1fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// Protocol Buffers - Google's data interchange format
2fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// Copyright 2008 Google Inc.  All rights reserved.
3afb4b72037e3f13db208590fc782c4bc8e27f862Jeff Davidson// https://developers.google.com/protocol-buffers/
4fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
5fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// Redistribution and use in source and binary forms, with or without
6fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// modification, are permitted provided that the following conditions are
7fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// met:
8fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
9fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//     * Redistributions of source code must retain the above copyright
10fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// notice, this list of conditions and the following disclaimer.
11fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//     * Redistributions in binary form must reproduce the above
12fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// copyright notice, this list of conditions and the following disclaimer
13fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// in the documentation and/or other materials provided with the
14fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// distribution.
15fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//     * Neither the name of Google Inc. nor the names of its
16fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// contributors may be used to endorse or promote products derived from
17fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// this software without specific prior written permission.
18fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
19fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
31fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// Author: brianolson@google.com (Brian Olson)
32fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
33fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// This file contains the definition for classes GzipInputStream and
34fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// GzipOutputStream.
35fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
36fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// GzipInputStream decompresses data from an underlying
37fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// ZeroCopyInputStream and provides the decompressed data as a
38fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// ZeroCopyInputStream.
39fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville//
40fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// GzipOutputStream is an ZeroCopyOutputStream that compresses data to
41fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// an underlying ZeroCopyOutputStream.
42fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
43fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville#ifndef GOOGLE_PROTOBUF_IO_GZIP_STREAM_H__
44fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville#define GOOGLE_PROTOBUF_IO_GZIP_STREAM_H__
45fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
46a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson#include <google/protobuf/stubs/common.h>
47fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville#include <google/protobuf/io/zero_copy_stream.h>
48b0575e93e4c39dec69365b850088a1eb7f82c5b3Tamas Berghammer#include <zlib.h>
49fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
50fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Savillenamespace google {
51fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Savillenamespace protobuf {
52fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Savillenamespace io {
53fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
54fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville// A ZeroCopyInputStream that reads compressed data through zlib
55fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Savilleclass LIBPROTOBUF_EXPORT GzipInputStream : public ZeroCopyInputStream {
56fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville public:
57fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Format key for constructor
58fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  enum Format {
59fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // zlib will autodetect gzip header or deflate stream
60fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    AUTO = 0,
61fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
62fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // GZIP streams have some extra header data for file attributes.
63fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    GZIP = 1,
64fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
65fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // Simpler zlib stream format.
66fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    ZLIB = 2,
67fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  };
68fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
69fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // buffer_size and format may be -1 for default of 64kB and GZIP format
70fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  explicit GzipInputStream(
71fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville      ZeroCopyInputStream* sub_stream,
72fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville      Format format = AUTO,
73fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville      int buffer_size = -1);
74fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  virtual ~GzipInputStream();
75fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
76fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Return last error message or NULL if no error.
77fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  inline const char* ZlibErrorMessage() const {
78fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    return zcontext_.msg;
79fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  }
80fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  inline int ZlibErrorCode() const {
81fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    return zerror_;
82fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  }
83fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
84fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // implements ZeroCopyInputStream ----------------------------------
85fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  bool Next(const void** data, int* size);
86fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void BackUp(int count);
87fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  bool Skip(int count);
88fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int64 ByteCount() const;
89fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
90fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville private:
91fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  Format format_;
92fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
93fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  ZeroCopyInputStream* sub_stream_;
94fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
95fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  z_stream zcontext_;
96fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int zerror_;
97fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
98fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void* output_buffer_;
99fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void* output_position_;
100fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  size_t output_buffer_length_;
101b0575e93e4c39dec69365b850088a1eb7f82c5b3Tamas Berghammer  int64 byte_count_;
102fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
103fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int Inflate(int flush);
104fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void DoNextOutput(const void** data, int* size);
105fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
106fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(GzipInputStream);
107fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville};
108fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
109fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
110fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Savilleclass LIBPROTOBUF_EXPORT GzipOutputStream : public ZeroCopyOutputStream {
111fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville public:
112fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Format key for constructor
113fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  enum Format {
114fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // GZIP streams have some extra header data for file attributes.
115fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    GZIP = 1,
116fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
117fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // Simpler zlib stream format.
118fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    ZLIB = 2,
119fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  };
120fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
121fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  struct Options {
122fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // Defaults to GZIP.
123fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    Format format;
124fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
125fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // What size buffer to use internally.  Defaults to 64kB.
126fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    int buffer_size;
127fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
128fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // A number between 0 and 9, where 0 is no compression and 9 is best
129fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // compression.  Defaults to Z_DEFAULT_COMPRESSION (see zlib.h).
130fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    int compression_level;
131fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
132fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // Defaults to Z_DEFAULT_STRATEGY.  Can also be set to Z_FILTERED,
133fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // Z_HUFFMAN_ONLY, or Z_RLE.  See the documentation for deflateInit2 in
134fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    // zlib.h for definitions of these constants.
135fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    int compression_strategy;
136fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
137fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    Options();  // Initializes with default values.
138fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  };
139fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
140fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Create a GzipOutputStream with default options.
141fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  explicit GzipOutputStream(ZeroCopyOutputStream* sub_stream);
142fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
143fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Create a GzipOutputStream with the given options.
144fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  GzipOutputStream(
145fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville      ZeroCopyOutputStream* sub_stream,
146fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville      const Options& options);
147fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
148fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  virtual ~GzipOutputStream();
149fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
150fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Return last error message or NULL if no error.
151fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  inline const char* ZlibErrorMessage() const {
152fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    return zcontext_.msg;
153fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  }
154fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  inline int ZlibErrorCode() const {
155fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville    return zerror_;
156fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  }
157fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
158fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Flushes data written so far to zipped data in the underlying stream.
159fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // It is the caller's responsibility to flush the underlying stream if
160fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // necessary.
161fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Compression may be less efficient stopping and starting around flushes.
162fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Returns true if no error.
163a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  //
164a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  // Please ensure that block size is > 6. Here is an excerpt from the zlib
165a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  // doc that explains why:
166a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  //
167a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  // In the case of a Z_FULL_FLUSH or Z_SYNC_FLUSH, make sure that avail_out
168a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  // is greater than six to avoid repeated flush markers due to
169a3b2a6da25a76f17c73d31def3952feb0fd2296eJeff Davidson  // avail_out == 0 on return.
170fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  bool Flush();
171fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
172fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Writes out all data and closes the gzip stream.
173fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // It is the caller's responsibility to close the underlying stream if
174fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // necessary.
175fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Returns true if no error.
176fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  bool Close();
177fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
178fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // implements ZeroCopyOutputStream ---------------------------------
179fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  bool Next(void** data, int* size);
180fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void BackUp(int count);
181fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int64 ByteCount() const;
182fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
183fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville private:
184fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  ZeroCopyOutputStream* sub_stream_;
185fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Result from calling Next() on sub_stream_
186fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void* sub_data_;
187fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int sub_data_size_;
188fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
189fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  z_stream zcontext_;
190fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int zerror_;
191fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void* input_buffer_;
192fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  size_t input_buffer_length_;
193fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
194fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Shared constructor code.
195fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  void Init(ZeroCopyOutputStream* sub_stream, const Options& options);
196fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
197fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Do some compression.
198fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Takes zlib flush mode.
199fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  // Returns zlib error code.
200fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  int Deflate(int flush);
201fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
202fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville  GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(GzipOutputStream);
203fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville};
204fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
205fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville}  // namespace io
206fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville}  // namespace protobuf
207fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville
208fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville}  // namespace google
209fbaaef999ba563838ebd00874ed8a1c01fbf286dWink Saville#endif  // GOOGLE_PROTOBUF_IO_GZIP_STREAM_H__
210