Publish rewritten technical article archive

2026-08-02 22:19:34 +03:00
parent 7c5b19c960
commit 2d914b543f
375 changed files with 2513 additions and 4 deletions
@@ -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;