Merge "Add an interface for bspatch reader"

am: 68540f685e

Change-Id: I5da1be9150aef4317084a177547ea97808a4c5d1

Android.bp

@@ -41,12 +41,15 @@ cc_library_static {
 
     srcs: [
         "bspatch.cc",
+        "bz2_decompressor.cc",
         "buffer_file.cc",
         "extents.cc",
         "extents_file.cc",
         "file.cc",
         "memory_file.cc",
+        "patch_reader.cc",
         "sink_file.cc",
+        "utils.cc",
     ],
 }
 
@@ -105,6 +108,7 @@ cc_test {
         "diff_encoder_unittest.cc",
         "extents_file_unittest.cc",
         "extents_unittest.cc",
+        "patch_reader_unittest.cc",
         "patch_writer_unittest.cc",
         "split_patch_writer_unittest.cc",
         "suffix_array_index_unittest.cc",

Makefile

@@ -43,12 +43,15 @@ bsdiff_src_files := \
 # "bspatch" program.
 bspatch_src_files := \
     bspatch.cc \
+    bz2_decompressor.cc \
     buffer_file.cc \
     extents.cc \
     extents_file.cc \
     file.cc \
     memory_file.cc \
-    sink_file.cc
+    patch_reader.cc \
+    sink_file.cc \
+    utils.cc
 
 # Unit test files.
 bsdiff_common_unittests := \
@@ -57,6 +60,7 @@ bsdiff_common_unittests := \
     diff_encoder_unittest.cc \
     extents_file_unittest.cc \
     extents_unittest.cc \
+    patch_reader_unittest.cc \
     patch_writer_unittest.cc \
     split_patch_writer_unittest.cc \
     suffix_array_index_unittest.cc \

bsdiff.gyp

@@ -31,12 +31,15 @@
   'variables': {
     'bspatch_sources': [
       'bspatch.cc',
+      'bz2_decompressor.cc',
       'buffer_file.cc',
       'extents.cc',
       'extents_file.cc',
       'file.cc',
       'memory_file.cc',
+      'patch_reader.cc',
       'sink_file.cc',
+      'utils.cc',
     ],
   },
   'targets': [
@@ -132,6 +135,7 @@
             'diff_encoder_unittest.cc',
             'extents_file_unittest.cc',
             'extents_unittest.cc',
+            'patch_reader_unittest.cc',
             'patch_writer_unittest.cc',
             'split_patch_writer_unittest.cc',
             'suffix_array_index_unittest.cc',

bspatch.cc

@@ -30,7 +30,6 @@ __FBSDID("$FreeBSD: src/usr.bin/bsdiff/bspatch/bspatch.c,v 1.1 2005/08/06 01:59:
 
 #include "bsdiff/bspatch.h"
 
-#include <bzlib.h>
 #include <errno.h>
 #include <fcntl.h>
 #include <inttypes.h>
@@ -41,75 +40,42 @@ __FBSDID("$FreeBSD: src/usr.bin/bsdiff/bspatch/bspatch.c,v 1.1 2005/08/06 01:59:
 #include <unistd.h>
 
 #include <algorithm>
-#include <limits>
 #include <memory>
 #include <vector>
 
 #include "bsdiff/buffer_file.h"
+#include "bsdiff/control_entry.h"
 #include "bsdiff/extents.h"
 #include "bsdiff/extents_file.h"
 #include "bsdiff/file.h"
 #include "bsdiff/file_interface.h"
 #include "bsdiff/memory_file.h"
+#include "bsdiff/patch_reader.h"
 #include "bsdiff/sink_file.h"
+#include "bsdiff/utils.h"
 
