mirror of
https://github.com/alam00000/bentopdf.git
synced 2026-11-05 20:02:47 +01:00
581 lines
17 KiB
JavaScript
581 lines
17 KiB
JavaScript
import fs from 'fs';
|
||
import path from 'path';
|
||
import { JSDOM } from 'jsdom';
|
||
import { fileURLToPath } from 'url';
|
||
import { serializeJsonLd } from './json-ld.mjs';
|
||
import { ENGLISH_ONLY_PAGES, processPage } from './content-segments.mjs';
|
||
import { completeDocsLanguages } from './docs-languages.mjs';
|
||
import { createTranslationChecker } from './page-translation.mjs';
|
||
import { canonicalBasePath } from './indexing-policy.mjs';
|
||
import { applyServerTranslations } from './server-translations.mjs';
|
||
|
||
const __filename = fileURLToPath(import.meta.url);
|
||
const __dirname = path.dirname(__filename);
|
||
|
||
const DIST_DIR = path.resolve(__dirname, '../dist');
|
||
const LOCALES_DIR = path.resolve(__dirname, '../public/locales');
|
||
const CONTENT_DIR = path.resolve(__dirname, '../translations');
|
||
const SITE_URL = (process.env.SITE_URL || 'https://www.bentopdf.com').replace(
|
||
/\/+$/,
|
||
''
|
||
);
|
||
const BASE_PATH = (process.env.BASE_URL || '/').replace(/\/$/, '');
|
||
|
||
const languages = fs.readdirSync(LOCALES_DIR).filter((file) => {
|
||
return fs.statSync(path.join(LOCALES_DIR, file)).isDirectory();
|
||
});
|
||
|
||
const OFFICIAL_DOCS_URL = 'https://www.bentopdf.com/docs';
|
||
|
||
const docsLanguages = completeDocsLanguages(
|
||
path.resolve(__dirname, '../docs'),
|
||
languages.filter((lang) => lang !== 'en'),
|
||
process.env.DOCS_LANGUAGES
|
||
).enabled;
|
||
|
||
const toCamelCase = (str) => {
|
||
return str.replace(/-([a-z])/g, (g) => g[1].toUpperCase());
|
||
};
|
||
|
||
const KEY_MAPPING = {
|
||
index: 'home',
|
||
404: 'notFound',
|
||
};
|
||
|
||
function loadAllTranslations() {
|
||
const translations = {};
|
||
for (const lang of languages) {
|
||
if (lang === 'en') continue;
|
||
const commonPath = path.join(LOCALES_DIR, `${lang}/common.json`);
|
||
const toolsPath = path.join(LOCALES_DIR, `${lang}/tools.json`);
|
||
const contentPath = path.join(CONTENT_DIR, lang, 'content.json');
|
||
translations[lang] = {
|
||
content: fs.existsSync(contentPath)
|
||
? JSON.parse(fs.readFileSync(contentPath, 'utf-8'))
|
||
: {},
|
||
common: fs.existsSync(commonPath)
|
||
? JSON.parse(fs.readFileSync(commonPath, 'utf-8'))
|
||
: {},
|
||
tools: fs.existsSync(toolsPath)
|
||
? JSON.parse(fs.readFileSync(toolsPath, 'utf-8'))
|
||
: {},
|
||
};
|
||
}
|
||
return translations;
|
||
}
|
||
|
||
function loadEnglishTools() {
|
||
const toolsPath = path.join(LOCALES_DIR, 'en/tools.json');
|
||
if (!fs.existsSync(toolsPath)) return {};
|
||
return JSON.parse(fs.readFileSync(toolsPath, 'utf-8'));
|
||
}
|
||
|
||
const ENGLISH_TOOLS = loadEnglishTools();
|
||
|
||
const contentTotals = {};
|
||
|
||
// TODO@ALAM: Let users build only a single language
|
||
function buildUrl(langPrefix, pagePath) {
|
||
const parts = [SITE_URL];
|
||
const canonicalBase = canonicalBasePath();
|
||
if (canonicalBase) parts.push(canonicalBase.replace(/^\//, ''));
|
||
if (langPrefix) parts.push(langPrefix);
|
