read-excel-file 9.1.1 → 9.2.0

package/modules/zip/unzipFromStream.fflate.js

@@ -0,0 +1,203 @@
+// `fflate` uses a javascript-only implementation of `.zip` compression/decompression.
+// This means that it could likely be less performant than Node.js's "native" `zlib` module`.
+
+// This code was originally submitted by Stian Jensen.
+// https://github.com/catamphetamine/read-excel-file/pull/122
+
+// A `*.zip` file consists of individual file entries with the "total" summary section
+// placed at the end of the file rather than at the start of it, which was originally done
+// to allow for easy append of data to a given `.zip` file.
+// https://en.wikipedia.org/wiki/ZIP_(file_format)
+//
+// But this also means that reading a `*.zip` file from a stream can't really be done
+// using the "officially recommended" way of first reading the "total" summary section
+// and only then reading the individual file entries specified in that summary section.
+//
+// So in order to be able to read a `*.zip` file from a stream, some corners have to be cut.
+// For example, the "total" summary section is completely ignored and instead the reader
+// should adopt "data recovery" software approach — it should proactively "scan" the input stream
+// for individual file entries and handle them one-by-one as they come.
+//
+// Such approach doesn't seem to contradict with the XLSX specification
+// because an `*.xlsx` files is supposed to be a normal `.zip` archive
+// without any "trickery" such as "deleted" files or "garbage" data
+// hiding under the hood.
+//
+// So when handling `*.xlsx` file, we assume that each such file must start
+// with an individual file entry followed by another individual file entry, etc.
+//
+// When the "summary" section is reached, we assume that the archive has ended.
+//
+// To read a `.zip` archive, the code uses `fflate`'s `Unzip` class
+// with `UnzipInflate` decompression implementation to decompress the data
+// that was previously compressed using `DEFLATE` compressing algorithm,
+// which is what `*.xlsx` files use.
+//
+// The `Unzip` class doesn't speak the Node.js stream interface, and `fflate`'s readme
+// doesn't include a clear "reading a `.zip` file from a Node.js stream" section.
+// https://github.com/101arrowz/fflate/issues/251
+// Instead, the `Unzip` class has its own `push(chunk)` / `onfile` / `entry.ondata` protocol.
+// This code reads the binary input stream and forwards each chunk of it to `unzip.push()`,
+// and then collects the decompressed file entries.
+//
+// P.S. In the comments to `UnzipInflate` in `fflate` package, it says:
+// "Streaming DEFLATE decompression for ZIP archives. Prefer AsyncZipInflate for better performance."
+// But there seems to be no `AsyncZipInflate` class in the `fflate` package.
+// https://github.com/101arrowz/fflate/issues/277
+// So just the regular `UnzipInflate` is used here.
+//
+import { Unzip, UnzipInflate } from 'fflate';
+import { Buffer } from 'node:buffer';
+
+/**
+ * Reads `*.zip` file contents.
+ * @param  {Stream} stream
+ * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.
+ */
+export default function unzipFromStream(stream) {
+  var _ref = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {},
+    filter = _ref.filter;
+  // The `files` object stores the files and their contents.
+  var files = {};
+  return new Promise(function (resolve, reject) {
+    var errored = false;
+    var onError = function onError(error) {
+      if (!errored) {
+        errored = true;
+        reject(error);
+      }
+    };
+    var _createZipFileValidat = createZipFileValidator(function (isValid) {
+        if (!isValid) {
+          onError(new Error('Invalid `.zip` archive'));
+        }
+      }),
+      validateChunk = _createZipFileValidat.validateChunk;
+
+    // `Unzip` discovers each individual file entry in the input data stream
+    // and then calls the callback function for each such entry.
+    var unzip = new Unzip(function (entry) {
+      // If there already was an error while reading this `.zip` file,
+      // ignore any follow-up entries.
+      if (errored) {
+        return;
+      }
+
+      // Skip directory entries (their names end with a slash).
+      // Only files are of any interest.
+      if (entry.name.endsWith('/')) {
+        return;
+      }
+
+      // See if this file should be ignored.
+      // If it should, this entry won't be processed, i.e. `Unzip` will not try
+      // to decompress its data, and will just discard it.
+      if (filter && !filter({
+        path: entry.name
+      })) {
+        return;
+      }
+      var chunks = [];
+
+      // `entry.ondata` is called with each decompressed chunk of the entry,
+      // and a final time with `isLast === true` once the entry is complete.
+      entry.ondata = function (error, chunk, isLast) {
+        if (error) {
+          return onError(error);
+        }
+        chunks.push(chunk);
+        if (isLast) {
+          files[entry.name] = Buffer.concat(chunks);
+        }
+      };
+
+      // Start decompressing this entry.
+      entry.start();
+    });
+
+    // Register the decompressor for the data that was compressed using
+    // `DEFLATE` compression algorithm (compression method `8`),
+    // which is what `.xlsx` files use.
+    unzip.register(UnzipInflate);
+    stream
+    // Catch errors emitted from the input stream (for example, a file read error).
+    .on('error', onError)
+    // When another chunk of data is read from the input stream.
+    .on('data', function (chunk) {
+      // If there already was an error while reading this `.zip` file,
+      // ignore any follow-up data chunks.
+      if (errored) {
+        return;
+      }
+      // Validate the `.zip` archive as its data comes through.
+      validateChunk(chunk);
+      // If the `.zip` archive is found to be invalid, stop any further
+      // processing of it.
+      if (errored) {
+        return;
+      }
+      // Push the next data chunk to `fflate`'s `Unzip` class instance.
+      // The `.push()` function is synchronous, meaning that by the time it returns,
+      // any complete files entries encountered so far have already been decompressed
+      // and populated in the `files` object.
+      try {
+        unzip.push(chunk, false);
+      } catch (error) {
+        onError(error);
+      }
+    })
+    // When there's no more data in the input stream to consume,
+    // finish reading the `.zip` archive.
+    .on('end', function () {
+      // If there were any errors when reading the `.zip` archive,
+      // don't `resolve()` with anything.
+      if (errored) {
+        return;
+      }
+      try {
+        // Signal the end of the archive to `fflate`'s `Unzip` class instance.
+        // It will flush any remaining state in it.
+        unzip.push(new Uint8Array(0), true);
+        // Resolve with the unzipped files.
+        resolve(files);
+      } catch (error) {
+        onError(error);
+      }
+    });
+  });
+}
+
+// Every section in a `.zip` archive is marked with 4 bytes, the first two of which
+// are `0x50` and `0x4B`, which reads "PK", referencing the initials of the inventor Phil Katz.
+//
+// It looks like `fflate`'s `Unzip` doesn't ever complain about whatever data is thrown at it.
+// Due to how `.zip` file format is defined, "garbage" data could be placed at various
+// places in it and it'd still be a valid `.zip` archive. It's likely that for this reason
+// `fflate` doesn't ever complain and simply emits no entries when fed any kind of invalid data.
+//
+// In order to introduce some basic validation, here we specifically demand
+// that a `.zip` archive must at least start with an individual file entry
+// because an `.xlsx` file creator softwared really shouldn't attempt doing
+// anything "funny" when writing a file, hence this adherence requirement.