 namespace {
-
-int64_t ParseInt64(const u_char* buf) {
-  int64_t y;
-
-  y = buf[7] & 0x7F;
-  y = y * 256;
-  y += buf[6];
-  y = y * 256;
-  y += buf[5];
-  y = y * 256;
-  y += buf[4];
-  y = y * 256;
-  y += buf[3];
-  y = y * 256;
-  y += buf[2];
-  y = y * 256;
-  y += buf[1];
-  y = y * 256;
-  y += buf[0];
-
-  if (buf[7] & 0x80)
-    y = -y;
-
-  return y;
-}
-
-bool ReadBZ2(bz_stream* stream, uint8_t* data, size_t size) {
-  stream->next_out = (char*)data;
+// Read the data in |stream| and write |size| decompressed data to |file|;
+// using the buffer in |buf| with size |buf_size|.
+// Returns 0 on success, 1 on I/O error and 2 on data error.
+int ReadStreamAndWriteAll(
+    const std::unique_ptr<bsdiff::FileInterface>& file,
+    size_t size,
+    uint8_t* buf,
+    size_t buf_size,
+    const std::function<bool(uint8_t*, size_t)>& read_func) {
   while (size > 0) {
-    unsigned int read_size = std::min(
-        static_cast<size_t>(std::numeric_limits<unsigned int>::max()), size);
-    stream->avail_out = read_size;
-    int bz2err = BZ2_bzDecompress(stream);
-    if (bz2err != BZ_OK && bz2err != BZ_STREAM_END)
-      return false;
-    size -= read_size - stream->avail_out;
-  }
-  return true;
-}
-
-int ReadBZ2AndWriteAll(const std::unique_ptr<bsdiff::FileInterface>& file,
-                       bz_stream* stream,
-                       size_t size,
-                       uint8_t* buf,
-                       size_t buf_size) {
-  while (size > 0) {
-    size_t bytes_to_read = std::min(size, buf_size);
-    if (!ReadBZ2(stream, buf, bytes_to_read)) {
-      fprintf(stderr, "Failed to read bzip stream.\n");
+    size_t bytes_to_output = std::min(size, buf_size);
+    if (!read_func(buf, bytes_to_output)) {
+      fprintf(stderr, "Failed to read stream.\n");
       return 2;
     }
-    if (!WriteAll(file, buf, bytes_to_read)) {
+
+    if (!WriteAll(file, buf, bytes_to_output)) {
       perror("WriteAll() failed");
       return 1;
     }
-    size -= bytes_to_read;
+    size -= bytes_to_output;
   }
   return 0;
 }
@@ -282,76 +248,14 @@ int bspatch(const std::unique_ptr<FileInterface>& old_file,
             const std::unique_ptr<FileInterface>& new_file,
             const uint8_t* patch_data,
             size_t patch_size) {
-  int bz2err;
-  u_char buf[8];
-  off_t ctrl[3];
-
-  // File format:
-  //   0       8    "BSDIFF40"
-  //   8       8    X
-  //   16      8    Y
-  //   24      8    sizeof(new_filename)
-  //   32      X    bzip2(control block)
-  //   32+X    Y    bzip2(diff block)
-  //   32+X+Y  ???  bzip2(extra block)
-  // with control block a set of triples (x,y,z) meaning "add x bytes
-  // from oldfile to x bytes from the diff block; copy y bytes from the
-  // extra block; seek forwards in oldfile by z bytes".
-
-  // Check for appropriate magic.
-  if (memcmp(patch_data, "BSDIFF40", 8) != 0) {
-    fprintf(stderr, "Not a bsdiff patch.\n");
+  BsdiffPatchReader patch_reader;
+  if (!patch_reader.Init(patch_data, patch_size)) {
+    fprintf(stderr, "Failed to initialize patch reader\n");
     return 2;
   }
 
