (id, prop) => lvl.hasProperty(id, prop),
move: (dir) => g.move(dir),
posOf: (id) => {
const o = lvl.board.find(id);
return o === undefined ? null : [o.x, o.y];
},
kindsAt: (x, y) =>
lvl.board
.at(x, y)
.map((o) => puzzle.Kind.format(o.kind))
.sort(),
stackAt: (x, y) =>
lvl.board
.at(x, y)
.map((o) => puzzle.Kind.format(o.kind)),
step: (dir) => g.step(dir),
won: () => lvl.won(),
lost: () => lvl.lost(),
// Part I has state (health, hands, the log) that the level's snapshot doesn't cover.
undo: () => {
throw new Error("undo is not supported in SpellGrid");
},
cast: (card, target) => g.cast(card, target),
play: (card, x, y) => g.play(card, x, y),
health: (id) => g.world.health(g.entity(id)),
hero: (player) => g.idOf(g.world.hero(player)),
nameOf: (id) => g.world.cardName(g.entity(id)),
minionAt: (x, y) => g.minionAt(x, y) ?? null,
hand: () => g.world.zone(0, "hand").map((id) => g.world.cardName(id)),
log: () =>
g.world
.log()
.flatMap((event) => toGameEvent(event) ?? [])
.map(withIds),
clearLog: () => g.world.clearLog(),
logTree: () =>
logTree(g.world, (event) => {
const shown = toGameEvent(event);
return shown === undefined ? undefined : withIds(shown);
}),
setChooser: (choose) => g.world.setChooser(choose),
// The naive engine reads the auras before the dead leave play: the same steps, reversed.
withNaiveAuraTiming: () => g.world.setEndOfPhase([...END_OF_PHASE].reverse()),
checkCast: (card, target) => g.checkCast(card, target) ?? null,
checkPlay: (card, x, y) => g.checkPlay(card, x, y) ?? null,
};
}
@@ -0,0 +1,70 @@
// The SpellGrid command line, played from a script (the reference's own check; not a course test).
import { readFileSync } from "node:fs";
import { expect, test } from "vitest";
import { run } from "../src/cli";
function play(level: string, input: string[]): { outcome: string; out: string } {
let out = "";
const outcome = run(readFileSync(`levels/${level}.json`, "utf8"), () => input.shift(), (s) => (out += s));
return { outcome, out };
}
test("cli_wins_level_1", () => {
const { outcome, out } = play("l1-double-tap", ["w", "w", "c Whirlwind", "s", "s", "a", "s", "s", "d", "d"]);
expect(outcome).toBe("won");
expect(out).toContain("DEA IS DOU");
expect(out).toContain("You win!");
});
test("cli_wins_level_2_with_choices_and_play", () => {
const input = ["c Graft", "Ember Moth", "Rune Scribe", "bottom", "c Ember Moth (Assembled) 4 3", "a", "a"];
const { outcome, out } = play("l2-spliced-corpse", input);
expect(outcome).toBe("won");
expect(out).toContain("Choose the base");
expect(out).toContain("Hand: Ember Moth (Assembled)");
});
test("cli_prints_the_log_tree_of_each_command", () => {
const { out } = play("l1-double-tap", ["w", "w", "c Whirlwind", "q"]);
// The moves run no Sequence, so only the cast prints a log: its tree, Deathrattles nested.
expect(out.split("Log:\n")).toHaveLength(2);
expect(out).toContain(
[
"Log:",
'SpellCast{card: "Whirlwind"}',
"Damaged{target: 32, amount: 1}",
"Damaged{target: 33, amount: 1}",
"Destroyed{id: 32}",
"DeathPhase{round: 1}",
'Triggered{source: 32, trigger: "deathrattle"}',
" Damaged{target: 33, amount: 1}",
'Triggered{source: 32, trigger: "deathrattle"}',
" Damaged{target: 33, amount: 1}",
"Destroyed{id: 33}",
"DeathPhase{round: 2}",
"",
].join("\n"),
);
});
test("cli_reports_errors_and_quits", () => {
const { outcome, out } = play("l3-simultaneous-paradox", ["c Spark", "x", "q"]);
expect(outcome).toBe("quit");
expect(out).toContain('"Spark" is not in the hand');
expect(out).toContain("Commands:");
});
// The two reference CLIs print the same text for the same input (tools/diff-tracks.mjs checks it).
test("cli_prints_card_names_as_typed", () => {
const { out } = play("l3-simultaneous-paradox", ['c a"b', "q"]);
expect(out).toContain('"a"b" is not in the hand');
});
test("cli_numbers_are_plain_digits", () => {
// "-1" is not a coordinate, so it is part of the card's name.
const cast = play("l1-double-tap", ["c Whirlwind -1 0", "q"]);
expect(cast.out).toContain('"Whirlwind -1 0" is not in the hand');
// "+2" is not a number, so the chooser takes the first option.
const graft = play("l2-spliced-corpse", ["c Graft", "+2", "1", "1", "q"]);
expect(graft.out).toContain("Hand: Clay Golem (Assembled)");
});
@@ -0,0 +1,13 @@
{
"compilerOptions": {
"target": "ES2022",
"module": "ESNext",
"moduleResolution": "Bundler",
"strict": true,
"noEmit": true,
"skipLibCheck": true,
"esModuleInterop": true,
"resolveJsonModule": true
},
"include": ["src", "tests"]
}
@@ -0,0 +1,5 @@
import { defineConfig } from "vitest/config";
export default defineConfig({
test: { include: ["tests/**/*.test.ts"], passWithNoTests: true },
});
+103
View File
@@ -0,0 +1,103 @@
# Module 11 scenarios: Conditions
Each heading is a test name. The Rust test lives in `tests/rust/test_11.rs` and the TypeScript test in `tests/ts/test_11.test.ts`; both state exactly what is written here.
The level format, `place`, `rules()`, `activeRules()`, `hasProperty`, `posOf`, `kindsAt`, `step`, `lost()` and the way a test writes a sentence (`BABA (0,0) IS (1,0) YOU (2,0)` is three text objects) are those of Modules 6 to 9. This module adds no harness function. It adds a word, `ON`, with a new grammatical role, `condition`, and it grows the grammar and the meaning of a rule:
- **Grammar.** `subjects [CONDITION noun {AND noun}] IS complements`. A condition word stands between the subjects and `IS`, followed by one or more nouns joined by `AND`. The condition belongs to every subject of the sentence. Stacked text still yields every possible rule: a tile holding two nouns after `ON` gives two rules, one with each noun.
- **Rule text.** `rules()` and `activeRules()` write a condition between the subject and `IS`, its nouns in the order they are written: `"KEY ON ROCK IS WIN"`, `"KEY ON ROCK AND FLAG IS WIN"`, `"BABA ON KEY IS NOT YOU"`.
- **Meaning.** A rule with a condition applies to an object only while the condition holds for *that object*. `X ON Y IS P`: the X has P while another object that `Y` names shares its tile. With `ON Y AND Z`, the X needs another object for each listed noun, a different one for each: `ON ROCK AND ROCK` needs two rocks. Float levels don't matter to `ON`. A group noun names what it names as a subject: `ON ALL` is satisfied by any other object that is not text.