+//
+function createZipFileValidator(onValidationResult) {
+  var firstBytesCount = 2;
+  var firstBytes = [];
+  var firstBytesCheckResult;
+  return {
+    validateChunk: function validateChunk(chunk) {
+      if (firstBytes.length < 2) {
+        var i = 0;
+        while (i < chunk.length && i < firstBytesCount) {
+          firstBytes.push(chunk[i]);
+          i++;
+        }
+        if (firstBytes.length === 2) {
+          var isValid = firstBytes[0] === 0x50 && firstBytes[1] === 0x4B;
+          onValidationResult(isValid);
+        }
+      }
+    }
+  };
+}
+//# sourceMappingURL=unzipFromStream.fflate.js.map

package/modules/zip/unzipFromStream.fflate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"unzipFromStream.fflate.js","names":["Unzip","UnzipInflate","Buffer","unzipFromStream","stream","_ref","arguments","length","undefined","filter","files","Promise","resolve","reject","errored","onError","error","_createZipFileValidat","createZipFileValidator","isValid","Error","validateChunk","unzip","entry","name","endsWith","path","chunks","ondata","chunk","isLast","push","concat","start","register","on","Uint8Array","onValidationResult","firstBytesCount","firstBytes","firstBytesCheckResult","i"],"sources":["../../source/zip/unzipFromStream.fflate.js"],"sourcesContent":["// `fflate` uses a javascript-only implementation of `.zip` compression/decompression.\r\n// This means that it could likely be less performant than Node.js's \"native\" `zlib` module`.\r\n\r\n// This code was originally submitted by Stian Jensen.\r\n// https://github.com/catamphetamine/read-excel-file/pull/122\r\n\r\n// A `*.zip` file consists of individual file entries with the \"total\" summary section\r\n// placed at the end of the file rather than at the start of it, which was originally done\r\n// to allow for easy append of data to a given `.zip` file.\r\n// https://en.wikipedia.org/wiki/ZIP_(file_format)\r\n//\r\n// But this also means that reading a `*.zip` file from a stream can't really be done\r\n// using the \"officially recommended\" way of first reading the \"total\" summary section\r\n// and only then reading the individual file entries specified in that summary section.\r\n//\r\n// So in order to be able to read a `*.zip` file from a stream, some corners have to be cut.\r\n// For example, the \"total\" summary section is completely ignored and instead the reader\r\n// should adopt \"data recovery\" software approach — it should proactively \"scan\" the input stream\r\n// for individual file entries and handle them one-by-one as they come.\r\n//\r\n// Such approach doesn't seem to contradict with the XLSX specification\r\n// because an `*.xlsx` files is supposed to be a normal `.zip` archive\r\n// without any \"trickery\" such as \"deleted\" files or \"garbage\" data\r\n// hiding under the hood.\r\n//\r\n// So when handling `*.xlsx` file, we assume that each such file must start\r\n// with an individual file entry followed by another individual file entry, etc.\r\n//\r\n// When the \"summary\" section is reached, we assume that the archive has ended.\r\n//\r\n// To read a `.zip` archive, the code uses `fflate`'s `Unzip` class\r\n// with `UnzipInflate` decompression implementation to decompress the data\r\n// that was previously compressed using `DEFLATE` compressing algorithm,\r\n// which is what `*.xlsx` files use.\r\n//\r\n// The `Unzip` class doesn't speak the Node.js stream interface, and `fflate`'s readme\r\n// doesn't include a clear \"reading a `.zip` file from a Node.js stream\" section.\r\n// https://github.com/101arrowz/fflate/issues/251\r\n// Instead, the `Unzip` class has its own `push(chunk)` / `onfile` / `entry.ondata` protocol.\r\n// This code reads the binary input stream and forwards each chunk of it to `unzip.push()`,\r\n// and then collects the decompressed file entries.\r\n//\r\n// P.S. In the comments to `UnzipInflate` in `fflate` package, it says:\r\n// \"Streaming DEFLATE decompression for ZIP archives. Prefer AsyncZipInflate for better performance.\"\r\n// But there seems to be no `AsyncZipInflate` class in the `fflate` package.\r\n// https://github.com/101arrowz/fflate/issues/277\r\n// So just the regular `UnzipInflate` is used here.\r\n//\r\nimport { Unzip, UnzipInflate } from 'fflate'\r\n\r\nimport { Buffer } from 'node:buffer'\r\n\r\n/**\r\n * Reads `*.zip` file contents.\r\n * @param  {Stream} stream\r\n * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.\r\n */\r\nexport default function unzipFromStream(stream, { filter } = {}) {\r\n\t// The `files` object stores the files and their contents.\r\n\tconst files = {}\r\n\r\n\treturn new Promise((resolve, reject) => {\r\n\t\tlet errored = false\r\n\r\n\t\tconst onError = (error) => {\r\n\t\t\tif (!errored) {\r\n\t\t\t\terrored = true\r\n\t\t\t\treject(error)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\tconst { validateChunk } = createZipFileValidator((isValid) => {\r\n\t\t\tif (!isValid) {\r\n\t\t\t\tonError(new Error('Invalid `.zip` archive'))\r\n\t\t\t}\r\n\t\t})\r\n\r\n\t\t// `Unzip` discovers each individual file entry in the input data stream\r\n\t\t// and then calls the callback function for each such entry.\r\n\t\tconst unzip = new Unzip((entry) => {\r\n\t\t\t// If there already was an error while reading this `.zip` file,\r\n\t\t\t// ignore any follow-up entries.\r\n\t\t\tif (errored) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\t// Skip directory entries (their names end with a slash).\r\n\t\t\t// Only files are of any interest.\r\n\t\t\tif (entry.name.endsWith('/')) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\t// See if this file should be ignored.\r\n\t\t\t// If it should, this entry won't be processed, i.e. `Unzip` will not try\r\n\t\t\t// to decompress its data, and will just discard it.\r\n\t\t\tif (filter && !filter({ path: entry.name })) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\tconst chunks = []\r\n\r\n\t\t\t// `entry.ondata` is called with each decompressed chunk of the entry,\r\n\t\t\t// and a final time with `isLast === true` once the entry is complete.\r\n\t\t\tentry.ondata = (error, chunk, isLast) => {\r\n\t\t\t\tif (error) {\r\n\t\t\t\t\treturn onError(error)\r\n\t\t\t\t}\r\n\t\t\t\tchunks.push(chunk)\r\n\t\t\t\tif (isLast) {\r\n\t\t\t\t\tfiles[entry.name] = Buffer.concat(chunks)\r\n\t\t\t\t}\r\n\t\t\t}\r\n\r\n\t\t\t// Start decompressing this entry.\r\n\t\t\tentry.start()\r\n\t\t})\r\n\r\n\t\t// Register the decompressor for the data that was compressed using\r\n\t\t// `DEFLATE` compression algorithm (compression method `8`),\r\n\t\t// which is what `.xlsx` files use.