-  // Read lengths from header.
-  uint64_t oldsize, newsize;
-  int64_t ctrl_len = ParseInt64(patch_data + 8);
-  int64_t data_len = ParseInt64(patch_data + 16);
-  int64_t signed_newsize = ParseInt64(patch_data + 24);
-  newsize = signed_newsize;
-  if ((ctrl_len < 0) || (data_len < 0) || (signed_newsize < 0) ||
-      (32 + ctrl_len + data_len > static_cast<int64_t>(patch_size))) {
-    fprintf(stderr, "Corrupt patch.\n");
-    return 2;
-  }
-
-  bz_stream cstream;
-  cstream.next_in = (char*)patch_data + 32;
-  cstream.avail_in = ctrl_len;
-  cstream.bzalloc = nullptr;
-  cstream.bzfree = nullptr;
-  cstream.opaque = nullptr;
-  if ((bz2err = BZ2_bzDecompressInit(&cstream, 0, 0)) != BZ_OK) {
-    fprintf(stderr, "Failed to bzinit control stream (%d)\n", bz2err);
-    return 2;
-  }
-
-  bz_stream dstream;
-  dstream.next_in = (char*)patch_data + 32 + ctrl_len;
-  dstream.avail_in = data_len;
-  dstream.bzalloc = nullptr;
-  dstream.bzfree = nullptr;
-  dstream.opaque = nullptr;
-  if ((bz2err = BZ2_bzDecompressInit(&dstream, 0, 0)) != BZ_OK) {
-    fprintf(stderr, "Failed to bzinit diff stream (%d)\n", bz2err);
-    return 2;
-  }
-
-  bz_stream estream;
-  estream.next_in = (char*)patch_data + 32 + ctrl_len + data_len;
-  estream.avail_in = patch_size - (32 + ctrl_len + data_len);
-  estream.bzalloc = nullptr;
-  estream.bzfree = nullptr;
-  estream.opaque = nullptr;
-  if ((bz2err = BZ2_bzDecompressInit(&estream, 0, 0)) != BZ_OK) {
-    fprintf(stderr, "Failed to bzinit extra stream (%d)\n", bz2err);
-    return 2;
-  }
-
-  uint64_t old_file_pos = 0;
-
-  if (!old_file->GetSize(&oldsize)) {
+  uint64_t old_file_size;
+  if (!old_file->GetSize(&old_file_size)) {
     fprintf(stderr, "Cannot obtain the size of old file.\n");
     return 1;
   }
@@ -359,26 +263,18 @@ int bspatch(const std::unique_ptr<FileInterface>& old_file,
   // The oldpos can be negative, but the new pos is only incremented linearly.
   int64_t oldpos = 0;
   uint64_t newpos = 0;
-  std::vector<uint8_t> old_buf(1024 * 1024), new_buf(1024 * 1024);
-  while (newpos < newsize) {
-    int64_t i;
-    // Read control data.
-    for (i = 0; i <= 2; i++) {
-      if (!ReadBZ2(&cstream, buf, 8)) {
-        fprintf(stderr, "Failed to read control stream.\n");
-        return 2;
-      }
-      ctrl[i] = ParseInt64(buf);
-    }
-
-    // Sanity-check.
-    if (ctrl[0] < 0 || ctrl[1] < 0) {
-      fprintf(stderr, "Corrupt patch.\n");
+  std::vector<uint8_t> old_buf(1024 * 1024);
+  std::vector<uint8_t> new_buf(1024 * 1024);
+  uint64_t old_file_pos = 0;
+  while (newpos < patch_reader.new_file_size()) {
+    ControlEntry control_entry(0, 0, 0);
+    if (!patch_reader.ParseControlEntry(&control_entry)) {
+      fprintf(stderr, "Failed to read control stream\n");
       return 2;
     }
 
     // Sanity-check.
-    if (newpos + ctrl[0] > newsize) {
+    if (newpos + control_entry.diff_size > patch_reader.new_file_size()) {
       fprintf(stderr, "Corrupt patch.\n");
       return 2;
     }
@@ -386,26 +282,30 @@ int bspatch(const std::unique_ptr<FileInterface>& old_file,
     int ret = 0;
     // Add old data to diff string. It is enough to fseek once, at
     // the beginning of the sequence, to avoid unnecessary overhead.
-    if ((i = oldpos) < 0) {
+    int64_t seek_offset = oldpos;
+    if (seek_offset < 0) {
       // Write diff block directly to new file without adding old data,
       // because we will skip part where |oldpos| < 0.
-      ret = ReadBZ2AndWriteAll(new_file, &dstream, -i, new_buf.data(),
-                               new_buf.size());
+      ret = ReadStreamAndWriteAll(
+          new_file, oldpos - old_file_size, new_buf.data(), new_buf.size(),
+          std::bind(&BsdiffPatchReader::ReadDiffStream, &patch_reader,
+                    std::placeholders::_1, std::placeholders::_2));
       if (ret)
         return ret;
-      i = 0;
+      seek_offset = 0;
     }
 
-    // We just checked that |i| is not negative.
-    if (static_cast<uint64_t>(i) != old_file_pos && !old_file->Seek(i)) {
-      fprintf(stderr, "Error seeking input file to offset %" PRId64 ": %s\n", i,
-              strerror(errno));
+    // We just checked that |seek_offset| is not negative.
+    if (static_cast<uint64_t>(seek_offset) != old_file_pos &&
+        !old_file->Seek(seek_offset)) {
+      fprintf(stderr, "Error seeking input file to offset %" PRId64 ": %s\n",
+              seek_offset, strerror(errno));
       return 1;
     }
-    if ((old_file_pos = oldpos + ctrl[0]) > oldsize)
-      old_file_pos = oldsize;
 