- **When.** A condition is checked every time a rule is about to be used, against the board as it is then: by `hasProperty`, and so by movement, by each hazard and by the win check, and by conversion. It is not decided when the rules are parsed.
- **Resolution.** A negation cancels a rule in `activeRules()` only if both have the same conditions (`BABA ON KEY IS NOT YOU` cancels `BABA ON KEY IS YOU`). A negation with different conditions doesn't cancel anything there; it is settled for each object when that object is asked about, among the rules whose conditions hold for it. The identity lock (`X IS X`) pairs the same way.
Words used: the nouns `BABA`, `ROCK`, `FLAG`, `LAVA`, `KEY`; the meta-nouns `TEXT` and `ALL`; the properties `YOU`, `WIN`, `PUSH`, `HOT`, `MELT`, `SINK`, `FLOAT`; the operators `IS`, `AND`, `NOT`; the condition `ON`; the base rule `TEXT IS PUSH`.
### test_11_condition_gates_property
- Setup: `newBaba` with an empty level 6 wide and 4 high. Place `KEY (0,0) ON (1,0) ROCK (2,0) IS (3,0) WIN (4,0)`. Place an object `key` at (0,2) (the key on a rock) and an object `rock` at (0,2), and an object `key` at (2,2) (the key alone).
- Assertions:
- `rules()` and `activeRules()` are both exactly `["KEY ON ROCK IS WIN", "TEXT IS PUSH"]`: one rule, with its condition. The sentence doesn't also spell `ROCK IS WIN` or `KEY IS WIN`.
- `hasProperty(onRock, "WIN")` is true.
- `hasProperty(alone, "WIN")` is false: the condition doesn't hold for it.
- `hasProperty(rock, "WIN")` is false: the rock is what the condition is about, not a subject.
### test_11_condition_follows_the_object
- Setup: `newBaba` with an empty level 6 wide and 5 high. Place `BABA (0,0) IS (1,0) YOU (2,0)`, `LAVA (0,1) IS (1,1) HOT (2,1)` and `BABA (0,2) ON (1,2) KEY (2,2) IS (3,2) MELT (4,2)`. Place an object `baba` at (0,4) (it will step onto a key), an object `key` and an object `lava` at (1,4), an object `baba` at (3,4) (it will step onto lava alone) and an object `lava` at (4,4). Record `activeRules()` as `before`.
- Assertions 1: `hasProperty(toKey, "MELT")` and `hasProperty(toLava, "MELT")` are both false: neither Baba is on a key.
- Action: `step("right")`.
- Assertions 2:
- `activeRules()` equals `before`: no text moved, and the rules are the same.
- `posOf(toKey)` is `null`, and `kindsAt(1,4)` is `["key", "lava"]`: the Baba that stepped onto the key was MELT when the hazards asked, and the lava melted it.
- `posOf(toLava)` is (4,4), and `kindsAt(4,4)` is `["baba", "lava"]`: the other Baba stands on lava, but on no key, so it is not MELT.
### test_11_and_requires_both
- Setup: `newBaba` with an empty level 7 wide and 4 high. Place `KEY (0,0) ON (1,0) ROCK (2,0) AND (3,0) FLAG (4,0) IS (5,0) WIN (6,0)`. Place an object `key` with an object `rock` at (0,2) (`withRock`), an object `key` with an object `flag` at (2,2) (`withFlag`), and an object `key` with an object `rock` and an object `flag` at (4,2) (`withBoth`).
- Assertions:
- `rules()` is exactly `["KEY ON ROCK AND FLAG IS WIN", "TEXT IS PUSH"]`: one rule with one condition on two nouns, not two rules.
- `hasProperty(withRock, "WIN")` and `hasProperty(withFlag, "WIN")` are false.
- `hasProperty(withBoth, "WIN")` is true.
### test_11_and_counts_objects
- Setup: `newBaba` with an empty level 7 wide and 4 high. Place `KEY (0,0) ON (1,0) ROCK (2,0) AND (3,0) ROCK (4,0) IS (5,0) WIN (6,0)`. Place an object `key` with one object `rock` at (0,2) (`oneRock`), and an object `key` with two objects `rock` at (2,2) (`twoRocks`).
- Assertions:
- `rules()` is exactly `["KEY ON ROCK AND ROCK IS WIN", "TEXT IS PUSH"]`.
- `hasProperty(oneRock, "WIN")` is false: one rock can't stand for both nouns.
- `hasProperty(twoRocks, "WIN")` is true. The wiki: "AND in the noun2 part requires at least as many objects as are mentioned."
### test_11_conditional_not_only_where_it_holds
- Setup: `newBaba` with an empty level 6 wide and 7 high. Place `BABA (0,0) IS (1,0) YOU (2,0)` and `BABA (0,2) ON (1,2) KEY (2,2) IS (3,2) NOT (4,2) YOU (5,2)`. Place an object `baba` with an object `key` at (0,6) (`onKey`), and an object `baba` at (2,6) (`free`).
- Assertions 1:
- `activeRules()` is exactly `["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]`: the conditions differ, so the resolution pass cancels neither.
- `hasProperty(onKey, "YOU")` is false: for this Baba the negation's condition holds, and the negation wins.
- `hasProperty(free, "YOU")` is true: for this Baba the negation doesn't apply.
- Action 1: place `BABA (0,4) ON (1,4) KEY (2,4) IS (3,4) YOU (4,4)`.
- Assertions 2:
- `rules()` contains `"BABA ON KEY IS YOU"`.
- `activeRules()` is still exactly `["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]`: a negation with the same condition cancels it in the resolution pass.
- Action 2: `step("right")`.
- Assertions 3: `posOf(onKey)` is (0,6), and `posOf(free)` is (3,6): only the Baba that is YOU moved.
### test_11_conditional_conversion
- Setup: `newBaba` with an empty level 6 wide and 5 high. Place `BABA (0,0) IS (1,0) YOU (2,0)`, `ROCK (0,1) IS (1,1) PUSH (2,1)` and `KEY (0,2) ON (1,2) ROCK (2,2) IS (3,2) FLAG (4,2)`. Place an object `key` with an object `rock` at (0,4) (`resting`), an object `baba` at (2,4), an object `rock` at (3,4) (`pushed`), an object `key` at (4,4) (`reached`) and an object `key` at (5,4) (`alone`).
- Action: `step("right")`.
- Assertions:
- `posOf(baba)` is (3,4) and `posOf(pushed)` is (4,4): Baba pushed the rock onto the key at (4,4).
- `kindsAt(0,4)` is `["flag", "rock"]` and `posOf(resting)` is (0,4): the key that was on a rock became a flag where it stood, keeping its id.
- `kindsAt(4,4)` is `["flag", "rock"]` and `posOf(reached)` is (4,4): conversion checked the condition on the board as movement left it, so the key the rock was pushed onto converted in the same turn.
- `kindsAt(5,4)` is `["key"]` and `posOf(alone)` is (5,4): a key on no rock stays a key.
