tractatus/scripts/migrate-licence-to-cc-by-4.js

#!/usr/bin/env node
/**
 * Migrate Document Licences — Apache 2.0 → CC BY 4.0
 *
 * Updates MongoDB documents: replaces Apache 2.0 licence text in content_html
 * and content_markdown for research papers. Sets the licence field on all documents.
 *
 * Usage:
 *   node scripts/migrate-licence-to-cc-by-4.js [--dry-run] [--db <name>]
 *
 * Defaults to tractatus_dev. Use --db tractatus for production.
 */

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

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

// --- Classification Map ---
// Research papers → CC BY 4.0. Everything else → Apache 2.0.
// Uses partial matching: if any of these strings appear in the slug, it's 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',
]);

function shouldBeCcBy(slug) {
  return CC_BY_SLUGS.has(slug);
}

// --- Replacement strings ---
// We use simple string search-and-replace. More reliable than regex on messy HTML.

const APACHE_STRINGS_TO_FIND = {
  en: [
    // Full licence block text (the body, not the heading)
    'Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License.',
    'Licensed under the Apache License, Version 2.0 (the "License");\nyou may not use this file except in compliance with the License.',
    // Inline metadata variants
    '**License:** Apache License 2.0',
    '<strong>License:</strong> Apache License 2.0',
    '<em>License: Apache License 2.0</em>',
    '*License: Apache License 2.0*',
    // Summary items
    'Apache License, Version 2.0, January 2004',
  ],
  de: [
    // Full German single-line block (as found in glossary-de)
    'Lizenziert unter der Apache License, Version 2.0 (die "Lizenz"); Sie d\u00fcrfen diese Datei nur in \u00dcbereinstimmung mit der Lizenz verwenden.',
    // Shorter variant
    'Lizenziert unter der Apache License, Version 2.0',
    'lizenziert unter der Apache License, Version 2.0',
    // Inline German metadata
    'Apache-Lizenz 2.0',
  ],
  fr: [
    'Sous licence Apache License, Version 2.0',
    'sous licence Apache License, Version 2.0',
    'Licencié sous la Licence Apache, Version 2.0',
    'Licence Apache 2.0',
    // French typography (space before colon)
    'Apache License 2.0</li>',
  ],
  mi: [
    'I raro i te Rāngai Apache, Putanga 2.0',
  ]
};

// What to check AFTER replacement — should not contain these (ignoring the dual-licence note)
function hasStrayApache(text) {
  if (!text) return false;
  // Remove the acceptable dual-licence note (various language forms)
  const cleaned = 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, '')
    // Also acceptable: the framework code reference in any context
    .replace(/source code is separately licensed under the Apache/g, '')
    .replace(/Quellcode.*?Apache License 2\.0/g, '')
    // Māori dual-licence note
    .replace(/kei raro anō i te Apache License 2\.0/g, '');
  return cleaned.includes('Apache License') || cleaned.includes('Apache-Lizenz');
}

async function main() {
  console.log(`\n=== Licence Migration: Apache 2.0 → CC BY 4.0 ===`);
  console.log(`Database: ${DB_NAME}`);
  console.log(`Mode: ${DRY_RUN ? 'DRY RUN' : 'LIVE'}\n`);

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

  try {
    await client.connect();
    const db = client.db(DB_NAME);
    const collection = db.collection('documents');

    const documents = await collection.find({}).toArray();
    console.log(`Found ${documents.length} documents in database\n`);

    let updated = 0;
    let warnings = 0;