-    size_t chunk_size = old_file_pos - i;
+    old_file_pos =
+        std::min<uint64_t>(oldpos + control_entry.diff_size, old_file_size);
+    size_t chunk_size = old_file_pos - seek_offset;
     while (chunk_size > 0) {
       size_t read_bytes;
       size_t bytes_to_read = std::min(chunk_size, old_buf.size());
@@ -418,7 +318,7 @@ int bspatch(const std::unique_ptr<FileInterface>& old_file,
         return 2;
       }
       // Read same amount of bytes from diff block
-      if (!ReadBZ2(&dstream, new_buf.data(), read_bytes)) {
+      if (!patch_reader.ReadDiffStream(new_buf.data(), read_bytes)) {
         fprintf(stderr, "Failed to read diff stream.\n");
         return 2;
       }
@@ -433,42 +333,46 @@ int bspatch(const std::unique_ptr<FileInterface>& old_file,
     }
 
     // Adjust pointers.
-    newpos += ctrl[0];
-    oldpos += ctrl[0];
+    newpos += control_entry.diff_size;
+    oldpos += control_entry.diff_size;
 
-    if (oldpos > static_cast<int64_t>(oldsize)) {
+    if (oldpos > static_cast<int64_t>(old_file_size)) {
       // Write diff block directly to new file without adding old data,
-      // because we skipped part where |oldpos| > oldsize.
-      ret = ReadBZ2AndWriteAll(new_file, &dstream, oldpos - oldsize,
-                               new_buf.data(), new_buf.size());
+      // because we skipped part where |oldpos| > old_file_size.
+      ret = ReadStreamAndWriteAll(
+          new_file, oldpos - old_file_size, new_buf.data(), new_buf.size(),
+          std::bind(&BsdiffPatchReader::ReadDiffStream, &patch_reader,
+                    std::placeholders::_1, std::placeholders::_2));
       if (ret)
         return ret;
     }
 
     // Sanity-check.
-    if (newpos + ctrl[1] > newsize) {
+    if (newpos + control_entry.extra_size > patch_reader.new_file_size()) {
       fprintf(stderr, "Corrupt patch.\n");
       return 2;
     }
 
     // Read extra block.
-    ret = ReadBZ2AndWriteAll(new_file, &estream, ctrl[1], new_buf.data(),
-                             new_buf.size());
+    ret = ReadStreamAndWriteAll(
+        new_file, control_entry.extra_size, new_buf.data(), new_buf.size(),
+        std::bind(&BsdiffPatchReader::ReadExtraStream, &patch_reader,
+                  std::placeholders::_1, std::placeholders::_2));
     if (ret)
       return ret;
 
     // Adjust pointers.
-    newpos += ctrl[1];
-    oldpos += ctrl[2];
+    newpos += control_entry.extra_size;
+    oldpos += control_entry.offset_increment;
   }
 
   // Close input file.
   old_file->Close();
 
-  // Clean up the bzip2 reads.
-  BZ2_bzDecompressEnd(&cstream);
-  BZ2_bzDecompressEnd(&dstream);
-  BZ2_bzDecompressEnd(&estream);
+  if (!patch_reader.Finish()) {
+    fprintf(stderr, "Failed to finish the patch reader\n");
+    return 2;
+  }
 
   if (!new_file->Close()) {
     perror("Error closing new file");

bz2_decompressor.cc

@@ -0,0 +1,65 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "bsdiff/bz2_decompressor.h"
+
+#include <algorithm>
+#include <limits>
+
+#include <bzlib.h>
+
+#include "bsdiff/logging.h"
+
+using std::endl;
+
+namespace bsdiff {
+
+bool BZ2Decompressor::SetInputData(const uint8_t* input_data, size_t size) {
+  // TODO(xunchang) update the avail_in for size > 2GB.
+  if (size > std::numeric_limits<unsigned int>::max()) {
+    LOG(ERROR) << "Oversized input data" << size << endl;
+    return false;
+  }
+
+  stream_.next_in = reinterpret_cast<char*>(const_cast<uint8_t*>(input_data));
+  stream_.avail_in = size;
+  stream_.bzalloc = nullptr;
+  stream_.bzfree = nullptr;
+  stream_.opaque = nullptr;
+  int bz2err = BZ2_bzDecompressInit(&stream_, 0, 0);
+  if (bz2err != BZ_OK) {
+    LOG(ERROR) << "Failed to bzinit control stream: " << bz2err << endl;
+    return false;
+  }
+  return true;
+}
+
+bool BZ2Decompressor::Read(uint8_t* output_data, size_t bytes_to_output) {
+  stream_.next_out = reinterpret_cast<char*>(output_data);
+  while (bytes_to_output > 0) {
+    size_t output_size = std::min<size_t>(
+        std::numeric_limits<unsigned int>::max(), bytes_to_output);
+    stream_.avail_out = output_size;
+
+    int bz2err = BZ2_bzDecompress(&stream_);
+    if (bz2err != BZ_OK && bz2err != BZ_STREAM_END) {
+      LOG(ERROR) << "Failed to decompress " << output_size
+                 << " bytes of data, error: " << bz2err << endl;
+      return false;
+    }
+    bytes_to_output -= (output_size - stream_.avail_out);
+  }
+  return true;
+}
+
+bool BZ2Decompressor::Close() {
+  int bz2err = BZ2_bzDecompressEnd(&stream_);
+  if (bz2err != BZ_OK) {
+    LOG(ERROR) << "BZ2_bzDecompressEnd returns with " << bz2err << endl;
+    return false;
+  }
+  return true;
+}
+
+}  // namespace bsdiff