### test_11_condition_and_stacked_text
- Setup: `newBaba` with an empty level 6 wide and 7 high. Place `BABA (0,0) ON (1,0)`, a text `KEY` and a text `ROCK` both at (2,0), then `IS (3,0) WIN (4,0)`. Place `FLAG (0,2)`, a text `ON` and a text `AND` both at (1,2), then `KEY (2,2) IS (3,2) PUSH (4,2)`. Place an object `baba` with an object `rock` at (0,4) (`onRock`), an object `baba` with an object `key` at (2,4) (`onKey`), and an object `baba` at (4,4) (`alone`).
- Assertions:
- `rules()` is exactly `["BABA ON KEY IS WIN", "BABA ON ROCK IS WIN", "FLAG IS PUSH", "FLAG ON KEY IS PUSH", "KEY IS PUSH", "TEXT IS PUSH"]`. Two nouns stacked after `ON` are two readings, one rule each, not `BABA ON KEY AND ROCK`. `ON` stacked on `AND` is read both ways: `FLAG ON KEY IS PUSH`, and `FLAG AND KEY IS PUSH`.
- `hasProperty(onRock, "WIN")` and `hasProperty(onKey, "WIN")` are true.
- `hasProperty(alone, "WIN")` is false.
### test_11_on_group_noun
- Setup: `newBaba` with an empty level 8 wide and 4 high. Place `ROCK (0,0) ON (1,0) ALL (2,0) IS (3,0) WIN (4,0)`. Place an object `rock` alone at (0,2) (`alone`); an object `rock` at (2,2) with a text `text:BABA` on the same tile (`onText`); an object `rock` with an object `key` at (4,2) (`onKey`); and two objects `rock` at (6,2) (`first`, then `second`).
- Assertions:
- `rules()` is exactly `["ROCK ON ALL IS WIN", "TEXT IS PUSH"]`.
- `hasProperty(alone, "WIN")` is false: `ALL` names the rock too, but `ON` needs another object.
- `hasProperty(onText, "WIN")` is false: `ALL` doesn't name text.
- `hasProperty(onKey, "WIN")` is true.
- `hasProperty(first, "WIN")` and `hasProperty(second, "WIN")` are both true: each rock is on the other.
### test_11_condition_rechecked_between_hazards
- Setup: `newBaba` with an empty level 6 wide and 8 high. Place `BABA (0,0) IS (1,0) YOU (2,0)`, `BABA (0,1) AND (1,1) LAVA (2,1) IS (3,1) FLOAT (4,1)`, `LAVA (0,2) IS (1,2) HOT (2,2)`, `KEY (0,3) IS (1,3) SINK (2,3)` and `BABA (0,4) ON (1,4) KEY (2,4) IS (3,4) MELT (4,4)`. Place an object `baba` at (2,6), and an object `lava`, an object `key` and an object `rock` at (3,6), in that order.
- Assertions 1:
- `rules()` contains `"BABA ON KEY IS MELT"`.
- `hasProperty(baba, "MELT")` is false: Baba is not on the key yet.
- Action: `step("right")`. Baba steps onto (3,6). After movement it is on the key, so when the rules are parsed it is MELT. Then the hazards run: SINK first. The key is SINK and shares its tile with the rock, on the same float level (Baba and the lava are FLOAT, and the key and the rock are not), so SINK destroys the key and the rock. Then HOT/MELT asks whether Baba is MELT. It is no longer on a key.
- Assertions 2:
- `posOf(key)` and `posOf(rock)` are `null`.
- `posOf(lava)` is (3,6) and `posOf(baba)` is (3,6), and `kindsAt(3,6)` is `["baba", "lava"]`: Baba survived the HOT lava, because the condition was checked again when HOT/MELT used the rule.
- `lost()` is false.
- An engine that decided each condition when it parsed the rules would have kept Baba MELT until the next parse, and the lava would have destroyed it.
+340
View File
@@ -0,0 +1,340 @@
// Module 11: Conditions. Each test matches a heading in scenarios.md.
use engine::harness::*;
/// An empty level `width` wide and `height` high.
fn empty_level(width: i32, height: i32) -> LevelDef {
LevelDef {
width,
height,
..Default::default()
}
}
/// Place one text object per word: `(word, x, y)`.
fn write(b: &mut Baba, words: &[(&str, i32, i32)]) {
for &(word, x, y) in words {
b.place(&format!("text:{word}"), x, y);
}
}
#[test]
fn test_11_condition_gates_property() {
let mut b = new_baba(empty_level(6, 4));
write(
&mut b,
&[
("KEY", 0, 0),
("ON", 1, 0),
("ROCK", 2, 0),
("IS", 3, 0),
("WIN", 4, 0),
],
);
let on_rock = b.place("key", 0, 2);
let rock = b.place("rock", 0, 2);
let alone = b.place("key", 2, 2);
assert_eq!(b.rules(), vec!["KEY ON ROCK IS WIN", "TEXT IS PUSH"]);
assert_eq!(b.active_rules(), vec!["KEY ON ROCK IS WIN", "TEXT IS PUSH"]);
assert!(b.has_property(on_rock, "WIN"));
assert!(!b.has_property(alone, "WIN"));
assert!(!b.has_property(rock, "WIN"));
}
#[test]
fn test_11_condition_follows_the_object() {
let mut b = new_baba(empty_level(6, 5));
write(&mut b, &[("BABA", 0, 0), ("IS", 1, 0), ("YOU", 2, 0)]);
write(&mut b, &[("LAVA", 0, 1), ("IS", 1, 1), ("HOT", 2, 1)]);
write(
&mut b,
&[
("BABA", 0, 2),
("ON", 1, 2),
("KEY", 2, 2),
("IS", 3, 2),
("MELT", 4, 2),
],
);
let to_key = b.place("baba", 0, 4);
b.place("key", 1, 4);
b.place("lava", 1, 4);
let to_lava = b.place("baba", 3, 4);
b.place("lava", 4, 4);
let before = b.active_rules();
assert!(!b.has_property(to_key, "MELT"));
assert!(!b.has_property(to_lava, "MELT"));
b.step(Dir::Right);
assert_eq!(b.active_rules(), before);
assert_eq!(b.pos_of(to_key), None);
assert_eq!(b.kinds_at(1, 4), vec!["key", "lava"]);
assert_eq!(b.pos_of(to_lava), Some((4, 4)));
assert_eq!(b.kinds_at(4, 4), vec!["baba", "lava"]);
}
#[test]
fn test_11_and_requires_both() {
let mut b = new_baba(empty_level(7, 4));
write(
&mut b,
&[
("KEY", 0, 0),
("ON", 1, 0),
("ROCK", 2, 0),
("AND", 3, 0),
("FLAG", 4, 0),
("IS", 5, 0),
("WIN", 6, 0),
],
);
let with_rock = b.place("key", 0, 2);
