1 /**************************************************************************
3 * Copyright 2011 Zack Rusin
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to deal
8 * in the Software without restriction, including without limitation the rights
9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 * copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
24 **************************************************************************/
27 #include "trace_snappyfile.hpp"
36 using namespace Trace;
42 * Snappy at its core is just a compressoin algorithm so we're
43 * creating a new file format which uses snappy compression
44 * to hold the trace data.
46 * The file is composed of a number of chunks, they are:
48 * uint32 - specifying the length of the compressed data
51 * File can contain any number of such chunks.
52 * The default size of an uncompressed chunk is specified in
56 * Currently the default size for a a to-be-compressed data is
57 * 1mb, meaning that the compressed data will be <= 1mb.
58 * The reason it's 1mb is because it seems
59 * to offer a pretty good compression/disk io speed ratio
60 * but that might change.
64 SnappyFile::SnappyFile(const std::string &filename,
71 m_compressedCache = new char[SNAPPY_CHUNK_SIZE];
74 SnappyFile::~SnappyFile()
76 delete [] m_compressedCache;
79 bool SnappyFile::rawOpen(const std::string &filename, File::Mode mode)
81 std::ios_base::openmode fmode = std::fstream::binary;
82 if (mode == File::Write) {
83 fmode |= (std::fstream::out | std::fstream::trunc);
84 createCache(SNAPPY_CHUNK_SIZE);
85 } else if (mode == File::Read) {
86 fmode |= std::fstream::in;
89 m_stream.open(filename.c_str(), fmode);
91 //read in the initial buffer if we're reading
92 if (m_stream.is_open() && mode == File::Read) {
93 // read the snappy file identifier
94 unsigned char byte1, byte2;
97 assert(byte1 == SNAPPY_BYTE1 && byte2 == SNAPPY_BYTE2);
100 } else if (m_stream.is_open() && mode == File::Write) {
101 // write the snappy file identifier
102 m_stream << SNAPPY_BYTE1;
103 m_stream << SNAPPY_BYTE2;
105 return m_stream.is_open();
108 bool SnappyFile::rawWrite(const void *buffer, int length)
110 if (freeCacheSize() > length) {
111 memcpy(m_cachePtr, buffer, length);
112 m_cachePtr += length;
113 } else if (freeCacheSize() == length) {
114 memcpy(m_cachePtr, buffer, length);
115 m_cachePtr += length;
118 int sizeToWrite = length;
120 while (sizeToWrite >= freeCacheSize()) {
121 int endSize = freeCacheSize();
122 int offset = length - sizeToWrite;
123 memcpy(m_cachePtr, (char*)buffer + offset, endSize);
124 sizeToWrite -= endSize;
125 m_cachePtr += endSize;
129 int offset = length - sizeToWrite;
130 memcpy(m_cachePtr, (char*)buffer + offset, sizeToWrite);
131 m_cachePtr += sizeToWrite;
138 bool SnappyFile::rawRead(void *buffer, int length)
144 if (freeCacheSize() >= length) {
145 memcpy(buffer, m_cachePtr, length);
146 m_cachePtr += length;
148 int sizeToRead = length;
151 int chunkSize = std::min(freeCacheSize(), sizeToRead);
152 offset = length - sizeToRead;
153 memcpy((char*)buffer + offset, m_cachePtr, chunkSize);
154 m_cachePtr += chunkSize;
155 sizeToRead -= chunkSize;
166 int SnappyFile::rawGetc()
174 void SnappyFile::rawClose()
183 void SnappyFile::rawFlush()
189 void SnappyFile::flushCache()
191 if (m_mode == File::Write) {
192 size_t compressedLength;
194 ::snappy::RawCompress(m_cache, SNAPPY_CHUNK_SIZE - freeCacheSize(),
195 m_compressedCache, &compressedLength);
197 writeCompressedLength(compressedLength);
198 m_stream.write(m_compressedCache, compressedLength);
199 m_cachePtr = m_cache;
200 } else if (m_mode == File::Read) {
203 //assert(m_cachePtr == m_cache + m_cacheSize);
204 m_currentOffset.chunk = m_stream.tellg();
205 size_t compressedLength;
206 compressedLength = readCompressedLength();
207 m_stream.read((char*)m_compressedCache, compressedLength);
209 * The reason we peek here is because the last read will
210 * read all the way until the last character, but that will not
211 * trigger m_stream.eof() to be set, so by calling peek
212 * we assure that if we in fact have read the entire stream
213 * then the m_stream.eof() is always set.
216 ::snappy::GetUncompressedLength(m_compressedCache, compressedLength,
220 createCache(m_cacheSize);
221 ::snappy::RawUncompress(m_compressedCache, compressedLength,
226 void SnappyFile::createCache(size_t size)
228 m_cache = new char[size];
229 m_cachePtr = m_cache;
233 void SnappyFile::writeCompressedLength(uint32_t value)
235 m_stream.write((const char*)&value, sizeof value);
238 uint32_t SnappyFile::readCompressedLength()
241 m_stream.read((char*)&len, sizeof len);
245 bool SnappyFile::supportsOffsets() const
250 File::Offset SnappyFile::currentOffset()
252 m_currentOffset.offsetInChunk = m_cachePtr - m_cache;
253 return m_currentOffset;
256 void SnappyFile::setCurrentOffset(const File::Offset &offset)
260 // seek to the start of a chunk
261 m_stream.seekg(offset.chunk, std::ios::beg);
264 assert(m_cacheSize >= offset.offsetInChunk);
265 // seek within our cache to the correct location within the chunk
266 m_cachePtr = m_cache + offset.offsetInChunk;