\r\n\t\tunzip.register(UnzipInflate)\r\n\r\n\t\tstream\r\n\t\t\t// Catch errors emitted from the input stream (for example, a file read error).\r\n\t\t\t.on('error', onError)\r\n\t\t\t// When another chunk of data is read from the input stream.\r\n\t\t\t.on('data', (chunk) => {\r\n\t\t\t\t// If there already was an error while reading this `.zip` file,\r\n\t\t\t\t// ignore any follow-up data chunks.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\t// Validate the `.zip` archive as its data comes through.\r\n\t\t\t\tvalidateChunk(chunk)\r\n\t\t\t\t// If the `.zip` archive is found to be invalid, stop any further\r\n\t\t\t\t// processing of it.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\t// Push the next data chunk to `fflate`'s `Unzip` class instance.\r\n\t\t\t\t// The `.push()` function is synchronous, meaning that by the time it returns,\r\n\t\t\t\t// any complete files entries encountered so far have already been decompressed\r\n\t\t\t\t// and populated in the `files` object.\r\n\t\t\t\ttry {\r\n\t\t\t\t\tunzip.push(chunk, false)\r\n\t\t\t\t} catch (error) {\r\n\t\t\t\t\tonError(error)\r\n\t\t\t\t}\r\n\t\t\t})\r\n\t\t\t// When there's no more data in the input stream to consume,\r\n\t\t\t// finish reading the `.zip` archive.\r\n\t\t\t.on('end', () => {\r\n\t\t\t\t// If there were any errors when reading the `.zip` archive,\r\n\t\t\t\t// don't `resolve()` with anything.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\ttry {\r\n\t\t\t\t\t// Signal the end of the archive to `fflate`'s `Unzip` class instance.\r\n\t\t\t\t\t// It will flush any remaining state in it.\r\n\t\t\t\t\tunzip.push(new Uint8Array(0), true)\r\n\t\t\t\t\t// Resolve with the unzipped files.\r\n\t\t\t\t\tresolve(files)\r\n\t\t\t\t} catch (error) {\r\n\t\t\t\t\tonError(error)\r\n\t\t\t\t}\r\n\t\t\t})\r\n\t})\r\n}\r\n\r\n// Every section in a `.zip` archive is marked with 4 bytes, the first two of which\r\n// are `0x50` and `0x4B`, which reads \"PK\", referencing the initials of the inventor Phil Katz.\r\n//\r\n// It looks like `fflate`'s `Unzip` doesn't ever complain about whatever data is thrown at it.\r\n// Due to how `.zip` file format is defined, \"garbage\" data could be placed at various\r\n// places in it and it'd still be a valid `.zip` archive. It's likely that for this reason\r\n// `fflate` doesn't ever complain and simply emits no entries when fed any kind of invalid data.\r\n//\r\n// In order to introduce some basic validation, here we specifically demand\r\n// that a `.zip` archive must at least start with an individual file entry\r\n// because an `.xlsx` file creator softwared really shouldn't attempt doing\r\n// anything \"funny\" when writing a file, hence this adherence requirement.\r\n//\r\nfunction createZipFileValidator(onValidationResult) {\r\n\tconst firstBytesCount = 2\r\n\tconst firstBytes = []\r\n\tlet firstBytesCheckResult\r\n\treturn {\r\n\t\tvalidateChunk(chunk) {\r\n\t\t\tif (firstBytes.length < 2) {\r\n\t\t\t\tlet i = 0\r\n\t\t\t\twhile (i < chunk.length && i < firstBytesCount) {\r\n\t\t\t\t\tfirstBytes.push(chunk[i])\r\n\t\t\t\t\ti++\r\n\t\t\t\t}\r\n\t\t\t\tif (firstBytes.length === 2) {\r\n\t\t\t\t\tconst isValid = firstBytes[0] === 0x50 && firstBytes[1] === 0x4B\r\n\t\t\t\t\tonValidationResult(isValid)\r\n\t\t\t\t}\r\n\t\t\t}\r\n\t\t}\r\n\t}\r\n}"],"mappings":"AAAA;AACA;;AAEA;AACA;;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASA,KAAK,EAAEC,YAAY,QAAQ,QAAQ;AAE5C,SAASC,MAAM,QAAQ,aAAa;;AAEpC;AACA;AACA;AACA;AACA;AACA,eAAe,SAASC,eAAeA,CAACC,MAAM,EAAmB;EAAA,IAAAC,IAAA,GAAAC,SAAA,CAAAC,MAAA,QAAAD,SAAA,QAAAE,SAAA,GAAAF,SAAA,MAAJ,CAAC,CAAC;IAAbG,MAAM,GAAAJ,IAAA,CAANI,MAAM;EACvD;EACA,IAAMC,KAAK,GAAG,CAAC,CAAC;EAEhB,OAAO,IAAIC,OAAO,CAAC,UAACC,OAAO,EAAEC,MAAM,EAAK;IACvC,IAAIC,OAAO,GAAG,KAAK;IAEnB,IAAMC,OAAO,GAAG,SAAVA,OAAOA,CAAIC,KAAK,EAAK;MAC1B,IAAI,CAACF,OAAO,EAAE;QACbA,OAAO,GAAG,IAAI;QACdD,MAAM,CAACG,KAAK,CAAC;MACd;IACD,CAAC;IAED,IAAAC,qBAAA,GAA0BC,sBAAsB,CAAC,UAACC,OAAO,EAAK;QAC7D,IAAI,CAACA,OAAO,EAAE;UACbJ,OAAO,CAAC,IAAIK,KAAK,CAAC,wBAAwB,CAAC,CAAC;QAC7C;MACD,CAAC,CAAC;MAJMC,aAAa,GAAAJ,qBAAA,CAAbI,aAAa;;IAMrB;IACA;IACA,IAAMC,KAAK,GAAG,IAAItB,KAAK,CAAC,UAACuB,KAAK,EAAK;MAClC;MACA;MACA,IAAIT,OAAO,EAAE;QACZ;MACD;;MAEA;MACA;MACA,IAAIS,KAAK,CAACC,IAAI,CAACC,QAAQ,CAAC,GAAG,CAAC,EAAE;QAC7B;MACD;;MAEA;MACA;MACA;MACA,IAAIhB,MAAM,IAAI,CAACA,MAAM,CAAC;QAAEiB,IAAI,EAAEH,KAAK,CAACC;MAAK,CAAC,CAAC,EAAE;QAC5C;MACD;MAEA,IAAMG,MAAM,GAAG,EAAE;;MAEjB;MACA;MACAJ,KAAK,CAACK,MAAM,GAAG,UAACZ,KAAK,EAAEa,KAAK,EAAEC,MAAM,EAAK;QACxC,IAAId,KAAK,EAAE;UACV,OAAOD,OAAO,CAACC,KAAK,CAAC;QACtB;QACAW,MAAM,CAACI,IAAI,CAACF,KAAK,CAAC;QAClB,IAAIC,MAAM,EAAE;UACXpB,KAAK,CAACa,KAAK,CAACC,IAAI,CAAC,GAAGtB,MAAM,CAAC8B,MAAM,CAACL,MAAM,CAAC;QAC1C;MACD,CAAC;;MAED;MACAJ,KAAK,CAACU,KAAK,CAAC,CAAC;IACd,CAAC,CAAC;;IAEF;IACA;IACA;IACAX,KAAK,CAACY,QAAQ,CAACjC,YAAY,CAAC;IAE5BG;IACC;IAAA,CACC+B,EAAE,CAAC,OAAO,EAAEpB,OAAO;IACpB;IAAA,CACCoB,EAAE,CAAC,MAAM,EAAE,UAACN,KAAK,EAAK;MACtB;MACA;MACA,IAAIf,OAAO,EAAE;QACZ;MACD;MACA;MACAO,aAAa,CAACQ,KAAK,CAAC;MACpB;MACA;MACA,IAAIf,OAAO,EAAE;QACZ;MACD;MACA;MACA;MACA;MACA;MACA,IAAI;QACHQ,KAAK,CAACS,IAAI,CAACF,KAAK,EAAE,KAAK,CAAC;MACzB,CAAC,CAAC,OAAOb,KAAK,EAAE;QACfD,OAAO,CAACC,KAAK,CAAC;MACf;IACD,CAAC;IACD;IACA;IAAA,CACCmB,EAAE,CAAC,KAAK,EAAE,YAAM;MAChB;MACA;MACA,IAAIrB,OAAO,EAAE;QACZ;MACD;MACA,IAAI;QACH;QACA;QACAQ,KAAK,CAACS,IAAI,CAAC,IAAIK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC;QACnC;QACAxB,OAAO,CAACF,KAAK,CAAC;MACf,CAAC,CAAC,OAAOM,KAAK,EAAE;QACfD,OAAO,CAACC,KAAK,CAAC;MACf;IACD,CAAC,CAAC;EACJ,CAAC,CAAC;AACH;;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASE,sBAAsBA,CAACmB,kBAAkB,EAAE;EACnD,IAAMC,eAAe,GAAG,CAAC;EACzB,IAAMC,UAAU,GAAG,EAAE;EACrB,IAAIC,qBAAqB;EACzB,OAAO;IACNnB,aAAa,WAAAA,cAACQ,KAAK,EAAE;MACpB,IAAIU,UAAU,CAAChC,MAAM,GAAG,CAAC,EAAE;QAC1B,IAAIkC,CAAC,GAAG,CAAC;QACT,OAAOA,CAAC,GAAGZ,KAAK,CAACtB,MAAM,IAAIkC,CAAC,GAAGH,eAAe,EAAE;UAC/CC,UAAU,CAACR,IAAI,CAACF,KAAK,CAACY,CAAC,CAAC,CAAC;UACzBA,CAAC,EAAE;QACJ;QACA,IAAIF,UAAU,CAAChC,MAAM,KAAK,CAAC,EAAE;UAC5B,IAAMY,OAAO,GAAGoB,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI,IAAIA,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI;UAChEF,kBAAkB,CAAClB,OAAO,CAAC;QAC5B;MACD;IACD;EACD,CAAC;AACF"}

package/modules/zip/unzipFromStream.js

@@ -1,200 +1,14 @@
-// This code was originally submitted by Stian Jensen.
-// https://github.com/catamphetamine/read-excel-file/pull/122
-
-// A `*.zip` file consists of individual file entries with the "total" summary section
-// placed at the end of the file rather than at the start of it, which was originally done
-// to allow for easy append of data to a given `.zip` file.