b.place("rock", 0, 2);
let with_flag = b.place("key", 2, 2);
b.place("flag", 2, 2);
let with_both = b.place("key", 4, 2);
b.place("rock", 4, 2);
b.place("flag", 4, 2);
assert_eq!(
b.rules(),
vec!["KEY ON ROCK AND FLAG IS WIN", "TEXT IS PUSH"]
);
assert!(!b.has_property(with_rock, "WIN"));
assert!(!b.has_property(with_flag, "WIN"));
assert!(b.has_property(with_both, "WIN"));
}
#[test]
fn test_11_and_counts_objects() {
let mut b = new_baba(empty_level(7, 4));
write(
&mut b,
&[
("KEY", 0, 0),
("ON", 1, 0),
("ROCK", 2, 0),
("AND", 3, 0),
("ROCK", 4, 0),
("IS", 5, 0),
("WIN", 6, 0),
],
);
let one_rock = b.place("key", 0, 2);
b.place("rock", 0, 2);
let two_rocks = b.place("key", 2, 2);
b.place("rock", 2, 2);
b.place("rock", 2, 2);
assert_eq!(
b.rules(),
vec!["KEY ON ROCK AND ROCK IS WIN", "TEXT IS PUSH"]
);
assert!(!b.has_property(one_rock, "WIN"));
assert!(b.has_property(two_rocks, "WIN"));
}
#[test]
fn test_11_conditional_not_only_where_it_holds() {
let mut b = new_baba(empty_level(6, 7));
write(&mut b, &[("BABA", 0, 0), ("IS", 1, 0), ("YOU", 2, 0)]);
write(
&mut b,
&[
("BABA", 0, 2),
("ON", 1, 2),
("KEY", 2, 2),
("IS", 3, 2),
("NOT", 4, 2),
("YOU", 5, 2),
],
);
let on_key = b.place("baba", 0, 6);
b.place("key", 0, 6);
let free = b.place("baba", 2, 6);
assert_eq!(
b.active_rules(),
vec!["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]
);
assert!(!b.has_property(on_key, "YOU"));
assert!(b.has_property(free, "YOU"));
write(
&mut b,
&[
("BABA", 0, 4),
("ON", 1, 4),
("KEY", 2, 4),
("IS", 3, 4),
("YOU", 4, 4),
],
);
assert!(b.rules().contains(&"BABA ON KEY IS YOU".to_string()));
assert_eq!(
b.active_rules(),
vec!["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]
);
b.step(Dir::Right);
assert_eq!(b.pos_of(on_key), Some((0, 6)));
assert_eq!(b.pos_of(free), Some((3, 6)));
}
#[test]
fn test_11_conditional_conversion() {
let mut b = new_baba(empty_level(6, 5));
write(&mut b, &[("BABA", 0, 0), ("IS", 1, 0), ("YOU", 2, 0)]);
write(&mut b, &[("ROCK", 0, 1), ("IS", 1, 1), ("PUSH", 2, 1)]);
write(
&mut b,
&[
("KEY", 0, 2),
("ON", 1, 2),
("ROCK", 2, 2),
("IS", 3, 2),
("FLAG", 4, 2),
],
);
let resting = b.place("key", 0, 4);
b.place("rock", 0, 4);
let baba = b.place("baba", 2, 4);
let pushed = b.place("rock", 3, 4);
let reached = b.place("key", 4, 4);
let alone = b.place("key", 5, 4);
b.step(Dir::Right);
assert_eq!(b.pos_of(baba), Some((3, 4)));
assert_eq!(b.pos_of(pushed), Some((4, 4)));
assert_eq!(b.kinds_at(0, 4), vec!["flag", "rock"]);
assert_eq!(b.pos_of(resting), Some((0, 4)));
assert_eq!(b.kinds_at(4, 4), vec!["flag", "rock"]);
assert_eq!(b.pos_of(reached), Some((4, 4)));
assert_eq!(b.kinds_at(5, 4), vec!["key"]);
assert_eq!(b.pos_of(alone), Some((5, 4)));
}
#[test]
fn test_11_condition_and_stacked_text() {
let mut b = new_baba(empty_level(6, 7));
write(
&mut b,
&[
("BABA", 0, 0),
("ON", 1, 0),
("KEY", 2, 0),
("ROCK", 2, 0),
("IS", 3, 0),
("WIN", 4, 0),
],
);
write(
&mut b,
&[
("FLAG", 0, 2),
("ON", 1, 2),
("AND", 1, 2),
("KEY", 2, 2),
("IS", 3, 2),
("PUSH", 4, 2),
],
);
let on_rock = b.place("baba", 0, 4);
b.place("rock", 0, 4);
let on_key = b.place("baba", 2, 4);
b.place("key", 2, 4);
let alone = b.place("baba", 4, 4);
assert_eq!(
b.rules(),
vec![
"BABA ON KEY IS WIN",
"BABA ON ROCK IS WIN",
"FLAG IS PUSH",
"FLAG ON KEY IS PUSH",
"KEY IS PUSH",
"TEXT IS PUSH"
]
);
assert!(b.has_property(on_rock, "WIN"));
assert!(b.has_property(on_key, "WIN"));
assert!(!b.has_property(alone, "WIN"));
}
#[test]
fn test_11_on_group_noun() {
let mut b = new_baba(empty_level(8, 4));
write(
&mut b,
&[
("ROCK", 0, 0),
("ON", 1, 0),
("ALL", 2, 0),
("IS", 3, 0),
("WIN", 4, 0),
],
);
let alone = b.place("rock", 0, 2);
let on_text = b.place("rock", 2, 2);
b.place("text:BABA", 2, 2);
let on_key = b.place("rock", 4, 2);
b.place("key", 4, 2);
let first = b.place("rock", 6, 2);
let second = b.place("rock", 6, 2);
assert_eq!(b.rules(), vec!["ROCK ON ALL IS WIN", "TEXT IS PUSH"]);
assert!(!b.has_property(alone, "WIN"));
assert!(!b.has_property(on_text, "WIN"));
assert!(b.has_property(on_key, "WIN"));
assert!(b.has_property(first, "WIN"));
assert!(b.has_property(second, "WIN"));
}
#[test]
fn test_11_condition_rechecked_between_hazards() {
let mut b = new_baba(empty_level(6, 8));
write(&mut b, &[("BABA", 0, 0), ("IS", 1, 0), ("YOU", 2, 0)]);
write(
&mut b,
&[
("BABA", 0, 1),
("AND", 1, 1),
("LAVA", 2, 1),
("IS", 3, 1),
("FLOAT", 4, 1),
],
);
write(&mut b, &[("LAVA", 0, 2), ("IS", 1, 2), ("HOT", 2, 2)]);
write(&mut b, &[("KEY", 0, 3), ("IS", 1, 3), ("SINK", 2, 3)]);
write(
&mut b,
&[
("BABA", 0, 4),
("ON", 1, 4),
("KEY", 2, 4),
("IS", 3, 4),
("MELT", 4, 4),
],
);
let baba = b.place("baba", 2, 6);
let lava = b.place("lava", 3, 6);
let key = b.place("key", 3, 6);
let rock = b.place("rock", 3, 6);
assert!(b.rules().contains(&"BABA ON KEY IS MELT".to_string()));
assert!(!b.has_property(baba, "MELT"));
b.step(Dir::Right);
assert_eq!(b.pos_of(key), None);
assert_eq!(b.pos_of(rock), None);
assert_eq!(b.pos_of(lava), Some((3, 6)));
assert_eq!(b.pos_of(baba), Some((3, 6)));
assert_eq!(b.kinds_at(3, 6), vec!["baba", "lava"]);
assert!(!b.lost());
}
@@ -0,0 +1,305 @@
// Module 11: Conditions. Each test matches a heading in scenarios.md.
