【神经网络与深度学习】Google Snappy - 一个高速压缩库

Snappy已经被Google开源，作为一个压缩库，它可以利用单颗Intel Corei7处理器内核处理至少每秒250MB~500MB的数据流。

Snappy的前身是Zippy。虽然只是一个数据压缩库，它却被Google用于许多内部项目程，其中就包括BigTable，MapReduce和RPC。Google宣称它在这个库本身及其算法做了数据处理速度上的优化，作为代价，并没有考虑输出大小以及和其他类似工具的兼容性问题。Snappy特地为64位x86处理器做了优化，在单个Intel
Core i7处理器内核上能够达到至少每秒250MB的压缩速率和每秒500MB的解压速率。

如果允许损失一些压缩率的话，那么可以达到更高的压缩速度，虽然生成的压缩文件可能会比其他库的要大上20%至100%，但是，相比其他的压缩库，Snappy却能够在特定的压缩率下拥有惊人的压缩速度，“压缩普通文本文件的速度是其他库的1.5-1.7倍，HTML能达到2-4倍，但是对于JPEG、PNG以及其他的已压缩的数据，压缩速度不会有明显改善”。

Google极力赞扬Snappy的各种优点，Snappy从一开始就被“设计为即便遇到损坏或者恶意的输入文件都不会崩溃”，而且被Google在生产环境中用于压缩PB级的数据。其健壮性和稳定程度可见一斑。

Snappy也可以用于和其他压缩库-zlib、LZO、LZF、FastLZ和QuickLZ-做对比测试，前提是你在机器上安装了这些压缩库。Snappy是一个C++的库，你可以在产品中使用，不过也有一些其他语言的版本，例如Haskell、Java、Perl、Python和Ruby。

Snappy采用新BSD协议开源。

近日需要在毕业设计中引入一个压缩库，要求压缩与解压缩速度快，但是压缩率可以不那么苛刻。查找资料发现Google的snappy库比较合适，而且该库开源，由C++写成。所以就拿来使用一下，下面权作记录。下面引出的任何涉及Google公司的源代码，版权归Google公司所有，我权作学习交流。文章安排如下，首先简要介绍Snappy，之后安装之，然后以实际例子介绍如何使用，接着bzip2和gzip做了性能比较，最后提出一些使用上面的疑问。

（一）简要介绍

去官网下载之http://code.google.com/p/snappy/。在Project Home处有这么一段英文，我想许多地方都引用和翻译了这段。我也尝试翻译一下。

Snappy is a compression/decompression library.

It does not aim for maximum compression,

or compatibility with any other compression library;

instead, it aims for veryhigh speeds and reasonable compression.

For instance, compared to the fastest mode of zlib,

Snappy is an order of magnitude faster for most inputs,

but the resulting compressed files are anywhere from 20% to 100% bigger.

On a single core of a Core i7 processor in 64-bit mode,

Snappy compresses at about 250 MB/sec or more and

decompresses at about 500 MB/sec or more.

Snappy is widely used inside Google, in everything from BigTable

and MapReduce to our internal RPC systems.

译文：Snappy是一个压缩/解压缩库。它不是以最大压缩率，或者与其他压缩库兼容为目标；它旨在获得高速的压缩和合理的压缩率。例如，Snappy对大多数的输入比zlib的最快模式要快几个数量级，但是其压缩过后的文件通常会比zlib大20%到100%。在Core i7的单核64位模式下，Snappy压缩速度大概可以达到250MB/s或者更快，解压缩可以达到大约500MB/s或更快。

Snappy在Google内部广泛使用，从BigTable，MapReduce到公司内部的RPC系统。

（二）安装过程

下面描述安装过程：

下载snappy-1.0.5.tar.gz，snappy的安装过程与传统的安装过程一样。解压后的INSTALL文件有详细的安装说明。

gunzip snappy-1.0.5.tar.gz

tar xf snappy-1.0.5.tar

cd snappy-1.0.5

./configure

make

make install

安装完成后，生成的动态库和静态库位于/usr/local/lib处，编程需要用到的头文件位于/usr/local/include处。注意需要将这些库文件cp至/usr/lib处，不然就算在链接的时候加上-L/usr/local/lib，在运行时也会报错。./main: error while loading shared libraries: libsnappy.so.1:

cannot open shared object file: No such file or directory

当然这是我的LD_LIBRARY_PATH环境变量的设置问题。

（三）使用snappy

解压出来的README文件介绍了一简单的使用方式。snappy是各种库标示符所在的命名空间。C++使用需要包含#include <snappy.h>头文件，C语言使用需要包含#include<snapyy-c.h>头文件。Snappy使用较为简单，我指的是跟bzip2的库比起来。所有的函数接口都暴露在上述两个头文件中，头文件中有详细的使用说明，并有简单的示例，而且英文通俗易懂。摘抄如下（Google公司版权所有）：