bz2_decompressor.h

@@ -0,0 +1,30 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef _BSDIFF_BZ2_DECOMPRESSOR_H_
+#define _BSDIFF_BZ2_DECOMPRESSOR_H_
+
+#include <bzlib.h>
+
+#include "bsdiff/decompressor_interface.h"
+
+namespace bsdiff {
+
+class BZ2Decompressor : public DecompressorInterface {
+ public:
+  BZ2Decompressor() = default;
+
+  bool SetInputData(const uint8_t* input_data, size_t size) override;
+
+  bool Read(uint8_t* output_data, size_t bytes_to_output) override;
+
+  bool Close() override;
+
+ private:
+  bz_stream stream_;
+};
+
+}  // namespace bsdiff
+
+#endif

decompressor_interface.h

@@ -0,0 +1,32 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef _BSDIFF_DECOMPRESSOR_INTERFACE_H_
+#define _BSDIFF_DECOMPRESSOR_INTERFACE_H_
+
+#include <stddef.h>
+#include <stdint.h>
+
+namespace bsdiff {
+
+class DecompressorInterface {
+ public:
+  virtual ~DecompressorInterface() {}
+
+  // Set the buffer starting from |input_data| with length |size| as the input
+  // compressed stream. This buffer is owned by the caller and should stay
+  // valid until the Close() function is called.
+  virtual bool SetInputData(const uint8_t* input_data, size_t size) = 0;
+
+  // Decompress the stream and output |bytes_to_output| bytes of data to
+  // |output_data|. Returns false if not all bytes_to_output can be
+  // decompressed from the stream due to a decompressor error or EOF.
+  virtual bool Read(uint8_t* output_data, size_t bytes_to_output) = 0;
+
+  // Close the decompression stream.
+  virtual bool Close() = 0;
+};
+
+}  // namespace bsdiff
+#endif  // _BSDIFF_DECOMPRESSOR_INTERFACE_H_

include/bsdiff/control_entry.h

@@ -0,0 +1,33 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef _BSDIFF_CONTROL_ENTRY_H_
+#define _BSDIFF_CONTROL_ENTRY_H_
+
+#include <stdint.h>
+
+struct ControlEntry {
+  ControlEntry(uint64_t diff_size,
+               uint64_t extra_size,
+               int64_t offset_increment)
+      : diff_size(diff_size),
+        extra_size(extra_size),
+        offset_increment(offset_increment) {}
+
+  // The number of bytes to copy from the source and diff stream.
+  uint64_t diff_size;
+
+  // The number of bytes to copy from the extra stream.
+  uint64_t extra_size;
+
+  // The value to add to the source pointer after patching from the diff stream.
+  int64_t offset_increment;
+
+  bool operator==(const ControlEntry& o) const {
+    return diff_size == o.diff_size && extra_size == o.extra_size &&
+           offset_increment == o.offset_increment;
+  }
+};
+
+#endif  // _BSDIFF_CONTROL_ENTRY_H_

include/bsdiff/patch_writer_interface.h

@@ -9,31 +9,10 @@
 
 #include <cstddef>
 
+#include "bsdiff/control_entry.h"
+
 namespace bsdiff {
 
-struct ControlEntry {
-  ControlEntry(uint64_t diff_size,
-               uint64_t extra_size,
-               int64_t offset_increment)
-      : diff_size(diff_size),
-        extra_size(extra_size),
-        offset_increment(offset_increment) {}
-
-  // The number of bytes to copy from the source and diff stream.
-  uint64_t diff_size;
-
-  // The number of bytes to copy from the extra stream.
-  uint64_t extra_size;
-
-  // The value to add to the source pointer after patching from the diff stream.
-  int64_t offset_increment;
-
-  bool operator==(const ControlEntry& o) const {
-    return diff_size == o.diff_size && extra_size == o.extra_size &&
-           offset_increment == o.offset_increment;
-  }
-};
-
 class PatchWriterInterface {
  public:
   virtual ~PatchWriterInterface() = default;

