Publish rewritten technical article archive

web/scripts/audit-agent-rewrites.mjs

@@ -0,0 +1,58 @@
+import articles from '../data/articles.json' with { type: 'json' };
+import { agentRewriteRevisions } from '../data/agent-rewrite-revisions.mjs';
+
+const archiveBySlug = new Map(articles.map((article) => [article.slug, article]));
+const forbidden = [
+  /synthetic-plan-hand-off/i,
+  /productionEffect/i,
+  /future-only/i,
+  /plan\/scenario/i,
+  /source cutoff/i,
+  /editorial date/i,
+  /planDate/i,
+  /not-collected/i,
+  /not-attempted/i,
+  /future owner/i,
+  /author trajectory/i,
+  /развитие автора/i,
+  /План на \d{4}/i,
+];
+
+function bodyWithoutSources(content) {
+  return content
+    .replace(/<h2>Проверяемые источники<\/h2>[\s\S]*$/, '')
+    .replace(/<[^>]+>/g, ' ')
+    .replace(/&(?:quot|amp|lt|gt|#039);/g, ' ')
+    .replace(/\s+/g, ' ')
+    .trim();
+}
+
+const failures = [];
+const seen = new Set();
+
+for (const revision of agentRewriteRevisions) {
+  const body = bodyWithoutSources(revision.contentHtml || '');
+  const issues = [];
+  if (!archiveBySlug.has(revision.slug)) issues.push('slug отсутствует в архиве');
+  if (seen.has(revision.slug)) issues.push('дубликат slug');
+  seen.add(revision.slug);
+  if (!revision.title || !revision.excerpt) issues.push('нет title/excerpt');
+  if (body.length < 5000 || body.length > 15000) issues.push('основной текст вне 5 000–15 000 знаков');
+  if (!/<figure>[\s\S]*?<img\b[^>]+alt=(?:"[^"<>]+"|'[^'<>]+')[\s\S]*?<\/figure>/.test(revision.contentHtml)) issues.push('нет figure с alt');
+  if (!/<table>[\s\S]*?<\/table>/.test(revision.contentHtml)) issues.push('нет table');
+  if (!/<pre><code>[\s\S]*?<\/code><\/pre>/.test(revision.contentHtml)) issues.push('нет pre/code');
+  if (!/<ol>[\s\S]*?<\/ol>/.test(revision.contentHtml)) issues.push('нет ol');
+  if (!/<h2>Проверяемые источники<\/h2>[\s\S]*?https?:\/\//.test(revision.contentHtml)) issues.push('нет проверяемого источника');
+  const meta = forbidden.find((pattern) => pattern.test(body));
+  if (meta) issues.push('мета-лексика: ' + meta);
+  if (issues.length) failures.push({ index: revision.index, slug: revision.slug, issues });
+}
+
+const report = {
+  drafts: agentRewriteRevisions.length,
+  uniqueSlugs: seen.size,
+  failures,
+  nextExpectedIndex: articles.findIndex((article) => !seen.has(article.slug)) + 1 || null,
+};
+console.log(JSON.stringify(report, null, 2));
+if (failures.length) process.exitCode = 1;

web/scripts/audit-quality-batch.mjs

@@ -82,7 +82,7 @@ for (const slug of slugs) {
 
   const content = cleanReaderHtml(article.contentHtml);
   const body = bodyText(content);
-  const imageSources = [...content.matchAll(/<img[^>]+src="([^"]+)"/g)].map((match) => match[1]);
+  const imageSources = [...content.matchAll(/<img[^>]+\bsrc=(['"])(.*?)\1/gi)].map((match) => match[2]);
   const figures = [...content.matchAll(/<figure>([\s\S]*?)<\/figure>/g)].map((match) => match[1]);
 
   if (article.readingMinutes < 8) issues.push('указано меньше 8 минут чтения');
@@ -98,7 +98,7 @@ for (const slug of slugs) {
   if (count(content, /<table>/g) < 1 || count(content, /<thead>/g) < 1) issues.push('нет доступной таблицы');
   if (count(content, /<pre><code>/g) < 1) issues.push('нет воспроизводимого примера');
   if (count(content, /<ol>/g) < 1) issues.push('нет последовательности проверки или действий');
-  if (count(content, /<a href="https?:\/\//g) < 2) issues.push('меньше двух внешних источников');
+  if (count(content, /<a\b[^>]*\bhref=(['"])https?:\/\/[^'"]+\1/gi) < 2) issues.push('меньше двух внешних источников');
   if (!content.includes('<h2>Проверяемые источники</h2>')) issues.push('нет отдельного раздела с источниками');
   const proseWithoutCode = content
     .replace(/<pre><code>[\s\S]*?<\/code><\/pre>/g, '')
@@ -114,8 +114,8 @@ for (const slug of slugs) {
 
   for (const figure of figures) {
     const image = figure.match(/<img\b[^>]*>/);
-    const alt = image?.[0].match(/\balt="([^"]*)"/);
-    if (!image || !alt || alt[1].trim() === '') {
+    const alt = image?.[0].match(/\balt=(['"])(.*?)\1/i);
+    if (!image || !alt || alt[2].trim() === '') {
       issues.push('у рисунка нет содержательного alt-текста');
       break;
     }