snappy.h

// Copyright 2005 and onwards Google Inc.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
//     * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// A light-weight compression algorithm.  It is designed for speed of
// compression and decompression, rather than for the utmost in space
// savings.
//
// For getting better compression ratios when you are compressing data
// with long repeated sequences or compressing data that is similar to
// other data, while still compressing fast, you might look at first
// using BMDiff and then compressing the output of BMDiff with
// Snappy.  

#ifndef UTIL_SNAPPY_SNAPPY_H__
#define UTIL_SNAPPY_SNAPPY_H__  

#include <stddef.h>
#include <string>  

#include "snappy-stubs-public.h"  

namespace snappy {
  class Source;
  class Sink;  

  // ------------------------------------------------------------------------
  // Generic compression/decompression routines.
  // ------------------------------------------------------------------------  

  // Compress the bytes read from "*source" and append to "*sink". Return the
  // number of bytes written.
  size_t Compress(Source* source, Sink* sink);  

  bool GetUncompressedLength(Source* source, uint32* result);  

  // ------------------------------------------------------------------------
  // Higher-level string based routines (should be sufficient for most users)
  // ------------------------------------------------------------------------  

  // Sets "*output" to the compressed version of "input[0,input_length-1]".
  // Original contents of *output are lost.
  //
  // REQUIRES: "input[]" is not an alias of "*output".
  size_t Compress(const char* input, size_t input_length, string* output);  

  // Decompresses "compressed[0,compressed_length-1]" to "*uncompressed".
  // Original contents of "*uncompressed" are lost.
  //
  // REQUIRES: "compressed[]" is not an alias of "*uncompressed".
  //
  // returns false if the message is corrupted and could not be decompressed
  bool Uncompress(const char* compressed, size_t compressed_length,
                  string* uncompressed);  

  // ------------------------------------------------------------------------
  // Lower-level character array based routines.  May be useful for
  // efficiency reasons in certain circumstances.
  // ------------------------------------------------------------------------  

  // REQUIRES: "compressed" must point to an area of memory that is at
  // least "MaxCompressedLength(input_length)" bytes in length.
  //
  // Takes the data stored in "input[0..input_length]" and stores
  // it in the array pointed to by "compressed".
  //
  // "*compressed_length" is set to the length of the compressed output.
  //
  // Example:
  //    char* output = new char[snappy::MaxCompressedLength(input_length)];
  //    size_t output_length;
  //    RawCompress(input, input_length, output, &output_length);
  //    ... Process(output, output_length) ...
  //    delete [] output;
  void RawCompress(const char* input,
                   size_t input_length,
                   char* compressed,
                   size_t* compressed_length);  

  // Given data in "compressed[0..compressed_length-1]" generated by
  // calling the Snappy::Compress routine, this routine
  // stores the uncompressed data to
  //    uncompressed[0..GetUncompressedLength(compressed)-1]
  // returns false if the message is corrupted and could not be decrypted
  bool RawUncompress(const char* compressed, size_t compressed_length,
                     char* uncompressed);  

  // Given data from the byte source 'compressed' generated by calling
  // the Snappy::Compress routine, this routine stores the uncompressed
  // data to
  //    uncompressed[0..GetUncompressedLength(compressed,compressed_length)-1]
  // returns false if the message is corrupted and could not be decrypted
  bool RawUncompress(Source* compressed, char* uncompressed);  

  // Returns the maximal size of the compressed representation of
  // input data that is "source_bytes" bytes in length;
  size_t MaxCompressedLength(size_t source_bytes);  

  // REQUIRES: "compressed[]" was produced by RawCompress() or Compress()
  // Returns true and stores the length of the uncompressed data in
  // *result normally.  Returns false on parsing error.
  // This operation takes O(1) time.
  bool GetUncompressedLength(const char* compressed, size_t compressed_length,
                             size_t* result);  

  // Returns true iff the contents of "compressed[]" can be uncompressed
  // successfully.  Does not return the uncompressed data.  Takes
  // time proportional to compressed_length, but is usually at least
  // a factor of four faster than actual decompression.
  bool IsValidCompressedBuffer(const char* compressed,
                               size_t compressed_length);  

