Add ability to fix zip corruption (#87)

`@turbowarp/[email protected]` now includes a recovery mode where it tries to find file entries by scanning the zip if the central directory is missing. The way that JSZip is really nice as it puts all the sizes and everything directly in the local header, so we don't need to do a lot of guesswork. If we had to support a bunch of random zip generators, this would be more work.

The sb3fix changes are to enable this mode, add logs, and remove files that have mismatched uncompressed data sizes so that regular Scratch will be able to load it.

These projects now get automatically fixed instead of waiting for me to run `zip -FF` manually:

 - All 3 projects in #82 (comment)
 - #80
 - #59
 - #32
 - #31
 - #30
 - #28
 - #7

It probably would've automatically fixed these but the links no longer work, so can't verify for certain:

 - #81
 - #50
 - #22
 - #5
 - #4
 - #3

Author: GarboMuffin

package-lock.json

@@ -21,7 +21,7 @@
   },
   "homepage": "https://github.com/TurboWarp/sb3fix#readme",
   "dependencies": {
-    "@turbowarp/jszip": "^3.11.1"
+    "@turbowarp/jszip": "^3.12.0"
   },
   "devDependencies": {
     "copy-webpack-plugin": "^13.0.0",

package.json

@@ -609,10 +609,46 @@ const fixJSON = (data, options = {}) => {
  * @returns {Promise<Uint8Array>} A promise that resolves to a fixed compressed .sb3 file.
  */
 const fixZip = async (data, options = {}) => {
+  /**
+   * @param {string} message
+   */
+  const log = (message) => {
+    if (options.logCallback) {
+      options.logCallback(message);
+    }
+  };
+
   // JSZip is not a small library, so we'll load it somewhat lazily.
   const JSZip = require('@turbowarp/jszip');
 
-  const zip = await JSZip.loadAsync(data);
+  let zip = await JSZip.loadAsync(data, {
+    recoverCorrupted: true,
+    onCorruptCentralDirectory: (error) => {
+      log(`zip had corrupt central directory: ${error}`);
+    },
+    onUnrecoverableFileEntry: (error) => {
+      log(`zip had unrecoverable file entry: ${error}`);
+    }
+  });
+
+  /** @type {Array<[string, import("@turbowarp/jszip").JSZipObject]>} */
+  const zipFiles = [];
+  zip.forEach((relativePath, file) => {
+    zipFiles.push([relativePath, file]);
+  });
+
+  // Remove any unreadable files from the zip. This can notably happen if the compressed data in the zip was
+  // corrupted, which would make the uncompressed data size field not match. Scratch/JSZip will refuse to
+  // keep loading the project if that happens. If we remove the asset, at least there's a chance it can now
+  // be downloaded from the asset server instead.
+  for (const [relativePath, file] of zipFiles) {
+    try {
+      await file.async('uint8array');
+    } catch (error) {
+      log(`zip had unreadable file ${relativePath}: ${error}`);
+      zip.remove(relativePath);
+    }
+  }
 
   // json is not guaranteed to be stored in the root.
   const jsonFile = zip.file(/(?:project|sprite)\.json/)[0];

src/sb3fix.js

@@ -0,0 +1,4 @@
+zip had corrupt central directory: Error: Corrupted zip: can't find end of central directory
+zip had unrecoverable file entry: Error: End of data reached (data length = 1000, asked index = 2194). Corrupted zip ?
+checking target 0
+checking target 1