import fs from 'fs';
import crypto from 'crypto';

const classification = JSON.parse(fs.readFileSync('docs/audit/28-full-repository-classification.json', 'utf8'));

const evidenceEntries = [];

for (const file of classification) {
  if (!file.includedInAudit) continue;

  try {
    const fileBytes = fs.readFileSync(file.path);
    const sha256 = crypto.createHash('sha256').update(fileBytes).digest('hex');
    const textContent = fileBytes.toString('utf8');
    const lines = textContent.split(/\r?\n/);
    const lineCount = lines.length;
    const byteCount = fileBytes.length;

    // Extract imports
    const importMatches = textContent.match(/^import\s+.*?from\s+['"].*?['"];?/gm) || [];
    const imports = importMatches.map(i => i.trim()).slice(0, 15);

    // Extract exports
    const exportMatches = textContent.match(/^export\s+.*?$/gm) || [];
    const exports = exportMatches.map(e => e.trim()).slice(0, 15);

    // Extract top-level symbols (functions, classes, consts)
    const symbolMatches = textContent.match(/export\s+(class|function|const|interface|type)\s+([A-Za-z0-9_]+)/g) || [];
    const topLevelSymbols = symbolMatches.map(s => s.replace(/export\s+(class|function|const|interface|type)\s+/, '')).slice(0, 15);

    // Extract decorators
    const decoratorMatches = textContent.match(/@[A-Za-z0-9_]+\s*\(/g) || [];
    const frameworkDecorators = [...new Set(decoratorMatches.map(d => d.replace(/[\s(]/g, '')))].slice(0, 15);

    // Extract test suites
    const suiteMatches = textContent.match(/(describe|it|test)\s*\(\s*['"](.*?)['"]/g) || [];
    const testSuiteNames = suiteMatches.map(s => s.replace(/(describe|it|test)\s*\(\s*['"]/, '').replace(/['"]$/, '')).slice(0, 15);

    // Extract JSON keys or config keys
    let configurationKeys = [];
    if (file.path.endsWith('.json') || file.path.endsWith('.yml') || file.path.endsWith('.toml')) {
      const keyMatches = textContent.match(/"([^"]+)":/g) || [];
      configurationKeys = [...new Set(keyMatches.map(k => k.replace(/"/g, '').replace(':', '')))].slice(0, 15);
    }

    evidenceEntries.push({
      path: file.path,
      sha256,
      byteCount,
      lineCount,
      language: file.path.endsWith('.ts') || file.path.endsWith('.tsx') ? 'TypeScript' : (file.path.endsWith('.json') ? 'JSON' : 'Plain text'),
      imports,
      exports,
      topLevelSymbols,
      frameworkDecorators,
      testSuiteNames,
      configurationKeys,
      readSucceeded: true,
      readError: null
    });
  } catch (err) {
    evidenceEntries.push({
      path: file.path,
      sha256: null,
      byteCount: 0,
      lineCount: 0,
      language: 'Unknown',
      imports: [],
      exports: [],
      topLevelSymbols: [],
      frameworkDecorators: [],
      testSuiteNames: [],
      configurationKeys: [],
      readSucceeded: false,
      readError: err.message
    });
  }
}

fs.writeFileSync('docs/audit/29-file-content-evidence.json', JSON.stringify(evidenceEntries, null, 2), 'utf8');
console.log(`Generated content evidence for ${evidenceEntries.length} files in 29-file-content-evidence.json`);