Learn_System/backend/scripts/gen-exam-textbook-sections.js

#!/usr/bin/env node
/**
 * gen-exam-textbook-sections.js
 *
 * Regenerates the §-section taxonomy of the grades 5-9 math-family textbooks,
 * used by tag-exam-textbook.js (the exam→textbook classifier).
 *
 * Outputs:
 *   backend/scripts/exam-textbook-sections.json   — machine-readable (the classifier reads this)
 *   plans/exam-textbook-links/taxonomy.md          — human-readable reference
 *
 * Re-run whenever a grade 5-9 algebra/geometry/math chapter gains or renames a §.
 * Note: math-5/6 are engine-rendered (math6_engine.js builds <section id="sec-<p.id>">
 * from window.M6.paras) — their §s are NOT extracted statically here (emitted with
 * engine:'math6' marker); the classifier links them at chapter level.
 *
 * Usage:  node backend/scripts/gen-exam-textbook-sections.js
 */
'use strict';
const fs = require('fs');
const path = require('path');

const DIR = path.join(__dirname, '../../frontend/textbooks');
const OUT_MD   = path.join(__dirname, '../../plans/exam-textbook-links/taxonomy.md');
const OUT_JSON = path.join(__dirname, 'exam-textbook-sections.json');

// chapter slug -> html file (from the textbooks table). Order = teaching order.
const CHAPTERS = [
  ['math-5-ch1', 'math_5_ch1.html'], ['math-5-ch2', 'math_5_ch2.html'], ['math-5-ch3', 'math_5_ch3.html'],
  ['math-6-ch1', 'math_6_ch1.html'], ['math-6-ch2', 'math_6_ch2.html'], ['math-6-ch3', 'math_6_ch3.html'],
  ['math-6-ch4', 'math_6_ch4.html'], ['math-6-ch5', 'math_6_ch5.html'], ['math-6-ch6', 'math_6_ch6.html'],
  ['algebra-7-ch1', 'algebra_7_ch1.html'], ['algebra-7-ch2', 'algebra_7_ch2.html'],
  ['algebra-7-ch3', 'algebra_7_ch3.html'], ['algebra-7-ch4', 'algebra_7_ch4.html'],
  ['geometry-7-ch1', 'geometry_7_ch1.html'], ['geometry-7-ch2', 'geometry_7_ch2.html'],
  ['geometry-7-ch3', 'geometry_7_ch3.html'], ['geometry-7-ch4', 'geometry_7_ch4.html'], ['geometry-7-ch5', 'geometry_7_ch5.html'],
  ['algebra-8-ch1', 'algebra_8.html'], ['algebra-8-ch2', 'algebra_8_ch2.html'], ['algebra-8-ch3', 'algebra_8_ch3.html'],
  ['geometry-8-ch1', 'geometry_8_ch1.html'], ['geometry-8-ch2', 'geometry_8_ch2.html'],
  ['geometry-8-ch3', 'geometry_8_ch3.html'], ['geometry-8-ch4', 'geometry_8_ch4.html'],
  ['algebra-9-ch1', 'algebra_9_ch1.html'], ['algebra-9-ch2', 'algebra_9_ch2.html'],
  ['algebra-9-ch3', 'algebra_9_ch3.html'], ['algebra-9-ch4', 'algebra_9_ch4.html'],
  ['geometry-9-ch1', 'geometry_9_ch1.html'], ['geometry-9-ch2', 'geometry_9_ch2.html'],
  ['geometry-9-ch3', 'geometry_9_ch3.html'], ['geometry-9-ch4', 'geometry_9_ch4.html'],
];

function strip(html) { return String(html).replace(/<[^>]+>/g, '').replace(/\s+/g, ' ').trim(); }

const lines = ['# §-таксономия учебников 5–9 (математика) — эталон для классификатора экзамена math9', ''];
const json = []; // [{book, chapter_slug, subject, grade, para_id, num, title}]
let prevBook = '';
for (const [slug, file] of CHAPTERS) {
  const book = slug.replace(/-ch\d+$/, '');
  const subject = book.replace(/-\d+$/, '');           // math|algebra|geometry
  const grade = Number((book.match(/-(\d+)$/) || [])[1]) || null;
  if (book !== prevBook) { lines.push(`\n## ${book}`); prevBook = book; }
  const p = path.join(DIR, file);
  if (!fs.existsSync(p)) { lines.push(`### ${slug}  (FILE MISSING: ${file})`); continue; }
  const html = fs.readFileSync(p, 'utf8');
  const tm = html.match(/<title>([^<]*)<\/title>/i);
  lines.push(`### ${slug} — ${tm ? strip(tm[1]) : file}`);

  const secRe = /<(?:section|div)\b[^>]*\sid="(sec-(?:p\d+|final\d*|[a-z0-9-]+))"[^>]*>/gi;
  let m; const secs = [];
  while ((m = secRe.exec(html)) !== null) secs.push({ id: m[1], start: m.index });
  if (!secs.length) {
    lines.push(`   (движок math6: статических sec[id] нет; якоря строятся из window.M6.paras → id="sec-<p.id>")`);
    json.push({ book, chapter_slug: slug, subject, grade, engine: 'math6', note: 'paras in window.M6 config; anchors sec-<p.id>' });
    continue;
  }
  for (let i = 0; i < secs.length; i++) {
    const seg = html.slice(secs[i].start, secs[i + 1] ? secs[i + 1].start : secs[i].start + 4000);
    const numM = seg.match(/class="sec-num"[^>]*>([\s\S]*?)<\//i);
    const hM   = seg.match(/class="sec-h"[^>]*>([\s\S]*?)<\//i);
    const paraId = secs[i].id.replace(/^sec-/, '');     // p10 | final3
    const num = numM ? strip(numM[1]) : '';
    const title = hM ? strip(hM[1]) : '';
    lines.push(`   ${secs[i].id.padEnd(12)} ${num ? '['+num+'] ' : ''}${title}`);
    if (/^p\d+$/.test(paraId)) {
      json.push({ book, chapter_slug: slug, subject, grade, para_id: paraId, num, title });
    }
  }
}

fs.mkdirSync(path.dirname(OUT_MD), { recursive: true });
fs.writeFileSync(OUT_MD, lines.join('\n'), 'utf8');
fs.writeFileSync(OUT_JSON, JSON.stringify(json, null, 2), 'utf8');
console.log('Wrote', OUT_MD);
console.log('Wrote', OUT_JSON, '(' + json.length + ' sections)');