patch_reader.cc

@@ -0,0 +1,141 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "bsdiff/patch_reader.h"
+
+#include <string.h>
+
+#include <limits>
+
+#include "bsdiff/bspatch.h"
+#include "bsdiff/bz2_decompressor.h"
+#include "bsdiff/logging.h"
+#include "bsdiff/utils.h"
+
+using std::endl;
+
+namespace bsdiff {
+
+const uint8_t kMagicHeader[] = "BSDIFF40";
+
+bool BsdiffPatchReader::Init(const uint8_t* patch_data, size_t patch_size) {
+  // File format:
+  //   0       8    "BSDIFF40"
+  //   8       8    X
+  //   16      8    Y
+  //   24      8    new_file_size
+  //   32      X    compressed control block
+  //   32+X    Y    compressed diff block
+  //   32+X+Y  ???  compressed extra block
+  // with control block a set of triples (x,y,z) meaning "add x bytes
+  // from oldfile to x bytes from the diff block; copy y bytes from the
+  // extra block; seek forwards in oldfile by z bytes".
+
+  // Check for appropriate magic.
+  if (memcmp(patch_data, kMagicHeader, 8) != 0) {
+    LOG(ERROR) << "Not a bsdiff patch." << endl;
+    return false;
+  }
+
+  // Read lengths from header.
+  int64_t ctrl_len = ParseInt64(patch_data + 8);
+  int64_t diff_len = ParseInt64(patch_data + 16);
+  int64_t signed_newsize = ParseInt64(patch_data + 24);
+  if ((ctrl_len < 0) || (diff_len < 0) || (signed_newsize < 0) ||
+      (32 + ctrl_len + diff_len > static_cast<int64_t>(patch_size))) {
+    LOG(ERROR) << "Corrupt patch.  ctrl_len: " << ctrl_len
+               << ", data_len: " << diff_len
+               << ", new_file_size: " << signed_newsize
+               << ", patch_size: " << patch_size;
+    return false;
+  }
+  new_file_size_ = signed_newsize;
+
+  // TODO(xunchang) set the correct decompressor based on the info in the
+  // header.
+  ctrl_stream_.reset(new BZ2Decompressor());
+  diff_stream_.reset(new BZ2Decompressor());
+  extra_stream_.reset(new BZ2Decompressor());
+
+  int64_t offset = 32;
+  if (!ctrl_stream_->SetInputData(const_cast<uint8_t*>(patch_data) + offset,
+                                  ctrl_len)) {
+    LOG(ERROR) << "Failed to init ctrl stream, ctrl_len: " << ctrl_len << endl;
+    return false;
+  }
+
+  offset += ctrl_len;
+  if (!diff_stream_->SetInputData(const_cast<uint8_t*>(patch_data) + offset,
+                                  diff_len)) {
+    LOG(ERROR) << "Failed to init ctrl stream, diff_len: " << diff_len << endl;
+    return false;
+  }
+
+  offset += diff_len;
+  if (!extra_stream_->SetInputData(const_cast<uint8_t*>(patch_data) + offset,
+                                   patch_size - offset)) {
+    LOG(ERROR) << "Failed to init extra stream, extra_offset: " << offset
+               << ", patch_size: " << patch_size << endl;
+    return false;
+  }
+  return true;
+}
+
+bool BsdiffPatchReader::ParseControlEntry(ControlEntry* control_entry) {
+  if (!control_entry)
+    return false;
+
+  uint8_t buf[8];
+  if (!ctrl_stream_->Read(buf, 8))
+    return false;
+  int64_t diff_size = ParseInt64(buf);
+
+  if (!ctrl_stream_->Read(buf, 8))
+    return false;
+  int64_t extra_size = ParseInt64(buf);
+
+  // Sanity check.
+  if (diff_size < 0 || extra_size < 0) {
+    LOG(ERROR) << "Corrupt patch; diff_size: " << diff_size
+               << ", extra_size: " << extra_size << endl;
+    return false;
+  }
+
+  control_entry->diff_size = diff_size;
+  control_entry->extra_size = extra_size;
+
+  if (!ctrl_stream_->Read(buf, 8))
+    return false;
+  control_entry->offset_increment = ParseInt64(buf);
+
+  return true;
+}
+
+bool BsdiffPatchReader::ReadDiffStream(uint8_t* buf, size_t size) {
+  return diff_stream_->Read(buf, size);
+}
+
+bool BsdiffPatchReader::ReadExtraStream(uint8_t* buf, size_t size) {
+  return extra_stream_->Read(buf, size);
+}
+
+bool BsdiffPatchReader::Finish() {
+  if (!ctrl_stream_->Close()) {
+    LOG(ERROR) << "Failed to close the control stream";
+    return false;
+  }
+
+  if (!diff_stream_->Close()) {
+    LOG(ERROR) << "Failed to close the diff stream";
+    return false;
+  }
+
+  if (!extra_stream_->Close()) {
+    LOG(ERROR) << "Failed to close the extra stream";
+    return false;
+  }
+  return true;
+}
+
+}  // namespace bsdiff