  // *** DO NOT CHANGE THE VALUE OF kBlockSize ***
  //
  // New Compression code chops up the input into blocks of at most
  // the following size.  This ensures that back-references in the
  // output never cross kBlockSize block boundaries.  This can be
  // helpful in implementing blocked decompression.  However the
  // decompression code should not rely on this guarantee since older
  // compression code may not obey it.
  static const int kBlockLog = 15;
  static const size_t kBlockSize = 1 << kBlockLog;  

  static const int kMaxHashTableBits = 14;
  static const size_t kMaxHashTableSize = 1 << kMaxHashTableBits;  

}  // end namespace snappy  

#endif  // UTIL_SNAPPY_SNAPPY_H__
[cpp] view plain copy 在CODE上查看代码片派生到我的代码片
// Copyright 2005 and onwards Google Inc.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
//     * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// A light-weight compression algorithm.  It is designed for speed of
// compression and decompression, rather than for the utmost in space
// savings.
//
// For getting better compression ratios when you are compressing data
// with long repeated sequences or compressing data that is similar to
// other data, while still compressing fast, you might look at first
// using BMDiff and then compressing the output of BMDiff with
// Snappy.  

#ifndef UTIL_SNAPPY_SNAPPY_H__
#define UTIL_SNAPPY_SNAPPY_H__  

#include <stddef.h>
#include <string>  

#include "snappy-stubs-public.h"  

namespace snappy {
  class Source;
  class Sink;  

  // ------------------------------------------------------------------------
  // Generic compression/decompression routines.
  // ------------------------------------------------------------------------  

  // Compress the bytes read from "*source" and append to "*sink". Return the
  // number of bytes written.
  size_t Compress(Source* source, Sink* sink);  

  bool GetUncompressedLength(Source* source, uint32* result);  

  // ------------------------------------------------------------------------
  // Higher-level string based routines (should be sufficient for most users)
  // ------------------------------------------------------------------------  

  // Sets "*output" to the compressed version of "input[0,input_length-1]".
  // Original contents of *output are lost.
  //
  // REQUIRES: "input[]" is not an alias of "*output".
  size_t Compress(const char* input, size_t input_length, string* output);  

  // Decompresses "compressed[0,compressed_length-1]" to "*uncompressed".
  // Original contents of "*uncompressed" are lost.
  //
  // REQUIRES: "compressed[]" is not an alias of "*uncompressed".
  //
  // returns false if the message is corrupted and could not be decompressed
  bool Uncompress(const char* compressed, size_t compressed_length,
                  string* uncompressed);  

  // ------------------------------------------------------------------------
  // Lower-level character array based routines.  May be useful for
  // efficiency reasons in certain circumstances.
  // ------------------------------------------------------------------------  

  // REQUIRES: "compressed" must point to an area of memory that is at
  // least "MaxCompressedLength(input_length)" bytes in length.
  //
  // Takes the data stored in "input[0..input_length]" and stores
  // it in the array pointed to by "compressed".
  //
  // "*compressed_length" is set to the length of the compressed output.
  //
  // Example:
  //    char* output = new char[snappy::MaxCompressedLength(input_length)];
  //    size_t output_length;
  //    RawCompress(input, input_length, output, &output_length);
  //    ... Process(output, output_length) ...
  //    delete [] output;
  void RawCompress(const char* input,
                   size_t input_length,
                   char* compressed,
                   size_t* compressed_length);  

  // Given data in "compressed[0..compressed_length-1]" generated by
  // calling the Snappy::Compress routine, this routine
  // stores the uncompressed data to
  //    uncompressed[0..GetUncompressedLength(compressed)-1]
  // returns false if the message is corrupted and could not be decrypted
  bool RawUncompress(const char* compressed, size_t compressed_length,
                     char* uncompressed);  

  // Given data from the byte source 'compressed' generated by calling
  // the Snappy::Compress routine, this routine stores the uncompressed
  // data to
  //    uncompressed[0..GetUncompressedLength(compressed,compressed_length)-1]
  // returns false if the message is corrupted and could not be decrypted
  bool RawUncompress(Source* compressed, char* uncompressed);  

  // Returns the maximal size of the compressed representation of
  // input data that is "source_bytes" bytes in length;
  size_t MaxCompressedLength(size_t source_bytes);  

  // REQUIRES: "compressed[]" was produced by RawCompress() or Compress()
  // Returns true and stores the length of the uncompressed data in
  // *result normally.  Returns false on parsing error.
  // This operation takes O(1) time.
  bool GetUncompressedLength(const char* compressed, size_t compressed_length,
                             size_t* result);  

