@zip.js/zip.js 2.8.30 → 2.8.31

package/dist/zip-core.js

@@ -113,6 +113,8 @@
 	const FUNCTION_TYPE = "function";
 	const OBJECT_TYPE = "object";
 
+	const EMPTY_UINT8_ARRAY = new Uint8Array();
+
 	/*
 	 Copyright (c) 2025 Gildas Lormeau. All rights reserved.
 
@@ -164,6 +166,22 @@
 		DecompressionStream: typeof DecompressionStream != UNDEFINED_TYPE && DecompressionStream
 	};
 
+	const CONFIGURABLE_PROPERTY_NAMES = [
+		"baseURI",
+		"wasmURI",
+		"workerURI",
+		"chunkSize",
+		"maxWorkers",
+		"terminateWorkerTimeout",
+		"workerStarvationTimeout",
+		"useCompressionStream",
+		"useWebWorkers",
+		"CompressionStream",
+		"DecompressionStream",
+		"CompressionStreamZlib",
+		"DecompressionStreamZlib"
+	];
+
 	const config = Object.assign({}, DEFAULT_CONFIGURATION);
 
 	function getConfiguration() {
@@ -175,39 +193,11 @@
 	}
 
 	function configure(configuration) {
-		const {
-			baseURI,
-			chunkSize,
-			maxWorkers,
-			terminateWorkerTimeout,
-			workerStarvationTimeout,
-			useCompressionStream,
-			useWebWorkers,
-			CompressionStream,
-			DecompressionStream,
-			CompressionStreamZlib,
-			DecompressionStreamZlib,
-			workerURI,
-			wasmURI
-		} = configuration;
-		setIfDefined("baseURI", baseURI);
-		setIfDefined("wasmURI", wasmURI);
-		setIfDefined("workerURI", workerURI);
-		setIfDefined("chunkSize", chunkSize);
-		setIfDefined("maxWorkers", maxWorkers);
-		setIfDefined("terminateWorkerTimeout", terminateWorkerTimeout);
-		setIfDefined("workerStarvationTimeout", workerStarvationTimeout);
-		setIfDefined("useCompressionStream", useCompressionStream);
-		setIfDefined("useWebWorkers", useWebWorkers);
-		setIfDefined("CompressionStream", CompressionStream);
-		setIfDefined("DecompressionStream", DecompressionStream);
-		setIfDefined("CompressionStreamZlib", CompressionStreamZlib);
-		setIfDefined("DecompressionStreamZlib", DecompressionStreamZlib);
-	}
-
-	function setIfDefined(propertyName, propertyValue) {
-		if (propertyValue !== UNDEFINED_VALUE) {
-			config[propertyName] = propertyValue;
+		for (const propertyName of CONFIGURABLE_PROPERTY_NAMES) {
+			const propertyValue = configuration[propertyName];
+			if (propertyValue !== UNDEFINED_VALUE) {
+				config[propertyName] = propertyValue;
+			}
 		}
 	}
 
@@ -1222,8 +1212,6 @@
 		if (GET_RANDOM_VALUES_SUPPORTED) {
 			return crypto.getRandomValues(array);
 		} else {
-			// all supported runtimes expose crypto.getRandomValues; failing hard beats
-			// silently generating predictable encryption salts
 			throw new Error(ERR_UNSUPPORTED_CRYPTO_API);
 		}
 	}
@@ -1286,12 +1274,7 @@
 		constructor({ password, rawPassword, encryptionStrength, checkPasswordOnly }) {
 			super({
 				start() {
-					Object.assign(this, {
-						ready: new Promise(resolve => this.resolveReady = resolve),
-						password: encodePassword(password, rawPassword),
-						strength: encryptionStrength - 1,
-						pending: new Uint8Array()
-					});
+					initAesCrypto(this, password, rawPassword, encryptionStrength);
 				},
 				async transform(chunk, controller) {
 					const aesCrypto = this;
@@ -1326,7 +1309,7 @@
 						await ready;
 						const chunkToDecrypt = subarray(pending, 0, pending.length - SIGNATURE_LENGTH);
 						const originalSignature = subarray(pending, pending.length - SIGNATURE_LENGTH);
-						let decryptedChunkArray = new Uint8Array();
+						let decryptedChunkArray = EMPTY_UINT8_ARRAY;
 						if (chunkToDecrypt.length) {
 							const encryptedChunk = toBits(codecBytes, chunkToDecrypt);
 							hmac.update(encryptedChunk);
@@ -1334,8 +1317,6 @@
 							decryptedChunkArray = fromBits(codecBytes, decryptedChunk);
 						}
 						const signature = subarray(fromBits(codecBytes, hmac.digest()), 0, SIGNATURE_LENGTH);
-						// branch-free comparison: accumulate the differences instead of
-						// short-circuiting on the first mismatching byte
 						let invalidSignature = pending.length < SIGNATURE_LENGTH ? 1 : 0;
 						for (let indexSignature = 0; indexSignature < SIGNATURE_LENGTH; indexSignature++) {
 							invalidSignature |= signature[indexSignature] ^ originalSignature[indexSignature];
@@ -1357,12 +1338,7 @@
 			let stream;
 			super({
 				start() {
-					Object.assign(this, {
-						ready: new Promise(resolve => this.resolveReady = resolve),
-						password: encodePassword(password, rawPassword),
-						strength: encryptionStrength - 1,
-						pending: new Uint8Array()
-					});
+					initAesCrypto(this, password, rawPassword, encryptionStrength);
 				},
 				async transform(chunk, controller) {
 					const aesCrypto = this;
@@ -1372,7 +1348,7 @@
 						resolveReady,
 						ready
 					} = aesCrypto;
-					let preamble = new Uint8Array();
+					let preamble = EMPTY_UINT8_ARRAY;
 					if (password) {
 						preamble = await createEncryptionKeys(aesCrypto, strength, password);
 						resolveReady();
@@ -1392,7 +1368,7 @@
 					} = this;
 					if (hmac && ctr) {
 						await ready;
-						let encryptedChunkArray = new Uint8Array();
+						let encryptedChunkArray = EMPTY_UINT8_ARRAY;
 						if (pending.length) {
 							const encryptedChunk = ctr.update(toBits(codecBytes, pending));
 							hmac.update(encryptedChunk);
@@ -1407,6 +1383,15 @@
 		}
 	}
 
+	function initAesCrypto(aesCrypto, password, rawPassword, encryptionStrength) {
+		Object.assign(aesCrypto, {
+			ready: new Promise(resolve => aesCrypto.resolveReady = resolve),
+			password: encodePassword(password, rawPassword),
+			strength: encryptionStrength - 1,
+			pending: EMPTY_UINT8_ARRAY
+		});
+	}
+
 	function append(aesCrypto, input, output, paddingStart, paddingEnd, verifySignature) {
 		const {
 			ctr,
@@ -1568,20 +1553,13 @@
 		constructor({ password, rawPassword, passwordVerification, checkPasswordOnly }) {
 			super({
 				start() {
-					Object.assign(this, {
-						password,
-						rawPassword,
-						passwordVerification
-					});
-					createKeys(this, password, rawPassword);
+					initZipCrypto(this, password, rawPassword, passwordVerification);
 				},
 				transform(chunk, controller) {
 					const zipCrypto = this;
 					if (zipCrypto.password || zipCrypto.rawPassword) {
 						const decryptedHeader = decrypt(zipCrypto, chunk.subarray(0, HEADER_LENGTH));
 						zipCrypto.password = zipCrypto.rawPassword = null;
-						// verify the password with a branch-free comparison of the verification byte, mirroring
-						// the constant-time AES HMAC signature check
 						if ((decryptedHeader.at(-1) ^ zipCrypto.passwordVerification) != 0) {
 							throw new Error(ERR_INVALID_PASSWORD);
 						}
@@ -1602,12 +1580,7 @@
 		constructor({ password, rawPassword, passwordVerification }) {
 			super({
 				start() {
-					Object.assign(this, {
-						password,
-						rawPassword,
-						passwordVerification
-					});
-					createKeys(this, password, rawPassword);
+					initZipCrypto(this, password, rawPassword, passwordVerification);
 				},
 				transform(chunk, controller) {
 					const zipCrypto = this;
@@ -1631,6 +1604,15 @@
 		}
 	}
 