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

      const updates = { licence };
      const changes = [];

      if (isCcBy) {
        const ccByHtml = `\n<p>Copyright \u00a9 2026 John Stroh.</p>\n<p>This work is licensed under the <a href="https://creativecommons.org/licenses/by/4.0/">Creative Commons Attribution 4.0 International Licence (CC BY 4.0)</a>.</p>\n<p>You are free to share, copy, redistribute, adapt, remix, transform, and build upon this material for any purpose, including commercially, provided you give appropriate attribution, provide a link to the licence, and indicate if changes were made.</p>\n<p><strong>Note:</strong> The Tractatus AI Safety Framework source code is separately licensed under the Apache License 2.0. This Creative Commons licence applies to the research paper text and figures only.</p>`;
        const ccByMd = `\n\nCopyright \u00a9 2026 John Stroh.\n\nThis work is licensed under the [Creative Commons Attribution 4.0 International Licence (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/).\n\nYou are free to share, copy, redistribute, adapt, remix, transform, and build upon this material for any purpose, including commercially, provided you give appropriate attribution, provide a link to the licence, and indicate if changes were made.\n\n**Note:** The Tractatus AI Safety Framework source code is separately licensed under the Apache License 2.0. This Creative Commons licence applies to the research paper text and figures only.\n`;

        // Process content_html
        if (doc.content_html) {
          let html = doc.content_html;
          let changed = false;
          // Apply ALL language needles (some documents mix languages)
          const allNeedles = [...APACHE_STRINGS_TO_FIND.en, ...APACHE_STRINGS_TO_FIND.de, ...APACHE_STRINGS_TO_FIND.fr, ...APACHE_STRINGS_TO_FIND.mi];
          for (const needle of allNeedles) {
            if (html.includes(needle)) {
              html = html.split(needle).join('');
              changed = true;
            }
          }
          // Replace the heading (may have id= attribute, e.g. <h2 id="license">)
          const licenseHeadingRe = /<h2[^>]*>(?:Document )?License<\/h2>/i;
          if (licenseHeadingRe.test(html)) {
            html = html.replace(licenseHeadingRe, '<h2>Licence</h2>');
            changed = true;
          }
          // Handle German/French headings — use [\s\S]*? to match through inner HTML elements
          const lizenzHeadingRe = /<h2[^>]*>Lizenz[\s\S]*?<\/h2>/i;
          if (lizenzHeadingRe.test(html)) {
            html = html.replace(lizenzHeadingRe, '<h2>Lizenz</h2>');
            changed = true;
          }
          const licenceHeadingRe = /<h2[^>]*>Licence[\s\S]*?<\/h2>/i;
          if (licenceHeadingRe.test(html)) {
            html = html.replace(licenceHeadingRe, '<h2>Licence</h2>');
            changed = true;
          }

          // Check if CC BY 4.0 text already present (from a previous run)
          const alreadyHasCcBy = html.includes('Creative Commons') || html.includes('CC BY 4.0');

          if (!alreadyHasCcBy) {
            if (changed) {
              // Apache text was found and removed — insert CC BY 4.0 after the Licence heading
              const licIdx = html.indexOf('<h2>Licence</h2>');
              const lizIdx = html.indexOf('<h2>Lizenz</h2>');
              const headingIdx = licIdx >= 0 ? licIdx : lizIdx;
              if (headingIdx >= 0) {
                const afterHeading = html.indexOf('</h2>', headingIdx) + 5;
                html = html.substring(0, afterHeading) + ccByHtml + html.substring(afterHeading);
              }
            } else {
              // No Apache text found AND no CC BY text present — append a licence section
              html = html.trimEnd() + '\n<h2>Licence</h2>' + ccByHtml + '\n';
              changed = true;
            }
          }

          if (changed) {
            updates.content_html = html;
            changes.push('content_html');
          }
          // Check for remaining Apache references AFTER all replacements
          if (hasStrayApache(updates.content_html || html)) {
            changes.push('content_html:WARNING_STRAY_APACHE');
            warnings++;
          }
        }