patch_reader.h

@@ -0,0 +1,68 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef _BSDIFF_PATCH_READER_H_
+#define _BSDIFF_PATCH_READER_H_
+
+#include <stddef.h>
+#include <stdint.h>
+
+#include <memory>
+
+#include "bsdiff/control_entry.h"
+#include "bsdiff/decompressor_interface.h"
+#include "bsdiff/file_interface.h"
+
+namespace bsdiff {
+
+extern const uint8_t kMagicHeader[];
+
+// A wrapper class to read and parse the data in a bsdiff patch. The reader
+// class contains the concatenated streams of control, diff, and extra data.
+// After initialization, this class provides the function to read the metadata
+// embedded in the control stream sequentially and decompresses the diff & extra
+// stream based on this metadata.
+
+class BsdiffPatchReader {
+ public:
+  BsdiffPatchReader() = default;
+
+  // Initialize the control stream, diff stream and extra stream from the
+  // corresponding offset of |patch_data|.
+  bool Init(const uint8_t* data_start, size_t patch_size);
+
+  // Read the control stream and parse the metadata of |diff_size_|,
+  // |extra_size_| and |offset_incremental_|.
+  bool ParseControlEntry(ControlEntry* control_entry);
+
+  // Read the data in |diff_stream_| and write |size| decompressed data to
+  // |buf|.
+  bool ReadDiffStream(uint8_t* buf, size_t size);
+
+  // Read the data in |extra_stream_| and write |size| decompressed data to
+  // |buf|.
+  bool ReadExtraStream(uint8_t* buf, size_t size);
+
+  uint64_t new_file_size() const { return new_file_size_; }
+
+  // Close the control/diff/extra stream. Return false if errors occur when
+  // closing any of these streams.
+  bool Finish();
+
+ private:
+  // The compressed stream that contains the control data; i.e. length of each
+  // diff/extra block and the corresponding offset to read in the source file.
+  std::unique_ptr<DecompressorInterface> ctrl_stream_{nullptr};
+  // The compressed stream that contains the concatenated diff blocks.
+  std::unique_ptr<DecompressorInterface> diff_stream_{nullptr};
+  // The compressed stream that contains the concatenated extra blocks.
+  std::unique_ptr<DecompressorInterface> extra_stream_{nullptr};
+
+  // Size of the target file.
+  uint64_t new_file_size_;
+};
+
+}  // namespace bsdiff
+
+#endif  // _BSDIFF_PATCH_READER_H_