+	function initZipCrypto(zipCrypto, password, rawPassword, passwordVerification) {
+		Object.assign(zipCrypto, {
+			password,
+			rawPassword,
+			passwordVerification
+		});
+		createKeys(zipCrypto, password, rawPassword);
+	}
+
 	function decrypt(target, input) {
 		const output = new Uint8Array(input.length);
 		for (let index = 0; index < input.length; index++) {
@@ -1735,14 +1717,6 @@
 			const stream = this;
 			let crc32Stream, encryptionStream, gzipCrc32Stream;
 			let readable = super.readable;
-			// The gzip trailer carries a CRC-32 of the uncompressed data (same polynomial as zip),
-			// computed in native code for free during compression. On the native CompressionStream,
-			// harvest it instead of running a separate CRC pass: compress as gzip, then strip the
-			// fixed 10-byte header and 8-byte trailer to recover the exact raw-deflate payload
-			// (byte-identical to a deflate-raw compression) while capturing the CRC. The CRC becomes
-			// available at end-of-stream, exactly like Crc32Stream, so nothing downstream (data
-			// descriptor, central directory, ZipCrypto) needs it any earlier. Not applied for the
-			// pure-JS/WASM ports (a separate slice-by-8 CRC pass is as fast or faster there).
 			const useGzipCrc32 = signed && compressed && !deflate64 && (!encrypted || zipCrypto) &&
 				Boolean(useCompressionStream && CompressionStream);
 			if ((!encrypted || zipCrypto) && signed && !useGzipCrc32) {
@@ -1779,12 +1753,6 @@
 		}
 	}
 
-	// Converts a gzip stream into its raw-deflate payload while capturing the CRC-32 from the gzip
-	// trailer. The CompressionStream gzip header is always the fixed 10-byte form (FLG=0, no optional
-	// fields) per the Compression Streams spec, so it is stripped by length; the trailer is the last
-	// 8 bytes (CRC-32 LE, then ISIZE LE). Bounded memory: at most GZIP_TRAILER_LENGTH bytes are held
-	// back across chunks. The signature is read big-endian-agnostically as a number, matching the
-	// value Crc32Stream produces, so the writer path is unchanged.
 	class GzipToRawDeflateStream extends TransformStream {
 
 		constructor() {
@@ -1810,10 +1778,6 @@
 						tail = pending;
 						return;
 					}
-					// Emit everything except the trailing GZIP_TRAILER_LENGTH bytes as a standalone,
-					// right-sized Uint8Array. Consumers may read chunk.buffer directly (e.g. custom
-					// writers), so an aliased subarray of a larger buffer would leak the held-back
-					// trailer bytes. Bytes are copied exactly once, into `output` or `tail`.
 					const emitLength = available - GZIP_TRAILER_LENGTH;
 					const output = new Uint8Array(emitLength);
 					const fromTail = Math.min(emitLength, tail.length);
@@ -1902,11 +1866,6 @@
 				throw error;
 			}
 		}
-		// The native CompressionStream/DecompressionStream and the pure-JS zlib port do not signal
-		// backpressure on their writable side (their `ready` never pends), so a plain pipeThrough drains
-		// the whole source into them and peak memory grows with the entry size. Awaiting each write()
-		// paces the source to the codec's consumption rate, keeping streaming memory bounded. The WASM
-		// codec already backpressures, so this is a no-op cost for it.
 		return pipeThroughBackpressured(readable, codecStream);
 	}
 
@@ -1914,10 +1873,6 @@
 		return readable.pipeThrough(transformStream);
 	}
 
-	// Like readable.pipeThrough(transformStream), but drives the writable side manually and awaits each
-	// write() so that a codec which under-reports backpressure cannot pull the whole source into memory.
-	// Errors propagate in both directions: a source error aborts the codec; a downstream cancel/error
-	// (surfaced through the codec's writable rejecting) cancels the source.
 	function pipeThroughBackpressured(readable, transformStream) {
 		const writer = transformStream.writable.getWriter();
 		const reader = readable.getReader();
@@ -1946,7 +1901,7 @@
 		try {
 			await writer.abort(error);
 		} catch {
-			// the writable may already be errored/closed
+			// ignored: the writable may already be errored/closed
 		}
 	}
 
@@ -1954,13 +1909,10 @@
 		try {
 			await reader.cancel(error);
 		} catch {
-			// the readable may already be errored/closed
+			// ignored: the readable may already be errored/closed
 		}
 	}
 
-	// DecompressionStream implementations can fail with a message-less TypeError on malformed
-	// input; give these errors an identifiable message (errors raised by the other stages of
-	// the pipeline, e.g. ERR_INVALID_PASSWORD, always carry a message and pass through as-is)
 	function mapInflateStreamError(readable) {
 		const reader = readable.getReader();
 		return new ReadableStream({
@@ -2285,9 +2237,6 @@
 				try {
 					await initModule(config);
 				} catch {
-					// The WASM module failed to load. Fall back to the native CompressionStream only if the
-					// zlib codec that would otherwise run depends on that module; a self-contained codec
-					// supplied through config (e.g. the pure-JS port) stays usable and is kept as requested.
 					const ZlibStream = options.codecType.startsWith(CODEC_DEFLATE) ?
 						config.CompressionStreamZlib :
 						config.DecompressionStreamZlib;
@@ -2365,7 +2314,6 @@
 			return resultValue;
 		} catch (error) {
 			await closeWritable();
-			// unblock the pipe watcher if the worker died before closing its end of the stream
 			abortPipe();
 			try {
 				await closed;
@@ -2390,8 +2338,6 @@
 		const abortController = new AbortController();
 		const { writable, readable } = new TransformStream();
 		const closed = readable.pipeTo(writableSource, { preventClose: true, preventAbort: true, signal: abortController.signal });
-		// the pipe can fail before `closed` is awaited (e.g. the destination write fails);
-		// mark the rejection as handled so it cannot kill the process in the meantime
 		closed.catch(() => { });
 		return { writable, closed, abortPipe: () => abortController.abort() };
 	}
@@ -2484,8 +2430,6 @@
 			event.preventDefault();
 		}
 		const { rejectResult, writer, onTaskFinished } = workerData;
-		// the worker is unusable (e.g. its script failed to load); terminate it and
-		// fail the current task instead of letting it hang forever
 		terminateWorker$1(workerData);
 		if (rejectResult) {
 			rejectResult(event.error || new Error(event.message || ERROR_EVENT_TYPE));
@@ -2537,8 +2481,6 @@
 	async function onMessage({ data }, workerData) {
 		const { type, value, messageId, result, error } = data;
 		const { reader, writer, resolveResult, rejectResult, onTaskFinished, generation } = workerData;
-		// continuations below can settle after the worker slot has been reassigned to a new
-		// task (e.g. a pending pull read failing late); they must not touch the new task
 		const stale = () => workerData.generation != generation;
 		try {
 			if (error) {
@@ -2794,26 +2736,24 @@
 	class Reader extends Stream {
 
 		get readable() {
+			return this.createReadable();
+		}
+
+		createReadable({ offset = 0, size, diskNumberStart, chunkSize = DEFAULT_CHUNK_SIZE } = {}) {
 			const reader = this;
-			const { chunkSize = DEFAULT_CHUNK_SIZE } = reader;
-			const readable = new ReadableStream({
-				start() {
-					this.chunkOffset = 0;
-				},
+			let chunkOffset = 0;
+			return new ReadableStream({
 				async pull(controller) {
-					const { offset = 0, size, diskNumberStart } = readable;
-					const { chunkOffset } = this;
 					const dataSize = size === UNDEFINED_VALUE ? chunkSize : Math.min(chunkSize, size - chunkOffset);
 					const data = await readUint8Array(reader, offset + chunkOffset, dataSize, diskNumberStart);
 					controller.enqueue(data);
 					if ((chunkOffset + chunkSize > size) || (size === UNDEFINED_VALUE && !data.length && dataSize)) {
 						controller.close();
 					} else {
-						this.chunkOffset += chunkSize;
+						chunkOffset += chunkSize;
 					}
 				}
 			});
-			return readable;
 		}
 	}
 
