import { readFile } from 'node:fs/promises';
import { cleanReaderHtml, readerBodyText } from '../lib/editorial-content.mjs';
import { editorialRevisions } from '../data/editorial-revisions.mjs';

const archive = JSON.parse(await readFile(new URL('../data/articles.json', import.meta.url), 'utf8'));
const revisionBySlug = new Map(editorialRevisions.map((revision) => [revision.slug, revision]));
const articles = archive.map((article) => ({ ...article, ...revisionBySlug.get(article.slug) }));
const hardPatterns = [
  /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,
  /В современном мире/i,
  /следует отметить/i,
  /нужно понимать, что/i,
];

function count(content, expression) {
  return (content.match(expression) || []).length;
}

function proseWithoutCode(content) {
  return cleanReaderHtml(content)
    .replace(/<h2>Проверяемые источники<\/h2>[\s\S]*?(?=<h2>|$)/, '')
    .replace(/<table>[\s\S]*?<\/table>/g, '')
    .replace(/<figure>[\s\S]*?<\/figure>/g, '')
    .replace(/<pre[\s\S]*?<\/pre>/g, '')
    .replace(/<ul>[\s\S]*?<\/ul>/g, '')
    .replace(/<ol>[\s\S]*?<\/ol>/g, '')
    .replace(/<\/(?:p|h2|h3)>/g, '. ')
    .replace(/<code>[\s\S]*?<\/code>/g, '')
    .replace(/```[\s\S]*?```/g, ' ')
    .replace(/<[^>]+>/g, ' ')
    .replace(/&(?:quot|amp|lt|gt|#039);/g, ' ')
    .replace(/\s+/g, ' ')
    .trim();
}

const failures = [];
const longSentenceWarnings = [];
let totalBodyCharacters = 0;

for (const article of articles) {
  const content = cleanReaderHtml(article.contentHtml || '');
  const body = readerBodyText(content);
  totalBodyCharacters += body.length;
  const issues = [];
  if (body.length < 5000 || body.length > 15000) issues.push('body вне 5 000–15 000 знаков');
  if (count(content, /<figure>/g) < 1) issues.push('нет рисунка');
  if (count(content, /<table>/g) < 1) issues.push('нет таблицы');
  if (count(content, /<pre><code>/g) < 1) issues.push('нет кода');
  if (count(content, /<ol>/g) < 1) issues.push('нет порядка действий');
  const meta = hardPatterns.find((pattern) => pattern.test(body));
  if (meta) issues.push('внутренняя мета-лексика: ' + meta);
  if (issues.length) failures.push({ slug: article.slug, issues });

  const sentences = proseWithoutCode(content)
    .split(/[.!?]+\s+/)
    .map((sentence) => sentence.trim())
    .filter(Boolean);
  const overlong = sentences
    .map((sentence) => ({ words: sentence.split(/\s+/).length, sample: sentence.slice(0, 180) }))
    .filter((sentence) => sentence.words > 45)
    .sort((left, right) => right.words - left.words);
  if (overlong.length) longSentenceWarnings.push({ slug: article.slug, count: overlong.length, maxWords: overlong[0].words, sample: overlong[0].sample });
}

console.log(JSON.stringify({
  articles: articles.length,
  totalBodyCharacters,
  failures,
  longSentenceWarnings: longSentenceWarnings.sort((left, right) => right.maxWords - left.maxWords).slice(0, 20),
  longSentenceWarningCount: longSentenceWarnings.reduce((sum, item) => sum + item.count, 0),
}, null, 2));

if (failures.length) process.exitCode = 1;