💄

README.md

@@ -23,3 +23,8 @@ You can try zip 64 format by running the following command:
 ```bash
 cd dduf-content && zip -r -0 -fz ../file-64.dduf .
 ```
+
+## Credits
+
+- https://en.wikipedia.org/wiki/ZIP_(file_format)
+- https://blog.yaakov.online/zip64-go-big-or-go-home/

src/lib/check-dduf.ts

@@ -1,10 +1,13 @@
 import { checkFilename } from './check-filename';
 import { WebBlob } from './WebBlob';
 
-export async function* checkDduf(url: string): AsyncGenerator<string> {
+export async function* checkDduf(
+	url: string,
+	opts?: { log?: (x: string) => void }
+): AsyncGenerator<{ type: 'file'; name: string; size: number; fileHeaderOffset: number }> {
 	const blob = await WebBlob.create(new URL(url));
 
-	yield 'File size: ' + blob.size;
+	opts?.log?.('File size: ' + blob.size);
 
 	// DDUF is a zip file, uncompressed.
 
@@ -28,25 +31,80 @@ export async function* checkDduf(url: string): AsyncGenerator<string> {
 		throw new Error('DDUF footer not found in last 100kB of file');
 	}
 
-	yield 'DDUF footer found at offset ' + (blob.size - last100kB.byteLength + index);
+	opts?.log?.('DDUF footer found at offset ' + (blob.size - last100kB.byteLength + index));
 
 	const diskNumber = view.getUint16(index + 4, true);
 
-	if (diskNumber === 0xffff) {
-		throw new Error('Spanned archives (ZIP64) not yet supported');
-	}
+	opts?.log?.('Disk number: ' + diskNumber);
 
-	if (diskNumber !== 0) {
+	if (diskNumber !== 0 && diskNumber !== 0xffff) {
 		throw new Error('Multi-disk archives not supported');
 	}
 
-	const fileCount = view.getUint16(index + 10, true);
-	const centralDirSize = view.getUint32(index + 12, true);
-	const centralDirOffset = view.getUint32(index + 16, true);
+	let fileCount = view.getUint16(index + 10, true);
+	let centralDirSize = view.getUint32(index + 12, true);
+	let centralDirOffset = view.getUint32(index + 16, true);
+	const isZip64 = centralDirOffset === 0xffffffff;
+
+	opts?.log?.('File count: ' + fileCount);
+
+	if (isZip64) {
+		opts?.log?.('Zip64 format detected');
+
+		index -= 20;
+		while (index >= 0) {
+			if (view.getUint32(index, true) === 0x07064b50) {
+				found = true;
+				break;
+			}
+
+			index--;
+		}
+
+		if (!found) {
+			throw new Error('Zip64 footer not found in last 100kB of file');
+		}
+
+		opts?.log?.('Zip64 footer found at offset ' + (blob.size - last100kB.byteLength + index));
+
+		const diskWithCentralDir = view.getUint32(index + 4, true);
+
+		if (diskWithCentralDir !== 0) {
+			throw new Error('Multi-disk archives not supported');
+		}
+
+		const endCentralDirOffset = Number(view.getBigUint64(index + 8, true));
+
+		index = endCentralDirOffset - (blob.size - last100kB.byteLength);
+
+		if (index < 0) {
+			throw new Error('Central directory offset is outside the last 100kB of the file');
+		}
+
+		if (view.getUint32(index, true) !== 0x06064b50) {
+			throw new Error('Invalid central directory header');
+		}
+
+		const thisDisk = view.getUint16(index + 16, true);
+		const centralDirDisk = view.getUint16(index + 20, true);
+
+		if (thisDisk !== 0) {
+			throw new Error('Multi-disk archives not supported');
+		}
+
+		if (centralDirDisk !== 0) {
+			throw new Error('Multi-disk archives not supported');
+		}
 
-	yield 'File count: ' + fileCount;
-	yield 'Central directory size: ' + centralDirSize;
-	yield 'Central directory offset: ' + centralDirOffset;
+		centralDirSize = Number(view.getBigUint64(index + 40, true));
+		centralDirOffset = Number(view.getBigUint64(index + 48, true));
+		fileCount = Number(view.getBigUint64(index + 32, true));
+
+		opts?.log?.('File count zip 64: ' + fileCount);
+	}
+
+	opts?.log?.('Central directory size: ' + centralDirSize);
+	opts?.log?.('Central directory offset: ' + centralDirOffset);
 
 	const centralDir =
 		centralDirOffset > blob.size - last100kB.byteLength
@@ -74,39 +132,82 @@ export async function* checkDduf(url: string): AsyncGenerator<string> {
 			throw new Error('Unsupported compression method: ' + compressionMethod);
 		}
 
-		const size = centralDirView.getUint32(offset + 24, true);
-		const compressedSize = centralDirView.getUint32(offset + 20, true);
-
-		if (size !== compressedSize) {
-			throw new Error('Compressed size and size differ');
-		}
-
 		const filenameLength = centralDirView.getUint16(offset + 28, true);
 		const fileName = new TextDecoder().decode(
 			new Uint8Array(centralDir, offset + 46, filenameLength)
 		);
 