@@ -2952,8 +2892,6 @@
 				}
 			});
 			writer.blob = new Response(transformStream.readable, { headers }).blob();
-			// mark early rejections (e.g. aborted pipe) as handled; awaiting getData() still
-			// surfaces them
 			writer.blob.catch(() => { });
 		}
 
@@ -3115,7 +3053,7 @@
 				return eocdCache;
 			}
 			if (index >= size || length === 0) {
-				return new Uint8Array();
+				return EMPTY_UINT8_ARRAY;
 			} else {
 				if (index + length > size) {
 					length = size - index;
@@ -3172,8 +3110,6 @@
 	async function getRequestData(httpReader, sendRequest) {
 		const response = await sendRequest(HTTP_METHOD_GET, httpReader, getHeaders(httpReader));
 		httpReader.data = new Uint8Array(await response.arrayBuffer());
-		// the downloaded content is authoritative, e.g. the Content-Length response header
-		// of the HEAD request can be the encoded size when the content is compressed
 		httpReader.size = httpReader.data.length;
 	}
 
@@ -3238,7 +3174,6 @@
 			super();
 			Object.assign(this, {
 				url,
-				// an explicit fetch implementation takes precedence over useXHR
 				reader: options.useXHR && !options.fetch ? new XHRReader(url, options) : new FetchReader(url, options)
 			});
 		}
@@ -3377,7 +3312,7 @@
 					}
 				}
 			} else {
-				result = new Uint8Array();
+				result = EMPTY_UINT8_ARRAY;
 			}
 			reader.lastDiskNumber = Math.max(currentDiskNumber, reader.lastDiskNumber);
 			return result;
@@ -3422,8 +3357,6 @@
 						writer.diskOffset += diskSourceWriter.size;
 						writer.diskNumber++;
 						diskWriter = null;
-						// the next disk is not opened yet but its space is available; leaving a
-						// stale value here would force a new disk for every subsequent write
 						writer.availableSize = writer.maxSize;
 						if (chunk.length > availableSize) {
 							await this.write(chunk.subarray(availableSize));
@@ -3603,7 +3536,6 @@
 		if (encoding && encoding.trim().toLowerCase() == "cp437") {
 			return decodeCP437(value);
 		} else {
-			// ignoreBOM keeps a leading U+FEFF in filenames/comments instead of dropping it
 			return new TextDecoder(encoding, { ignoreBOM: true }).decode(value);
 		}
 	}
@@ -3851,8 +3783,6 @@
 	const CHARSET_UTF8 = "utf-8";
 	const PROPERTY_NAME_UTF8_SUFFIX = "UTF8";
 	const CHARSET_CP437 = "cp437";
-	// bits of the general purpose flag that change how an entry is interpreted; the level bits (1-2) are excluded
-	// because they are informative only
 	const BITFLAG_AMBIGUITY_MASK = BITFLAG_ENCRYPTED | BITFLAG_DATA_DESCRIPTOR | BITFLAG_LANG_ENCODING_FLAG;
 	const ZIP64_PROPERTIES = [
 		[PROPERTY_NAME_UNCOMPRESSED_SIZE, MAX_32_BITS],
@@ -3894,7 +3824,6 @@
 			if (reader.size < END_OF_CENTRAL_DIR_LENGTH) {
 				throw new Error(ERR_BAD_FORMAT);
 			}
-			reader.chunkSize = getChunkSize(config);
 			const strictness = getStrictness(getOptionValue$1(zipReader, options, OPTION_STRICTNESS), getOptionValue$1(zipReader, options, OPTION_CHECK_AMBIGUITY));
 			const checkAmbiguity = strictness == STRICTNESS_STRICT;
 			const rejectAmbiguousEndOfDirectory = strictness != STRICTNESS_TOLERANT;
@@ -3909,8 +3838,6 @@
 					throw new Error(ERR_EOCDR_NOT_FOUND);
 				}
 			}
-			// two or more end-anchored records that each dereference to a valid central directory cannot be told
-			// apart (see comment_length ⟺ EOF): the archive is genuinely ambiguous, so refuse rather than guess
 			if (rejectAmbiguousEndOfDirectory && endOfDirectoryReachingEndCount > 1) {
 				throwAmbiguousArchive("multiple end of central directory records");
 			}
@@ -3933,7 +3860,7 @@
 			if (directoryDataOffset == MAX_32_BITS || directoryDataLength == MAX_32_BITS || filesLength == MAX_16_BITS || diskNumber == MAX_16_BITS) {
 				const endOfDirectoryLocatorArray = endOfDirectoryInfo.offset >= ZIP64_END_OF_CENTRAL_DIR_LOCATOR_LENGTH ?
 					await readUint8Array(reader, endOfDirectoryInfo.offset - ZIP64_END_OF_CENTRAL_DIR_LOCATOR_LENGTH, ZIP64_END_OF_CENTRAL_DIR_LOCATOR_LENGTH) :
-					new Uint8Array();
+					EMPTY_UINT8_ARRAY;
 				const endOfDirectoryLocatorView = getDataView$1(endOfDirectoryLocatorArray);
 				if (endOfDirectoryLocatorArray.length == ZIP64_END_OF_CENTRAL_DIR_LOCATOR_LENGTH &&
 					getUint32(endOfDirectoryLocatorView, 0) == ZIP64_END_OF_CENTRAL_DIR_LOCATOR_SIGNATURE) {
@@ -4000,12 +3927,6 @@
 				}
 				const expectedDirectoryDataOffset = centralDirectoryEndOffset - directoryDataLength - (reader.lastDiskOffset || 0);
 				if (directoryDataOffset != expectedDirectoryDataOffset && diskNumber == lastDiskNumber) {
-					// the reconciled offset (the directory ends exactly where the canonical record begins) is anchored
-					// to the unforgeable end of the archive; the stored offset is not. Prefer the reconciled offset
-					// unless the stored one points at a directory and the reconciled one does not — that exception
-					// keeps a corrupt declared directory length from sending the read astray, while still moving off a
-					// stored offset that only looks valid because an append remnant left the previous (identically
-					// laid out) directory sitting there.
 					const storedPointsAtDirectory = getUint32(directoryView, offset) == CENTRAL_FILE_HEADER_SIGNATURE;
 					let reconcile = !storedPointsAtDirectory;
 					if (!reconcile && expectedDirectoryDataOffset >= 0 && expectedDirectoryDataOffset + 4 <= reader.size) {
@@ -4162,11 +4083,11 @@
 			const extractPrependedData = getOptionValue$1(zipReader, options, OPTION_EXTRACT_PREPENDED_DATA);
 			const extractAppendedData = getOptionValue$1(zipReader, options, OPTION_EXTRACT_APPENDED_DATA);
 			if (extractPrependedData) {
-				zipReader.prependedData = startOffset > 0 ? await readUint8Array(reader, 0, startOffset) : new Uint8Array();
+				zipReader.prependedData = startOffset > 0 ? await readUint8Array(reader, 0, startOffset) : EMPTY_UINT8_ARRAY;
 			}
-			zipReader.comment = commentLength ? await readUint8Array(reader, commentOffset + END_OF_CENTRAL_DIR_LENGTH, commentLength) : new Uint8Array();
+			zipReader.comment = commentLength ? await readUint8Array(reader, commentOffset + END_OF_CENTRAL_DIR_LENGTH, commentLength) : EMPTY_UINT8_ARRAY;
 			if (extractAppendedData) {
-				zipReader.appendedData = appendedDataOffset < reader.size ? await readUint8Array(reader, appendedDataOffset, reader.size - appendedDataOffset) : new Uint8Array();
+				zipReader.appendedData = appendedDataOffset < reader.size ? await readUint8Array(reader, appendedDataOffset, reader.size - appendedDataOffset) : EMPTY_UINT8_ARRAY;
 			}
 			return true;
 		}
@@ -4278,7 +4199,7 @@
 				filenameLength
 			} = localDirectory;
 			const checkAmbiguity = getStrictness(getOptionValue$1(zipEntry, options, OPTION_STRICTNESS), getOptionValue$1(zipEntry, options, OPTION_CHECK_AMBIGUITY)) == STRICTNESS_STRICT;
-			let rawLocalFilename = new Uint8Array();
+			let rawLocalFilename = EMPTY_UINT8_ARRAY;
 			if (checkAmbiguity && (filenameLength || extraFieldLength)) {
 				const trailingDataArray = await readUint8Array(reader, offset + HEADER_SIZE, filenameLength + extraFieldLength, diskNumberStart);
 				rawLocalFilename = trailingDataArray.subarray(0, filenameLength);
@@ -4286,7 +4207,7 @@
 			} else {
 				localDirectory.rawExtraField = extraFieldLength ?
 					await readUint8Array(reader, offset + HEADER_SIZE + filenameLength, extraFieldLength, diskNumberStart) :
-					new Uint8Array();
+					EMPTY_UINT8_ARRAY;
 			}
 			readCommonFooter(zipEntry, localDirectory, dataView, 4, true);
 			if (checkAmbiguity) {
@@ -4313,12 +4234,7 @@
 			}
 			const dataOffset = offset + HEADER_SIZE + filenameLength + extraFieldLength;
 			const size = compressedSize;
-			const readable = reader.readable;
-			Object.assign(readable, {
-				diskNumberStart,
-				offset: dataOffset,
-				size
-			});
+			const readable = reader.createReadable({ offset: dataOffset, size, diskNumberStart, chunkSize: getChunkSize(config) });
 			const signal = getOptionValue$1(zipEntry, options, OPTION_SIGNAL);
 			const checkPasswordOnly = getOptionValue$1(zipEntry, options, OPTION_CHECK_PASSWORD_ONLY);
 			let checkOverlappingEntry = getOptionValue$1(zipEntry, options, OPTION_CHECK_OVERLAPPING_ENTRY);
@@ -4589,7 +4505,6 @@
 			const view = getDataView$1(new Uint8Array(extraField.data));
 			let uid, gid;
 			if (isInfoZip) {
-				// Info-ZIP New Unix "ux" field (0x7875): version + variable-length uid/gid
 				let offset = 0;
 				const version = getUint8(view, offset++);
 				const uidSize = getUint8(view, offset++);
@@ -4599,8 +4514,6 @@
 				gid = unpackUnixId(extraField.data.subarray(offset, offset + gidSize));
 				Object.assign(extraField, { version, uid, gid });
 			} else if (extraField.data.length >= 4) {
-				// Info-ZIP Unix "Ux" field (0x7855): fixed 2-byte uid + gid (the file mode, if any, is
-				// carried in the external file attributes, not here)
 				uid = getUint16(view, 0);
 				gid = getUint16(view, 2);
 				Object.assign(extraField, { uid, gid });
@@ -4652,7 +4565,6 @@
 		timeProperties.forEach((propertyName, indexProperty) => {
 			if (extraFieldExtendedTimestamp.data.length >= offset + 4) {
 				const time = getUint32(extraFieldView, offset);
-				// the extended timestamp is a signed 32-bit time_t value
 				directory[propertyName] = extraFieldExtendedTimestamp[propertyName] = new Date((time | 0) * 1000);
 				const rawPropertyName = timeRawProperties[indexProperty];
 				extraFieldExtendedTimestamp[rawPropertyName] = time;
@@ -4729,7 +4641,6 @@
 
