Feature: correct invalid extensions when raw-copying archives (#1128)

src/modules/candidateGenerator.ts

@@ -185,16 +185,14 @@ export default class CandidateGenerator extends Module {
           && !this.options.shouldExtract()
         ) {
           try {
-            if (this.options.shouldTest() || this.options.getOverwriteInvalid()) {
-              // If we're testing, then we need to calculate the archive's checksums
-              inputFile = await FileFactory.fileFrom(
-                inputFile.getFilePath(),
-                inputFile.getChecksumBitmask(),
-              );
-            } else {
-              // Otherwise, we can skip calculating checksums for efficiency
-              inputFile = await FileFactory.fileFrom(inputFile.getFilePath(), ChecksumBitmask.NONE);
-            }
+            inputFile = await FileFactory.archiveFileFrom(
+              inputFile.getArchive(),
+              // If we're testing, then we need to calculate the archive's checksums, otherwise we
+              // can skip calculating checksums for efficiency
+              this.options.shouldTest() || this.options.getOverwriteInvalid()
+                ? inputFile.getChecksumBitmask()
+                : ChecksumBitmask.NONE,
+            );
           } catch (error) {
             this.progressBar.logWarn(`${dat.getNameShort()}: ${game.getName()}: ${error}`);
             return [rom, undefined];

src/types/files/archives/archive.ts

@@ -15,6 +15,8 @@ export default abstract class Archive {
 
   protected abstract new(filePath: string): Archive;
 
+  abstract getExtension(): string;
+
   getFilePath(): string {
     return this.filePath;
   }

src/types/files/archives/archiveFile.ts

@@ -0,0 +1,15 @@
+import File, { FileProps } from '../file.js';
+import Archive from './archive.js';
+
+export default class ArchiveFile extends File {
+  private readonly archive: Archive;
+
+  public constructor(archive: Archive, fileProps: FileProps) {
+    super(fileProps);
+    this.archive = archive;
+  }
+
+  getArchive(): Archive {
+    return this.archive;
+  }
+}

src/types/files/archives/gzip.ts

@@ -0,0 +1,21 @@
+import SevenZip from './sevenZip.js';
+
+export default class Gzip extends SevenZip {
+  // eslint-disable-next-line class-methods-use-this
+  protected new(filePath: string): SevenZip {
+    return new Gzip(filePath);
+  }
+
+  static getExtensions(): string[] {
+    return ['.gz', '.gzip'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return Gzip.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [Buffer.from('1F8B', 'hex')];
+  }
+}

src/types/files/archives/rar.ts

@@ -9,20 +9,29 @@ import Archive from './archive.js';
 import ArchiveEntry from './archiveEntry.js';
 
 export default class Rar extends Archive {
-  static readonly SUPPORTED_FILES: [string[], Buffer[]][] = [
-    [['.rar'], [
-      Buffer.from('526172211A0700', 'hex'), // v1.50+
-      Buffer.from('526172211A070100', 'hex'), // v5.00+
-    ]],
-  ];
-
   private static readonly EXTRACT_MUTEX = new Mutex();
 
   // eslint-disable-next-line class-methods-use-this
   protected new(filePath: string): Archive {
     return new Rar(filePath);
   }
 
+  static getExtensions(): string[] {
+    return ['.rar'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return Rar.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [
+      Buffer.from('526172211A0700', 'hex'), // v1.50+
+      Buffer.from('526172211A070100', 'hex'), // v5.00+
+    ];
+  }
+
   async getArchiveEntries(checksumBitmask: number): Promise<ArchiveEntry<this>[]> {
     const rar = await unrar.createExtractorFromFile({
       filepath: this.getFilePath(),

src/types/files/archives/sevenZip.ts

@@ -10,70 +10,26 @@ import Archive from './archive.js';
 import ArchiveEntry from './archiveEntry.js';
 
 export default class SevenZip extends Archive {
-  // p7zip `7za i`
-  // WARNING: tar+compression doesn't work, you'll be left with a tar file output
-  static readonly SUPPORTED_FILES: [string[], Buffer[]][] = [
-    [['.7z'], [
-      Buffer.from('377ABCAF271C', 'hex'),
-    ]],
-    // [['.bz2', '.bzip2'], [
-    //   'BZh',
-    // ]],
-    // [['.cab'], [
-    //   'MSCF',
-    // ]],
-    [['.gz', '.gzip'], [
-      Buffer.from('1F8B', 'hex'),
-    ]],
-    // [['.lzma'], [
-    //   // ???
-    // ]],
-    // [['.lzma86'], [
-    //   // ???
-    // ]],
-    // [['.pmd'], [
-    //   // ???
-    // ]],
-    // [['.tar', '.ova'], [
-    //   Buffer.from('7573746172003030', 'hex'),
-    //   Buffer.from('7573746172202000', 'hex'),
-    // ]],
-    // [['.xz'], [
-    //   Buffer.from('FD377A585A00', 'hex'), // LZMA2 compression
-    // ]],
-    [['.z'], [
-      Buffer.from('1F9D', 'hex'), // LZW compression
-      Buffer.from('1FA0', 'hex'), // LZH compression
-    ]],
-    [['.zip', '.zip.001', '.z01'], [
-      Buffer.from('504B0304', 'hex'),
-      Buffer.from('504B0506', 'hex'), // empty archive
-      Buffer.from('504B0708', 'hex'), // spanned archive
-    ]],
-    [['.zipx', '.zx01'], [
-      // ???
-    ]],
-    // [['.zst'], [
-    //   Buffer.from('28B52FFD', 'hex'),
-    // ]],
-    // [['.lz4'], [
-    //   Buffer.from('04224D18', 'hex'),
-    // ]],
-    // [['.lz5'], [
-    //   // ???
-    // ]],
-    // [['.liz'], [
-    //   // ???
-    // ]],
-  ];
-
   private static readonly LIST_MUTEX = new Mutex();
 
