Function main

scripts/docs-validation/extract.ts:428–544

()

}

async function main() {
  console.log("📖 Extracting code blocks from documentation...\n");

  // Clean output directory
  if (fs.existsSync(OUTPUT_DIR)) {
    fs.rmSync(OUTPUT_DIR, { recursive: true });
  }
  fs.mkdirSync(OUTPUT_DIR, { recursive: true });

  // Create language subdirectories
  for (const lang of ["typescript", "python", "go", "csharp", "java"]) {
    fs.mkdirSync(path.join(OUTPUT_DIR, lang), { recursive: true });
  }

  // Find all markdown files
  const mdFiles = await glob("**/*.md", {
    cwd: DOCS_DIR,
    ignore: [".validation/**", "node_modules/**", "IMPROVEMENT_PLAN.md"],
  });

  console.log(`Found ${mdFiles.length} markdown files\n`);

  const manifest: ExtractionManifest = {
    extractedAt: new Date().toISOString(),
    blocks: [],
  };

  const langCounts = new Map<string, number>();
  let totalBlocks = 0;
  let skippedBlocks = 0;
  let hiddenBlocks = 0;

  for (const mdFile of mdFiles) {
    const fullPath = path.join(DOCS_DIR, mdFile);
    const content = fs.readFileSync(fullPath, "utf-8");
    const blocks = parseMarkdownCodeBlocks(content, mdFile);

    for (const block of blocks) {
      if (block.skip) {
        skippedBlocks++;
        continue;
      }

      if (block.hidden) {
        hiddenBlocks++;
      }

      // Skip empty or trivial blocks
      if (block.code.trim().length < 10) {
        continue;
      }

      // Skip incomplete code fragments that can't be validated standalone
      if (shouldSkipFragment(block)) {
        skippedBlocks++;
        continue;
      }