 	function getStrictness(strictness, checkAmbiguity) {
 		if (strictness === UNDEFINED_VALUE) {
-			// `checkAmbiguity: true` is kept as a backward-compatible alias for the strictest mode
 			return checkAmbiguity ? STRICTNESS_STRICT : STRICTNESS_BALANCED;
 		}
 		return strictness;
@@ -4740,81 +4651,45 @@
 			return maxAppendedDataSize;
 		}
 		if (strictness == STRICTNESS_STRICT) {
-			// the comment must reach the end of the file: no trailing data tolerated
 			return 0;
 		}
 		if (strictness == STRICTNESS_TOLERANT) {
 			return Infinity;
 		}
-		// balanced: tolerate up to a legal comment's worth of trailing data (a 16-bit length); beyond that the
-		// trailing bytes cannot hide inside a comment, so they are treated as a second thing bolted onto the file
 		return MAX_16_BITS;
 	}
 
-	// cap on how many out-of-window central directory probes the scan below may issue. A legitimate archive needs
-	// at most a couple (the canonical record, plus one more to detect genuine ambiguity); every other reachability
-	// check is served from the tail window already in memory. The cap keeps an archive whose comment is stuffed
-	// with unreachable end-anchored records from forcing an unbounded number of (potentially remote) reads.
 	const MAX_END_OF_CENTRAL_DIR_PROBES = 64;
 
-	// reachability rankings for an end-anchored candidate record (see getCentralDirectoryReachability)
 	const CENTRAL_DIRECTORY_UNREACHABLE = 0;
 	const CENTRAL_DIRECTORY_PLAUSIBLE = 1;
 	const CENTRAL_DIRECTORY_REACHABLE = 2;
 
