tractatus/scripts/analyze-doc-violations.js

/**
 * Analyze which of the 22 public documents have inst_016/017/018 violations
 */
const { MongoClient } = require('mongodb');

const PUBLIC_SLUGS = [
  // Getting Started (6)
  'introduction', 'core-concepts', 'executive-summary-tractatus-inflection-point',
  'implementation-guide-v1.1', 'implementation-guide', 'implementation-guide-python-examples',

  // Research & Theory (7)
  'tractatus-framework-research', 'pluralistic-values-research-foundations',
  'the-27027-incident-a-case-study-in-pattern-recognition-bias',
  'real-world-ai-governance-a-case-study-in-framework-failure-and-recovery',
  'llm-integration-feasibility-research-scope',
  'research-topic-concurrent-session-architecture',
  'research-topic-rule-proliferation-transactional-overhead',

  // Technical Reference (5)
  'technical-architecture', 'api-reference-complete', 'api-javascript-examples',
  'api-python-examples', 'openapi-specification',

  // Advanced Topics (3)
  'value-pluralism-faq', 'tractatus-ai-safety-framework-core-values-and-principles',
  'organizational-theory-foundations',

  // Business Leadership (1)
  'business-case-tractatus-framework'
];

// Prohibited patterns
const PATTERNS = {
  inst_016: [
    /\b\d+%\b(?!.*\[NEEDS VERIFICATION\])/g,  // Percentages without verification
    /\b\d+\s*(million|billion|thousand)\b(?!.*\[NEEDS VERIFICATION\])/gi
  ],
  inst_017: [
    /\b(guarantee|guarantees|guaranteed|ensuring|ensures)\b/gi,
    /\b(completely safe|totally secure|absolutely)\b/gi,
    /\b(eliminate all|prevent all|never fail)\b/gi
  ],
  inst_018: [
    /\b(production-ready|production ready|battle-tested|proven solution|mature)\b/gi,
    /\b(enterprise-grade|industry-leading|world-class)(?!.*evidence)\b/gi
  ]
};

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

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

  console.log('Analyzing 22 public documents for inst_016/017/018 violations...\n');

  const results = [];

  for (const slug of PUBLIC_SLUGS) {
    const doc = await collection.findOne({ slug });
    if (!doc) {
      console.log(`⚠️  NOT FOUND: ${slug}`);
      continue;
    }

    const content = (doc.content_markdown || '') + ' ' + (doc.content_html || '');
    const violations = { inst_016: 0, inst_017: 0, inst_018: 0 };

    // Check each pattern
    for (const [rule, patterns] of Object.entries(PATTERNS)) {
      for (const pattern of patterns) {
        const matches = content.match(pattern);
        if (matches) {
          violations[rule] += matches.length;
        }
      }
    }

    const totalViolations = violations.inst_016 + violations.inst_017 + violations.inst_018;

    if (totalViolations > 0) {
      results.push({
        slug,
        title: doc.title,
        category: doc.category,
        violations,
        total: totalViolations
      });
    }
  }

  // Sort by total violations
  results.sort((a, b) => b.total - a.total);

  console.log('═══════════════════════════════════════════════════════════');
  console.log('  VIOLATION SUMMARY');
  console.log('═══════════════════════════════════════════════════════════\n');

  let grandTotal = 0;
  for (const r of results) {
    console.log(`${r.title}`);
    console.log(`  Slug: ${r.slug}`);
    console.log(`  Category: ${r.category}`);
    console.log(`  inst_016 (stats): ${r.violations.inst_016}`);
    console.log(`  inst_017 (assurance): ${r.violations.inst_017}`);
    console.log(`  inst_018 (maturity): ${r.violations.inst_018}`);
    console.log(`  TOTAL: ${r.total}\n`);
    grandTotal += r.total;
  }

  console.log('═══════════════════════════════════════════════════════════');
  console.log(`Documents with violations: ${results.length}/${PUBLIC_SLUGS.length}`);
  console.log(`Total violations: ${grandTotal}\n`);

  await client.close();
}

run().catch(console.error);