STTracker/plugins/csv/index.ts

import { createHash } from "node:crypto";
import { parseCsv } from "@sttracker/plugin-sdk/csv";
import { definePlugin } from "@sttracker/plugin-sdk/types";
import type { EngagementFields, MediaType, EntryFields, SourceRecord, WorkHint } from "@sttracker/plugin-sdk/types";
import {
  cleanId, defaultScale, guessMapping, normalizeRating, parseDate, parseMapping, parseMediaType, parseStatus, parseTags, rowsToFieldRows,
  type FieldRow,
} from "./src/mapping.ts";

const str = (v: string | string[] | undefined): string | undefined => (typeof v === "string" && v.trim() !== "" ? v.trim() : undefined);

function jsonRows(text: string): FieldRow[] {
  let v: unknown;
  try {
    v = JSON.parse(text);
  } catch {
    throw new Error("csv plugin: file is not valid JSON");
  }
  if (!Array.isArray(v)) throw new Error("csv plugin: JSON must be an array of objects");
  return v.map((o) => {
    const r: Record<string, string | string[]> = {};
    if (typeof o === "object" && o !== null) {
      for (const [k, x] of Object.entries(o)) {
        if (x == null || x === "") continue;
        r[k] = Array.isArray(x) ? x.map(String) : String(x);
      }
    }
    return r as FieldRow;
  });
}

function* toRecords(rows: FieldRow[], rawRows: unknown[], defaultMedia: MediaType): Generator<SourceRecord> {
  const scale = defaultScale(rows);
  const seen = new Map<string, number>();
  for (const [i, r] of rows.entries()) {
    const title = str(r.title);
    if (!title) continue;
    const mediaType = parseMediaType(str(r.mediaType), defaultMedia);
    const finished = parseDate(str(r.finishedAt));
    const started = parseDate(str(r.startedAt));
    const hash = createHash("sha1").update([title, mediaType, finished?.value ?? ""].join("\u0000")).digest("hex");
    const n = (seen.get(hash) ?? 0) + 1;
    seen.set(hash, n);
    const key = n === 1 ? `row:${hash}` : `row:${hash}#${n}`;

    const externalIds: Record<string, string> = {};
    for (const f of ["isbn13", "imdb", "tmdb_movie", "igdb"] as const) {
      const id = cleanId(r[f]);
      if (id) externalIds[f] = id;
    }
    const author = str(r.author);
    const work: WorkHint = { title, mediaType, externalIds, ...(author ? { creators: [{ name: author }] } : {}) };

    const entry: EntryFields = {};
    const status = parseStatus(str(r.status));
    if (status) entry.status = status;
    const rating = normalizeRating(r, scale);
    if (rating !== null) entry.rating = rating;
    const review = str(r.review);
    if (review) entry.review = review;
    const notes = str(r.notes);
    if (notes) entry.notes = notes;
    const tags = parseTags(r.tags);
    if (tags) entry.tags = tags;
    const raw = rawRows[i];
    yield { kind: "entry", key, raw, work, entry };

    if (finished || started) {
      const engagement: EngagementFields = {};
      if (started) engagement.startedAt = started;
      if (finished) engagement.finishedAt = finished;
      yield { kind: "engagement", key: `${key}:engagement`, raw, work, engagement };
    }
  }
}

export default definePlugin({
  manifest: {
    id: "csv",
    name: "CSV / JSON import",
    version: "0.1.0",
    sdkVersion: 1,
    mediaTypes: ["book", "comic", "manga", "game", "movie", "show", "podcast", "audio_drama", "album", "video", "other"],
    capabilities: ["importFile"],
    config: [
      {
        key: "mapping",
        label: "Column mapping",
        type: "string",
        help: 'JSON { "column": "field" }. Fields: title mediaType status rating ratingScale(5|10|100) review finishedAt startedAt tags isbn13 imdb tmdb_movie igdb notes. Leave empty to guess from header names (Goodreads exports work out of the box). Ignored for .json.',
      },
    ],
  },
  async *importFile(ctx, file) {
    const text = new TextDecoder("utf-8").decode(file.bytes);
    const lower = file.name.toLowerCase();
    if (lower.endsWith(".json")) {
      const clean = text.replace(/^\uFEFF/, "");
      const rows = jsonRows(clean);
      const raw = JSON.parse(clean) as unknown[];
      const bookish = raw.some((o) => typeof o === "object" && o !== null && ("author" in o || "isbn13" in o));
      yield* toRecords(rows, raw, bookish ? "book" : "other");
    } else if (lower.endsWith(".csv")) {
      const table = parseCsv(text);
      const header = table[0] ?? [];
      const mapping = parseMapping(ctx.config.mapping) ?? guessMapping(header);
      const rows = rowsToFieldRows(table, mapping);
      const bookish = header.some((h) => mapping[h] === "author" || mapping[h] === "isbn13");
      yield* toRecords(rows, table.slice(1).map((cells) => Object.fromEntries(header.map((h, i) => [h, cells[i] ?? ""]))), bookish ? "book" : "other");
    } else {
      throw new Error(`csv plugin: unsupported file "${file.name}" (expected .csv or .json)`);
    }
  },
});