-	// Locates the authoritative end of central directory record. The canonical record is the last one whose
-	// declared comment reaches exactly the end of the file ("end-anchored") and that points to a central directory
-	// ("reachable"); earlier signatures (stale append remnants, bytes embedded in a comment) are ignored. When two
-	// or more end-anchored records point to a central directory the archive is ambiguous, and the count is returned
-	// so the caller can refuse it. A record that reaches the end of the file but points to no central directory (an
-	// empty archive) is only "plausible": it cannot be corroborated, so it never counts towards ambiguity and is
-	// chosen only when no reachable record exists — otherwise an empty record forged in a comment could outrank the
-	// genuine directory. When no record is end-anchored (appended data or an oversized comment), falls back to the
-	// last record within the tolerated window that points to a directory (see seekEndOfCentralDirectory).
 	async function findEndOfCentralDirectory(reader, rejectAmbiguous, maxAppendedDataSize) {
 		const { size } = reader;
-		// an end-anchored record can only live within the last END_OF_CENTRAL_DIR_LENGTH + MAX_16_BITS bytes,
-		// since the comment length that ties it to the end of the file is a 16-bit field
 		const anchoredLength = Math.min(size, END_OF_CENTRAL_DIR_LENGTH + MAX_16_BITS);
-		const anchoredOffset = size - anchoredLength;
-		const anchoredArray = await readUint8Array(reader, anchoredOffset, anchoredLength);
-		const anchoredView = getDataView$1(anchoredArray);
 		const remoteProbeBudget = { count: MAX_END_OF_CENTRAL_DIR_PROBES };
 		let endOfDirectoryInfo;
 		let plausibleEndOfDirectoryInfo;
 		let endOfDirectoryReachingEndCount = 0;
-		for (let indexByte = anchoredArray.length - END_OF_CENTRAL_DIR_LENGTH; indexByte >= 0; indexByte--) {
-			if (getUint32(anchoredView, indexByte) == END_OF_CENTRAL_DIR_SIGNATURE) {
-				const offset = anchoredOffset + indexByte;
-				const commentLength = getUint16(anchoredView, indexByte + 20);
-				// end-anchored: the declared comment extends exactly to the end of the file. The comment length is
-				// attacker-controlled, but the reconciliation against the (unforgeable) end of file is not.
-				if (offset + END_OF_CENTRAL_DIR_LENGTH + commentLength == size) {
-					const reachability = await getCentralDirectoryReachability(reader, anchoredView, anchoredOffset, indexByte, offset, size, remoteProbeBudget);
-					if (reachability == CENTRAL_DIRECTORY_REACHABLE) {
-						if (!endOfDirectoryInfo) {
-							endOfDirectoryInfo = {
-								offset,
-								buffer: anchoredArray.slice(indexByte, indexByte + END_OF_CENTRAL_DIR_LENGTH).buffer
-							};
-						}
-						endOfDirectoryReachingEndCount++;
-						// the canonical record (highest offset) is found first; a second one is enough to flag ambiguity
-						if (!rejectAmbiguous || endOfDirectoryReachingEndCount > 1) {
-							break;
-						}
-					} else if (reachability == CENTRAL_DIRECTORY_PLAUSIBLE && !plausibleEndOfDirectoryInfo) {
-						plausibleEndOfDirectoryInfo = {
-							offset,
-							buffer: anchoredArray.slice(indexByte, indexByte + END_OF_CENTRAL_DIR_LENGTH).buffer
-						};
+		for await (const [anchoredView, anchoredOffset, anchoredArray, indexByte, offset] of scanEndOfCentralDirectory(reader, anchoredLength)) {
+			const commentLength = getUint16(anchoredView, indexByte + 20);
+			if (offset + END_OF_CENTRAL_DIR_LENGTH + commentLength == size) {
+				const reachability = await getCentralDirectoryReachability(reader, anchoredView, anchoredOffset, indexByte, offset, size, remoteProbeBudget);
+				if (reachability == CENTRAL_DIRECTORY_REACHABLE) {
+					if (!endOfDirectoryInfo) {
+						endOfDirectoryInfo = getEndOfCentralDirectoryInfo(anchoredArray, indexByte, offset);
 					}
+					endOfDirectoryReachingEndCount++;
+					if (!rejectAmbiguous || endOfDirectoryReachingEndCount > 1) {
+						break;
+					}
+				} else if (reachability == CENTRAL_DIRECTORY_PLAUSIBLE && !plausibleEndOfDirectoryInfo) {
+					plausibleEndOfDirectoryInfo = getEndOfCentralDirectoryInfo(anchoredArray, indexByte, offset);
 				}
 			}
 		}
 		if (!endOfDirectoryInfo) {
-			// no record points to a directory: prefer a plausible (empty) end-anchored record before scanning the
-			// tolerated window for a bare signature
 			endOfDirectoryInfo = plausibleEndOfDirectoryInfo;
 		}
 		if (!endOfDirectoryInfo) {
@@ -4823,69 +4698,53 @@
 		return { endOfDirectoryInfo, endOfDirectoryReachingEndCount };
 	}
 
-	// Fallback for findEndOfCentralDirectory when no record is end-anchored: appended data or an oversized comment
-	// pushed the record's declared comment past the end of the file, so its position can no longer be reconciled
-	// against the end of the file. Scans the tolerated window from the end and returns the first record that points
-	// to a central directory, so a stray signature embedded in appended data cannot hijack discovery (which the
-	// anchored scan already prevents for anchored records). Falls back to the first plausible (empty) record, then
-	// to the last bare signature, so a degenerate archive still opens as before. The window ends at the file, so the
-	// same in-window / metered reads as the anchored scan apply.
 	async function seekEndOfCentralDirectory(reader, maxAppendedDataSize, remoteProbeBudget) {
 		const { size } = reader;
 		const searchLength = Math.min(size, maxAppendedDataSize == Infinity ? size :
 			END_OF_CENTRAL_DIR_LENGTH + MAX_16_BITS + maxAppendedDataSize);
-		const searchOffset = size - searchLength;
-		const searchArray = await readUint8Array(reader, searchOffset, searchLength);
-		const searchView = getDataView$1(searchArray);
 		let firstSignatureInfo, plausibleInfo;
-		for (let indexByte = searchArray.length - END_OF_CENTRAL_DIR_LENGTH; indexByte >= 0; indexByte--) {
-			if (getUint32(searchView, indexByte) == END_OF_CENTRAL_DIR_SIGNATURE) {
-				const offset = searchOffset + indexByte;
-				const record = { offset, buffer: searchArray.slice(indexByte, indexByte + END_OF_CENTRAL_DIR_LENGTH).buffer };
-				if (!firstSignatureInfo) {
-					firstSignatureInfo = record;
-				}
-				const reachability = await getCentralDirectoryReachability(reader, searchView, searchOffset, indexByte, offset, size, remoteProbeBudget);
-				if (reachability == CENTRAL_DIRECTORY_REACHABLE) {
-					return record;
-				}
-				if (reachability == CENTRAL_DIRECTORY_PLAUSIBLE && !plausibleInfo) {
-					plausibleInfo = record;
-				}
+		for await (const [searchView, searchOffset, searchArray, indexByte, offset] of scanEndOfCentralDirectory(reader, searchLength)) {
+			const record = getEndOfCentralDirectoryInfo(searchArray, indexByte, offset);
+			if (!firstSignatureInfo) {
+				firstSignatureInfo = record;
+			}
+			const reachability = await getCentralDirectoryReachability(reader, searchView, searchOffset, indexByte, offset, size, remoteProbeBudget);
+			if (reachability == CENTRAL_DIRECTORY_REACHABLE) {
+				return record;
+			}
+			if (reachability == CENTRAL_DIRECTORY_PLAUSIBLE && !plausibleInfo) {
+				plausibleInfo = record;
 			}
 		}
 		return plausibleInfo || firstSignatureInfo;
 	}
 
-	// Ranks how strongly an end-anchored candidate record is backed by an actual central directory, used to decide
-	// which record is canonical and which count towards ambiguity. It is intentionally conservative and does not
-	// fully parse the directory (the caller does that for the canonical record):
-	//   - REACHABLE: the record dereferences a central directory (a central file header signature at the
-	//     reconciled offset), or, for a zip64 record, the zip64 locator that a genuine zip64 archive places
-	//     immediately before it. Only these count towards ambiguity.
-	//   - PLAUSIBLE: the record describes an empty archive, which has no central directory to dereference. Its
-	//     fields are indistinguishable from an empty end of central directory record forged inside a comment, so it
-	//     is accepted only as a last resort and never inflates the ambiguity count.
-	//   - UNREACHABLE: neither — stale append remnant or signature bytes embedded in a comment.
-	// Signatures are read from the tail window (`view`, spanning `[anchoredOffset, size)`) when they fall inside it;
-	// only a target below the window costs a (possibly remote) read, metered by `remoteProbeBudget` so a stuffed
-	// comment cannot amplify into an unbounded number of requests.
+	async function* scanEndOfCentralDirectory(reader, scanLength) {
+		const scanOffset = reader.size - scanLength;
+		const scanArray = await readUint8Array(reader, scanOffset, scanLength);
+		const scanView = getDataView$1(scanArray);
+		for (let indexByte = scanArray.length - END_OF_CENTRAL_DIR_LENGTH; indexByte >= 0; indexByte--) {
+			if (getUint32(scanView, indexByte) == END_OF_CENTRAL_DIR_SIGNATURE) {
+				yield [scanView, scanOffset, scanArray, indexByte, scanOffset + indexByte];
+			}
+		}
+	}
+
+	function getEndOfCentralDirectoryInfo(scanArray, indexByte, offset) {
+		return { offset, buffer: scanArray.slice(indexByte, indexByte + END_OF_CENTRAL_DIR_LENGTH).buffer };
+	}
+
 	async function getCentralDirectoryReachability(reader, view, anchoredOffset, indexByte, offset, size, remoteProbeBudget) {
 		const filesLength = getUint16(view, indexByte + 10);
 		const directoryDataLength = getUint32(view, indexByte + 12);
 		const directoryDataOffset = getUint32(view, indexByte + 16);
 		if (filesLength == MAX_16_BITS || directoryDataLength == MAX_32_BITS || directoryDataOffset == MAX_32_BITS) {
-			// saturated fields imply a zip64 record; corroborate it with the zip64 end of central directory locator
-			// that sits immediately before it, so saturated bytes inside a comment cannot masquerade as one
 			const locatorSignature = await readSignature(reader, view, anchoredOffset, offset - ZIP64_END_OF_CENTRAL_DIR_LOCATOR_LENGTH, size, remoteProbeBudget);
 			return locatorSignature == ZIP64_END_OF_CENTRAL_DIR_LOCATOR_SIGNATURE ? CENTRAL_DIRECTORY_REACHABLE : CENTRAL_DIRECTORY_UNREACHABLE;
 		}
 		if (!filesLength && !directoryDataLength) {
-			// a valid empty archive has no central directory to dereference
 			return CENTRAL_DIRECTORY_PLAUSIBLE;
 		}
-		// the central directory ends where this record starts; reconcile the stored offset against that actual
-		// position (see stored_cd_offset ⟺ eocd_pos − cd_size) so prepended data does not hide the directory
 		for (const centralDirectoryOffset of [offset - directoryDataLength, directoryDataOffset]) {
 			if (await readSignature(reader, view, anchoredOffset, centralDirectoryOffset, size, remoteProbeBudget) == CENTRAL_FILE_HEADER_SIGNATURE) {
 				return CENTRAL_DIRECTORY_REACHABLE;
@@ -4894,15 +4753,11 @@
 		return CENTRAL_DIRECTORY_UNREACHABLE;
 	}
 
