tractatus/scripts/validate-licences.js

#!/usr/bin/env node
/**
 * Validate Document Licences — All Delivery Channels
 *
 * Checks MongoDB, HTML downloads, and markdown source files to verify
 * correct licence assignment (CC BY 4.0 for research, Apache 2.0 for code).
 *
 * Usage:
 *   node scripts/validate-licences.js [--db <name>]
 *
 * Defaults to tractatus_dev. Use --db tractatus for production.
 */

const { MongoClient } = require('mongodb');
const fs = require('fs').promises;
const path = require('path');

const dbArg = process.argv.indexOf('--db');
const DB_NAME = dbArg !== -1 ? process.argv[dbArg + 1] : 'tractatus_dev';

// --- Classification: slugs that MUST be CC BY 4.0 ---
const CC_BY_SLUGS = new Set([
  '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',
  'research-topic-concurrent-session-architecture',
  'research-topic-rule-proliferation-transactional-overhead',
  'executive-summary-tractatus-inflection-point',
  'value-pluralism-faq',
  'value-pluralism-in-tractatus-frequently-asked-questions',
  'tractatus-ai-safety-framework-core-values-and-principles',
  'organizational-theory-foundations',
  'glossary',
  'glossary-de',
  'glossary-fr',
  'business-case-tractatus-framework',
  'case-studies',
  'steering-vectors-mechanical-bias-sovereign-ai',
  'steering-vectors-and-mechanical-bias-inference-time-debiasing-for-sovereign-small-language-models',
  'taonga-centred-steering-governance-polycentric-ai',
  'taonga-centred-steering-governance-polycentric-authority-for-sovereign-small-language-models',
  'pattern-bias-from-code-to-conversation',
  'architectural-alignment-academic',
  'philosophical-foundations-village-project',
  'research-timeline',
  'architectural-safeguards-against-llm-hierarchical-dominance-prose',
  'case-studies-real-world-llm-failure-modes-appendix',
]);

// HTML download files that MUST be CC BY 4.0
const CC_BY_HTML_FILES = [
  'steering-vectors-mechanical-bias-sovereign-ai.html',
  'steering-vectors-mechanical-bias-sovereign-ai-de.html',
  'steering-vectors-mechanical-bias-sovereign-ai-fr.html',
  'steering-vectors-mechanical-bias-sovereign-ai-mi.html',
  'taonga-centred-steering-governance-polycentric-ai.html',
  'taonga-centred-steering-governance-polycentric-ai-de.html',
  'taonga-centred-steering-governance-polycentric-ai-fr.html',
  'taonga-centred-steering-governance-polycentric-ai-mi.html',
  'architectural-alignment-academic-de.html',
  'architectural-alignment-academic-fr.html',
  'architectural-alignment-academic-mi.html',
  'philosophical-foundations-village-project-de.html',
  'philosophical-foundations-village-project-fr.html',
  'philosophical-foundations-village-project-mi.html',
];

// Markdown files that MUST be CC BY 4.0
const CC_BY_MARKDOWN_FILES = [
  'docs/markdown/tractatus-framework-research.md',
  'docs/markdown/business-case-tractatus-framework.md',
  'docs/markdown/organizational-theory-foundations.md',
  'docs/markdown/tractatus-ai-safety-framework-core-values-and-principles.md',
  'docs/markdown/GLOSSARY.md',
  'docs/markdown/GLOSSARY-DE.md',
  'docs/markdown/GLOSSARY-FR.md',
  'docs/markdown/case-studies.md',
  'docs/research/pluralistic-values-research-foundations.md',
  'docs/research/executive-summary-tractatus-inflection-point.md',
  'docs/research/rule-proliferation-and-transactional-overhead.md',
  'docs/research/concurrent-session-architecture-limitations.md',
  'docs/research/ARCHITECTURAL-SAFEGUARDS-Against-LLM-Hierarchical-Dominance-Prose.md',
];