||
if (pagePath) parts.push(pagePath.replace(/^\//, ''));
|
||
return parts.filter(Boolean).join('/').replace(/\/+$/, '') || SITE_URL;
|
||
}
|
||
|
||
const ORGANIZATION_LD_MARKER = 'data-bentopdf-organization';
|
||
|
||
function injectOrganizationLd(document) {
|
||
if (document.querySelector(`script[${ORGANIZATION_LD_MARKER}]`)) return;
|
||
const existing = document.querySelectorAll(
|
||
'script[type="application/ld+json"]'
|
||
);
|
||
for (const node of existing) {
|
||
try {
|
||
const parsed = JSON.parse(node.textContent || '');
|
||
if (parsed && parsed['@type'] === 'Organization') return;
|
||
} catch {
|
||
continue;
|
||
}
|
||
}
|
||
const data = {
|
||
'@context': 'https://schema.org',
|
||
'@type': 'Organization',
|
||
name: 'BentoPDF',
|
||
url: SITE_URL,
|
||
logo: `${SITE_URL}/images/favicon.svg`,
|
||
sameAs: [
|
||
'https://github.com/alam00000/bentopdf',
|
||
'https://x.com/BentoPDF',
|
||
'https://www.linkedin.com/company/bentopdf/',
|
||
'https://www.instagram.com/thebentopdf/',
|
||
],
|
||
};
|
||
const script = document.createElement('script');
|
||
script.setAttribute('type', 'application/ld+json');
|
||
script.setAttribute(ORGANIZATION_LD_MARKER, '');
|
||
script.textContent = serializeJsonLd(data);
|
||
document.body.appendChild(script);
|
||
}
|
||
|
||
const BREADCRUMB_MARKER = 'data-bentopdf-breadcrumb';
|
||
const BRAND_NAME = process.env.VITE_BRAND_NAME || 'BentoPDF';
|
||
|
||
function buildLocalHomeHref(lang) {
|
||
const langSegment = lang === 'en' ? '/' : `/${lang}`;
|
||
return `${BASE_PATH}${langSegment}`.replace(/\/+/g, '/');
|
||
}
|
||
|
||
function removeStaticBreadcrumbLd(document) {
|
||
for (const script of document.querySelectorAll(
|
||
`script[type="application/ld+json"]:not([${BREADCRUMB_MARKER}])`
|
||
)) {
|
||
try {
|
||
const parsed = JSON.parse(script.textContent || '');
|
||
if (parsed && parsed['@type'] === 'BreadcrumbList') script.remove();
|
||
} catch {
|
||
continue;
|
||
}
|
||
}
|
||
}
|
||
|
||
function injectToolBreadcrumb(
|
||
document,
|
||
lang,
|
||
toolName,
|
||
toolUrl,
|
||
label = 'Breadcrumb'
|
||
) {
|
||
const h1 = document.querySelector('h1[data-i18n^="tools:"]');
|
||
if (!h1) return;
|
||
if (document.querySelector(`[${BREADCRUMB_MARKER}]`)) return;
|
||
removeStaticBreadcrumbLd(document);
|
||
|
||
const homeUrl = buildUrl(lang === 'en' ? '' : lang, '');
|
||
|
||
const nav = document.createElement('nav');
|
||
nav.setAttribute('aria-label', label);
|
||
nav.setAttribute(BREADCRUMB_MARKER, '');
|
||
nav.setAttribute('data-no-i18n', '');
|
||
nav.className = 'text-sm text-gray-400 mb-4';
|
||
|
||
const homeLink = document.createElement('a');
|
||
homeLink.href = buildLocalHomeHref(lang);
|
||
homeLink.className = 'hover:text-indigo-300';
|
||
homeLink.textContent = BRAND_NAME;
|
||
|
||
const sep = document.createElement('span');
|
||
sep.setAttribute('aria-hidden', 'true');
|
||
sep.className = 'mx-2';
|
||
sep.textContent = '›';
|
||
|
||
const current = document.createElement('span');
|
||
current.className = 'text-gray-300';
|
||
current.setAttribute('aria-current', 'page');
|
||
current.textContent = toolName;