-	// Reads a 4-byte little-endian signature at `signatureOffset`, served from the tail window (`view`, spanning
-	// `[anchoredOffset, size)`) when it falls inside it and otherwise from a metered (possibly remote) read. Returns
-	// UNDEFINED_VALUE when the position is out of range or the out-of-window read budget is exhausted.
 	async function readSignature(reader, view, anchoredOffset, signatureOffset, size, remoteProbeBudget) {
 		if (signatureOffset < 0 || signatureOffset + 4 > size) {
 			return UNDEFINED_VALUE;
 		}
 		if (signatureOffset >= anchoredOffset) {
-			// the target sits inside the tail window already in memory: no extra read needed
 			return getUint32(view, signatureOffset - anchoredOffset);
 		}
 		if (remoteProbeBudget.count > 0) {
@@ -4910,7 +4765,6 @@
 			const signatureArray = await readUint8Array(reader, signatureOffset, 4);
 			return getUint32(getDataView$1(signatureArray), 0);
 		}
-		// out of budget: leave this candidate unverified rather than issue another read
 		return UNDEFINED_VALUE;
 	}
 
@@ -4926,8 +4780,6 @@
 		if (localDirectory.compressionMethod != zipEntry.compressionMethod) {
 			throwAmbiguousArchive("mismatched local file header (compression method)");
 		}
-		// when the data descriptor flag is set, the local signature and sizes are zeroed out; all-zero values are
-		// also tolerated when the flag is clear because some tools produce them (streaming writers predating bit 3)
 		if (!localDirectory.bitFlag.dataDescriptor &&
 			(localDirectory.signature || localDirectory.compressedSize || localDirectory.uncompressedSize) &&
 			(localDirectory.signature != zipEntry.signature ||
@@ -5017,7 +4869,6 @@
 	const ERR_INVALID_ENCRYPTION_STRENGTH = "The strength must equal 1, 2, or 3";
 	const ERR_INVALID_EXTRAFIELD_TYPE = "Extra field type exceeds 65535";
 	const ERR_INVALID_EXTRAFIELD_DATA = "Extra field data exceeds 64KB";
-	// same message as the constant exported by zip-reader.js
 	const ERR_UNSUPPORTED_COMPRESSION = "Compression method not supported";
 	const MIN_UNIX_TIME = -2147483648;
 	const MAX_UNIX_TIME = 2147483647;
@@ -5100,12 +4951,12 @@
 					rawExtraField,
 				} = entry;
 				const { level, languageEncodingFlag, dataDescriptor } = bitFlag;
-				rawExtraFieldZip64 = rawExtraFieldZip64 || new Uint8Array();
-				rawExtraFieldAES = rawExtraFieldAES || new Uint8Array();
-				rawExtraFieldExtendedTimestamp = rawExtraFieldExtendedTimestamp || new Uint8Array();
-				rawExtraFieldNTFS = rawExtraFieldNTFS || new Uint8Array();
-				rawExtraFieldUnix = rawExtraFieldUnix || new Uint8Array();
-				rawExtraField = rawExtraField || new Uint8Array();
+				rawExtraFieldZip64 = rawExtraFieldZip64 || EMPTY_UINT8_ARRAY;
+				rawExtraFieldAES = rawExtraFieldAES || EMPTY_UINT8_ARRAY;
+				rawExtraFieldExtendedTimestamp = rawExtraFieldExtendedTimestamp || EMPTY_UINT8_ARRAY;
+				rawExtraFieldNTFS = rawExtraFieldNTFS || EMPTY_UINT8_ARRAY;
+				rawExtraFieldUnix = rawExtraFieldUnix || EMPTY_UINT8_ARRAY;
+				rawExtraField = rawExtraField || EMPTY_UINT8_ARRAY;
 				if (entry.extraFieldAES) {
 					compressionMethod = COMPRESSION_METHOD_AES;
 				}
@@ -5211,7 +5062,7 @@
 			return false;
 		}
 
