tractatus/scripts/migrate-doc-categories.js

#!/usr/bin/env node
/**
 * Migrate documents to new granular category structure
 *
 * New Categories:
 * - getting-started: Introduction, core concepts, quick start, FAQ
 * - technical-reference: API docs, implementation guides, code examples
 * - research-theory: Research papers, theoretical foundations
 * - case-studies: Real-world examples, failure modes, success stories
 * - deployment-operations: Deployment guides, architecture, troubleshooting
 * - business-leadership: Business cases, executive briefs, ROI analysis
 * - downloads-resources: PDFs, sample configs, external links
 */

// Load environment variables
require('dotenv').config();

const { getDb } = require('../src/utils/db.util');

// Slug-to-category mapping (precise, no fuzzy matching)
const slugToCategory = {
  // Getting Started
  'introduction-to-the-tractatus-framework': 'getting-started',
  'core-concepts-of-the-tractatus-framework': 'getting-started',

  // Technical Reference
  'api-reference-complete': 'technical-reference',
  'openapi-specification': 'technical-reference',
  'implementation-guide': 'technical-reference',
  'tractatus-framework-implementation-guide': 'technical-reference',
  'api-python-examples': 'technical-reference',
  'api-javascript-examples': 'technical-reference',
  'technical-architecture-diagram': 'technical-reference',
  'technical-architecture': 'technical-reference',
  'tractatus-agentic-governance-system-glossary-of-terms': 'technical-reference',
  'comparison-matrix-claude-code-claudemd-and-tractatus-framework': 'technical-reference',

  // Research & Theory
  'organizational-theory-foundations-of-the-tractatus-framework': 'research-theory',
  'research-foundations-scholarly-review-and-context': 'research-theory',
  'executive-brief-tractatus-based-llm-architecture-for-ai-safety': 'research-theory',
  'structural-governance-for-agentic-ai-the-tractatus-inflection-point': 'research-theory',
  'tractatus-ai-safety-framework-core-values-and-principles': 'research-theory',
  'architectural-overview-and-research-status': 'research-theory',

  // Advanced Topics
  'value-pluralism-faq': 'advanced-topics',
  'pluralistic-values-research-foundations': 'advanced-topics',
  'pluralistic-values-deliberation-plan-v2': 'advanced-topics',
  'research-scope-feasibility-of-llm-integrated-tractatus-framework': 'advanced-topics',
  'research-topic-rule-proliferation-and-transactional-overhead-in-ai-governance': 'advanced-topics',
  'research-topic-concurrent-session-architecture-limitations-in-claude-code-governance': 'advanced-topics',

  // Case Studies
  'case-studies-real-world-llm-failure-modes': 'case-studies',
  'the-27027-incident-a-case-study-in-pattern-recognition-bias': 'case-studies',
  'our-framework-in-action-detecting-and-correcting-ai-fabrications': 'case-studies',
  'framework-governance-in-action-pre-publication-security-audit': 'case-studies',
  'when-frameworks-fail-and-why-thats-ok': 'case-studies',
  'real-world-ai-governance-a-case-study-in-framework-failure-and-recovery': 'case-studies',

  // Business & Leadership
  'ai-governance-business-case-template-tractatus-framework': 'business-leadership',
  'business-case-for-ai-governance-tractatus-framework': 'business-leadership',
  'executive-summary-tractatus-inflection-point': 'business-leadership',
  'implementation-roadmap-24-month-deployment-plan': 'business-leadership',

  // Archived
  'phase-2': 'archived',
  'phase-3': 'archived',
  'phase-5-poc-session-1-summary': 'archived',
  'phase-5-poc-session-2-summary': 'archived'
};

async function migrateCategories() {
  console.log('🔄 Migrating document categories to new structure...\n');

  const db = await getDb();
  const collection = db.collection('documents');

  // Get all documents
  const documents = await collection.find({}).toArray();
  console.log(`📚 Found ${documents.length} documents\n`);

  let updated = 0;
  let errors = 0;

  for (const doc of documents) {
    try {
      // Use slug-based mapping (precise)
      const newCategory = slugToCategory[doc.slug] || 'technical-reference';

      // Update document if category changed
      if (doc.category !== newCategory) {
        await collection.updateOne(
          { _id: doc._id },
          {
            $set: {
              category: newCategory,
              'metadata.date_updated': new Date()
            }
          }
        );

        console.log(`✅ Updated: "${doc.title}"`);
        console.log(`   Category: ${doc.category || 'none'} → ${newCategory}\n`);
        updated++;
      }
    } catch (error) {
      console.error(`❌ Error updating "${doc.title}":`, error.message);
      errors++;
    }
  }

  console.log('\n' + '='.repeat(60));
  console.log('📊 Migration Summary:');
  console.log('='.repeat(60));
  console.log(`Total documents: ${documents.length}`);
  console.log(`✅ Updated: ${updated}`);
  console.log(`⏭️  Unchanged: ${documents.length - updated - errors}`);
  console.log(`❌ Errors: ${errors}`);
  console.log('='.repeat(60));

  // Show category breakdown
  console.log('\n📂 Category Breakdown:');
  const categoryCounts = await collection.aggregate([
    { $group: { _id: '$category', count: { $sum: 1 } } },
    { $sort: { count: -1 } }
  ]).toArray();

  categoryCounts.forEach(({ _id, count }) => {
    console.log(`   ${_id || 'none'}: ${count}`);
  });

  process.exit(0);
}

// Run migration
migrateCategories().catch(error => {
  console.error('❌ Migration failed:', error);
  process.exit(1);
});