anyclip-video-manager/scripts/extract-sources.ts

#!/usr/bin/env bun
/**
 * Extract source files from sourcemaps into proper directory structure
 *
 * Usage:
 *   bun scripts/extract-sources.ts [options]
 *
 * Options:
 *   --output, -o <dir>   Output directory (default: anyclip)
 *   --input, -i <dir>    Sourcemaps directory (default: sourcemaps)
 *   --verbose, -v        Verbose output
 *   --no-clean           Don't delete output directory first
 */

import { mkdir, rm, stat, unlink } from "fs/promises";
import { dirname, join, normalize, basename } from "path";
import { parseArgs } from "util";

// Parse CLI arguments
const { values: args } = parseArgs({
  args: Bun.argv.slice(2),
  options: {
    output: { type: "string", short: "o", default: "anyclip" },
    input: { type: "string", short: "i", default: "sourcemaps" },
    verbose: { type: "boolean", short: "v", default: false },
    "no-clean": { type: "boolean", default: false },
  },
});

const OUTPUT_DIR = args.output!;
const INPUT_DIR = args.input!;
const VERBOSE = args.verbose!;
const NO_CLEAN = args["no-clean"]!;

interface SourceMap {
  version: number;
  sources: string[];
  sourcesContent: (string | null)[];
  names?: string[];
  mappings?: string;
}

// Collect all files first, then write (to handle conflicts)
const filesToWrite = new Map<string, string>();

/**
 * Sanitize a source path from a sourcemap
 * Inspired by sourcemapper and shuji
 */
function sanitizePath(source: string): string | null {
  let cleanPath = source
    // Remove webpack:// prefix (webpack://package-name/path or webpack:///path)
    .replace(/^webpack:\/\/[^/]*\//, "")
    // Remove leading ./
    .replace(/^\.\//, "")
    // Remove query strings (?v=123, ?module, etc)
    .replace(/\?[^/]*$/, "")
    // Remove Windows-illegal characters
    .replace(/[?%*|:"<>]/g, "")
    // Replace spaces with hyphens
    .replace(/ /g, "-");

  // Skip empty paths
  if (!cleanPath || cleanPath === "") return null;

  // Skip node_modules and external dependencies
  if (cleanPath.includes("node_modules") || cleanPath.startsWith("external ")) {
    return null;
  }

  // Handle (webpack) paths - put in webpack/ directory
  if (cleanPath.includes("(webpack)")) {
    cleanPath = "webpack/" + cleanPath.replace(/\(webpack\)\//g, "");
  }

  // Normalize path: resolve ../ sequences safely
  // Prepend a fake root to prevent escaping, then normalize and strip it
  const withRoot = "/" + cleanPath;
  const normalized = normalize(withRoot).slice(1); // Remove leading /

  // Handle webpack runtime files (no extension) - add .js
  if (normalized.startsWith("webpack/") && !basename(normalized).includes(".")) {
    return normalized + ".js";
  }

  // Skip paths without file extension (likely directories)
  if (!basename(normalized).includes(".")) {
    return null;
  }

  return normalized;
}

async function extractSourcemap(mapFile: string): Promise<number> {
  if (VERBOSE) console.log(`Processing ${mapFile}...`);

  const fileContent = await Bun.file(mapFile).text();
  let sourcemap: SourceMap;

  try {
    sourcemap = JSON.parse(fileContent);
  } catch (e) {
    if (VERBOSE) console.log(`  Failed to parse: ${e}`);
    return 0;
  }

  if (!sourcemap.sources || !sourcemap.sourcesContent) {
    if (VERBOSE) console.log(`  No sources found`);
    return 0;
  }

  let collected = 0;

  for (let i = 0; i < sourcemap.sources.length; i++) {
    const source = sourcemap.sources[i];
    const content = sourcemap.sourcesContent[i];

    if (!content) continue;

    const cleanPath = sanitizePath(source);
    if (!cleanPath) continue;

    const outputPath = join(OUTPUT_DIR, cleanPath);

    // Keep the larger file if there's a conflict (more complete source)
    if (!filesToWrite.has(outputPath) || content.length > filesToWrite.get(outputPath)!.length) {
      filesToWrite.set(outputPath, content);
      collected++;
    }
  }

  if (VERBOSE) console.log(`  Collected ${collected} files`);
  return collected;
}

// Delete existing output directory (unless --no-clean)
if (!NO_CLEAN) {
  console.log(`Cleaning ${OUTPUT_DIR}/...`);
  await rm(OUTPUT_DIR, { recursive: true, force: true });
}

// Process all sourcemaps
const sourcemaps = await Array.fromAsync(new Bun.Glob("*.map").scan(INPUT_DIR));
console.log(`Found ${sourcemaps.length} sourcemaps in ${INPUT_DIR}/`);

for (const map of sourcemaps) {
  await extractSourcemap(join(INPUT_DIR, map));
}

// Write all files
console.log(`\nWriting ${filesToWrite.size} files to ${OUTPUT_DIR}/...`);
let written = 0;
let skipped = 0;
let errors = 0;

for (const [outputPath, content] of filesToWrite) {
  try {
    const dir = dirname(outputPath);

    // Handle file/directory conflicts
    const parts = dir.split("/");
    let currentPath = "";
    for (const part of parts) {
      currentPath = currentPath ? `${currentPath}/${part}` : part;
      try {
        const s = await stat(currentPath);
        if (s.isFile()) {
          await unlink(currentPath);
        }
      } catch {
        // Path doesn't exist yet
      }
    }

    await mkdir(dir, { recursive: true, mode: 0o755 });
    await Bun.write(outputPath, content);
    written++;

    if (VERBOSE) console.log(`  ✓ ${outputPath}`);
  } catch (e) {
    errors++;
    if (VERBOSE || errors <= 3) {
      console.log(`  ✗ ${outputPath}: ${e}`);
    }
  }
}

if (!VERBOSE && errors > 3) {
  console.log(`  ... and ${errors - 3} more errors`);
}

console.log(`\n✅ Extracted ${written} files to ${OUTPUT_DIR}/`);
if (errors > 0) console.log(`⚠️  ${errors} errors`);

// Show directory summary
const dirs = new Set([...filesToWrite.keys()].map(p => p.split("/").slice(0, 2).join("/")));
console.log(`\nDirectories: ${[...dirs].sort().join(", ")}`);