feat(archive): add tar.bz2 converter via 7za

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

package-lock.json

@@ -10,6 +10,7 @@
       "dependencies": {
         "@phosphor-icons/react": "^2.1.10",
         "@prisma/client": "^6.19.3",
+        "7zip-min": "^3.0.1",
         "adm-zip": "^0.6.0",
         "archiver": "^8.0.0",
         "docx": "^9.7.1",
@@ -1821,6 +1822,21 @@
         "node": ">=10.0.0"
       }
     },
+    "node_modules/7zip-bin": {
+      "version": "5.1.1",
+      "resolved": "https://registry.npmjs.org/7zip-bin/-/7zip-bin-5.1.1.tgz",
+      "integrity": "sha512-sAP4LldeWNz0lNzmTird3uWfFDWWTeg6V/MsmyyLR9X1idwKBWIgt/ZvinqQldJm3LecKEs1emkbquO6PCiLVQ==",
+      "license": "MIT"
+    },
+    "node_modules/7zip-min": {
+      "version": "3.0.1",
+      "resolved": "https://registry.npmjs.org/7zip-min/-/7zip-min-3.0.1.tgz",
+      "integrity": "sha512-WB4VCA/KSKzxhj+BAp8fI3ZYMMAftclkXlUTckuiDacsqyubQxxG3lGcpBcgzWWuJqnfQncEq1xrJpPLSxqsxw==",
+      "license": "MIT",
+      "dependencies": {
+        "7zip-bin": "5.1.1"
+      }
+    },
     "node_modules/abort-controller": {
       "version": "3.0.0",
       "resolved": "https://registry.npmjs.org/abort-controller/-/abort-controller-3.0.0.tgz",

package.json

@@ -17,6 +17,7 @@
   "dependencies": {
     "@phosphor-icons/react": "^2.1.10",
     "@prisma/client": "^6.19.3",
+    "7zip-min": "^3.0.1",
     "adm-zip": "^0.6.0",
     "archiver": "^8.0.0",
     "docx": "^9.7.1",

src/converters/archive.js

@@ -5,6 +5,7 @@ import os from 'node:os';
 import AdmZip from 'adm-zip';
 import { ZipArchive } from 'archiver';
 import * as tar from 'tar';
+import _7z from '7zip-min';
 import { register } from './registry.js';
 
 const MAX_EXTRACTED_BYTES = 2 * 1024 * 1024 * 1024;
@@ -70,8 +71,45 @@ async function createTarGz(srcDir, outputPath, quality) {
   await tar.create({ file: outputPath, cwd: srcDir, gzip: { level: quality ?? 6 } }, entries);
 }
 
-const EXTRACTORS = { zip: extractZip, tar: extractTarLike, 'tar.gz': extractTarLike };
-const CREATORS = { zip: createZip, tar: createTar, 'tar.gz': createTarGz };
+function bzip2Level(quality) {
+  return Math.max(1, quality ?? 9);
+}
+
+async function extractTarBz2(inputPath, destDir) {
+  const decompressDir = await fs.mkdtemp(path.join(os.tmpdir(), 'archive-bz2-'));
+  try {
+    await _7z.unpack(inputPath, decompressDir);
+    const [tarName] = await fs.readdir(decompressDir);
+    await tar.extract({ file: path.join(decompressDir, tarName), cwd: destDir });
+  } finally {
+    await fs.rm(decompressDir, { recursive: true, force: true });
+  }
+}
+
+async function createTarBz2(srcDir, outputPath, quality) {
+  const buildDir = await fs.mkdtemp(path.join(os.tmpdir(), 'archive-bz2-'));
+  try {
+    const tarPath = path.join(buildDir, 'archive.tar');
+    const entries = await fs.readdir(srcDir);
+    await tar.create({ file: tarPath, cwd: srcDir }, entries);
+    await _7z.cmd(['a', '-tbzip2', `-mx=${bzip2Level(quality)}`, outputPath, tarPath]);
+  } finally {
+    await fs.rm(buildDir, { recursive: true, force: true });
+  }
+}
+
+const EXTRACTORS = {
+  zip: extractZip,
+  tar: extractTarLike,
+  'tar.gz': extractTarLike,
+  'tar.bz2': extractTarBz2,
+};
+const CREATORS = {
+  zip: createZip,
+  tar: createTar,
+  'tar.gz': createTarGz,
+  'tar.bz2': createTarBz2,
+};
 
 async function convert(inputPath, outputPath, options, sourceFormat, targetFormat) {
   const { quality } = options ?? {};

test/converters/archive.test.js

@@ -133,3 +133,31 @@ describe('archive converters — tar / tar.gz', () => {
     await fs.rm(listDir, { recursive: true, force: true });
   });
 });
+
+describe('archive converters — tar.bz2', () => {
+  it('registers tar.bz2 as a target for tar and vice versa', () => {
+    expect(listTargetFormats('tar')).toContain('tar.bz2');
+    expect(listTargetFormats('tar.bz2')).toContain('tar');
+  });
+
+  it(
+    'converts zip -> tar.bz2 -> tar, preserving content',
+    async () => {
+      const zipPath = path.join(tmpDir, 'for-bz2.zip');
+      buildZipFixture(zipPath);
+
+      const bz2Path = path.join(tmpDir, 'output.tar.bz2');
+      await resolve('zip', 'tar.bz2').convert(zipPath, bz2Path, { quality: 9 });
+
+      const backToTarPath = path.join(tmpDir, 'from-bz2.tar');
+      await resolve('tar.bz2', 'tar').convert(bz2Path, backToTarPath);
+
+      const listDir = await fs.mkdtemp(path.join(os.tmpdir(), 'bz2-check-'));
+      await tar.extract({ file: backToTarPath, cwd: listDir });
+      expect(await fs.readFile(path.join(listDir, 'hello.txt'), 'utf8')).toBe('hello world');
+      expect(await fs.readFile(path.join(listDir, 'nested', 'inner.txt'), 'utf8')).toBe('nested content');
+      await fs.rm(listDir, { recursive: true, force: true });
+    },
+    20000
+  );
+});