-// https://en.wikipedia.org/wiki/ZIP_(file_format)
+// Currently, there're two implementations:
+// * `fflate` — a pure-javascript implementation that uses `fflate` package.
+// * `unzipper` — a "native" Node.js module that uses Node's `zlib` which is written in C.
 //
-// But this also means that reading a `*.zip` file from a stream can't really be done
-// using the "officially recommended" way of first reading the "total" summary section
-// and only then reading the individual file entries specified in that summary section.
+// The implementations are compared in a benchmark:
 //
-// So in order to be able to read a `*.zip` file from a stream, some corners have to be cut.
-// For example, the "total" summary section is completely ignored and instead the reader
-// should adopt "data recovery" software approach — it should proactively "scan" the input stream
-// for individual file entries and handle them one-by-one as they come.
+// ```
+// npm run test:benchmark:unzipFromStream
+// ```
 //
-// Such approach doesn't seem to contradict with the XLSX specification
-// because an `*.xlsx` files is supposed to be a normal `.zip` archive
-// without any "trickery" such as "deleted" files or "garbage" data
-// hiding under the hood.
+// The benchmark tells that `unzipper` is 2x faster than `fflate`.
 //
-// So when handling `*.xlsx` file, we assume that each such file must start
-// with an individual file entry followed by another individual file entry, etc.
-//
-// When the "summary" section is reached, we assume that the archive has ended.
-//
-// To read a `.zip` archive, the code uses `fflate`'s `Unzip` class
-// with `UnzipInflate` decompression implementation to decompress the data
-// that was previously compressed using `DEFLATE` compressing algorithm,
-// which is what `*.xlsx` files use.
-//
-// The `Unzip` class doesn't speak the Node.js stream interface, and `fflate`'s readme
-// doesn't include a clear "reading a `.zip` file from a Node.js stream" section.
-// https://github.com/101arrowz/fflate/issues/251
-// Instead, the `Unzip` class has its own `push(chunk)` / `onfile` / `entry.ondata` protocol.
-// This code reads the binary input stream and forwards each chunk of it to `unzip.push()`,
-// and then collects the decompressed file entries.
-//
-// P.S. In the comments to `UnzipInflate` in `fflate` package, it says:
-// "Streaming DEFLATE decompression for ZIP archives. Prefer AsyncZipInflate for better performance."
-// But there seems to be no `AsyncZipInflate` class in the `fflate` package.
-// https://github.com/101arrowz/fflate/issues/277
-// So just the regular `UnzipInflate` is used here.
-//
-import { Unzip, UnzipInflate } from 'fflate';
-import { Buffer } from 'node:buffer';
-
-/**
- * Reads `*.zip` file contents.
- * @param  {Stream} stream
- * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.
- */
-export default function unzipFromStream(stream) {
-  var _ref = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {},
-    filter = _ref.filter;
-  // The `files` object stores the files and their contents.
-  var files = {};
-  return new Promise(function (resolve, reject) {
-    var errored = false;
-    var onError = function onError(error) {
-      if (!errored) {
-        errored = true;
-        reject(error);
-      }
-    };
-    var _createZipFileValidat = createZipFileValidator(function (isValid) {
-        if (!isValid) {
-          onError(new Error('Invalid `.zip` archive'));
-        }
-      }),
-      validateChunk = _createZipFileValidat.validateChunk;
-
-    // `Unzip` discovers each individual file entry in the input data stream
-    // and then calls the callback function for each such entry.
-    var unzip = new Unzip(function (entry) {
-      // If there already was an error while reading this `.zip` file,
-      // ignore any follow-up entries.
-      if (errored) {
-        return;
-      }
-
-      // Skip directory entries (their names end with a slash).
-      // Only files are of any interest.
-      if (entry.name.endsWith('/')) {
-        return;
-      }
-
-      // See if this file should be ignored.
-      // If it should, this entry won't be processed, i.e. `Unzip` will not try
-      // to decompress its data, and will just discard it.
-      if (filter && !filter({
-        path: entry.name
-      })) {
-        return;
-      }
-      var chunks = [];
-
-      // `entry.ondata` is called with each decompressed chunk of the entry,
-      // and a final time with `isLast === true` once the entry is complete.
-      entry.ondata = function (error, chunk, isLast) {
-        if (error) {
-          return onError(error);
-        }
-        chunks.push(chunk);
-        if (isLast) {
-          files[entry.name] = Buffer.concat(chunks);
-        }
-      };
-
-      // Start decompressing this entry.
-      entry.start();
-    });
-
-    // Register the decompressor for the data that was compressed using
-    // `DEFLATE` compression algorithm (compression method `8`),
-    // which is what `.xlsx` files use.
-    unzip.register(UnzipInflate);
-    stream
-    // Catch errors emitted from the input stream (for example, a file read error).
-    .on('error', onError)
-    // When another chunk of data is read from the input stream.
-    .on('data', function (chunk) {
-      // If there already was an error while reading this `.zip` file,
-      // ignore any follow-up data chunks.
-      if (errored) {
-        return;
-      }
-      // Validate the `.zip` archive as its data comes through.
-      validateChunk(chunk);
-      // If the `.zip` archive is found to be invalid, stop any further
-      // processing of it.
-      if (errored) {
-        return;
-      }
-      // Push the next data chunk to `fflate`'s `Unzip` class instance.
-      // The `.push()` function is synchronous, meaning that by the time it returns,
-      // any complete files entries encountered so far have already been decompressed
-      // and populated in the `files` object.
-      try {
-        unzip.push(chunk, false);
-      } catch (error) {
-        onError(error);
-      }
-    })
-    // When there's no more data in the input stream to consume,
-    // finish reading the `.zip` archive.
-    .on('end', function () {
-      // If there were any errors when reading the `.zip` archive,
-      // don't `resolve()` with anything.
-      if (errored) {
-        return;
-      }
-      try {
-        // Signal the end of the archive to `fflate`'s `Unzip` class instance.
-        // It will flush any remaining state in it.
-        unzip.push(new Uint8Array(0), true);
-        // Resolve with the unzipped files.
-        resolve(files);
-      } catch (error) {
-        onError(error);
-      }
-    });
-  });
-}
-
-// Every section in a `.zip` archive is marked with 4 bytes, the first two of which
-// are `0x50` and `0x4B`, which reads "PK", referencing the initials of the inventor Phil Katz.
-//
-// It looks like `fflate`'s `Unzip` doesn't ever complain about whatever data is thrown at it.
-// Due to how `.zip` file format is defined, "garbage" data could be placed at various
-// places in it and it'd still be a valid `.zip` archive. It's likely that for this reason
-// `fflate` doesn't ever complain and simply emits no entries when fed any kind of invalid data.
-//
-// In order to introduce some basic validation, here we specifically demand
-// that a `.zip` archive must at least start with an individual file entry
-// because an `.xlsx` file creator softwared really shouldn't attempt doing
-// anything "funny" when writing a file, hence this adherence requirement.