import { expect, test } from "vitest";
import { newBaba, type Baba } from "../src/harness";
/** An empty level `width` wide and `height` high. */
const emptyLevel = (width: number, height: number) => ({ width, height, objects: [] });
/** Place one text object per word: `[word, x, y]`. */
function write(b: Baba, words: [string, number, number][]): void {
for (const [word, x, y] of words) b.place(`text:${word}`, x, y);
}
test("test_11_condition_gates_property", () => {
const b = newBaba(emptyLevel(6, 4));
write(b, [
["KEY", 0, 0],
["ON", 1, 0],
["ROCK", 2, 0],
["IS", 3, 0],
["WIN", 4, 0],
]);
const onRock = b.place("key", 0, 2);
const rock = b.place("rock", 0, 2);
const alone = b.place("key", 2, 2);
expect(b.rules()).toEqual(["KEY ON ROCK IS WIN", "TEXT IS PUSH"]);
expect(b.activeRules()).toEqual(["KEY ON ROCK IS WIN", "TEXT IS PUSH"]);
expect(b.hasProperty(onRock, "WIN")).toBe(true);
expect(b.hasProperty(alone, "WIN")).toBe(false);
expect(b.hasProperty(rock, "WIN")).toBe(false);
});
test("test_11_condition_follows_the_object", () => {
const b = newBaba(emptyLevel(6, 5));
write(b, [
["BABA", 0, 0],
["IS", 1, 0],
["YOU", 2, 0],
]);
write(b, [
["LAVA", 0, 1],
["IS", 1, 1],
["HOT", 2, 1],
]);
write(b, [
["BABA", 0, 2],
["ON", 1, 2],
["KEY", 2, 2],
["IS", 3, 2],
["MELT", 4, 2],
]);
const toKey = b.place("baba", 0, 4);
b.place("key", 1, 4);
b.place("lava", 1, 4);
const toLava = b.place("baba", 3, 4);
b.place("lava", 4, 4);
const before = b.activeRules();
expect(b.hasProperty(toKey, "MELT")).toBe(false);
expect(b.hasProperty(toLava, "MELT")).toBe(false);
b.step("right");
expect(b.activeRules()).toEqual(before);
expect(b.posOf(toKey)).toBeNull();
expect(b.kindsAt(1, 4)).toEqual(["key", "lava"]);
expect(b.posOf(toLava)).toEqual([4, 4]);
expect(b.kindsAt(4, 4)).toEqual(["baba", "lava"]);
});
test("test_11_and_requires_both", () => {
const b = newBaba(emptyLevel(7, 4));
write(b, [
["KEY", 0, 0],
["ON", 1, 0],
["ROCK", 2, 0],
["AND", 3, 0],
["FLAG", 4, 0],
["IS", 5, 0],
["WIN", 6, 0],
]);
const withRock = b.place("key", 0, 2);
b.place("rock", 0, 2);
const withFlag = b.place("key", 2, 2);
b.place("flag", 2, 2);
const withBoth = b.place("key", 4, 2);
b.place("rock", 4, 2);
b.place("flag", 4, 2);
expect(b.rules()).toEqual(["KEY ON ROCK AND FLAG IS WIN", "TEXT IS PUSH"]);
expect(b.hasProperty(withRock, "WIN")).toBe(false);
expect(b.hasProperty(withFlag, "WIN")).toBe(false);
expect(b.hasProperty(withBoth, "WIN")).toBe(true);
});
test("test_11_and_counts_objects", () => {
const b = newBaba(emptyLevel(7, 4));
write(b, [
["KEY", 0, 0],
["ON", 1, 0],
["ROCK", 2, 0],
["AND", 3, 0],
["ROCK", 4, 0],
["IS", 5, 0],
["WIN", 6, 0],
]);
const oneRock = b.place("key", 0, 2);
b.place("rock", 0, 2);
const twoRocks = b.place("key", 2, 2);
b.place("rock", 2, 2);
b.place("rock", 2, 2);
expect(b.rules()).toEqual(["KEY ON ROCK AND ROCK IS WIN", "TEXT IS PUSH"]);
expect(b.hasProperty(oneRock, "WIN")).toBe(false);
expect(b.hasProperty(twoRocks, "WIN")).toBe(true);
});
test("test_11_conditional_not_only_where_it_holds", () => {
const b = newBaba(emptyLevel(6, 7));
write(b, [
["BABA", 0, 0],
["IS", 1, 0],
["YOU", 2, 0],
]);
write(b, [
["BABA", 0, 2],
["ON", 1, 2],
["KEY", 2, 2],
["IS", 3, 2],
["NOT", 4, 2],
["YOU", 5, 2],
]);
const onKey = b.place("baba", 0, 6);
b.place("key", 0, 6);
const free = b.place("baba", 2, 6);
expect(b.activeRules()).toEqual(["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]);
expect(b.hasProperty(onKey, "YOU")).toBe(false);
expect(b.hasProperty(free, "YOU")).toBe(true);
write(b, [
["BABA", 0, 4],
["ON", 1, 4],
["KEY", 2, 4],
["IS", 3, 4],
["YOU", 4, 4],
]);
expect(b.rules()).toContain("BABA ON KEY IS YOU");
expect(b.activeRules()).toEqual(["BABA IS YOU", "BABA ON KEY IS NOT YOU", "TEXT IS PUSH"]);
b.step("right");
expect(b.posOf(onKey)).toEqual([0, 6]);
expect(b.posOf(free)).toEqual([3, 6]);
});
test("test_11_conditional_conversion", () => {
const b = newBaba(emptyLevel(6, 5));
write(b, [
["BABA", 0, 0],
["IS", 1, 0],
["YOU", 2, 0],
]);
write(b, [
["ROCK", 0, 1],
["IS", 1, 1],
["PUSH", 2, 1],
]);
write(b, [
["KEY", 0, 2],
["ON", 1, 2],
["ROCK", 2, 2],
["IS", 3, 2],
["FLAG", 4, 2],
]);
const resting = b.place("key", 0, 4);
b.place("rock", 0, 4);
const baba = b.place("baba", 2, 4);
const pushed = b.place("rock", 3, 4);
const reached = b.place("key", 4, 4);
const alone = b.place("key", 5, 4);
b.step("right");
expect(b.posOf(baba)).toEqual([3, 4]);
expect(b.posOf(pushed)).toEqual([4, 4]);
expect(b.kindsAt(0, 4)).toEqual(["flag", "rock"]);
expect(b.posOf(resting)).toEqual([0, 4]);
expect(b.kindsAt(4, 4)).toEqual(["flag", "rock"]);
expect(b.posOf(reached)).toEqual([4, 4]);
expect(b.kindsAt(5, 4)).toEqual(["key"]);
expect(b.posOf(alone)).toEqual([5, 4]);
});
test("test_11_condition_and_stacked_text", () => {
const b = newBaba(emptyLevel(6, 7));
write(b, [
["BABA", 0, 0],
["ON", 1, 0],
["KEY", 2, 0],
["ROCK", 2, 0],
["IS", 3, 0],
["WIN", 4, 0],
]);
write(b, [
["FLAG", 0, 2],
["ON", 1, 2],
["AND", 1, 2],
["KEY", 2, 2],
["IS", 3, 2],
["PUSH", 4, 2],
]);
const onRock = b.place("baba", 0, 4);
b.place("rock", 0, 4);
const onKey = b.place("baba", 2, 4);
b.place("key", 2, 4);
const alone = b.place("baba", 4, 4);
expect(b.rules()).toEqual([
"BABA ON KEY IS WIN",
"BABA ON ROCK IS WIN",
"FLAG IS PUSH",
"FLAG ON KEY IS PUSH",
"KEY IS PUSH",
"TEXT IS PUSH",
]);
expect(b.hasProperty(onRock, "WIN")).toBe(true);
expect(b.hasProperty(onKey, "WIN")).toBe(true);
expect(b.hasProperty(alone, "WIN")).toBe(false);
});
test("test_11_on_group_noun", () => {
const b = newBaba(emptyLevel(8, 4));
write(b, [
["ROCK", 0, 0],
["ON", 1, 0],
["ALL", 2, 0],