|
||
|
||
nav.appendChild(homeLink);
|
||
nav.appendChild(sep);
|
||
nav.appendChild(current);
|
||
|
||
h1.parentNode.insertBefore(nav, h1);
|
||
|
||
const ld = {
|
||
'@context': 'https://schema.org',
|
||
'@type': 'BreadcrumbList',
|
||
itemListElement: [
|
||
{
|
||
'@type': 'ListItem',
|
||
position: 1,
|
||
name: BRAND_NAME,
|
||
item: homeUrl,
|
||
},
|
||
{
|
||
'@type': 'ListItem',
|
||
position: 2,
|
||
name: toolName,
|
||
item: toolUrl,
|
||
},
|
||
],
|
||
};
|
||
|
||
const script = document.createElement('script');
|
||
script.setAttribute('type', 'application/ld+json');
|
||
script.setAttribute(BREADCRUMB_MARKER, '');
|
||
script.textContent = serializeJsonLd(ld);
|
||
document.body.appendChild(script);
|
||
}
|
||
|
||
function resolveToolName(translationKey, langTools) {
|
||
const langEntry = langTools && langTools[translationKey];
|
||
if (langEntry && langEntry.name) return langEntry.name;
|
||
const enEntry = ENGLISH_TOOLS[translationKey];
|
||
return enEntry && enEntry.name ? enEntry.name : null;
|
||
}
|
||
|
||
function translationKeyForFile(filenameNoExt) {
|
||
return KEY_MAPPING[filenameNoExt] || toCamelCase(filenameNoExt);
|
||
}
|
||
|
||
let isPageTranslated = null;
|
||
|
||
function translatedLangsForFile(filenameNoExt, translations) {
|
||
if (!isPageTranslated) {
|
||
const toolsByLang = {};
|
||
for (const lang of languages) {
|
||
toolsByLang[lang] = translations[lang] ? translations[lang].tools : {};
|
||
}
|
||
isPageTranslated = createTranslationChecker(
|
||
toolsByLang,
|
||
translationKeyForFile
|
||
);
|
||
}
|
||
return languages.filter((l) => isPageTranslated(filenameNoExt, l));
|
||
}
|
||
|
||
function processFileForLanguage(
|
||
originalContent,
|
||
file,
|
||
lang,
|
||
translations,
|
||
langDir,
|
||
translatedLangs
|
||
) {
|
||
const filenameNoExt = file.replace('.html', '');
|
||
const translationKey = translationKeyForFile(filenameNoExt);
|
||
const isTranslated = translatedLangs.includes(lang);
|
||
|
||
const { tools } = translations[lang];
|
||
const dom = new JSDOM(originalContent);
|
||
const document = dom.window.document;
|
||
|
||
document.documentElement.lang = lang;
|
||
document.documentElement.dir = ['ar', 'fa'].includes(lang) ? 'rtl' : 'ltr';
|
||
document.querySelector('script[data-language-redirect]')?.remove();
|
||
if (ENGLISH_ONLY_PAGES.has(filenameNoExt)) {
|
||
const englishContent = document.getElementById('app');
|
||
englishContent?.setAttribute('lang', 'en');
|
||
englishContent?.setAttribute('dir', 'ltr');
|
||
}
|
||
|
||
const metaContent = (selector) =>
|
||
document.querySelector(selector)?.getAttribute('content') ?? null;
|
||
const englishMeta = {
|
||
title: document.title,
|
||
description: metaContent('meta[name="description"]'),
|
||
};
|
||
|
||
applyServerTranslations(document, translations[lang]);
|
||
const contentStats = processPage(document, filenameNoExt, {
|
||
translate: (id) => translations[lang].content[id] || null,
|
||
});
|
||
contentTotals[lang] ??= { found: 0, translated: 0, invalid: 0 };
|
||
contentTotals[lang].found += contentStats.found;
|
||
contentTotals[lang].translated += contentStats.translated;