-		async close(comment = new Uint8Array(), options = {}) {
+		async close(comment = EMPTY_UINT8_ARRAY, options = {}) {
 			const zipWriter = this;
 			const { pendingAddFileCalls, writer } = this;
 			const { writable } = writer;
@@ -5510,7 +5361,7 @@
 		if (!zipCrypto && (password !== UNDEFINED_VALUE || rawPassword !== UNDEFINED_VALUE) && !(encryptionStrength >= 1 && encryptionStrength <= 3)) {
 			throw new Error(ERR_INVALID_ENCRYPTION_STRENGTH);
 		}
-		let rawExtraField = new Uint8Array();
+		let rawExtraField = EMPTY_UINT8_ARRAY;
 		const extraField = options[PROPERTY_NAME_EXTRA_FIELD];
 		if (extraField) {
 			let extraFieldSize = 0;
@@ -5571,8 +5422,6 @@
 		let maximumCompressedSize = 0;
 		let uncompressedSize = 0;
 		if (passThrough) {
-			// the headers of a passThrough entry describe the payload verbatim: without a
-			// reader they would declare content that is not there
 			if (!reader) {
 				throw new Error(ERR_UNDEFINED_READER);
 			}
@@ -5583,8 +5432,6 @@
 		}
 		const zip64Enabled = zip64 === true;
 		const encrypted = getOptionValue(zipWriter, options, PROPERTY_NAME_ENCRYPTED);
-		// entries without a reader have no payload: write them as empty stored files instead of
-		// declaring a compressed or encrypted content that does not exist
 		const encryptedEntry = Boolean(reader) && (Boolean((password && getLength(password)) || (rawPassword && getLength(rawPassword))) || (passThrough && encrypted));
 		if (!reader) {
 			level = 0;
@@ -5722,7 +5569,6 @@
 				fileEntry.diskNumberStart = writer.diskNumber;
 				fileEntry.offset = getSegmentOffset(zipWriter, writer);
 				if (usdz) {
-					// the padding depends on the entry offset, only known now that the writer is locked
 					const previousMetadataSize = entryInfo.metadataSize;
 					appendExtraFieldUSDZ(entryInfo, zipWriter.offset - writer.diskOffset);
 					fileEntry.size += entryInfo.metadataSize - previousMetadataSize;
@@ -5765,10 +5611,6 @@
 			if (releaseLockWriter) {
 				releaseLockWriter();
 			}
-			// A buffered entry uses a temporary stream (see `createTempStream`). When that stream is
-			// backed by a resource (a file, an OPFS handle, ...) it must be released on every exit path,
-			// including the error path where the stream is otherwise abandoned without being closed or
-			// cancelled. `dispose()` is optional and best-effort so it never masks the original outcome.
 			if (bufferedWrite && fileWriter && fileWriter.dispose) {
 				try {
 					await fileWriter.dispose();
@@ -5795,7 +5637,7 @@
 		async function skipDiskIfNeeded() {
 			if (getLength(headerInfo.localHeaderArray) > writer.availableSize) {
 				writer.availableSize = 0;
-				await writeData(writer, new Uint8Array());
+				await writeData(writer, EMPTY_UINT8_ARRAY);
 			}
 		}
 	}
@@ -5906,8 +5748,7 @@
 		}
 		const { writable } = writer;
 		if (reader) {
-			reader.chunkSize = getChunkSize(config);
-			const readable = reader.readable;
+			const readable = reader.createReadable ? reader.createReadable({ chunkSize: getChunkSize(config) }) : reader.readable;
 			const size = reader.size;
 			const workerOptions = {
 				options: {
@@ -6007,13 +5848,9 @@
 			: compressionMethod !== COMPRESSION_METHOD_STORE);
 		let rawLocalExtraFieldZip64;
 		const uncompressedFile = passThrough || !compressed;
-		// entries without a data descriptor are always buffered (cf. getFileEntry), so the
-		// zip64 extra field values can be patched into the local header before flushing it
 		const zip64ExtraFieldComplete = zip64 && (options.bufferedWrite || !dataDescriptor || ((!zip64UncompressedSize && !zip64CompressedSize) || uncompressedFile));
 		const writeLocalExtraFieldZip64 = zip64ExtraFieldComplete || (zip64 && dataDescriptor && (zip64UncompressedSize || zip64CompressedSize));
 		if (zip64 && (zip64UncompressedSize || zip64CompressedSize)) {
-			// APPNOTE 4.5.3: the zip64 extra field of a local header must include both the
-			// uncompressed and compressed size fields
 			const length = 4 + 16;
 			const extraFieldZip64 = createRecordWriter(length);
 			extraFieldZip64.uint16(EXTRAFIELD_TYPE_ZIP64);
@@ -6027,7 +5864,7 @@
 				}
 			}
 		} else {
-			rawLocalExtraFieldZip64 = new Uint8Array();
+			rawLocalExtraFieldZip64 = EMPTY_UINT8_ARRAY;
 		}
 		let rawExtraFieldAES;
 		if (encrypted && !zipCrypto) {
@@ -6037,7 +5874,7 @@
 			rawExtraFieldAES = extraFieldAES.array;
 			rawExtraFieldAES[8] = encryptionStrength;
 		} else {
-			rawExtraFieldAES = new Uint8Array();
+			rawExtraFieldAES = EMPTY_UINT8_ARRAY;
 		}
 		let rawExtraFieldNTFS;
 		let rawExtraFieldExtendedTimestamp;
@@ -6060,8 +5897,7 @@
 				}
 				rawExtraFieldExtendedTimestamp = extraFieldTimestamp.array;
 			} else {
-				// the date cannot be stored as a 32-bit time_t; the NTFS field replaces it
-				rawExtraFieldExtendedTimestamp = new Uint8Array();
+				rawExtraFieldExtendedTimestamp = EMPTY_UINT8_ARRAY;
 			}
 			try {
 				const lastModTimeNTFS = getTimeNTFS(lastModDate);
@@ -6076,10 +5912,10 @@
 				extraFieldNTFS.uint64(getTimeNTFS(creationDate) || lastModTimeNTFS);
 				rawExtraFieldNTFS = extraFieldNTFS.array;
 			} catch {
-				rawExtraFieldNTFS = new Uint8Array();
+				rawExtraFieldNTFS = EMPTY_UINT8_ARRAY;
 			}
 		} else {
-			rawExtraFieldNTFS = rawExtraFieldExtendedTimestamp = new Uint8Array();
+			rawExtraFieldNTFS = rawExtraFieldExtendedTimestamp = EMPTY_UINT8_ARRAY;
 		}
 		let rawExtraFieldUnix;
 		try {
@@ -6105,10 +5941,10 @@
 				extraFieldUnix.uint16((gid === UNDEFINED_VALUE ? 0 : gid) & MAX_16_BITS);
 				rawExtraFieldUnix = extraFieldUnix.array;
 			} else {
-				rawExtraFieldUnix = new Uint8Array();
+				rawExtraFieldUnix = EMPTY_UINT8_ARRAY;
 			}
 		} catch {
-			rawExtraFieldUnix = new Uint8Array();
+			rawExtraFieldUnix = EMPTY_UINT8_ARRAY;
 		}
 		if (compressionMethod === UNDEFINED_VALUE) {
 			compressionMethod = compressed ? COMPRESSION_METHOD_DEFLATE : COMPRESSION_METHOD_STORE;
@@ -6175,7 +6011,7 @@
 			version,
 			compressionMethod,
 			extraFieldExtendedTimestampFlag,
-			rawExtraFieldZip64: new Uint8Array(),
+			rawExtraFieldZip64: EMPTY_UINT8_ARRAY,
 			localExtraFieldZip64Length,
 			rawExtraFieldExtendedTimestamp,
 			rawExtraFieldNTFS,
@@ -6188,7 +6024,6 @@
 	function appendExtraFieldUSDZ(entryInfo, zipWriterOffset) {
 		const { headerInfo } = entryInfo;
 		let { localHeaderArray, extraFieldLength } = headerInfo;
-		let localHeaderArrayView;
 		let extraBytesLength = 64 - ((zipWriterOffset + getLength(localHeaderArray)) % 64);
 		if (extraBytesLength < 4) {
 			extraBytesLength += 64;
@@ -6201,7 +6036,7 @@
 		headerInfo.localHeaderArray = localHeaderArray = new Uint8Array(getLength(previousLocalHeaderArray) + extraBytesLength);
 		arraySet(localHeaderArray, previousLocalHeaderArray);
 		arraySet(localHeaderArray, rawExtraFieldUSDZ, getLength(previousLocalHeaderArray));
-		localHeaderArrayView = getDataView(localHeaderArray);
+		const localHeaderArrayView = getDataView(localHeaderArray);
 		setUint16(localHeaderArrayView, 28, extraFieldLength + extraBytesLength);
 		headerInfo.localHeaderView = localHeaderArrayView;
 		entryInfo.metadataSize += extraBytesLength;
@@ -6209,7 +6044,7 @@
 
