tractatus/scripts/fix-document-violations.js

/**
 * Fix inst_016/017/018 violations in documents
 *
 * Approach:
 * - inst_016: Add [NEEDS VERIFICATION] to statistics without citations
 * - inst_017: Replace absolute assurance terms with evidence-based language
 * - inst_018: Remove maturity claims or replace with evidence-based terms
 */
const { MongoClient } = require('mongodb');

// Replacements for inst_017 (absolute assurance → evidence-based)
const ASSURANCE_REPLACEMENTS = {
  'guarantees': 'provides strong safeguards for',
  'guaranteed': 'designed to support',
  'guarantee': 'provide strong safeguards for',
  'ensuring': 'supporting',
  'ensures': 'supports',
  'completely safe': 'designed with multiple safety layers',
  'totally secure': 'implements defense-in-depth security',
  'absolutely': 'with high confidence',
  'eliminate all': 'significantly reduce',
  'prevent all': 'mitigate',
  'never fail': 'are resilient and designed to recover from failures'
};

// Replacements for inst_018 (maturity claims → evidence-based)
const MATURITY_REPLACEMENTS = {
  'production-ready': 'under active development',
  'production ready': 'under active development',
  'battle-tested': 'tested in real-world scenarios',
  'proven solution': 'approach validated through',
  'mature': 'established',
  'enterprise-grade': 'designed for organizational use',
  'industry-leading': 'implementing modern approaches to',
  'world-class': 'high-quality'
};

async function fixDocument(collection, slug) {
  const doc = await collection.findOne({ slug });
  if (!doc) {
    console.log(`  ⚠️  NOT FOUND: ${slug}`);
    return { fixed: 0, errors: 1 };
  }

  let markdown = doc.content_markdown || '';
  let html = doc.content_html || '';
  let searchIndex = doc.search_index || '';
  let excerpt = doc.excerpt || '';
  let fixCount = 0;

  // Fix inst_016: Add [NEEDS VERIFICATION] to statistics
  const statPatterns = [
    /(\b\d+%)\b(?!.*\[NEEDS VERIFICATION\])/g,
    /(\b\d+\s*(?:million|billion|thousand))\b(?!.*\[NEEDS VERIFICATION\])/gi
  ];

  for (const pattern of statPatterns) {
    const markdownBefore = markdown;
    markdown = markdown.replace(pattern, (match) => {
      // Skip if already has citation or verification marker
      if (markdown.includes(`${match} [NEEDS VERIFICATION]`)) return match;
      fixCount++;
      return `${match} [NEEDS VERIFICATION]`;
    });

    const htmlBefore = html;
    html = html.replace(pattern, (match) => {
      if (html.includes(`${match} [NEEDS VERIFICATION]`)) return match;
      return `${match} [NEEDS VERIFICATION]`;
    });
  }

  // Fix inst_017: Replace assurance terms (case-insensitive)
  for (const [bad, good] of Object.entries(ASSURANCE_REPLACEMENTS)) {
    const pattern = new RegExp(`\\b${bad}\\b`, 'gi');
    const matches = markdown.match(pattern);
    if (matches) {
      markdown = markdown.replace(pattern, (match) => {
        // Preserve original casing
        if (match[0] === match[0].toUpperCase()) {
          return good.charAt(0).toUpperCase() + good.slice(1);
        }
        return good;
      });
      html = html.replace(pattern, (match) => {
        if (match[0] === match[0].toUpperCase()) {
          return good.charAt(0).toUpperCase() + good.slice(1);
        }
        return good;
      });
      fixCount += matches.length;
    }
  }

  // Fix inst_018: Replace maturity claims
  for (const [bad, good] of Object.entries(MATURITY_REPLACEMENTS)) {
    const pattern = new RegExp(`\\b${bad}\\b`, 'gi');
    const matches = markdown.match(pattern);
    if (matches) {
      markdown = markdown.replace(pattern, (match) => {
        if (match[0] === match[0].toUpperCase()) {
          return good.charAt(0).toUpperCase() + good.slice(1);
        }
        return good;
      });
      html = html.replace(pattern, (match) => {
        if (match[0] === match[0].toUpperCase()) {
          return good.charAt(0).toUpperCase() + good.slice(1);
        }
        return good;
      });
      fixCount += matches.length;
    }
  }