|
||
contentTotals[lang].invalid += contentStats.invalid.length;
|
||
|
||
let title = null;
|
||
let description = null;
|
||
|
||
if (tools[translationKey]) {
|
||
title =
|
||
tools[translationKey].pageTitle ||
|
||
(tools[translationKey].name
|
||
? `${tools[translationKey].name} - BentoPDF`
|
||
: null);
|
||
description = tools[translationKey].subtitle;
|
||
}
|
||
|
||
const titleTranslated = document.title !== englishMeta.title;
|
||
const descriptionTranslated =
|
||
metaContent('meta[name="description"]') !== englishMeta.description;
|
||
|
||
if (title && !titleTranslated) {
|
||
document.title = title;
|
||
const metaTitle = document.querySelector('meta[property="og:title"]');
|
||
if (metaTitle) metaTitle.content = title;
|
||
const metaTwitterTitle = document.querySelector(
|
||
'meta[name="twitter:title"]'
|
||
);
|
||
if (metaTwitterTitle) metaTwitterTitle.content = title;
|
||
}
|
||
|
||
if (description && !descriptionTranslated) {
|
||
const metaDesc = document.querySelector('meta[name="description"]');
|
||
if (metaDesc) metaDesc.content = description;
|
||
const metaOgDesc = document.querySelector(
|
||
'meta[property="og:description"]'
|
||
);
|
||
if (metaOgDesc) metaOgDesc.content = description;
|
||
const metaTwitterDesc = document.querySelector(
|
||
'meta[name="twitter:description"]'
|
||
);
|
||
if (metaTwitterDesc) metaTwitterDesc.content = description;
|
||
}
|
||
|
||
document
|
||
.querySelectorAll('link[rel="alternate"][hreflang]')
|
||
.forEach((el) => el.remove());
|
||
|
||
const pagePath = filenameNoExt === 'index' ? '' : filenameNoExt;
|
||
|
||
if (isTranslated) {
|
||
translatedLangs.forEach((l) => {
|
||
const link = document.createElement('link');
|
||
link.rel = 'alternate';
|
||
link.hreflang = l;
|
||
link.href = buildUrl(l === 'en' ? '' : l, pagePath);
|
||
document.head.appendChild(link);
|
||
});
|
||
|
||
const defaultLink = document.createElement('link');
|
||
defaultLink.rel = 'alternate';
|
||
defaultLink.hreflang = 'x-default';
|
||
defaultLink.href = buildUrl('', pagePath);
|
||
document.head.appendChild(defaultLink);
|
||
}
|
||
|
||
const englishUrl = buildUrl('', pagePath);
|
||
const localizedUrl = buildUrl(lang, pagePath);
|
||
const canonicalUrl = isTranslated ? localizedUrl : englishUrl;
|
||
let canonical = document.querySelector('link[rel="canonical"]');
|
||
if (!canonical) {
|
||
canonical = document.createElement('link');
|
||
canonical.rel = 'canonical';
|
||
document.head.appendChild(canonical);
|
||
}
|
||
canonical.href = canonicalUrl;
|
||
|
||
const ogUrl = document.querySelector('meta[property="og:url"]');
|
||
if (ogUrl) ogUrl.content = canonicalUrl;
|
||
const twitterUrl = document.querySelector('meta[name="twitter:url"]');
|
||
if (twitterUrl) twitterUrl.content = canonicalUrl;
|
||
|
||
injectOrganizationLd(document);
|
||
|
||
const localizedToolName = resolveToolName(translationKey, tools);
|
||
if (localizedToolName) {
|
||
injectToolBreadcrumb(
|
||
document,
|
||
lang,
|
||
localizedToolName,
|
||
localizedUrl,
|
||
translations[lang].common?.nav?.breadcrumb || 'Breadcrumb'
|
||
);
|
||
}
|
||
|