  // Returns true iff the contents of "compressed[]" can be uncompressed
  // successfully.  Does not return the uncompressed data.  Takes
  // time proportional to compressed_length, but is usually at least
  // a factor of four faster than actual decompression.
  bool IsValidCompressedBuffer(const char* compressed,
                               size_t compressed_length);  

  // *** DO NOT CHANGE THE VALUE OF kBlockSize ***
  //
  // New Compression code chops up the input into blocks of at most
  // the following size.  This ensures that back-references in the
  // output never cross kBlockSize block boundaries.  This can be
  // helpful in implementing blocked decompression.  However the
  // decompression code should not rely on this guarantee since older
  // compression code may not obey it.
  static const int kBlockLog = 15;
  static const size_t kBlockSize = 1 << kBlockLog;  

  static const int kMaxHashTableBits = 14;
  static const size_t kMaxHashTableSize = 1 << kMaxHashTableBits;  

}  // end namespace snappy  

#endif  // UTIL_SNAPPY_SNAPPY_H__  

snapp-c.h

[cpp] view plaincopy
/*
 * Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are
 * met:
 *
 *     * Redistributions of source code must retain the above copyright
 * notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above
 * copyright notice, this list of conditions and the following disclaimer
 * in the documentation and/or other materials provided with the
 * distribution.
 *     * Neither the name of Google Inc. nor the names of its
 * contributors may be used to endorse or promote products derived from
 * this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 * Plain C interface (a wrapper around the C++ implementation).
 */  

#ifndef UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_
#define UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_  

#ifdef __cplusplus
extern "C" {
#endif  

#include <stddef.h>  

/*
 * Return values; see the documentation for each function to know
 * what each can return.
 */
typedef enum {
  SNAPPY_OK = 0,
  SNAPPY_INVALID_INPUT = 1,
  SNAPPY_BUFFER_TOO_SMALL = 2,
} snappy_status;  

/*
 * Takes the data stored in "input[0..input_length-1]" and stores
 * it in the array pointed to by "compressed".
 *
 * <compressed_length> signals the space available in "compressed".
 * If it is not at least equal to "snappy_max_compressed_length(input_length)",
 * SNAPPY_BUFFER_TOO_SMALL is returned. After successful compression,
 * <compressed_length> contains the true length of the compressed output,
 * and SNAPPY_OK is returned.
 *
 * Example:
 *   size_t output_length = snappy_max_compressed_length(input_length);
 *   char* output = (char*)malloc(output_length);
 *   if (snappy_compress(input, input_length, output, &output_length)
 *       == SNAPPY_OK) {
 *     ... Process(output, output_length) ...
 *   }
 *   free(output);
 */
snappy_status snappy_compress(const char* input,
                              size_t input_length,
                              char* compressed,
                              size_t* compressed_length);  

/*
 * Given data in "compressed[0..compressed_length-1]" generated by
 * calling the snappy_compress routine, this routine stores
 * the uncompressed data to
 *   uncompressed[0..uncompressed_length-1].
 * Returns failure (a value not equal to SNAPPY_OK) if the message
 * is corrupted and could not be decrypted.
 *
 * <uncompressed_length> signals the space available in "uncompressed".
 * If it is not at least equal to the value returned by
 * snappy_uncompressed_length for this stream, SNAPPY_BUFFER_TOO_SMALL
 * is returned. After successful decompression, <uncompressed_length>
 * contains the true length of the decompressed output.
 *
 * Example:
 *   size_t output_length;
 *   if (snappy_uncompressed_length(input, input_length, &output_length)
 *       != SNAPPY_OK) {
 *     ... fail ...
 *   }
 *   char* output = (char*)malloc(output_length);
 *   if (snappy_uncompress(input, input_length, output, &output_length)
 *       == SNAPPY_OK) {
 *     ... Process(output, output_length) ...
 *   }
 *   free(output);
 */
snappy_status snappy_uncompress(const char* compressed,
                                size_t compressed_length,
                                char* uncompressed,
                                size_t* uncompressed_length);  

/*
 * Returns the maximal size of the compressed representation of
 * input data that is "source_length" bytes in length.
 */
size_t snappy_max_compressed_length(size_t source_length);  

/*
 * REQUIRES: "compressed[]" was produced by snappy_compress()
 * Returns SNAPPY_OK and stores the length of the uncompressed data in
 * *result normally. Returns SNAPPY_INVALID_INPUT on parsing error.
 * This operation takes O(1) time.
 */
snappy_status snappy_uncompressed_length(const char* compressed,
                                         size_t compressed_length,
                                         size_t* result);  