-//
-function createZipFileValidator(onValidationResult) {
-  var firstBytesCount = 2;
-  var firstBytes = [];
-  var firstBytesCheckResult;
-  return {
-    validateChunk: function validateChunk(chunk) {
-      if (firstBytes.length < 2) {
-        var i = 0;
-        while (i < chunk.length && i < firstBytesCount) {
-          firstBytes.push(chunk[i]);
-          i++;
-        }
-        if (firstBytes.length === 2) {
-          var isValid = firstBytes[0] === 0x50 && firstBytes[1] === 0x4B;
-          onValidationResult(isValid);
-        }
-      }
-    }
-  };
-}
+export { default } from './unzipFromStream.unzipper.js';
 //# sourceMappingURL=unzipFromStream.js.map

package/modules/zip/unzipFromStream.js.map

@@ -1 +1 @@
-{"version":3,"file":"unzipFromStream.js","names":["Unzip","UnzipInflate","Buffer","unzipFromStream","stream","_ref","arguments","length","undefined","filter","files","Promise","resolve","reject","errored","onError","error","_createZipFileValidat","createZipFileValidator","isValid","Error","validateChunk","unzip","entry","name","endsWith","path","chunks","ondata","chunk","isLast","push","concat","start","register","on","Uint8Array","onValidationResult","firstBytesCount","firstBytes","firstBytesCheckResult","i"],"sources":["../../source/zip/unzipFromStream.js"],"sourcesContent":["// This code was originally submitted by Stian Jensen.\r\n// https://github.com/catamphetamine/read-excel-file/pull/122\r\n\r\n// A `*.zip` file consists of individual file entries with the \"total\" summary section\r\n// placed at the end of the file rather than at the start of it, which was originally done\r\n// to allow for easy append of data to a given `.zip` file.\r\n// https://en.wikipedia.org/wiki/ZIP_(file_format)\r\n//\r\n// But this also means that reading a `*.zip` file from a stream can't really be done\r\n// using the \"officially recommended\" way of first reading the \"total\" summary section\r\n// and only then reading the individual file entries specified in that summary section.\r\n//\r\n// So in order to be able to read a `*.zip` file from a stream, some corners have to be cut.\r\n// For example, the \"total\" summary section is completely ignored and instead the reader\r\n// should adopt \"data recovery\" software approach — it should proactively \"scan\" the input stream\r\n// for individual file entries and handle them one-by-one as they come.\r\n//\r\n// Such approach doesn't seem to contradict with the XLSX specification\r\n// because an `*.xlsx` files is supposed to be a normal `.zip` archive\r\n// without any \"trickery\" such as \"deleted\" files or \"garbage\" data\r\n// hiding under the hood.\r\n//\r\n// So when handling `*.xlsx` file, we assume that each such file must start\r\n// with an individual file entry followed by another individual file entry, etc.\r\n//\r\n// When the \"summary\" section is reached, we assume that the archive has ended.\r\n//\r\n// To read a `.zip` archive, the code uses `fflate`'s `Unzip` class\r\n// with `UnzipInflate` decompression implementation to decompress the data\r\n// that was previously compressed using `DEFLATE` compressing algorithm,\r\n// which is what `*.xlsx` files use.\r\n//\r\n// The `Unzip` class doesn't speak the Node.js stream interface, and `fflate`'s readme\r\n// doesn't include a clear \"reading a `.zip` file from a Node.js stream\" section.\r\n// https://github.com/101arrowz/fflate/issues/251\r\n// Instead, the `Unzip` class has its own `push(chunk)` / `onfile` / `entry.ondata` protocol.\r\n// This code reads the binary input stream and forwards each chunk of it to `unzip.push()`,\r\n// and then collects the decompressed file entries.\r\n//\r\n// P.S. In the comments to `UnzipInflate` in `fflate` package, it says:\r\n// \"Streaming DEFLATE decompression for ZIP archives. Prefer AsyncZipInflate for better performance.\"\r\n// But there seems to be no `AsyncZipInflate` class in the `fflate` package.\r\n// https://github.com/101arrowz/fflate/issues/277\r\n// So just the regular `UnzipInflate` is used here.\r\n//\r\nimport { Unzip, UnzipInflate } from 'fflate'\r\n\r\nimport { Buffer } from 'node:buffer'\r\n\r\n/**\r\n * Reads `*.zip` file contents.\r\n * @param  {Stream} stream\r\n * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.\r\n */\r\nexport default function unzipFromStream(stream, { filter } = {}) {\r\n\t// The `files` object stores the files and their contents.\r\n\tconst files = {}\r\n\r\n\treturn new Promise((resolve, reject) => {\r\n\t\tlet errored = false\r\n\r\n\t\tconst onError = (error) => {\r\n\t\t\tif (!errored) {\r\n\t\t\t\terrored = true\r\n\t\t\t\treject(error)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\tconst { validateChunk } = createZipFileValidator((isValid) => {\r\n\t\t\tif (!isValid) {\r\n\t\t\t\tonError(new Error('Invalid `.zip` archive'))\r\n\t\t\t}\r\n\t\t})\r\n\r\n\t\t// `Unzip` discovers each individual file entry in the input data stream\r\n\t\t// and then calls the callback function for each such entry.\r\n\t\tconst unzip = new Unzip((entry) => {\r\n\t\t\t// If there already was an error while reading this `.zip` file,\r\n\t\t\t// ignore any follow-up entries.\r\n\t\t\tif (errored) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\t// Skip directory entries (their names end with a slash).\r\n\t\t\t// Only files are of any interest.\r\n\t\t\tif (entry.name.endsWith('/')) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\t// See if this file should be ignored.\r\n\t\t\t// If it should, this entry won't be processed, i.e. `Unzip` will not try\r\n\t\t\t// to decompress its data, and will just discard it.\r\n\t\t\tif (filter && !filter({ path: entry.name })) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\r\n\t\t\tconst chunks = []\r\n\r\n\t\t\t// `entry.ondata` is called with each decompressed chunk of the entry,\r\n\t\t\t// and a final time with `isLast === true` once the entry is complete.\r\n\t\t\tentry.ondata = (error, chunk, isLast) => {\r\n\t\t\t\tif (error) {\r\n\t\t\t\t\treturn onError(error)\r\n\t\t\t\t}\r\n\t\t\t\tchunks.push(chunk)\r\n\t\t\t\tif (isLast) {\r\n\t\t\t\t\tfiles[entry.name] = Buffer.concat(chunks)\r\n\t\t\t\t}\r\n\t\t\t}\r\n\r\n\t\t\t// Start decompressing this entry.\r\n\t\t\tentry.start()\r\n\t\t})\r\n\r\n\t\t// Register the decompressor for the data that was compressed using\r\n\t\t// `DEFLATE` compression algorithm (compression method `8`),\r\n\t\t// which is what `.xlsx` files use.\r\n\t\tunzip.register(UnzipInflate)\r\n\r\n\t\tstream\r\n\t\t\t// Catch errors emitted from the input stream (for example, a file read error).\r\n\t\t\t.on('error', onError)\r\n\t\t\t// When another chunk of data is read from the input stream.