 	function packUnixId(id) {
 		if (id === UNDEFINED_VALUE) {
-			return new Uint8Array();
+			return EMPTY_UINT8_ARRAY;
 		} else {
 			const dataArray = new Uint8Array(4);
 			const dataView = getDataView(dataArray);
@@ -6252,7 +6087,7 @@
 		dataDescriptor,
 		dataDescriptorSignature
 	}) {
-		let dataDescriptorArray = new Uint8Array();
+		let dataDescriptorArray = EMPTY_UINT8_ARRAY;
 		let dataDescriptorView, dataDescriptorOffset = 0;
 		let dataDescriptorLength = zip64 ? DATA_DESCRIPTOR_RECORD_ZIP_64_LENGTH : DATA_DESCRIPTOR_RECORD_LENGTH;
 		if (dataDescriptorSignature) {
@@ -6336,7 +6171,6 @@
 			}
 		}
 		if (zip64 && localExtraFieldZip64Length) {
-			// the local zip64 extra field always contains both size slots
 			const localHeaderOffset = HEADER_SIZE + getLength(rawFilename) + 4;
 			setBigUint64(localHeaderView, localHeaderOffset, BigInt(uncompressedSize));
 			setBigUint64(localHeaderView, localHeaderOffset + 8, BigInt(compressedSize));
@@ -6390,7 +6224,7 @@
 				}
 				rawExtraFieldZip64 = extraFieldZip64.array;
 			} else {
-				rawExtraFieldZip64 = new Uint8Array();
+				rawExtraFieldZip64 = EMPTY_UINT8_ARRAY;
 			}
 			fileEntry.rawExtraFieldZip64 = rawExtraFieldZip64;
 			fileEntry.zip64Offset = zip64Offset;
@@ -6405,7 +6239,7 @@
 				extraFieldTimestamp.uint32(lastModTimeUnix);
 				rawExtraFieldTimestamp = extraFieldTimestamp.array;
 			} else {
-				rawExtraFieldTimestamp = new Uint8Array();
+				rawExtraFieldTimestamp = EMPTY_UINT8_ARRAY;
 			}
 			fileEntry.rawExtraFieldExtendedTimestamp = rawExtraFieldTimestamp;
 			const extraFieldLength = getLength(
@@ -6461,7 +6295,7 @@
 				await writeData(writer, directoryArray.slice(directoryDiskOffset, offset));
 				directoryDiskOffset = offset;
 				writer.availableSize = 0;
-				await writeData(writer, new Uint8Array());
+				await writeData(writer, EMPTY_UINT8_ARRAY);
 			}
 			if (indexFileEntry == 0) {
 				cdStartDiskNumber = writer.diskNumber;
@@ -6534,7 +6368,7 @@
 		const endOfdirectoryRecord = createRecordWriter(zip64 ? ZIP64_END_OF_CENTRAL_DIR_TOTAL_LENGTH : END_OF_CENTRAL_DIR_LENGTH);
 		if (getLength(endOfdirectoryRecord.array) + commentLength > writer.availableSize) {
 			writer.availableSize = 0;
-			await writeData(writer, new Uint8Array());
+			await writeData(writer, EMPTY_UINT8_ARRAY);
 		}
 		lastDiskNumber = writer.diskNumber;
 		if (zip64) {
@@ -6631,7 +6465,6 @@
 		return Math.floor(date.getTime() / 1000);
 	}
 
-	// the extended timestamp extra field stores dates as signed 32-bit time_t values
 	function inUnixTimeRange(timeUnix) {
 		return timeUnix >= MIN_UNIX_TIME && timeUnix <= MAX_UNIX_TIME;
 	}
@@ -6816,36 +6649,12 @@
 	const DEFAULT_THRESHOLD = 1024 * 1024;
 	const DEFAULT_DIRECTORY_NAME = ".zip.js-temp";
 
-	// Builds a `createTempStream` factory (see `ZipWriter`'s option of the same name) that spills the
-	// data of a buffered entry to the Origin Private File System (OPFS) instead of keeping it in memory.
-	//
-	// It is meant for the buffered-write path (keep-order concurrent `add()`, non-seekable output, ...),
-	// where the compressed data of an entry has to be held until the writer is free. In memory that data
-	// grows with the entry size; here it is written to an OPFS file and streamed back, so peak memory
-	// stays bounded.
-	//
-	// It is hybrid on purpose: an entry stays fully in memory until it exceeds `thresholdBytes`, and only
-	// then spills to a file. Small entries never touch the disk, so a workload made of many tiny entries
-	// keeps the speed of the in-memory default; only genuinely large entries pay for (and benefit from)
-	// the file round-trip.
-	//
-	// OPFS is a browser/worker feature; there is no fallback here. Feature-detect
-	// `navigator.storage.getDirectory` (or inject `getDirectory`) before using it, and let the writer use
-	// its in-memory default elsewhere.
-	//
-	// Options:
-	//   thresholdBytes  spill to a file once a buffered entry exceeds this size (default 1 MiB).
-	//   directoryName   name of the OPFS sub-directory holding the temp files (default ".zip.js-temp").
-	//   getDirectory    returns (or resolves to) the root `FileSystemDirectoryHandle`. Defaults to
-	//                   `navigator.storage.getDirectory()`. Inject it to run inside a worker with a
-	//                   pre-obtained handle, or to test against a mock.
 	function createOPFSTempStream(options = {}) {
 		const {
 			thresholdBytes = DEFAULT_THRESHOLD,
 			directoryName = DEFAULT_DIRECTORY_NAME,
 			getDirectory = () => navigator.storage.getDirectory()
 		} = options;
-		// The temp directory is resolved once and shared by every entry of this factory.
 		let directoryHandlePromise;
 		function getTempDirectory() {
 			if (!directoryHandlePromise) {
@@ -6854,15 +6663,12 @@
 			}
 			return directoryHandlePromise;
 		}
-		return async function () {
+		return function () {
 			const memoryChunks = [];
 			let bufferedSize = 0;
 			let spilled = false;
 			let fileName, fileHandle, fileWriter, fileReader;
 
-			// Move whatever is buffered in memory to a fresh OPFS file, then keep writing to that file.
-			// `FileSystemWritableFileStream` is a `WritableStream`, so it is driven through the standard
-			// writer API rather than its non-standard `write()`/`seek()` convenience methods.
 			async function spillToFile() {
 				const directoryHandle = await getTempDirectory();
 				fileName = crypto.randomUUID();
@@ -6872,7 +6678,6 @@
 				for (const chunk of memoryChunks) {
 					await fileWriter.write(chunk);
 				}
-				// Release the in-memory copy; from now on data lives in the file.
 				memoryChunks.length = 0;
 			}
 
@@ -6896,9 +6701,6 @@
 				}
 			});
 
-			// The writer always closes `writable` before reading `readable`, so `spilled` is final by the
-			// time `pull` runs. It is read lazily here (not in `start`, which runs at construction time,
-			// before anything is written and before `spilled` is known).
 			let memoryIndex = 0;
 			const readable = new ReadableStream({
 				async pull(controller) {
@@ -6925,11 +6727,6 @@
 					}
 				}
 			}, { highWaterMark: 0 });
-			// highWaterMark 0: do not pull (and possibly close on an empty buffer) until the consumer
-			// actually reads, which the writer only does once the data has been fully written.
-
-			// Called by the writer on every exit path (success, error, abort). Best-effort: it must never
-			// throw. Closes any still-open file writer, then deletes the temp file.
 			async function dispose() {
 				if (fileWriter) {
 					try {
@@ -7014,6 +6811,7 @@
 	exports.ERR_INVALID_SIGNATURE = ERR_INVALID_SIGNATURE;
 	exports.ERR_INVALID_UNCOMPRESSED_SIZE = ERR_INVALID_UNCOMPRESSED_SIZE;
 	exports.ERR_INVALID_VERSION = ERR_INVALID_VERSION;
+	exports.ERR_ITERATOR_COMPLETED_TOO_SOON = ERR_ITERATOR_COMPLETED_TOO_SOON;
 	exports.ERR_LOCAL_FILE_HEADER_NOT_FOUND = ERR_LOCAL_FILE_HEADER_NOT_FOUND;
 	exports.ERR_OVERLAPPING_ENTRY = ERR_OVERLAPPING_ENTRY;
 	exports.ERR_SPLIT_ZIP_FILE = ERR_SPLIT_ZIP_FILE;
@@ -7022,6 +6820,7 @@
 	exports.ERR_UNSUPPORTED_COMPRESSION = ERR_UNSUPPORTED_COMPRESSION$1;
 	exports.ERR_UNSUPPORTED_ENCRYPTION = ERR_UNSUPPORTED_ENCRYPTION;
 	exports.ERR_UNSUPPORTED_FORMAT = ERR_UNSUPPORTED_FORMAT;
+	exports.ERR_WRITER_NOT_INITIALIZED = ERR_WRITER_NOT_INITIALIZED;
 	exports.ERR_ZIP_NOT_EMPTY = ERR_ZIP_NOT_EMPTY;
 	exports.HttpRangeReader = HttpRangeReader;
 	exports.HttpReader = HttpReader;