        // Process content_markdown
        if (doc.content_markdown) {
          let md = doc.content_markdown;
          let changed = false;
          const allNeedles = [...APACHE_STRINGS_TO_FIND.en, ...APACHE_STRINGS_TO_FIND.de, ...APACHE_STRINGS_TO_FIND.fr, ...APACHE_STRINGS_TO_FIND.mi];
          for (const needle of allNeedles) {
            if (md.includes(needle)) {
              md = md.split(needle).join('');
              changed = true;
            }
          }
          if (md.includes('## License') || md.includes('## Document License') || md.includes('## Lizenz')) {
            md = md.replace(/## (?:Document )?License/, '## Licence');
            md = md.replace(/## Lizenz(?:\s+Copyright)/, '## Licence\n\nCopyright');
            changed = true;
          }

          // Check if CC BY 4.0 text already present
          const alreadyHasCcBy = md.includes('Creative Commons') || md.includes('CC BY 4.0');

          if (!alreadyHasCcBy) {
            if (changed && md.includes('## Licence')) {
              md = md.replace('## Licence\n', `## Licence${ccByMd}`);
            } else if (!changed) {
              // No Apache text found AND no CC BY text present — append a licence section
              md = md.trimEnd() + '\n\n## Licence' + ccByMd;
              changed = true;
            }
          }

          if (changed) {
            updates.content_markdown = md;
            changes.push('content_markdown');
          }
          // Check AFTER all replacements
          if (hasStrayApache(updates.content_markdown || md)) {
            changes.push('content_markdown:WARNING_STRAY_APACHE');
            warnings++;
          }
        }

        // Process translations
        if (doc.translations) {
          for (const [lang, translation] of Object.entries(doc.translations)) {
            const needles = APACHE_STRINGS_TO_FIND[lang] || APACHE_STRINGS_TO_FIND.en;

            if (translation.content_html) {
              let html = translation.content_html;
              let changed = false;
              // Apply both language-specific and English needles (some translations mix)
              const allNeedles = [...needles, ...APACHE_STRINGS_TO_FIND.en];
              for (const needle of allNeedles) {
                if (html.includes(needle)) {
                  html = html.split(needle).join('');
                  changed = true;
                }
              }
              if (changed) {
                // Replace heading variants
                html = html.replace(/<h2>Lizenz[^<]*<\/h2>/, '<h2>Lizenz</h2>');
                html = html.replace(/<h2>Licence[^<]*<\/h2>/, '<h2>Licence</h2>');
                html = html.replace(/<h2>License<\/h2>/, '<h2>Licence</h2>');
                html = html.replace(/<h2>R\u0101ngai[^<]*<\/h2>/, '<h2>R\u0101ngai</h2>');

                updates[`translations.${lang}.content_html`] = html;
                changes.push(`translations.${lang}.content_html`);
              }
              if (hasStrayApache(html)) {
                changes.push(`translations.${lang}:WARNING_STRAY_APACHE`);
                warnings++;
              }
            }
          }
        }
      }

      // Only log if there are actual changes or it's a CC BY doc
      if (changes.length > 0 || isCcBy) {
        const status = changes.length > 0 ? changes.join(', ') : (isCcBy ? 'already correct or no licence block' : '');
        console.log(`[${slug}] → ${licence} ${status ? '| ' + status : ''}`);
      }

      if (!DRY_RUN && Object.keys(updates).length > 0) {
        await collection.updateOne({ _id: doc._id }, { $set: updates });
        updated++;
      } else if (Object.keys(updates).length > 0) {
        updated++;
      }
    }

    console.log(`\n--- Summary ---`);
    console.log(`Total documents: ${documents.length}`);
    console.log(`Updated: ${updated}`);
    console.log(`Warnings (stray Apache text): ${warnings}`);

    if (warnings > 0) {
      console.log('\nWARNING: Some documents still contain Apache text after replacement.');
      console.log('These may need manual review — the text may be in an unusual format.');
    }

    if (DRY_RUN) {
      console.log('\nRe-run without --dry-run to apply changes.');
    }

  } finally {
    await client.close();
  }
}

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