  // Also fix search_index and excerpt
  for (const [bad, good] of Object.entries(ASSURANCE_REPLACEMENTS)) {
    const pattern = new RegExp(`\\b${bad}\\b`, 'gi');
    searchIndex = searchIndex.replace(pattern, (match) => {
      if (match[0] === match[0].toUpperCase()) {
        return good.charAt(0).toUpperCase() + good.slice(1);
      }
      return good;
    });
    excerpt = excerpt.replace(pattern, (match) => {
      if (match[0] === match[0].toUpperCase()) {
        return good.charAt(0).toUpperCase() + good.slice(1);
      }
      return good;
    });
  }

  for (const [bad, good] of Object.entries(MATURITY_REPLACEMENTS)) {
    const pattern = new RegExp(`\\b${bad}\\b`, 'gi');
    searchIndex = searchIndex.replace(pattern, (match) => {
      if (match[0] === match[0].toUpperCase()) {
        return good.charAt(0).toUpperCase() + good.slice(1);
      }
      return good;
    });
    excerpt = excerpt.replace(pattern, (match) => {
      if (match[0] === match[0].toUpperCase()) {
        return good.charAt(0).toUpperCase() + good.slice(1);
      }
      return good;
    });
  }

  // Update document
  await collection.updateOne(
    { slug },
    {
      $set: {
        content_markdown: markdown,
        content_html: html,
        search_index: searchIndex,
        excerpt: excerpt,
        updated_at: new Date()
      }
    }
  );

  return { fixed: fixCount, errors: 0 };
}

async function run() {
  const args = process.argv.slice(2);
  const dryRun = args.includes('--dry-run');

  const DOCUMENTS_TO_FIX = [
    'real-world-ai-governance-a-case-study-in-framework-failure-and-recovery',
    'executive-summary-tractatus-inflection-point',
    'tractatus-framework-research',
    'research-topic-concurrent-session-architecture',
    'introduction',
    'pluralistic-values-research-foundations',
    'core-concepts',
    'research-topic-rule-proliferation-transactional-overhead',
    'business-case-tractatus-framework',
    'implementation-guide-v1.1',
    'the-27027-incident-a-case-study-in-pattern-recognition-bias',
    'llm-integration-feasibility-research-scope',
    'tractatus-ai-safety-framework-core-values-and-principles'
  ];

  const client = new MongoClient('mongodb://localhost:27017');
  await client.connect();

  const db = client.db('tractatus_dev');
  const collection = db.collection('documents');

  console.log('═══════════════════════════════════════════════════════════');
  console.log('  FIXING DOCUMENT VIOLATIONS');
  console.log('═══════════════════════════════════════════════════════════\n');
  console.log(`Mode: ${dryRun ? 'DRY RUN (no changes)' : 'LIVE FIX'}`);
  console.log(`Documents to process: ${DOCUMENTS_TO_FIX.length}\n`);

  const stats = { fixed: 0, errors: 0 };

  for (const slug of DOCUMENTS_TO_FIX) {
    const doc = await collection.findOne({ slug });
    if (!doc) {
      console.log(`⚠️  NOT FOUND: ${slug}\n`);
      stats.errors++;
      continue;
    }

    console.log(`Processing: ${doc.title}`);

    if (dryRun) {
      console.log(`  [DRY RUN] Would fix violations in this document\n`);
    } else {
      const result = await fixDocument(collection, slug);
      stats.fixed += result.fixed;
      stats.errors += result.errors;
      console.log(`  ✓ Fixed ${result.fixed} violations\n`);
    }
  }

  console.log('═══════════════════════════════════════════════════════════');
  console.log('  SUMMARY');
  console.log('═══════════════════════════════════════════════════════════\n');
  console.log(`  Total fixes applied: ${stats.fixed}`);
  console.log(`  Errors: ${stats.errors}\n`);

  if (!dryRun) {
    console.log('  Next steps:');
    console.log('  1. Review changes: node scripts/analyze-doc-violations.js');
    console.log('  2. Regenerate PDFs: node scripts/generate-public-pdfs.js');
    console.log('  3. Export for production: node scripts/export-for-production.js\n');
  }

  await client.close();
}

run().catch(console.error);