Add some unit tests for unarchivers. Provide a way to disconnect the impl from the host (for unit tests).

archive/common.js

@@ -10,6 +10,12 @@
 
 // Requires the following JavaScript features: MessageChannel, MessagePort, and dynamic imports.
 
+/**
+ * @typedef Implementation
+ * @property {MessagePort} hostPort The port the host uses to communicate with the implementation.
+ * @property {Function} disconnectFn A function to call when the port has been disconnected.
+ */
+
 /**
  * Connects a host to a compress/decompress implementation via MessagePorts. The implementation must
  * have an exported connect() function that accepts a MessagePort. If the runtime support Workers
@@ -17,8 +23,9 @@
  * dynamically imports the implementation inside the current JS context (node, bun).
  * @param {string} implFilename The compressor/decompressor implementation filename relative to this
  *     path (e.g. './unzip.js').
- * @returns {Promise<MessagePort>} The Promise resolves to the MessagePort connected to the
- *     implementation that the host should use.
+ * @param {Function} disconnectFn A function to run when the port is disconnected.
+ * @returns {Promise<Implementation>} The Promise resolves to the Implementation, which includes the
+ *     MessagePort connected to the implementation that the host should use.
  */
 export async function getConnectedPort(implFilename) {
   const messageChannel = new MessageChannel();
@@ -28,13 +35,19 @@ export async function getConnectedPort(implFilename) {
   if (typeof Worker === 'undefined') {
     const implModule = await import(`${implFilename}`);
     await implModule.connect(implPort);
-    return hostPort;
+    return {
+      hostPort,
+      disconnectFn: () => implModule.disconnect(),
+    };
   }
   
   return new Promise((resolve, reject) => {
     const workerScriptPath = new URL(`./webworker-wrapper.js`, import.meta.url).href;
     const worker = new Worker(workerScriptPath, { type: 'module' });
     worker.postMessage({ implSrc: implFilename }, [implPort]);
-    resolve(hostPort);
+    resolve({
+      hostPort,
+      disconnectFn: () => worker.postMessage({ disconnect: true }),
+    });
   });
 }

archive/decompress.js

@@ -56,6 +56,13 @@ export class Unarchiver extends EventTarget {
    */
   port_;
 
+  /**
+   * A function to call to disconnect the implementation from the host.
+   * @type {Function}
+   * @private
+   */
+  disconnectFn_;
+
   /**
    * @param {ArrayBuffer} arrayBuffer The Array Buffer. Note that this ArrayBuffer must not be
    *     referenced once it is sent to the Unarchiver, since it is marked as Transferable and sent
@@ -87,6 +94,16 @@ export class Unarchiver extends EventTarget {
     this.debugMode_ = !!(options.debug);
   }
 
+  /**
+   * Overridden so that the type hints for eventType are specific.
+   * @param {'progress'|'extract'|'finish'} eventType 
+   * @param {EventListenerOrEventListenerObject} listener 
+   * @override
+   */
+  addEventListener(eventType, listener) {
+    super.addEventListener(eventType, listener);
+  }
+
   /**
    * This method must be overridden by the subclass to return the script filename.
    * @returns {string} The MIME type of the archive.
@@ -164,7 +181,9 @@ export class Unarchiver extends EventTarget {
    *     using the update() method.
    */
   async start() {
-    this.port_ = await getConnectedPort(this.getScriptFileName());
+    const impl = await getConnectedPort(this.getScriptFileName());
+    this.port_ = impl.hostPort;
+    this.disconnectFn_ = impl.disconnectFn;
     return new Promise((resolve, reject) => {
       this.port_.onerror = (evt) => {
         console.log('Impl error: message = ' + evt.message);
@@ -221,7 +240,9 @@ export class Unarchiver extends EventTarget {
   stop() {
     if (this.port_) {
       this.port_.close();
+      this.disconnectFn_();
       this.port_ = null;
+      this.disconnectFn_ = null;
     }
   }
 }