||
if (docsLanguages.includes(lang)) {
|
||
const localDocs = `${BASE_PATH}/docs`;
|
||
document.querySelectorAll('a[href]').forEach((link) => {
|
||
const href = link.getAttribute('href');
|
||
for (const docsUrl of [OFFICIAL_DOCS_URL, localDocs]) {
|
||
if (href === docsUrl || href?.startsWith(`${docsUrl}/`)) {
|
||
link.setAttribute(
|
||
'href',
|
||
href.replace(docsUrl, `${docsUrl}/${lang}`)
|
||
);
|
||
break;
|
||
}
|
||
}
|
||
});
|
||
}
|
||
|
||
const links = document.querySelectorAll('a[href]');
|
||
links.forEach((link) => {
|
||
const href = link.getAttribute('href');
|
||
if (!href) return;
|
||
|
||
if (
|
||
href.startsWith('http') ||
|
||
href.startsWith('//') ||
|
||
href.startsWith('#') ||
|
||
href.startsWith('mailto:') ||
|
||
href.startsWith('tel:') ||
|
||
href.startsWith('javascript:') ||
|
||
href.startsWith('data:') ||
|
||
href.startsWith('vbscript:')
|
||
) {
|
||
return;
|
||
}
|
||
|
||
if (href.startsWith('/assets/') || href.includes('/assets/')) return;
|
||
|
||
const hrefWithoutBase =
|
||
BASE_PATH && href.startsWith(`${BASE_PATH}/`)
|
||
? href.slice(BASE_PATH.length)
|
||
: href;
|
||
if (/^\/(blog|docs)(\/|$)/.test(hrefWithoutBase)) {
|
||
return;
|
||
}
|
||
|
||
const langPrefixRegex = new RegExp(
|
||
`^(${BASE_PATH})?/(${languages.join('|')})(/|$)`
|
||
);
|
||
if (langPrefixRegex.test(href)) return;
|
||
|
||
let newHref;
|
||
if (href.startsWith('/')) {
|
||
const pathWithoutBase = href.startsWith(BASE_PATH)
|
||
? href.slice(BASE_PATH.length)
|
||
: href;
|
||
newHref = `${BASE_PATH}/${lang}${pathWithoutBase}`;
|
||
} else {
|
||
newHref = `${BASE_PATH}/${lang}/${href}`;
|
||
}
|
||
|
||
link.setAttribute(
|
||
'href',
|
||
newHref.replace(/^([^?#]*[^/?#])\/(?=$|[?#])/, '$1')
|
||
);
|
||
});
|
||
|
||
const result = dom.serialize();
|
||
|
||
dom.window.close();
|
||
|
||
fs.writeFileSync(path.join(langDir, file), result);
|
||
}
|
||
|
||
function updateEnglishFile(filePath, originalContent, translatedLangs) {
|
||
const filenameNoExt = path.basename(filePath, '.html');
|
||
const dom = new JSDOM(originalContent);
|
||
const document = dom.window.document;
|
||
|
||
document
|
||
.querySelectorAll('link[rel="alternate"][hreflang]')
|
||
.forEach((el) => el.remove());
|
||
|
||
const pagePath = filenameNoExt === 'index' ? '' : filenameNoExt;
|
||
const canonicalUrl = buildUrl('', pagePath);
|
||
|
||
translatedLangs.forEach((l) => {
|
||
const link = document.createElement('link');
|
||
link.rel = 'alternate';
|
||
link.hreflang = l;
|
||
link.href = buildUrl(l === 'en' ? '' : l, pagePath);
|
||
document.head.appendChild(link);
|
||
});
|
||
|
||
const defaultLink = document.createElement('link');
|
||
defaultLink.rel = 'alternate';
|
||
defaultLink.hreflang = 'x-default';
|
||
defaultLink.href = canonicalUrl;
|
||
document.head.appendChild(defaultLink);
|
||
|
||
let canonical = document.querySelector('link[rel="canonical"]');
|
||
if (!canonical) {
|
||
canonical = document.createElement('link');
|
||
canonical.rel = 'canonical';
|
||
document.head.appendChild(canonical);
|
||
}
|
||
canonical.href = canonicalUrl;
|
||
|
||