\r\n\t\t\t.on('data', (chunk) => {\r\n\t\t\t\t// If there already was an error while reading this `.zip` file,\r\n\t\t\t\t// ignore any follow-up data chunks.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\t// Validate the `.zip` archive as its data comes through.\r\n\t\t\t\tvalidateChunk(chunk)\r\n\t\t\t\t// If the `.zip` archive is found to be invalid, stop any further\r\n\t\t\t\t// processing of it.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\t// Push the next data chunk to `fflate`'s `Unzip` class instance.\r\n\t\t\t\t// The `.push()` function is synchronous, meaning that by the time it returns,\r\n\t\t\t\t// any complete files entries encountered so far have already been decompressed\r\n\t\t\t\t// and populated in the `files` object.\r\n\t\t\t\ttry {\r\n\t\t\t\t\tunzip.push(chunk, false)\r\n\t\t\t\t} catch (error) {\r\n\t\t\t\t\tonError(error)\r\n\t\t\t\t}\r\n\t\t\t})\r\n\t\t\t// When there's no more data in the input stream to consume,\r\n\t\t\t// finish reading the `.zip` archive.\r\n\t\t\t.on('end', () => {\r\n\t\t\t\t// If there were any errors when reading the `.zip` archive,\r\n\t\t\t\t// don't `resolve()` with anything.\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\t\t\t\ttry {\r\n\t\t\t\t\t// Signal the end of the archive to `fflate`'s `Unzip` class instance.\r\n\t\t\t\t\t// It will flush any remaining state in it.\r\n\t\t\t\t\tunzip.push(new Uint8Array(0), true)\r\n\t\t\t\t\t// Resolve with the unzipped files.\r\n\t\t\t\t\tresolve(files)\r\n\t\t\t\t} catch (error) {\r\n\t\t\t\t\tonError(error)\r\n\t\t\t\t}\r\n\t\t\t})\r\n\t})\r\n}\r\n\r\n// Every section in a `.zip` archive is marked with 4 bytes, the first two of which\r\n// are `0x50` and `0x4B`, which reads \"PK\", referencing the initials of the inventor Phil Katz.\r\n//\r\n// It looks like `fflate`'s `Unzip` doesn't ever complain about whatever data is thrown at it.\r\n// Due to how `.zip` file format is defined, \"garbage\" data could be placed at various\r\n// places in it and it'd still be a valid `.zip` archive. It's likely that for this reason\r\n// `fflate` doesn't ever complain and simply emits no entries when fed any kind of invalid data.\r\n//\r\n// In order to introduce some basic validation, here we specifically demand\r\n// that a `.zip` archive must at least start with an individual file entry\r\n// because an `.xlsx` file creator softwared really shouldn't attempt doing\r\n// anything \"funny\" when writing a file, hence this adherence requirement.\r\n//\r\nfunction createZipFileValidator(onValidationResult) {\r\n\tconst firstBytesCount = 2\r\n\tconst firstBytes = []\r\n\tlet firstBytesCheckResult\r\n\treturn {\r\n\t\tvalidateChunk(chunk) {\r\n\t\t\tif (firstBytes.length < 2) {\r\n\t\t\t\tlet i = 0\r\n\t\t\t\twhile (i < chunk.length && i < firstBytesCount) {\r\n\t\t\t\t\tfirstBytes.push(chunk[i])\r\n\t\t\t\t\ti++\r\n\t\t\t\t}\r\n\t\t\t\tif (firstBytes.length === 2) {\r\n\t\t\t\t\tconst isValid = firstBytes[0] === 0x50 && firstBytes[1] === 0x4B\r\n\t\t\t\t\tonValidationResult(isValid)\r\n\t\t\t\t}\r\n\t\t\t}\r\n\t\t}\r\n\t}\r\n}"],"mappings":"AAAA;AACA;;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASA,KAAK,EAAEC,YAAY,QAAQ,QAAQ;AAE5C,SAASC,MAAM,QAAQ,aAAa;;AAEpC;AACA;AACA;AACA;AACA;AACA,eAAe,SAASC,eAAeA,CAACC,MAAM,EAAmB;EAAA,IAAAC,IAAA,GAAAC,SAAA,CAAAC,MAAA,QAAAD,SAAA,QAAAE,SAAA,GAAAF,SAAA,MAAJ,CAAC,CAAC;IAAbG,MAAM,GAAAJ,IAAA,CAANI,MAAM;EACvD;EACA,IAAMC,KAAK,GAAG,CAAC,CAAC;EAEhB,OAAO,IAAIC,OAAO,CAAC,UAACC,OAAO,EAAEC,MAAM,EAAK;IACvC,IAAIC,OAAO,GAAG,KAAK;IAEnB,IAAMC,OAAO,GAAG,SAAVA,OAAOA,CAAIC,KAAK,EAAK;MAC1B,IAAI,CAACF,OAAO,EAAE;QACbA,OAAO,GAAG,IAAI;QACdD,MAAM,CAACG,KAAK,CAAC;MACd;IACD,CAAC;IAED,IAAAC,qBAAA,GAA0BC,sBAAsB,CAAC,UAACC,OAAO,EAAK;QAC7D,IAAI,CAACA,OAAO,EAAE;UACbJ,OAAO,CAAC,IAAIK,KAAK,CAAC,wBAAwB,CAAC,CAAC;QAC7C;MACD,CAAC,CAAC;MAJMC,aAAa,GAAAJ,qBAAA,CAAbI,aAAa;;IAMrB;IACA;IACA,IAAMC,KAAK,GAAG,IAAItB,KAAK,CAAC,UAACuB,KAAK,EAAK;MAClC;MACA;MACA,IAAIT,OAAO,EAAE;QACZ;MACD;;MAEA;MACA;MACA,IAAIS,KAAK,CAACC,IAAI,CAACC,QAAQ,CAAC,GAAG,CAAC,EAAE;QAC7B;MACD;;MAEA;MACA;MACA;MACA,IAAIhB,MAAM,IAAI,CAACA,MAAM,CAAC;QAAEiB,IAAI,EAAEH,KAAK,CAACC;MAAK,CAAC,CAAC,EAAE;QAC5C;MACD;MAEA,IAAMG,MAAM,GAAG,EAAE;;MAEjB;MACA;MACAJ,KAAK,CAACK,MAAM,GAAG,UAACZ,KAAK,EAAEa,KAAK,EAAEC,MAAM,EAAK;QACxC,IAAId,KAAK,EAAE;UACV,OAAOD,OAAO,CAACC,KAAK,CAAC;QACtB;QACAW,MAAM,CAACI,IAAI,CAACF,KAAK,CAAC;QAClB,IAAIC,MAAM,EAAE;UACXpB,KAAK,CAACa,KAAK,CAACC,IAAI,CAAC,GAAGtB,MAAM,CAAC8B,MAAM,CAACL,MAAM,CAAC;QAC1C;MACD,CAAC;;MAED;MACAJ,KAAK,CAACU,KAAK,CAAC,CAAC;IACd,CAAC,CAAC;;IAEF;IACA;IACA;IACAX,KAAK,CAACY,QAAQ,CAACjC,YAAY,CAAC;IAE5BG;IACC;IAAA,CACC+B,EAAE,CAAC,OAAO,EAAEpB,OAAO;IACpB;IAAA,CACCoB,EAAE,CAAC,MAAM,EAAE,UAACN,KAAK,EAAK;MACtB;MACA;MACA,IAAIf,OAAO,EAAE;QACZ;MACD;MACA;MACAO,aAAa,CAACQ,KAAK,CAAC;MACpB;MACA;MACA,IAAIf,OAAO,EAAE;QACZ;MACD;MACA;MACA;MACA;MACA;MACA,IAAI;QACHQ,KAAK,CAACS,IAAI,CAACF,KAAK,EAAE,KAAK,CAAC;MACzB,CAAC,CAAC,OAAOb,KAAK,EAAE;QACfD,OAAO,CAACC,KAAK,CAAC;MACf;IACD,CAAC;IACD;IACA;IAAA,CACCmB,EAAE,CAAC,KAAK,EAAE,YAAM;MAChB;MACA;MACA,IAAIrB,OAAO,EAAE;QACZ;MACD;MACA,IAAI;QACH;QACA;QACAQ,KAAK,CAACS,IAAI,CAAC,IAAIK,UAAU,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC;QACnC;QACAxB,OAAO,CAACF,KAAK,CAAC;MACf,CAAC,CAAC,OAAOM,KAAK,EAAE;QACfD,OAAO,CAACC,KAAK,CAAC;MACf;IACD,CAAC,CAAC;EACJ,CAAC,CAAC;AACH;;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASE,sBAAsBA,CAACmB,kBAAkB,EAAE;EACnD,IAAMC,eAAe,GAAG,CAAC;EACzB,IAAMC,UAAU,GAAG,EAAE;EACrB,IAAIC,qBAAqB;EACzB,OAAO;IACNnB,aAAa,WAAAA,cAACQ,KAAK,EAAE;MACpB,IAAIU,UAAU,CAAChC,MAAM,GAAG,CAAC,EAAE;QAC1B,IAAIkC,CAAC,GAAG,CAAC;QACT,OAAOA,CAAC,GAAGZ,KAAK,CAACtB,MAAM,IAAIkC,CAAC,GAAGH,eAAe,EAAE;UAC/CC,UAAU,CAACR,IAAI,CAACF,KAAK,CAACY,CAAC,CAAC,CAAC;UACzBA,CAAC,EAAE;QACJ;QACA,IAAIF,UAAU,CAAChC,MAAM,KAAK,CAAC,EAAE;UAC5B,IAAMY,OAAO,GAAGoB,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI,IAAIA,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI;UAChEF,kBAAkB,CAAClB,OAAO,CAAC;QAC5B;MACD;IACD;EACD,CAAC;AACF"}
+{"version":3,"file":"unzipFromStream.js","names":["default"],"sources":["../../source/zip/unzipFromStream.js"],"sourcesContent":["// Currently, there're two implementations:\r\n// * `fflate` — a pure-javascript implementation that uses `fflate` package.\r\n// * `unzipper` — a \"native\" Node.js module that uses Node's `zlib` which is written in C.\r\n//\r\n// The implementations are compared in a benchmark:\r\n//\r\n// ```\r\n// npm run test:benchmark:unzipFromStream\r\n// ```\r\n//\r\n// The benchmark tells that `unzipper` is 2x faster than `fflate`.\r\n//\r\nexport { default as default } from './unzipFromStream.unzipper.js'"],"mappings":"AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASA,OAAkB,QAAQ,+BAA+B"}

package/modules/zip/unzipFromStream.unzipper.js

@@ -0,0 +1,121 @@
+// `unzipper` uses Node.js native `zlib` module.