archive/unrar.js

@@ -1466,8 +1466,28 @@ const onmessage = function (event) {
  */
 export function connect(port) {
   if (hostPort) {
-    throw `hostPort already connected`;
+    throw `hostPort already connected in unrar.js`;
   }
   hostPort = port;
   port.onmessage = onmessage;
 }
+
+export function disconnect() {
+  if (!hostPort) {
+    throw `hostPort was not connected in unzip.js`;
+  }
+
+  hostPort = null;
+
+  unarchiveState = UnarchiveState.NOT_STARTED;
+  bytestream = null;
+  allLocalFiles = null;
+  logToConsole = false;
+  
+  currentFilename = '';
+  currentFileNumber = 0;
+  currentBytesUnarchivedInFile = 0;
+  currentBytesUnarchived = 0;
+  totalUncompressedBytesInArchive = 0;
+  totalFilesInArchive = 0;  
+}

archive/untar.js

@@ -221,8 +221,28 @@ const onmessage = function (event) {
  */
 export function connect(port) {
   if (hostPort) {
-    throw `hostPort already connected`;
+    throw `hostPort already connected in untar.js`;
   }
   hostPort = port;
   port.onmessage = onmessage;
 }
+
+export function disconnect() {
+  if (!hostPort) {
+    throw `hostPort was not connected in unzip.js`;
+  }
+
+  hostPort = null;
+
+  unarchiveState = UnarchiveState.NOT_STARTED;
+  bytestream = null;
+  allLocalFiles = null;
+  logToConsole = false;
+  
+  currentFilename = '';
+  currentFileNumber = 0;
+  currentBytesUnarchivedInFile = 0;
+  currentBytesUnarchived = 0;
+  totalUncompressedBytesInArchive = 0;
+  totalFilesInArchive = 0;
+}

archive/unzip.js

@@ -781,8 +781,30 @@ const onmessage = function (event) {
  */
 export function connect(port) {
   if (hostPort) {
-    throw `hostPort already connected`;
+    throw `hostPort already connected in unzip.js`;
   }
+
   hostPort = port;
   port.onmessage = onmessage;
 }
+
+export function disconnect() {
+  if (!hostPort) {
+    throw `hostPort was not connected in unzip.js`;
+  }
+
+  hostPort = null;
+
+  unarchiveState = UnarchiveState.NOT_STARTED;
+  bytestream = null;
+  allLocalFiles = null;
+  logToConsole = false;
+  
+  // Progress variables.
+  currentFilename = '';
+  currentFileNumber = 0;
+  currentBytesUnarchivedInFile = 0;
+  currentBytesUnarchived = 0;
+  totalUncompressedBytesInArchive = 0;
+  totalFilesInArchive = 0;  
+}

archive/webworker-wrapper.js

@@ -16,6 +16,10 @@
 let implPort;
 
 onmessage = async (evt) => {
+  if (evt.data.implSrc) {
     const module = await import(evt.data.implSrc);
     module.connect(evt.ports[0]);
+  } else if (evt.data.disconnect) {
+    module.disconnect();
+  }
 };

archive/zip.js

@@ -80,7 +80,6 @@ const CompressorState = {
   FINISHED: 3,
 };
 let state = CompressorState.NOT_STARTED;
-let lastFileReceived = false;
 const crc32Table = createCRC32Table();
 
 /** Helper functions. */
@@ -280,8 +279,21 @@ const onmessage = function(evt) {
  */
 export function connect(port) {
   if (hostPort) {
-    throw `hostPort already connected`;
+    throw `hostPort already connected in zip.js`;
   }
   hostPort = port;
   port.onmessage = onmessage;
 }
+
+export function disconnect() {
+  if (!hostPort) {
+    throw `hostPort was not connected in unzip.js`;
+  }
+
+  hostPort = null;
+
+  centralDirectoryInfos = [];
+  numBytesWritten = 0;
+  state = CompressorState.NOT_STARTED;
+  lastFileReceived = false;
+}