   // eslint-disable-next-line class-methods-use-this
   protected new(filePath: string): Archive {
     return new SevenZip(filePath);
   }
 
+  static getExtensions(): string[] {
+    return ['.7z'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return SevenZip.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [Buffer.from('377ABCAF271C', 'hex')];
+  }
+
   async getArchiveEntries(checksumBitmask: number): Promise<ArchiveEntry<this>[]> {
     /**
      * WARN(cemmer): even with the above mutex, {@link _7z.list} will still sometimes return no

src/types/files/archives/tar.ts

@@ -10,21 +10,30 @@ import Archive from './archive.js';
 import ArchiveEntry from './archiveEntry.js';
 
 export default class Tar extends Archive {
-  static readonly SUPPORTED_FILES: [string[], Buffer[]][] = [
-    [['.tar'], [
-      Buffer.from('7573746172003030', 'hex'),
-      Buffer.from('7573746172202000', 'hex'),
-    ]],
-    [['.tar.gz', '.tgz'], [
-      Buffer.from('1F8B', 'hex'),
-    ]],
-  ];
-
   // eslint-disable-next-line class-methods-use-this
   protected new(filePath: string): Archive {
     return new Tar(filePath);
   }
 
+  static getExtensions(): string[] {
+    return ['.tar', '.tar.gz', '.tgz'];
+  }
+
+  getExtension(): string {
+    // We can't reliably know the extension
+    return path.parse(this.getFilePath()).ext;
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [
+      // .tar
+      Buffer.from('7573746172003030', 'hex'),
+      Buffer.from('7573746172202000', 'hex'),
+      // .tar.gz / .tgz
+      Buffer.from('1F8B', 'hex'),
+    ];
+  }
+
   async getArchiveEntries(checksumBitmask: number): Promise<ArchiveEntry<this>[]> {
     const archiveEntryPromises: Promise<ArchiveEntry<this>>[] = [];
 

src/types/files/archives/z.ts

@@ -0,0 +1,24 @@
+import SevenZip from './sevenZip.js';
+
+export default class Z extends SevenZip {
+  // eslint-disable-next-line class-methods-use-this
+  protected new(filePath: string): SevenZip {
+    return new Z(filePath);
+  }
+
+  static getExtensions(): string[] {
+    return ['.z'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return Z.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [
+      Buffer.from('1F9D', 'hex'), // LZW compression
+      Buffer.from('1FA0', 'hex'), // LZH compression
+    ];
+  }
+}

src/types/files/archives/zip.ts

@@ -16,18 +16,27 @@ import Archive from './archive.js';
 import ArchiveEntry from './archiveEntry.js';
 
 export default class Zip extends Archive {
-  static readonly SUPPORTED_FILES: [string[], Buffer[]][] = [
-    [['.zip'], [
-      Buffer.from('504B0304', 'hex'),
-      Buffer.from('504B0506', 'hex'), // empty archive
-    ]],
-  ];
-
   // eslint-disable-next-line class-methods-use-this
   protected new(filePath: string): Archive {
     return new Zip(filePath);
   }
 
+  static getExtensions(): string[] {
+    return ['.zip'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return Zip.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [
+      Buffer.from('504B0304', 'hex'),
+      Buffer.from('504B0506', 'hex'), // empty archive
+    ];
+  }
+
   async getArchiveEntries(checksumBitmask: number): Promise<ArchiveEntry<this>[]> {
     // https://github.com/ZJONSSON/node-unzipper/issues/280
     // UTF-8 entry names are not decoded correctly

src/types/files/archives/zipSpanned.ts

@@ -0,0 +1,21 @@
+import SevenZip from './sevenZip.js';
+
+export default class ZipSpanned extends SevenZip {
+  // eslint-disable-next-line class-methods-use-this
+  protected new(filePath: string): SevenZip {
+    return new ZipSpanned(filePath);
+  }
+
+  static getExtensions(): string[] {
+    return ['.zip.001', '.z01'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return ZipSpanned.getExtensions()[0];
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [Buffer.from('504B0708', 'hex')];
+  }
+}

src/types/files/archives/zipX.ts

@@ -0,0 +1,23 @@
+import path from 'node:path';
+
+import SevenZip from './sevenZip.js';
+
+export default class ZipX extends SevenZip {
+  // eslint-disable-next-line class-methods-use-this
+  protected new(filePath: string): SevenZip {
+    return new ZipX(filePath);
+  }
+
+  static getExtensions(): string[] {
+    return ['.zipx', '.zx01'];
+  }
+
+  // eslint-disable-next-line class-methods-use-this
+  getExtension(): string {
+    return path.parse(this.getFilePath()).ext;
+  }
+
+  static getFileSignatures(): Buffer[] {
+    return [];
+  }
+}