["IS", 3, 0],
["WIN", 4, 0],
]);
const alone = b.place("rock", 0, 2);
const onText = b.place("rock", 2, 2);
b.place("text:BABA", 2, 2);
const onKey = b.place("rock", 4, 2);
b.place("key", 4, 2);
const first = b.place("rock", 6, 2);
const second = b.place("rock", 6, 2);
expect(b.rules()).toEqual(["ROCK ON ALL IS WIN", "TEXT IS PUSH"]);
expect(b.hasProperty(alone, "WIN")).toBe(false);
expect(b.hasProperty(onText, "WIN")).toBe(false);
expect(b.hasProperty(onKey, "WIN")).toBe(true);
expect(b.hasProperty(first, "WIN")).toBe(true);
expect(b.hasProperty(second, "WIN")).toBe(true);
});
test("test_11_condition_rechecked_between_hazards", () => {
const b = newBaba(emptyLevel(6, 8));
write(b, [
["BABA", 0, 0],
["IS", 1, 0],
["YOU", 2, 0],
]);
write(b, [
["BABA", 0, 1],
["AND", 1, 1],
["LAVA", 2, 1],
["IS", 3, 1],
["FLOAT", 4, 1],
]);
write(b, [
["LAVA", 0, 2],
["IS", 1, 2],
["HOT", 2, 2],
]);
write(b, [
["KEY", 0, 3],
["IS", 1, 3],
["SINK", 2, 3],
]);
write(b, [
["BABA", 0, 4],
["ON", 1, 4],
["KEY", 2, 4],
["IS", 3, 4],
["MELT", 4, 4],
]);
const baba = b.place("baba", 2, 6);
const lava = b.place("lava", 3, 6);
const key = b.place("key", 3, 6);
const rock = b.place("rock", 3, 6);
expect(b.rules()).toContain("BABA ON KEY IS MELT");
expect(b.hasProperty(baba, "MELT")).toBe(false);
b.step("right");
expect(b.posOf(key)).toBeNull();
expect(b.posOf(rock)).toBeNull();
expect(b.posOf(lava)).toEqual([3, 6]);
expect(b.posOf(baba)).toEqual([3, 6]);
expect(b.kindsAt(3, 6)).toEqual(["baba", "lava"]);
expect(b.lost()).toBe(false);
});
+14
View File
@@ -296,6 +296,20 @@
"text": "negates the property or noun after IS",
"source": "https://github.com/utilForever/baba-is-auto/blob/main/Documents/rule-language.md"
},
{
"name": "ON",
"kind": "word",
"role": "condition",
"text": "the rule applies to an object while it shares its tile with an object of each listed noun",
"source": "https://babaiswiki.fandom.com/wiki/ON"
},
{
"name": "NEAR",
"kind": "word",
"role": "condition",
"text": "the rule applies to an object while an object of each listed noun is on its tile or one of the eight around it",
"source": "https://babaiswiki.fandom.com/wiki/NEAR"
},
{
"name": "YOU",
"kind": "word",
+3
View File
@@ -82,6 +82,7 @@ In TypeScript each event is a plain object with a `type` field, for example `{ t
- M7: `hasProperty(id, prop: string): boolean`; `activeRules(): string[]` (after NOT/identity post-pass; same format as `rules()`). From M7, `rules()` also lists negated rules (`"BABA IS NOT YOU"`).
- M8: `move(dir: "up"|"down"|"left"|"right"): void` (movement only, then reparse); `posOf(id): [number, number] | null` (null once destroyed); `kindsAt(x, y): string[]` (sorted); `stackAt(x, y): string[]` (the same kinds, oldest first).
- M9 (requires the M7 Expansion Drop: the meta-noun `ALL` is in the word list): `step(dir): void` (full simplified turn); `won(): boolean`; `lost(): boolean` (no YOU objects remain); `undo(): boolean` (back to before the last `step` or `move`; false if there is none).
- M11: no new function; the contract of `rules()`, `activeRules()`, `hasProperty` and `step` grows. A word with role `condition` (`ON`) may stand between a sentence's subjects and `IS`, followed by nouns joined by `AND`: `KEY ON ROCK AND FLAG IS WIN`. `rules()` and `activeRules()` write the condition between the subject and `IS`, its nouns in the order written. A rule with a condition applies to an object only while the condition holds for that object, checked against the board every time the rule is used (`hasProperty`, and so movement, each hazard and the win check; conversion). Resolution pairs a negation or an identity lock only with rules whose conditions are the same; otherwise negation is settled per object.
Grid coordinates: `x` grows right, `y` grows down; vertical sentences read top-to-bottom.
@@ -141,6 +142,8 @@ pub enum Dir {
From M9, `step(dir)` plays one whole turn, in the course's simplified order: movement (M8's, without its reparse), parse, conversion, parse, hazards, win, and a last parse so that `rules()` matches the board. Conversion: every object named by an active `X IS Y` (`Y` a noun naming its own things) becomes a `Y` in place, keeping its id, unless an active `… IS NOT Y` also names the object (negation is settled per object, as in `hasProperty`); with several `Y`s it becomes the first and each other `Y` is a new object on its tile; `X IS X` blocks it; an object named by `X IS NOT X` is deleted (its other `Y`s still appear, as new objects). All conversions are decided from the board as movement left it. Hazards run in content's order (the reference: `SINK`, `HOT`/`MELT`, `DEFEAT`); each sees only the objects that share a tile on the same float level (both `FLOAT` or both not), and destroys everything it finds at once before the next runs. Hazards and the win check use the rules from the parse after conversion. `won()` becomes `true` when the win check finds a `YOU` object that is `WIN`, or shares its tile with a same-float `WIN` object, and stays `true`. `lost()` is `true` whenever no object is `YOU`. `move(dir)` keeps its M8 meaning. `undo()` puts the level back as it was before the last `step` or `move` and returns `true`; repeated calls walk further back; with nothing to undo it returns `false` and changes nothing. Afterwards every `posOf`, `kindsAt`, `stackAt`, `rules()`, `activeRules()` and `won()` is as it was, and the next new object gets the id it would have got then: a snapshot holds every object's id, kind and position in age order, the id counter and `won` (the rules are recompiled, not saved). `place` clears the history.