+// This means that most likely it's more performant than an equivalent javascript-only implementation.
+
+// `unzipper` has a bug — it is written in CommonJS and uses a conditional `require()` call
+// to create an AWS S3 client instance: `require('@aws-sdk/client-s3')`.
+// https://github.com/ZJONSSON/node-unzipper/issues/330
+// The author of the package refuses to acknowledge the issue and apply a simple fix.
+// The package, therefore, looks like an "abandonware".
+// Someone could easily fix the issue though by forking the package.
+//
+// When not forking to apply the fix, a workaround could be to install "@aws-sdk/client-s3"
+// package manually, which introduces unnecessary complexity and "esoteric knowledge" to the application setup.
+//
+// Another workaround could be to "alias" "@aws-sdk/client-s3" package in a "bundler" configuration file
+// with a path to a `*.js` file containing just "export default null". But that kind of a workaround would also
+// result in errors when using other packages that happen to `import` anything from "@aws-sdk/client-s3" package,
+// so it's not really a workaround but more of a ticking bomb. Not to mention that it introduces
+// unnecessary complexity and "esoteric knowledge" to the application setup.
+//
+import { Parse } from 'unzipper-esm';
+import { Buffer } from 'buffer';
+
+/**
+ * Reads `*.zip` file contents.
+ * @param  {Stream} stream
+ * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.
+ */
+export default function unzipFromStream(stream) {
+  var _ref = arguments.length > 1 && arguments[1] !== undefined ? arguments[1] : {},
+    filter = _ref.filter;
+  // The `files` object stores the files and their contents.
+  var files = {};
+  return new Promise(function (resolve, reject) {
+    var promises = [];
+    var errored = false;
+    var onError = function onError(error) {
+      if (!errored) {
+        errored = true;
+        reject(error);
+      }
+    };
+    stream
+    // This first "error" listener catches the original stream errors.
+    //
+    // That's because the .pipe() method does not automatically propagate errors
+    // from a source (input) stream to the destination stream or the end of the pipeline.
+    // You would need to attach an 'error' event handler to each stream in the chain.
+    //
+    // A more convenient alternative would be to use `stream.pipeline()` function:
+    // `pipeline(stream1, stream2, (error) => { ... })`
+    //
+    .on('error', onError)
+    // Pipe the input stream through the unzipper stream.
+    .pipe(Parse())
+    // This second "error" listener catches the unzipper stream errors.
+    //
+    // That's because the .pipe() method does not automatically propagate errors
+    // from a source (input) stream to the destination stream or the end of the pipeline.
+    // You would need to attach an 'error' event handler to each stream in the chain.
+    //
+    // A more convenient alternative would be to use `stream.pipeline()` function:
+    // `pipeline(stream1, stream2, (error) => { ... })`
+    //
+    .on('error', onError)
+    // The unzipper stream is closed when all `entries` have been reported.
+    .on('finish', function () {
+      if (!errored) {
+        // Wait for all `entries` to be read.
+        // The second argument of `.then()` function is not required
+        // but I didn't remove it just to potentially prevent any potential silly bugs
+        // in case of some potential changes in some potential future.
+        Promise.all(promises).then(function () {
+          resolve(files);
+        }, onError);
+      }
+    }).on('entry', function (entry) {
+      // See if this file should be ignored.
+      var ignore = false;
+      // `entry.type` could be 'Directory' or 'File'.
+      // Ignore anything except files.
+      if (entry.type === 'Directory') {
+        ignore = true;
+      }
+      if (errored) {
+        ignore = true;
+      }
+      if (filter) {
+        if (!filter({
+          path: entry.path
+        })) {
+          ignore = true;
+        }
+      }
+
+      // If this file should be ignored.
+      if (ignore) {
+        // Call `entry.autodrain()` when you do not intend to process a specific `entry`'s raw data.
+        // Otherwise, if an `entry` is not consumed (via .pipe(), .buffer(), or .autodrain()),
+        // the stream will halt, preventing further file processing.
+        entry.autodrain().on('error', onError);
+        return;
+      }
+      var chunks = [];
+      promises.push(new Promise(function (resolve) {
+        // `entry` seems to be a generic Node.js stream.
+        // `entry.pipe()` pipes the file contents to a stream.
+        // `entry.stream()` returns a readable stream.
+        // `entry.buffer()` returns a promise that resolves to a `Buffer` with the file contents.
+        entry.on('data', function (data) {
+          chunks.push(data);
+        }).on('error', function (error) {
+          onError(error);
+        }).on('finish', function () {
+          files[entry.path] = Buffer.concat(chunks);
+          resolve();
+        });
+      }));
+    });
+  });
+}
+//# sourceMappingURL=unzipFromStream.unzipper.js.map

package/modules/zip/unzipFromStream.unzipper.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"unzipFromStream.unzipper.js","names":["Parse","Buffer","unzipFromStream","stream","_ref","arguments","length","undefined","filter","files","Promise","resolve","reject","promises","errored","onError","error","on","pipe","all","then","entry","ignore","type","path","autodrain","chunks","push","data","concat"],"sources":["../../source/zip/unzipFromStream.unzipper.js"],"sourcesContent":["// `unzipper` uses Node.js native `zlib` module.\r\n// This means that most likely it's more performant than an equivalent javascript-only implementation.\r\n\r\n// `unzipper` has a bug — it is written in CommonJS and uses a conditional `require()` call\r\n// to create an AWS S3 client instance: `require('@aws-sdk/client-s3')`.\r\n// https://github.com/ZJONSSON/node-unzipper/issues/330\r\n// The author of the package refuses to acknowledge the issue and apply a simple fix.\r\n// The package, therefore, looks like an \"abandonware\".\r\n// Someone could easily fix the issue though by forking the package.\r\n//\r\n// When not forking to apply the fix, a workaround could be to install \"@aws-sdk/client-s3\"\r\n// package manually, which introduces unnecessary complexity and \"esoteric knowledge\" to the application setup.