/*
 * Check if the contents of "compressed[]" can be uncompressed successfully.
 * Does not return the uncompressed data; if so, returns SNAPPY_OK,
 * or if not, returns SNAPPY_INVALID_INPUT.
 * Takes time proportional to compressed_length, but is usually at least a
 * factor of four faster than actual decompression.
 */
snappy_status snappy_validate_compressed_buffer(const char* compressed,
                                                size_t compressed_length);  

#ifdef __cplusplus
}  // extern "C"
#endif  

#endif  /* UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_ */
[cpp] view plain copy 在CODE上查看代码片派生到我的代码片
/*
 * Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are
 * met:
 *
 *     * Redistributions of source code must retain the above copyright
 * notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above
 * copyright notice, this list of conditions and the following disclaimer
 * in the documentation and/or other materials provided with the
 * distribution.
 *     * Neither the name of Google Inc. nor the names of its
 * contributors may be used to endorse or promote products derived from
 * this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 * Plain C interface (a wrapper around the C++ implementation).
 */  

#ifndef UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_
#define UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_  

#ifdef __cplusplus
extern "C" {
#endif  

#include <stddef.h>  

/*
 * Return values; see the documentation for each function to know
 * what each can return.
 */
typedef enum {
  SNAPPY_OK = 0,
  SNAPPY_INVALID_INPUT = 1,
  SNAPPY_BUFFER_TOO_SMALL = 2,
} snappy_status;  

/*
 * Takes the data stored in "input[0..input_length-1]" and stores
 * it in the array pointed to by "compressed".
 *
 * <compressed_length> signals the space available in "compressed".
 * If it is not at least equal to "snappy_max_compressed_length(input_length)",
 * SNAPPY_BUFFER_TOO_SMALL is returned. After successful compression,
 * <compressed_length> contains the true length of the compressed output,
 * and SNAPPY_OK is returned.
 *
 * Example:
 *   size_t output_length = snappy_max_compressed_length(input_length);
 *   char* output = (char*)malloc(output_length);
 *   if (snappy_compress(input, input_length, output, &output_length)
 *       == SNAPPY_OK) {
 *     ... Process(output, output_length) ...
 *   }
 *   free(output);
 */
snappy_status snappy_compress(const char* input,
                              size_t input_length,
                              char* compressed,
                              size_t* compressed_length);  

/*
 * Given data in "compressed[0..compressed_length-1]" generated by
 * calling the snappy_compress routine, this routine stores
 * the uncompressed data to
 *   uncompressed[0..uncompressed_length-1].
 * Returns failure (a value not equal to SNAPPY_OK) if the message
 * is corrupted and could not be decrypted.
 *
 * <uncompressed_length> signals the space available in "uncompressed".
 * If it is not at least equal to the value returned by
 * snappy_uncompressed_length for this stream, SNAPPY_BUFFER_TOO_SMALL
 * is returned. After successful decompression, <uncompressed_length>
 * contains the true length of the decompressed output.
 *
 * Example:
 *   size_t output_length;
 *   if (snappy_uncompressed_length(input, input_length, &output_length)
 *       != SNAPPY_OK) {
 *     ... fail ...
 *   }
 *   char* output = (char*)malloc(output_length);
 *   if (snappy_uncompress(input, input_length, output, &output_length)
 *       == SNAPPY_OK) {
 *     ... Process(output, output_length) ...
 *   }
 *   free(output);
 */
snappy_status snappy_uncompress(const char* compressed,
                                size_t compressed_length,
                                char* uncompressed,
                                size_t* uncompressed_length);  

/*
 * Returns the maximal size of the compressed representation of
 * input data that is "source_length" bytes in length.
 */
size_t snappy_max_compressed_length(size_t source_length);  

/*
 * REQUIRES: "compressed[]" was produced by snappy_compress()
 * Returns SNAPPY_OK and stores the length of the uncompressed data in
 * *result normally. Returns SNAPPY_INVALID_INPUT on parsing error.
 * This operation takes O(1) time.
 */
snappy_status snappy_uncompressed_length(const char* compressed,
                                         size_t compressed_length,
                                         size_t* result);  

/*
 * Check if the contents of "compressed[]" can be uncompressed successfully.
 * Does not return the uncompressed data; if so, returns SNAPPY_OK,
 * or if not, returns SNAPPY_INVALID_INPUT.
 * Takes time proportional to compressed_length, but is usually at least a
 * factor of four faster than actual decompression.
 */
snappy_status snappy_validate_compressed_buffer(const char* compressed,
                                                size_t compressed_length);  

#ifdef __cplusplus
}  // extern "C"
#endif  

#endif  /* UTIL_SNAPPY_OPENSOURCE_SNAPPY_C_H_ */