From M11, `ON` with nouns `N1 … Nk` holds for an object when the other objects on its tile can supply a different object for each `Ni` that `Ni` names (`ON ROCK AND ROCK` needs two rocks; `ON ALL` is satisfied by any other object that isn't text); float levels don't matter. Two rules have the same conditions when they have the same condition words, each with the same nouns in any order. Stacked text yields every reading: a stack of nouns after `ON` gives one rule per noun, and `ON` stacked on `AND` is read both ways. In SpellGrid, a rule with a condition gives no aura (seam 1).
## Part III (`SpellGrid`)
- M10: `newSpellGrid(level: LevelDef, cards: CardDef[]): SpellGrid` exposing all of Part II (except `undo`, which throws `not supported`: Part I has state the level's snapshot doesn't cover) plus `cast(card, target?)`, `play(card, x, y): Id`, `health(id)`, `hero(player)`, `nameOf(id)`, `minionAt(x, y): Id | null`, `hand(): string[]`, `log()`, `clearLog()`, `logTree()`, `setChooser(choose)`, `won()`, and the test hook `withNaiveAuraTiming()`. Minions are grid objects whose noun is their card's `"noun"` field.
+8
View File
@@ -262,6 +262,14 @@ All five repos exist. I shallow-cloned each on 2026-10-28 and read the files lis
**Searched but not included:** XMage ability architecture (not fetched), the Forge "AbilityFactory/Triggers" subpages (not fetched, since the two pages above were enough), and general ECS-events results (gamedev.net 2013 thread, evenio crate; not read in depth).
### 10d. Wikipedia: Guarded Command Language and Guard (computer science) (fetched 2026-11-11, course v2 Task 8)
- URLs: https://en.wikipedia.org/wiki/Guarded_Command_Language and https://en.wikipedia.org/wiki/Guard_(computer_science), loaded with curl as `index.php?title=<page>&action=raw` (HTTP 200).
- Verbatim quotes (wikitext links removed):
- Guarded Command Language: "A guarded command consists of a boolean condition or guard, and a statement "guarded" by it."
- Guarded Command Language: "Since guards cannot have side effects and the choice of clause is arbitrary, an implementation may evaluate the guards in any sequence and choose the first true clause, for example."
- Guard (computer science): "a guard is a Boolean expression that must evaluate to true if the execution of the program is to continue in the branch in question."
- **Course use:** Module 11's Name it. A Baba Is You condition is a guard on a rule: `BABA ON KEY IS WIN` is the rule `BABA IS WIN` guarded by "on a key", evaluated per object at the moment of use.
---
## Best source per pattern
+15
View File
@@ -120,6 +120,21 @@ All pages were loaded as wikitext via `https://babaiswiki.fandom.com/api.php?act
- **Existence confirmed** via YouTube oEmbed (https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=Jf5O8S5GiOo&format=json): `"title":"Reading the Rules of Baba Is You","author_name":"GDC Festival of Gaming"` (the GDC channel). WebFetch of the watch page also returned the title "Reading the Rules of Baba Is You."
- **Speaker** (confirmed 2026-10-29, Task 17): the Game Developer post [Video: Understanding the rules of Baba Is You](https://www.gamedeveloper.com/design/video-understanding-the-rules-of-i-baba-is-you-i-) (April 14, 2020), loaded with curl, says: "In this GDC 2020 virtual talk Arvi Teikari, developer of the innovative and award-winning puzzle game Baba Is You, describes how the game's unique sentence-based rule system was implemented." The post points to the official GDC YouTube channel; its HTML does not contain the video id, so the match to `Jf5O8S5GiOo` rests on the title and channel. The gdconf.com news page still returns 404. The video itself was not watched, and no claim in this document relies on it. No lesson cites the talk.
### 7. Baba Is You Wiki: conditions (fetched 2026-11-11, course v2 Task 8, for Module 11)
Loaded with curl through the MediaWiki API, as in §5: `https://babaiswiki.fandom.com/api.php?action=parse&page=<NAME>&prop=wikitext&format=json&redirects=1` (HTTP 200 for each). Revision ids from `action=query&prop=revisions`. Templates rendered by hand as before: `{{O|FLAG|ON|BABA|IS|WIN}}` is written FLAG ON BABA IS WIN, `{{ON}}` is ON, `{{O|flag}}` is flag.