patch_reader_unittest.cc

@@ -0,0 +1,96 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include <unistd.h>
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+#include <gtest/gtest.h>
+
+#include "bsdiff/bz2_compressor.h"
+#include "bsdiff/patch_reader.h"
+#include "bsdiff/utils.h"
+
+namespace {
+
+void EncodeInt64(int64_t x, uint8_t* buf) {
+  uint64_t y = x < 0 ? (1ULL << 63ULL) - x : x;
+  for (int i = 0; i < 8; ++i) {
+    buf[i] = y & 0xff;
+    y /= 256;
+  }
+}
+
+}  // namespace
+
+namespace bsdiff {
+
+TEST(PatchReaderTest, BZ2PatchReaderSmoke) {
+  BZ2Compressor ctrl_stream;
+  BZ2Compressor diff_stream;
+  BZ2Compressor extra_stream;
+
+  // Write 3 blocks of data to diff stream & extra stream respectively.
+  std::vector<std::string> diff_data = {"HelloWorld", "BspatchPatchTest",
+                                        "BspatchDiffData"};
+  std::vector<std::string> extra_data = {"HelloWorld!", "BZ2PatchReaderSmoke",
+                                         "BspatchExtraData"};
+  std::vector<int64_t> offset_increment = {100, 200, 300};
+  for (size_t i = 0; i < diff_data.size(); i++) {
+    uint8_t buf[24];
+    EncodeInt64(diff_data[i].size(), buf);
+    EncodeInt64(extra_data[i].size(), buf + 8);
+    EncodeInt64(offset_increment[i], buf + 16);
+    EXPECT_TRUE(ctrl_stream.Write(buf, sizeof(buf)));
+    EXPECT_TRUE(
+        diff_stream.Write(reinterpret_cast<const uint8_t*>(diff_data[i].data()),
+                          diff_data[i].size()));
+    EXPECT_TRUE(extra_stream.Write(
+        reinterpret_cast<const uint8_t*>(extra_data[i].data()),
+        extra_data[i].size()));
+  }
+  EXPECT_TRUE(ctrl_stream.Finish());
+  EXPECT_TRUE(diff_stream.Finish());
+  EXPECT_TRUE(extra_stream.Finish());
+
+  // Encode the header
+  uint8_t buf[24];
+  EncodeInt64(ctrl_stream.GetCompressedData().size(), buf);
+  EncodeInt64(diff_stream.GetCompressedData().size(), buf + 8);
+  EncodeInt64(500, buf + 16);
+
+  std::vector<uint8_t> patch_data;
+  std::copy(kMagicHeader, kMagicHeader + 8, std::back_inserter(patch_data));
+  std::copy(buf, buf + sizeof(buf), std::back_inserter(patch_data));
+
+  // Concatenate the three streams into one patch.
+  std::copy(ctrl_stream.GetCompressedData().begin(),
+            ctrl_stream.GetCompressedData().end(),
+            std::back_inserter(patch_data));
+  std::copy(diff_stream.GetCompressedData().begin(),
+            diff_stream.GetCompressedData().end(),
+            std::back_inserter(patch_data));
+  std::copy(extra_stream.GetCompressedData().begin(),
+            extra_stream.GetCompressedData().end(),
+            std::back_inserter(patch_data));
+
+  BsdiffPatchReader patch_reader;
+  EXPECT_TRUE(patch_reader.Init(patch_data.data(), patch_data.size()));
+  EXPECT_EQ(500ull, patch_reader.new_file_size());
+  // Check that the decompressed data matches what we wrote.
+  for (size_t i = 0; i < diff_data.size(); i++) {
+    ControlEntry control_entry(0, 0, 0);
+    EXPECT_TRUE(patch_reader.ParseControlEntry(&control_entry));
+    uint8_t buffer[128] = {};
+    EXPECT_TRUE(patch_reader.ReadDiffStream(buffer, diff_data[i].size()));
+    EXPECT_EQ(0, memcmp(buffer, diff_data[i].data(), diff_data[i].size()));
+    EXPECT_TRUE(patch_reader.ReadExtraStream(buffer, extra_data[i].size()));
+    EXPECT_EQ(0, memcmp(buffer, extra_data[i].data(), extra_data[i].size()));
+  }
+  EXPECT_TRUE(patch_reader.Finish());
+}
+
+}  // namespace bsdiff

patch_writer.cc

@@ -6,6 +6,7 @@
 
 #include <string.h>
 
+#include "bsdiff/control_entry.h"
 #include "bsdiff/logging.h"
 
 using std::endl;

utils.cc

@@ -0,0 +1,20 @@
+#include <stdint.h>
+
+#include "bsdiff/utils.h"
+
+namespace bsdiff {
+
+
+int64_t ParseInt64(const uint8_t* buf) {
+  int64_t result = buf[7] & 0x7F;
+  for (int i = 6; i >= 0; i--) {
+    result <<= 8;
+    result |= buf[i];
+  }
+
+  if (buf[7] & 0x80)
+    result = -result;
+  return result;
+}
+
+}  // namespace bsdiff

utils.h

@@ -0,0 +1,16 @@
+// Copyright 2017 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef _BSDIFF_UTILS_H_
+#define _BSDIFF_UTILS_H_
+
+#include <stdint.h>
+
+namespace bsdiff {
+
+int64_t ParseInt64(const uint8_t* buf);
+
+}  // namespace bsdiff
+
+#endif