tractatus/scripts/audit-section-categories.js

#!/usr/bin/env node

const { MongoClient } = require('mongodb');

(async () => {
  const client = await MongoClient.connect('mongodb://localhost:27017/tractatus_dev');
  const db = client.db();

  const docs = await db.collection('documents')
    .find({visibility: 'public'}, {projection: {title: 1, slug: 1, sections: 1}})
    .sort({order: 1})
    .toArray();

  console.log('═══════════════════════════════════════════════════════════');
  console.log('  SECTION CATEGORY AUDIT - 22 Public Documents');
  console.log('═══════════════════════════════════════════════════════════\n');

  let totalDocs = 0;
  let docsWithSections = 0;
  const categoryStats = {};

  docs.forEach(doc => {
    totalDocs++;

    if (!doc.sections || doc.sections.length === 0) {
      console.log(`${doc.title}:`);
      console.log('  ⚠️  NO SECTIONS (traditional view)\n');
      return;
    }

    docsWithSections++;
    const categories = {};
    doc.sections.forEach(s => {
      const cat = s.category || 'uncategorized';
      categories[cat] = (categories[cat] || 0) + 1;
      categoryStats[cat] = (categoryStats[cat] || 0) + 1;
    });

    console.log(`${doc.title}:`);
    console.log(`  Sections: ${doc.sections.length}`);
    Object.entries(categories).sort((a,b) => b[1] - a[1]).forEach(([cat, count]) => {
      const percent = Math.round(count / doc.sections.length * 100);
      console.log(`  - ${cat}: ${count} (${percent}%)`);
    });
    console.log('');
  });

  console.log('═══════════════════════════════════════════════════════════');
  console.log('  OVERALL STATISTICS');
  console.log('═══════════════════════════════════════════════════════════\n');
  console.log(`Total documents: ${totalDocs}`);
  console.log(`Documents with sections: ${docsWithSections}`);
  console.log(`Documents without sections: ${totalDocs - docsWithSections}\n`);

  console.log('Category distribution across ALL sections:');
  const sortedStats = Object.entries(categoryStats).sort((a,b) => b[1] - a[1]);
  const totalSections = sortedStats.reduce((sum, [,count]) => sum + count, 0);

  sortedStats.forEach(([cat, count]) => {
    const percent = Math.round(count / totalSections * 100);
    console.log(`  - ${cat}: ${count} sections (${percent}%)`);
  });

  console.log('\n');
  await client.close();
})();