refactor(image-stream): parse xz and gzip metadata using a custom read function (#1590)

This commit refactors the xz and gzip image handlers to pass/use a
custom read function to be able to determine the uncompressed size, and
other needed metadata.

By using this function (which currently only uses the `fs` module), we
can implement support for getting the uncompressed size of compressed
files using HTTP Ranges.

Change-Type: patch
Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

lib/image-stream/gzip.js

@@ -16,9 +16,6 @@
 
 'use strict';
 
-const Bluebird = require('bluebird');
-const fs = Bluebird.promisifyAll(require('fs'));
-
 /**
  * @summary The byte length of ISIZE
  * @type {Number}
@@ -29,46 +26,47 @@ const fs = Bluebird.promisifyAll(require('fs'));
 const ISIZE_LENGTH = 4;
 
 /**
- * @summary Get a gzip file uncompressed size
+ * @summary Get the estimated uncompressed size of a gzip file
  * @function
  * @public
  *
  * @description
  * This function determines the uncompressed size of the gzip file
- * by reading its `ISIZE`. The specification clarifies that this
- * value is just an estimation.
+ * by reading its `ISIZE` field at the end of the file. The specification
+ * clarifies that this value is just an estimation.
  *
- * @param {String} file - path to gzip file
+ * @param {Object} options - options
+ * @param {Number} options.size - file size
+ * @param {Function} options.read - read function (position, count)
  * @fulfil {Number} - uncompressed size
  * @returns {Promise}
  *
  * @example
- * gzip.getUncompressedSize('path/to/file.gz').then((uncompressedSize) => {
+ * const fd = fs.openSync('path/to/image', 'r');
+ *
+ * gzip.getUncompressedSize({
+ *   size: fs.statSync('path/to/image.gz').size,
+ *   read: (position, count) => {
+ *     const buffer = Buffer.alloc(count);
+ *     return new Promise((resolve, reject) => {
+ *       fs.read(fd, buffer, 0, count, position, (error) => {
+ *         if (error) {
+ *           return reject(error);
+ *         }
+ *
+ *         resolve(buffer);
+ *       });
+ *     });
+ *   }
+ * }).then((uncompressedSize) => {
  *   console.log(`The uncompressed size is: ${uncompressedSize}`);
+ *   fs.closeSync(fd);
  * });
  */
-exports.getUncompressedSize = (file) => {
-  return Bluebird.using(fs.openAsync(file, 'r').disposer((fileDescriptor) => {
-    return fs.closeAsync(fileDescriptor);
-  }), (fileDescriptor) => {
-    return fs.fstatAsync(fileDescriptor).then((stats) => {
-      const ISIZE_BUFFER_FILL_VALUE = 0;
-      const ISIZE_BUFFER_START = 0;
-      const isizeBuffer = Buffer.alloc(ISIZE_LENGTH, ISIZE_BUFFER_FILL_VALUE);
-
-      return fs.readAsync(
-        fileDescriptor,
-        isizeBuffer,
-        ISIZE_BUFFER_START,
-        ISIZE_LENGTH,
-        stats.size - ISIZE_LENGTH
-      ).then((bytesRead) => {
-        if (bytesRead !== ISIZE_LENGTH) {
-          throw new Error(`Bytes read mismatch: ${bytesRead} != ${ISIZE_LENGTH}`);
-        }
-
-        return isizeBuffer.readUInt32LE(ISIZE_BUFFER_START);
-      });
-    });
+exports.getUncompressedSize = (options) => {
+  const ISIZE_BUFFER_START = 0;
+  const ISIZE_POSITION = options.size - ISIZE_LENGTH;
+  return options.read(ISIZE_POSITION, ISIZE_LENGTH).then((buffer) => {
+    return buffer.readUInt32LE(ISIZE_BUFFER_START);
   });
 };