// --- Helpers ---

// Acceptable Apache references (dual-licence notes)
function stripAcceptableApache(text) {
  return text
    .replace(/separately licensed under the Apache License 2\.0/g, '')
    .replace(/separat unter der Apache License 2\.0 lizenziert/g, '')
    .replace(/séparément sous la Licence Apache 2\.0/g, '')
    .replace(/Apache License 2\.0\. This Creative Commons/g, '')
    .replace(/Apache License 2\.0\. Diese Creative-Commons/g, '')
    .replace(/Apache License 2\.0\. Cette licence Creative/g, '')
    .replace(/source code is separately licensed under the Apache/g, '')
    .replace(/Quellcode.*?Apache License 2\.0/g, '')
    .replace(/licencié séparément sous la Licence Apache/g, '')
    // Māori dual-licence note
    .replace(/kei raro anō i te Apache License 2\.0/g, '');
}

function hasUnwantedApache(text) {
  if (!text) return false;
  const cleaned = stripAcceptableApache(text);
  return cleaned.includes('Apache License') || cleaned.includes('Apache-Lizenz');
}

function hasCcBy(text) {
  if (!text) return false;
  return text.includes('Creative Commons') || text.includes('CC BY 4.0') || text.includes('CC BY');
}

// --- Channel 1: MongoDB ---

async function validateMongoDB(client) {
  console.log('\n══════════════════════════════════════════');
  console.log('  CHANNEL 1: MongoDB Documents');
  console.log('══════════════════════════════════════════\n');

  const db = client.db(DB_NAME);
  const collection = db.collection('documents');
  const documents = await collection.find({}).toArray();

  console.log(`Found ${documents.length} documents\n`);

  const errors = [];
  let checkedCcBy = 0;
  let checkedApache = 0;

  for (const doc of documents) {
    const slug = doc.slug;
    const isCcBy = CC_BY_SLUGS.has(slug);
    const expectedLicence = isCcBy ? 'CC-BY-4.0' : 'Apache-2.0';

    // Check 1: licence field exists and is correct
    if (!doc.licence) {
      errors.push({ slug, channel: 'mongodb', issue: 'MISSING licence field' });
    } else if (doc.licence !== expectedLicence) {
      errors.push({ slug, channel: 'mongodb', issue: `WRONG licence field: ${doc.licence} (expected ${expectedLicence})` });
    }

    if (isCcBy) {
      checkedCcBy++;

      // Check 2: content_html should NOT have stray Apache
      if (hasUnwantedApache(doc.content_html)) {
        errors.push({ slug, channel: 'mongodb:content_html', issue: 'Contains stray Apache licence text' });
      }

      // Check 3: content_html SHOULD have CC BY
      if (doc.content_html && !hasCcBy(doc.content_html)) {
        errors.push({ slug, channel: 'mongodb:content_html', issue: 'MISSING CC BY 4.0 text' });
      }

      // Check 4: content_markdown should NOT have stray Apache
      if (hasUnwantedApache(doc.content_markdown)) {
        errors.push({ slug, channel: 'mongodb:content_markdown', issue: 'Contains stray Apache licence text' });
      }

      // Check 5: translations
      if (doc.translations) {
        for (const [lang, translation] of Object.entries(doc.translations)) {
          if (hasUnwantedApache(translation.content_html)) {
            errors.push({ slug, channel: `mongodb:translations.${lang}`, issue: 'Contains stray Apache licence text' });
          }
        }
      }
    } else {
      checkedApache++;
    }
  }

  console.log(`  CC BY 4.0 documents checked: ${checkedCcBy}`);
  console.log(`  Apache 2.0 documents checked: ${checkedApache}`);
  console.log(`  Errors found: ${errors.length}`);

  for (const err of errors) {
    console.log(`  ❌ [${err.slug}] ${err.channel}: ${err.issue}`);
  }

  if (errors.length === 0) {
    console.log('  ✓ All MongoDB documents have correct licences');
  }

  return errors;
}

// --- Channel 2: HTML Downloads ---

async function validateHtmlDownloads() {
  console.log('\n══════════════════════════════════════════');
  console.log('  CHANNEL 2: HTML Download Files');
  console.log('══════════════════════════════════════════\n');

  const downloadsDir = path.resolve(__dirname, '..', 'public', 'downloads');
  const errors = [];
  let checked = 0;

  for (const filename of CC_BY_HTML_FILES) {
    const fullPath = path.join(downloadsDir, filename);
    let content;
    try {
      content = await fs.readFile(fullPath, 'utf-8');
    } catch {
      errors.push({ file: filename, issue: 'FILE NOT FOUND' });
      continue;
    }

    checked++;