io/bytestream.js

@@ -22,7 +22,6 @@ export class ByteStream {
    */
   constructor(ab, opt_offset, opt_length) {
     if (!(ab instanceof ArrayBuffer)) {
-      console.error(typeof ab);
       throw 'Error! ByteStream constructed with an invalid ArrayBuffer object';
     }
 

tests/archive-testfiles/sample-1.txt

@@ -0,0 +1,20 @@
+This is a sample text file. This text file is large enough to make creating
+zip files more interesting, since compression has a chance to work on a larger
+sample file that has duplicate words in it.
+
+This is a sample text file. This text file is large enough to make creating
+zip files more interesting, since compression has a chance to work on a larger
+sample file that has duplicate words in it.
+
+This is a sample text file. This text file is large enough to make creating
+zip files more interesting, since compression has a chance to work on a larger
+sample file that has duplicate words in it.
+
+This is a sample text file. This text file is large enough to make creating
+zip files more interesting, since compression has a chance to work on a larger
+sample file that has duplicate words in it.
+
+This is a sample text file. This text file is large enough to make creating
+zip files more interesting, since compression has a chance to work on a larger
+sample file that has duplicate words in it.
+

tests/archive-testfiles/sample-2.csv

@@ -0,0 +1,4 @@
+filetype,extension
+"text file","txt"
+"JSON file","json"
+"CSV file","csv"

tests/archive-testfiles/sample-3.json

@@ -0,0 +1,6 @@
+{
+  "file formats": ["csv", "json", "txt"],
+  "tv shows": {
+    "it's": ["monty", "python's", "flying", "circus"]
+  }
+}

tests/decompress.spec.js

@@ -0,0 +1,57 @@
+import * as fs from 'node:fs';
+import 'mocha';
+import { expect } from 'chai';
+
+import { Unarchiver, Unrarrer, Untarrer, Unzipper, getUnarchiver } from '../archive/decompress.js';
+
+const PATH = `tests/archive-testfiles/`;
+
+const INPUT_FILES = [
+  'sample-1.txt',
+  'sample-2.csv',
+  'sample-3.json',
+];
+
+const ARCHIVE_FILES = [
+  'archive-rar.rar',
+  'archive-tar.tar',
+  'archive-zip-store.zip',
+  'archive-zip-faster.zip',
+  'archive-zip-smaller.zip',
+];
+
+describe('bitjs.archive.decompress', () => {
+  /** @type {Map<string, ArrayBuffer>} */
+  let inputArrayBuffers = new Map();
+
+  before(() => {
+    for (const inputFile of INPUT_FILES) {
+      const nodeBuf = fs.readFileSync(`${PATH}${inputFile}`);
+      const ab = nodeBuf.buffer.slice(nodeBuf.byteOffset, nodeBuf.byteOffset + nodeBuf.length);
+      inputArrayBuffers.set(inputFile, ab);
+    }
+  });
+
+  for (const outFile of ARCHIVE_FILES) {
+    it(outFile, (done) => {
+      const bufs = new Map(inputArrayBuffers);
+      const nodeBuf = fs.readFileSync(`${PATH}${outFile}`);
+      const ab = nodeBuf.buffer.slice(nodeBuf.byteOffset, nodeBuf.byteOffset + nodeBuf.length);
+      let unarchiver = getUnarchiver(ab);
+      expect(unarchiver instanceof Unarchiver).equals(true);
+  
+      unarchiver.addEventListener('extract', evt => {
+        const {filename, fileData} = evt.unarchivedFile;
+        expect(bufs.has(filename)).equals(true);
+        const ab = bufs.get(filename);
+        expect(fileData.byteLength).equals(ab.byteLength);
+        for (let b = 0; b < fileData.byteLength; ++b) {
+          expect(fileData[b] === ab[b]);
+        }
+        // Remove the value from the map so that it is only used once.
+        bufs.delete(filename);
+      });
+      unarchiver.start().then(() => { done() });
+    });
+  }
+});