\r\n//\r\n// Another workaround could be to \"alias\" \"@aws-sdk/client-s3\" package in a \"bundler\" configuration file\r\n// with a path to a `*.js` file containing just \"export default null\". But that kind of a workaround would also\r\n// result in errors when using other packages that happen to `import` anything from \"@aws-sdk/client-s3\" package,\r\n// so it's not really a workaround but more of a ticking bomb. Not to mention that it introduces\r\n// unnecessary complexity and \"esoteric knowledge\" to the application setup.\r\n//\r\nimport { Parse } from 'unzipper-esm'\r\n\r\nimport { Buffer } from 'buffer'\r\n\r\n/**\r\n * Reads `*.zip` file contents.\r\n * @param  {Stream} stream\r\n * @return {Promise<Record<string,Buffer>>} Resolves to an object holding `*.zip` file entries. P.S. `Buffer` is a `Uint8Array`.\r\n */\r\nexport default function unzipFromStream(stream, { filter } = {}) {\r\n\t// The `files` object stores the files and their contents.\r\n\tconst files = {}\r\n\r\n\treturn new Promise((resolve, reject) => {\r\n\t\tconst promises = []\r\n\r\n\t\tlet errored = false\r\n\r\n\t\tconst onError = (error) => {\r\n\t\t\tif (!errored) {\r\n\t\t\t\terrored = true\r\n\t\t\t\treject(error)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\tstream\r\n\t\t\t// This first \"error\" listener catches the original stream errors.\r\n\t\t\t//\r\n\t\t\t// That's because the .pipe() method does not automatically propagate errors\r\n\t\t\t// from a source (input) stream to the destination stream or the end of the pipeline.\r\n\t\t\t// You would need to attach an 'error' event handler to each stream in the chain.\r\n\t\t\t//\r\n\t\t\t// A more convenient alternative would be to use `stream.pipeline()` function:\r\n\t\t\t// `pipeline(stream1, stream2, (error) => { ... })`\r\n\t\t\t//\r\n\t\t\t.on('error', onError)\r\n\t\t\t// Pipe the input stream through the unzipper stream.\r\n\t\t\t.pipe(Parse())\r\n\t\t\t// This second \"error\" listener catches the unzipper stream errors.\r\n\t\t\t//\r\n\t\t\t// That's because the .pipe() method does not automatically propagate errors\r\n\t\t\t// from a source (input) stream to the destination stream or the end of the pipeline.\r\n\t\t\t// You would need to attach an 'error' event handler to each stream in the chain.\r\n\t\t\t//\r\n\t\t\t// A more convenient alternative would be to use `stream.pipeline()` function:\r\n\t\t\t// `pipeline(stream1, stream2, (error) => { ... })`\r\n\t\t\t//\r\n\t\t\t.on('error', onError)\r\n\t\t\t// The unzipper stream is closed when all `entries` have been reported.\r\n\t\t\t.on('finish', () => {\r\n\t\t\t\tif (!errored) {\r\n\t\t\t\t\t// Wait for all `entries` to be read.\r\n\t\t\t\t\t// The second argument of `.then()` function is not required\r\n\t\t\t\t\t// but I didn't remove it just to potentially prevent any potential silly bugs\r\n\t\t\t\t\t// in case of some potential changes in some potential future.\r\n\t\t\t\t\tPromise.all(promises).then(() => {\r\n\t\t\t\t\t\tresolve(files)\r\n\t\t\t\t\t}, onError)\r\n\t\t\t\t}\r\n\t\t\t})\r\n\t\t\t.on('entry', (entry) => {\r\n\t\t\t\t// See if this file should be ignored.\r\n\t\t\t\tlet ignore = false\r\n\t\t\t\t// `entry.type` could be 'Directory' or 'File'.\r\n\t\t\t\t// Ignore anything except files.\r\n\t\t\t\tif (entry.type === 'Directory') {\r\n\t\t\t\t\tignore = true\r\n\t\t\t\t}\r\n\t\t\t\tif (errored) {\r\n\t\t\t\t\tignore = true\r\n\t\t\t\t}\r\n\t\t\t\tif (filter) {\r\n\t\t\t\t\tif (!filter({ path: entry.path })) {\r\n\t\t\t\t\t\tignore = true\r\n\t\t\t\t\t}\r\n\t\t\t\t}\r\n\r\n\t\t\t\t// If this file should be ignored.\r\n\t\t\t\tif (ignore) {\r\n\t\t\t\t\t// Call `entry.autodrain()` when you do not intend to process a specific `entry`'s raw data.\r\n\t\t\t\t\t// Otherwise, if an `entry` is not consumed (via .pipe(), .buffer(), or .autodrain()),\r\n\t\t\t\t\t// the stream will halt, preventing further file processing.\r\n\t\t\t\t\tentry.autodrain().on('error', onError)\r\n\t\t\t\t\treturn\r\n\t\t\t\t}\r\n\r\n\t\t\t\tconst chunks = []\r\n\r\n\t\t\t\tpromises.push(new Promise((resolve) => {\r\n\t\t\t\t\t// `entry` seems to be a generic Node.js stream.\r\n\t\t\t\t\t// `entry.pipe()` pipes the file contents to a stream.\r\n\t\t\t\t\t// `entry.stream()` returns a readable stream.\r\n\t\t\t\t\t// `entry.buffer()` returns a promise that resolves to a `Buffer` with the file contents.\r\n\t\t\t\t\tentry\r\n\t\t\t\t\t\t.on('data', (data) => {\r\n\t\t\t\t\t\t\tchunks.push(data)\r\n\t\t\t\t\t\t})\r\n\t\t\t\t\t\t.on('error', (error) => {\r\n\t\t\t\t\t\t\tonError(error)\r\n\t\t\t\t\t\t})\r\n\t\t\t\t\t\t.on('finish', () => {\r\n\t\t\t\t\t\t\tfiles[entry.path] = Buffer.concat(chunks)\r\n\t\t\t\t\t\t\tresolve()\r\n\t\t\t\t\t\t})\r\n\t\t\t\t}))\r\n\t\t\t})\r\n\t})\r\n}"],"mappings":"AAAA;AACA;;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASA,KAAK,QAAQ,cAAc;AAEpC,SAASC,MAAM,QAAQ,QAAQ;;AAE/B;AACA;AACA;AACA;AACA;AACA,eAAe,SAASC,eAAeA,CAACC,MAAM,EAAmB;EAAA,IAAAC,IAAA,GAAAC,SAAA,CAAAC,MAAA,QAAAD,SAAA,QAAAE,SAAA,GAAAF,SAAA,MAAJ,CAAC,CAAC;IAAbG,MAAM,GAAAJ,IAAA,CAANI,MAAM;EACvD;EACA,IAAMC,KAAK,GAAG,CAAC,CAAC;EAEhB,OAAO,IAAIC,OAAO,CAAC,UAACC,OAAO,EAAEC,MAAM,EAAK;IACvC,IAAMC,QAAQ,GAAG,EAAE;IAEnB,IAAIC,OAAO,GAAG,KAAK;IAEnB,IAAMC,OAAO,GAAG,SAAVA,OAAOA,CAAIC,KAAK,EAAK;MAC1B,IAAI,CAACF,OAAO,EAAE;QACbA,OAAO,GAAG,IAAI;QACdF,MAAM,CAACI,KAAK,CAAC;MACd;IACD,CAAC;IAEDb;IACC;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IAAA,CACCc,EAAE,CAAC,OAAO,EAAEF,OAAO;IACpB;IAAA,CACCG,IAAI,CAAClB,KAAK,CAAC,CAAC;IACb;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IAAA,CACCiB,EAAE,CAAC,OAAO,EAAEF,OAAO;IACpB;IAAA,CACCE,EAAE,CAAC,QAAQ,EAAE,YAAM;MACnB,IAAI,CAACH,OAAO,EAAE;QACb;QACA;QACA;QACA;QACAJ,OAAO,CAACS,GAAG,CAACN,QAAQ,CAAC,CAACO,IAAI,CAAC,YAAM;UAChCT,OAAO,CAACF,KAAK,CAAC;QACf,CAAC,EAAEM,OAAO,CAAC;MACZ;IACD,CAAC,CAAC,CACDE,EAAE,CAAC,OAAO,EAAE,UAACI,KAAK,EAAK;MACvB;MACA,IAAIC,MAAM,GAAG,KAAK;MAClB;MACA;MACA,IAAID,KAAK,CAACE,IAAI,KAAK,WAAW,EAAE;QAC/BD,MAAM,GAAG,IAAI;MACd;MACA,IAAIR,OAAO,EAAE;QACZQ,MAAM,GAAG,IAAI;MACd;MACA,IAAId,MAAM,EAAE;QACX,IAAI,CAACA,MAAM,CAAC;UAAEgB,IAAI,EAAEH,KAAK,CAACG;QAAK,CAAC,CAAC,EAAE;UAClCF,MAAM,GAAG,IAAI;QACd;MACD;;MAEA;MACA,IAAIA,MAAM,EAAE;QACX;QACA;QACA;QACAD,KAAK,CAACI,SAAS,CAAC,CAAC,CAACR,EAAE,CAAC,OAAO,EAAEF,OAAO,CAAC;QACtC;MACD;MAEA,IAAMW,MAAM,GAAG,EAAE;MAEjBb,QAAQ,CAACc,IAAI,CAAC,IAAIjB,OAAO,CAAC,UAACC,OAAO,EAAK;QACtC;QACA;QACA;QACA;QACAU,KAAK,CACHJ,EAAE,CAAC,MAAM,EAAE,UAACW,IAAI,EAAK;UACrBF,MAAM,CAACC,IAAI,CAACC,IAAI,CAAC;QAClB,CAAC,CAAC,CACDX,EAAE,CAAC,OAAO,EAAE,UAACD,KAAK,EAAK;UACvBD,OAAO,CAACC,KAAK,CAAC;QACf,CAAC,CAAC,CACDC,EAAE,CAAC,QAAQ,EAAE,YAAM;UACnBR,KAAK,CAACY,KAAK,CAACG,IAAI,CAAC,GAAGvB,MAAM,CAAC4B,MAAM,CAACH,MAAM,CAAC;UACzCf,OAAO,CAAC,CAAC;QACV,CAAC,CAAC;MACJ,CAAC,CAAC,CAAC;IACJ,CAAC,CAAC;EACJ,CAAC,CAAC;AACH"}