    // Should NOT have stray Apache
    if (hasUnwantedApache(content)) {
      errors.push({ file: filename, issue: 'Contains stray Apache licence text' });
    }

    // SHOULD have CC BY
    if (!hasCcBy(content)) {
      errors.push({ file: filename, issue: 'MISSING CC BY 4.0 text' });
    }
  }

  console.log(`  CC BY 4.0 HTML files checked: ${checked}`);
  console.log(`  Errors found: ${errors.length}`);

  for (const err of errors) {
    console.log(`  ❌ [${err.file}] ${err.issue}`);
  }

  if (errors.length === 0) {
    console.log('  ✓ All HTML download files have correct licences');
  }

  return errors;
}

// --- Channel 3: Markdown Source Files ---

async function validateMarkdownFiles() {
  console.log('\n══════════════════════════════════════════');
  console.log('  CHANNEL 3: Markdown Source Files');
  console.log('══════════════════════════════════════════\n');

  const projectRoot = path.resolve(__dirname, '..');
  const errors = [];
  let checked = 0;

  for (const relPath of CC_BY_MARKDOWN_FILES) {
    const fullPath = path.join(projectRoot, relPath);
    let content;
    try {
      content = await fs.readFile(fullPath, 'utf-8');
    } catch {
      errors.push({ file: relPath, issue: 'FILE NOT FOUND' });
      continue;
    }

    checked++;

    // Should NOT have stray Apache (outside dual-licence note)
    if (hasUnwantedApache(content)) {
      errors.push({ file: relPath, issue: 'Contains stray Apache licence text' });
    }

    // SHOULD have CC BY
    if (!hasCcBy(content)) {
      errors.push({ file: relPath, issue: 'MISSING CC BY 4.0 text' });
    }
  }

  console.log(`  CC BY 4.0 markdown files checked: ${checked}`);
  console.log(`  Errors found: ${errors.length}`);

  for (const err of errors) {
    console.log(`  ❌ [${err.file}] ${err.issue}`);
  }

  if (errors.length === 0) {
    console.log('  ✓ All markdown source files have correct licences');
  }

  return errors;
}

// --- Main ---

async function main() {
  console.log('╔══════════════════════════════════════════╗');
  console.log('║  Licence Validation — All Channels       ║');
  console.log('╠══════════════════════════════════════════╣');
  console.log(`║  Database: ${DB_NAME.padEnd(29)}║`);
  console.log('╚══════════════════════════════════════════╝');

  const client = new MongoClient('mongodb://localhost:27017');
  let allErrors = [];

  try {
    await client.connect();

    const mongoErrors = await validateMongoDB(client);
    allErrors = allErrors.concat(mongoErrors.map(e => ({ ...e, channel_type: 'mongodb' })));
  } finally {
    await client.close();
  }

  const htmlErrors = await validateHtmlDownloads();
  allErrors = allErrors.concat(htmlErrors.map(e => ({ ...e, channel_type: 'html' })));

  const mdErrors = await validateMarkdownFiles();
  allErrors = allErrors.concat(mdErrors.map(e => ({ ...e, channel_type: 'markdown' })));

  // --- Final Summary ---
  console.log('\n╔══════════════════════════════════════════╗');
  console.log('║  FINAL SUMMARY                           ║');
  console.log('╚══════════════════════════════════════════╝\n');

  if (allErrors.length === 0) {
    console.log('  ✓ ALL CHANNELS PASS — zero licence mismatches\n');
    process.exit(0);
  } else {
    console.log(`  ✗ ${allErrors.length} ERRORS FOUND:\n`);
    for (const err of allErrors) {
      const loc = err.slug || err.file;
      const chan = err.channel || err.channel_type;
      console.log(`    ❌ [${loc}] (${chan}) ${err.issue}`);
    }
    console.log('');
    process.exit(1);
  }
}

main().catch(err => {
  console.error('Fatal error:', err);
  process.exit(1);
});