| Page | Revision | Key verbatim quote(s) |
|---|---|---|
| [ON](https://babaiswiki.fandom.com/wiki/ON) | 17403 (2025-10-05) | "ON is an infix condition that makes a rule apply when object is overlapping another object." / "FLAG ON BABA IS WIN makes flag WIN only when it's on a baba." / "FLAG ON BABA AND BABA IS WIN makes flag WIN only when it's on two babas." / "FLAG NOT ON BABA IS WIN makes flag WIN only when it's not on a baba." / "ON ignores FLOAT levels: FLAG ON BABA IS WIN makes flag WIN when it's on a baba even if BABA IS FLOAT and FLAG IS NOT FLOAT or vice versa." / "Objects are parsed in order of object priority; earlier on the list goes first." |
| [NEAR](https://babaiswiki.fandom.com/wiki/NEAR) | 18131 (2026-04-11) | "NEAR is an infix condition that makes a rule apply when an object is on a target or in one of the eight spaces around it" / "PILLAR NEAR FLAG IS PINK makes Pillar turn PINK if a Flag is ON the Pillar, directly adjacent, or diagonally adjacent." / "FLAG NEAR BABA AND KEKE IS WIN makes flag WIN if NEAR both a baba and a keke." / "NEAR can be stacked: ROCK NEAR GRASS AND GRASS IS WIN requires at least 2 grass NEAR the Rock for it to be WIN." / "PILLAR NOT NEAR FLAG IS GREEN makes Pillar turn GREEN if there is no Flag ON the Pillar, directly adjacent, or diagonally adjacent." The page says nothing about FLOAT. |
| [Conditions](https://babaiswiki.fandom.com/wiki/Category:Conditions) (redirects to Category:Conditions) | 18040 (2026-02-02) | "Conditions are operators that allow only some of a particular object to be impacted by a rule." / "They are parsed as noun1 condition noun2 verb thing, where noun1 verb thing only if the condition is true." / "Conditions can be negated by prefixing with NOT." / "AND in the noun1 part works like normal. For example, BABA AND KEKE ON SKULL IS DEFEAT expands into BABA ON SKULL IS DEFEAT and KEKE ON SKULL IS DEFEAT" / "AND in the noun2 part requires at least as many objects as are mentioned." / "BABA ON KEKE AND KEKE IS WIN makes a Baba WIN if it is ON two (stacked) Kekes." / "Multiple conditions can share objects between the conditions." / "It is not possible to write something like noun1 condition1 noun2 condition2 noun3 verb thing. This will parse as noun2 condition2 noun3 verb thing." / On matching AND-ed nouns with NOT: "it doesn't backtract to try matching" (sic) and "the game first claims objects for all non-NOT cases, then claims objects for the NOT cases in order with the first object that matches." Prefix conditions (LONELY, IDLE, POWERED, ...) are "parsed as condition noun verb thing". |
| [NOT](https://babaiswiki.fandom.com/wiki/NOT) (re-fetched) | 18314 | "When added after a verb it makes the corresponding rule not apply, overriding other instances of the rule it disables." / "ROCK ON BABA IS NOT PUSH makes ROCK that is on BABA not PUSHable." / "When added before a condition, it reverses the condition." / "FLAG ON NOT BABA IS WIN makes flag WIN when it is ON something other than baba" |
| [Order of Operations](https://babaiswiki.fandom.com/wiki/Order_of_Operations) (re-fetched) | 17942 (2026-01-03) | Same text as §3 for the quotes recorded there. On conditions: "Conditions process when the game checks for what objects have a rule, which is usually right before it uses it." / "it completely finishes moving for a particular move stage before checking which units are in the next, so conditions (e.g. IT ON BABA IS YOU AND MOVE) reparse in between." / Under "Other considerations": "conditions on these rules are rechecked at the time of destruction, even if multiple destructions occur in the same turn." (about HAS). |
- **What the pages do not say.** None of them says in so many words how a conditional `NOT` rule meets an unconditional one (`BABA IS YOU` with `BABA ON KEY IS NOT YOU`). The NOT page says a negation overrides "other instances of the rule it disables", and its example "ROCK ON BABA IS NOT PUSH makes ROCK that is on BABA not PUSHable" names only the rocks on a Baba. Module 11 reads the two together: the negation wins for each object its condition holds for, and only for those. That reading is the course's.
- **Whether an object can satisfy its own condition.** ON says "overlapping another object"; NEAR says "on a target or in one of the eight spaces around it". Neither says whether `ROCK ON ROCK` can be satisfied by the rock itself. The course counts only other objects.
- **Course use (Module 11):** ON ignores float levels (stated). AND in the condition needs one distinct object per listed noun (stated: "at least as many objects as are mentioned"). The course's grammar has one condition per sentence, no `NOT` before a condition, and no prefix conditions; NEAR is the Expansion Drop.
---
## Answers
+8 -1
View File
@@ -84,6 +84,7 @@ export function generateCase(seed, vocab) {
const nouns = vocab.words.filter((w) => w.role === "noun").map((w) => w.name);
const objectNouns = vocab.words.filter((w) => w.role === "noun" && !w.selects).map((w) => w.name);
const properties = vocab.words.filter((w) => w.role === "property").map((w) => w.name);
const conditions = vocab.words.filter((w) => w.role === "condition").map((w) => w.name);
const placeable = vocab.cards.filter((c) => c.type === "minion" && c.noun);
const holdable = vocab.cards.filter((c) => c.type === "spell" || (c.type === "minion" && c.noun));
@@ -106,12 +107,18 @@ export function generateCase(seed, vocab) {
const cardNouns = cards.map((c) => vocab.cards.find((v) => v.name === c.card).noun);
const subjects = [...objectKinds, ...cardNouns];
// Whole sentences, in a row or a column: NOUN [AND NOUN] IS [NOT] (PROPERTY | NOUN) [AND ...].
// Whole sentences, in a row or a column: NOUN [AND NOUN] [CONDITION NOUN [AND NOUN]] IS [NOT]
// (PROPERTY | NOUN) [AND ...]. A condition only with a vocabulary that has condition words
// (Module 11 on), so the cases of earlier modules' vocabularies don't change.
const sentence = (first) => {
// The first is the player's: the first object noun IS YOU (that noun always has an object).
if (first) return [objectKinds[0], "IS", "YOU"];
const line = [chance(0.7) ? pick(subjects) : pick(nouns)];
if (chance(0.15)) line.push("AND", pick(subjects));
if (conditions.length && chance(0.4)) {
line.push(pick(conditions), pick(subjects));
if (chance(0.2)) line.push("AND", pick(nouns));
}
line.push("IS");
if (chance(0.15)) line.push("NOT");
line.push(chance(0.75) ? pick(properties) : pick(nouns));
+20
View File
@@ -89,6 +89,26 @@ test("diff_tracks_levels_form_sentences", () => {
assert.ok(withSentence.length > SEEDS.length * 0.8, `${withSentence.length} of ${SEEDS.length}`);
});
test("diff_tracks_writes_conditions_from_module_11", () => {
// Module 11's vocabulary has a condition word (ON): some sentences carry a condition.
// Module 10's has none, so its cases have no condition word in a sentence.
const vocab11 = loadVocab(join(course, "11-conditions", "reference", "ts", "src", "content", "words.json"), cardDirs("11"));
const conditions = new Set(vocab11.words.filter((w) => w.role === "condition").map((w) => w.name));
assert.ok(conditions.has("ON"));
// A condition word followed by a noun in a row or column, as a sentence writes it.
const withCondition = (v) => SEEDS.filter((seed) => {
const { level } = generateCase(seed, v);
const at = new Map(level.objects.filter((o) => o.kind.startsWith("text:")).map((o) => [`${o.x},${o.y}`, o.kind.slice(5)]));
const noun = (w) => w !== undefined && v.words.some((x) => x.name === w && x.role === "noun");
return [...at].some(([key, word]) => {
const [x, y] = key.split(",").map(Number);
return conditions.has(word) && (noun(at.get(`${x + 1},${y}`)) || noun(at.get(`${x},${y + 1}`)));
});
});
assert.ok(withCondition(vocab11).length > SEEDS.length * 0.3, `${withCondition(vocab11).length} of ${SEEDS.length}`);
assert.deepEqual(withCondition(vocab), []);
});
test("diff_tracks_verdict", () => {
const ok = (out, code = 1) => ({ out, err: "", code, signal: null });
assert.equal(verdict(ok("a\n"), ok("a\n")), "agree");
+1 -1
View File
@@ -20,7 +20,7 @@ const cardEntry = {
required: ["type", "cost"],
properties: { type: { enum: ["minion", "spell"] }, cost: { type: "integer", minimum: 0 }, attack: { type: "integer" }, health: { type: "integer" } },
};
const wordEntry = { type: "object", required: ["role"], properties: { role: { enum: ["noun", "operator", "property", "meta"] } } };
const wordEntry = { type: "object", required: ["role"], properties: { role: { enum: ["noun", "operator", "property", "meta", "condition"] } } };
test("catalog_valid", () => {
const cat = loadCatalog();