const ogUrl = document.querySelector('meta[property="og:url"]');
|
||
if (ogUrl) ogUrl.content = canonicalUrl;
|
||
const twitterUrl = document.querySelector('meta[name="twitter:url"]');
|
||
if (twitterUrl) twitterUrl.content = canonicalUrl;
|
||
|
||
injectOrganizationLd(document);
|
||
|
||
const enTranslationKey =
|
||
KEY_MAPPING[filenameNoExt] || toCamelCase(filenameNoExt);
|
||
const enToolName = resolveToolName(enTranslationKey, ENGLISH_TOOLS);
|
||
if (enToolName) {
|
||
injectToolBreadcrumb(document, 'en', enToolName, canonicalUrl);
|
||
}
|
||
|
||
const result = dom.serialize();
|
||
|
||
dom.window.close();
|
||
|
||
fs.writeFileSync(filePath, result);
|
||
}
|
||
|
||
async function generateI18nPages() {
|
||
console.log('🌍 Generating i18n pages...');
|
||
console.log(` SITE_URL: ${SITE_URL}`);
|
||
console.log(` BASE_PATH: ${BASE_PATH || '/'}`);
|
||
console.log(` Languages: ${languages.length} (${languages.join(', ')})`);
|
||
|
||
if (!fs.existsSync(DIST_DIR)) {
|
||
console.error('❌ dist directory not found. Please run build first.');
|
||
process.exit(1);
|
||
}
|
||
|
||
console.log(' Loading translations...');
|
||
const translations = loadAllTranslations();
|
||
|
||
const htmlFiles = fs
|
||
.readdirSync(DIST_DIR)
|
||
.filter((file) => file.endsWith('.html'));
|
||
|
||
console.log(` Processing ${htmlFiles.length} HTML files...`);
|
||
|
||
for (const lang of languages) {
|
||
if (lang === 'en') continue;
|
||
const langDir = path.join(DIST_DIR, lang);
|
||
if (!fs.existsSync(langDir)) {
|
||
fs.mkdirSync(langDir, { recursive: true });
|
||
}
|
||
}
|
||
|
||
let processed = 0;
|
||
const total = htmlFiles.length * (languages.length - 1);
|
||
|
||
for (const file of htmlFiles) {
|
||
const filePath = path.join(DIST_DIR, file);
|
||
const originalContent = fs.readFileSync(filePath, 'utf-8');
|
||
const translatedLangs = translatedLangsForFile(
|
||
file.replace('.html', ''),
|
||
translations
|
||
);
|
||
|
||
for (const lang of languages) {
|
||
if (lang === 'en') continue;
|
||
|
||
const langDir = path.join(DIST_DIR, lang);
|
||
|
||
processFileForLanguage(
|
||
originalContent,
|
||
file,
|
||
lang,
|
||
translations,
|
||
langDir,
|
||
translatedLangs
|
||
);
|
||
|
||
processed++;
|
||
if (processed % 10 === 0 || processed === total) {
|
||
console.log(` Progress: ${processed}/${total} pages`);
|
||
}
|
||
|
||
// Clean up JSDOM instances
|
||
await new Promise((resolve) => setImmediate(resolve));
|
||
}
|
||
|
||
updateEnglishFile(filePath, originalContent, translatedLangs);
|
||
}
|
||
|
||
const covered = Object.entries(contentTotals).filter(
|
||
([, totals]) => totals.translated > 0
|
||
);
|
||
if (covered.length > 0) {
|
||
console.log(
|
||
` Page content translated: ${covered
|
||
.map(
|
||
([lang, totals]) =>
|
||
`${lang} ${Math.round((totals.translated / totals.found) * 100)}%${totals.invalid ? ` (${totals.invalid} rejected)` : ''}`
|
||
)
|
||
.join(', ')}`
|
||
);
|
||
}
|
||
|
||
console.log('✅ i18n pages generated successfully!');
|
||
}
|
||
|
||
generateI18nPages().catch((err) => {
|
||
console.error('❌ i18n page generation failed:', err);
|
||
process.exit(1);
|
||
});
|