-		yield 'File ' + i;
-		yield 'File name: ' + fileName;
-		yield 'File size: ' + size;
+		opts?.log?.('File ' + i);
+		opts?.log?.('File name: ' + fileName);
 
 		checkFilename(fileName);
 
 		const fileDiskNumber = centralDirView.getUint16(34, true);
 
-		if (fileDiskNumber !== 0) {
+		if (fileDiskNumber !== 0 && fileDiskNumber !== 0xffff) {
 			throw new Error('Multi-disk archives not supported');
 		}
 
+		let size = centralDirView.getUint32(offset + 24, true);
+		let compressedSize = centralDirView.getUint32(offset + 20, true);
+		let filePosition = centralDirView.getUint32(offset + 42, true);
+
 		const extraFieldLength = centralDirView.getUint16(offset + 30, true);
-		const commentLength = centralDirView.getUint16(offset + 32, true);
 
-		const filePosition = centralDirView.getUint32(offset + 42, true);
+		if (size === 0xffffffff || compressedSize === 0xffffffff || filePosition === 0xffffffff) {
+			opts?.log?.('File size is in zip64 format');
+
+			const extraFields = new DataView(centralDir, offset + 46 + filenameLength, extraFieldLength);
+
+			let extraFieldOffset = 0;
+
+			while (extraFieldOffset < extraFieldLength) {
+				const headerId = extraFields.getUint16(extraFieldOffset, true);
+				const extraFieldSize = extraFields.getUint16(extraFieldOffset + 2, true);
+				if (headerId !== 0x0001) {
+					extraFieldOffset += 4 + extraFieldSize;
+					continue;
+				}
+
+				const zip64ExtraField = new DataView(
+					centralDir,
+					offset + 46 + filenameLength + extraFieldOffset + 4,
+					extraFieldSize
+				);
+				let zip64ExtraFieldOffset = 0;
+
+				if (size === 0xffffffff) {
+					size = Number(zip64ExtraField.getBigUint64(zip64ExtraFieldOffset, true));
+					zip64ExtraFieldOffset += 8;
+				}
+
+				if (compressedSize === 0xffffffff) {
+					compressedSize = Number(zip64ExtraField.getBigUint64(zip64ExtraFieldOffset, true));
+					zip64ExtraFieldOffset += 8;
+				}
+
+				if (filePosition === 0xffffffff) {
+					filePosition = Number(zip64ExtraField.getBigUint64(zip64ExtraFieldOffset, true));
+					zip64ExtraFieldOffset += 8;
+				}
+
+				break;
+			}
+		}
+
+		if (size !== compressedSize) {
+			throw new Error('Compressed size and size differ: ' + compressedSize + ' vs ' + size);
+		}
+		opts?.log?.('File size: ' + size);
 
-		yield 'File position in archive: ' + filePosition;
+		const commentLength = centralDirView.getUint16(offset + 32, true);
+
+		opts?.log?.('File header position in archive: ' + filePosition);
 
 		offset += 46 + filenameLength + extraFieldLength + commentLength;
+
+		yield { type: 'file', name: fileName, size, fileHeaderOffset: filePosition };
 	}
 
-	yield 'All files checked';
+	opts?.log?.('All files checked');
 }

src/routes/+page.svelte

@@ -2,9 +2,10 @@
 	import { checkDduf } from '$lib/check-dduf';
 	import { tick } from 'svelte';
 
-	let url = 'https://huggingface.co/spaces/coyotte508/dduf-check/resolve/main/file.dduf';
-	let output = '';
-	let error = '';
+	let url = $state('https://huggingface.co/spaces/coyotte508/dduf-check/resolve/main/file.dduf');
+	let output = $state('');
+	let error = $state('');
+	let files: Array<{ position: number; size: number; name: string }> = $state([]);
 
 	async function handleSubmit(event: Event) {
 		event.preventDefault();
@@ -12,10 +13,18 @@
 		error = '';
 
 		try {
-			for await (const str of checkDduf(url)) {
-				output += '\n' + str;
-				tick().then(() => {
-					textarea.scrollTop = textarea.scrollHeight;
+			for await (const file of checkDduf(url, {
+				log: (s) => {
+					output += '\n' + s;
+					tick().then(() => {
+						textarea.scrollTop = textarea.scrollHeight;
+					});
+				}
+			})) {
+				files.push({
+					position: file.fileHeaderOffset,
+					size: file.size,
+					name: file.name
 				});
 			}
 		} catch (e) {
@@ -54,5 +63,16 @@
 		{#if error}
 			<p class="text-red-500">{error}</p>
 		{/if}
+
+		{#if files.length > 0}
+			<h2 class="text-lg font-bold">Files</h2>
+			<ul class="ml-4 list-disc">
+				{#each files as file}
+					<li style="margin-left: {file.name.slice(0, -1).split('/').length - 1}rem">
+						{file.name} - {file.size} B
+					</li>
+				{/each}
+			</ul>
+		{/if}
 	</form>
 </div>