lib/image-stream/handlers.js

@@ -27,6 +27,7 @@ const unbzip2Stream = require('unbzip2-stream');
 const gzip = require('./gzip');
 const udif = Bluebird.promisifyAll(require('udif'));
 const archive = require('./archive');
+const utils = require('./utils');
 const zipArchiveHooks = require('./archive-hooks/zip');
 const fileExtensions = require('../shared/file-extensions');
 const path = require('path');
@@ -83,7 +84,16 @@ module.exports = {
    * @returns {Promise}
    */
   'application/gzip': (imagePath, options) => {
-    return gzip.getUncompressedSize(imagePath).then((uncompressedSize) => {
+    return Bluebird.using(fs.openAsync(imagePath, 'r').disposer((fileDescriptor) => {
+      return fs.closeAsync(fileDescriptor);
+    }), (fileDescriptor) => {
+      return gzip.getUncompressedSize({
+        size: options.size,
+        read: (position, count) => {
+          return utils.readBufferFromImageFileDescriptor(fileDescriptor, position, count);
+        }
+      });
+    }).then((uncompressedSize) => {
       return Bluebird.props({
         path: imagePath,
         archiveExtension: fileExtensions.getLastFileExtension(imagePath),
@@ -115,9 +125,14 @@ module.exports = {
    * @returns {Promise}
    */
   'application/x-xz': (imagePath, options) => {
-    return fs.openAsync(imagePath, 'r').then((fileDescriptor) => {
-      return lzma.parseFileIndexFDAsync(fileDescriptor).tap(() => {
-        return fs.closeAsync(fileDescriptor);
+    return Bluebird.using(fs.openAsync(imagePath, 'r').disposer((fileDescriptor) => {
+      return fs.closeAsync(fileDescriptor);
+    }), (fileDescriptor) => {
+      return lzma.parseFileIndexAsync({
+        fileSize: options.size,
+        read: (count, position, callback) => {
+          utils.readBufferFromImageFileDescriptor(fileDescriptor, position, count).asCallback(callback);
+        }
       });
     }).then((metadata) => {
       return {

lib/image-stream/mime.js

@@ -18,9 +18,10 @@
 
 const _ = require('lodash');
 const Bluebird = require('bluebird');
+const fs = Bluebird.promisifyAll(require('fs'));
 const fileType = require('file-type');
 const mime = require('mime-types');
-const fs = require('fs');
+const utils = require('./utils');
 
 /**
  * @summary The default MIME type
@@ -50,15 +51,13 @@ exports.getMimeTypeFromFileName = (filename) => {
     return Bluebird.resolve(mimeType);
   }
 
+  const FILE_TYPE_ID_START = 0;
   const FILE_TYPE_ID_BYTES = 262;
 
   return Bluebird.using(fs.openAsync(filename, 'r').disposer((fileDescriptor) => {
     return fs.closeAsync(fileDescriptor);
   }), (fileDescriptor) => {
-    const BUFFER_START = 0;
-    const buffer = Buffer.alloc(FILE_TYPE_ID_BYTES);
-
-    return fs.readAsync(fileDescriptor, buffer, BUFFER_START, FILE_TYPE_ID_BYTES, null).then(() => {
+    return utils.readBufferFromImageFileDescriptor(fileDescriptor, FILE_TYPE_ID_START, FILE_TYPE_ID_BYTES).then((buffer) => {
       return _.get(fileType(buffer), [ 'mime' ], exports.DEFAULT_MIME_TYPE);
     });
   });

lib/image-stream/utils.js

@@ -17,6 +17,41 @@
 'use strict';
 
 const Bluebird = require('bluebird');
+const fs = Bluebird.promisifyAll(require('fs'));
+const errors = require('../shared/errors');
+
+/**
+ * @summary Read a buffer from an image file descriptor
+ * @function
+ * @private
+ *
+ * @param {Number} fileDescriptor - file descriptor
+ * @param {Number} position - image position to start reading from
+ * @param {Number} count - number of bytes to read
+ * @fulfil {Buffer} - buffer
+ * @returns {Promise}
+ *
+ * @example
+ * fs.openAsync('path/to/image.img', 'r').then((fileDescriptor) => {
+ *   return utils.readBufferFromImageFileDescriptor(fileDescriptor, 0, 512);
+ * }).then((buffer) => {
+ *   console.log(buffer);
+ * });
+ */
+exports.readBufferFromImageFileDescriptor = (fileDescriptor, position, count) => {
+  const BUFFER_FILL_VALUE = 0;
+  const BUFFER_START_POSITION = 0;
+  const buffer = Buffer.alloc(count, BUFFER_FILL_VALUE);
+
+  return fs.readAsync(fileDescriptor, buffer, BUFFER_START_POSITION, count, position).tap((bytesRead) => {
+    if (bytesRead !== count) {
+      throw errors.createUserError({
+        title: 'Looks like the image is truncated',
+        description: `We tried to read ${count} bytes at ${position}, but got ${bytesRead} bytes instead`
+      });
+    }
+  }).return(buffer);
+};
 
 /**
  * @summary Extract the data of a readable stream