1 // ***************************************************************************
\r
2 // BGZF.h (c) 2009 Derek Barnett, Michael Str�mberg
\r
3 // Marth Lab, Department of Biology, Boston College
\r
4 // All rights reserved.
\r
5 // ---------------------------------------------------------------------------
\r
6 // Last modified: 20 October 2010 (DB)
\r
7 // ---------------------------------------------------------------------------
\r
8 // BGZF routines were adapted from the bgzf.c code developed at the Broad
\r
10 // ---------------------------------------------------------------------------
\r
11 // Provides the basic functionality for reading & writing BGZF files
\r
12 // ***************************************************************************
\r
23 // Platform-specific large-file support
\r
24 #ifndef BAMTOOLS_LFS
\r
25 #define BAMTOOLS_LFS
\r
27 #define ftell64(a) _ftelli64(a)
\r
28 #define fseek64(a,b,c) _fseeki64(a,b,c)
\r
30 #define ftell64(a) ftello(a)
\r
31 #define fseek64(a,b,c) fseeko(a,b,c)
\r
33 #endif // BAMTOOLS_LFS
\r
35 // Platform-specific type definitions
\r
36 #ifndef BAMTOOLS_TYPES
\r
37 #define BAMTOOLS_TYPES
\r
39 typedef char int8_t;
\r
40 typedef unsigned char uint8_t;
\r
41 typedef short int16_t;
\r
42 typedef unsigned short uint16_t;
\r
43 typedef int int32_t;
\r
44 typedef unsigned int uint32_t;
\r
45 typedef long long int64_t;
\r
46 typedef unsigned long long uint64_t;
\r
50 #endif // BAMTOOLS_TYPES
\r
52 namespace BamTools {
\r
55 const int GZIP_ID1 = 31;
\r
56 const int GZIP_ID2 = 139;
\r
57 const int CM_DEFLATE = 8;
\r
58 const int FLG_FEXTRA = 4;
\r
59 const int OS_UNKNOWN = 255;
\r
60 const int BGZF_XLEN = 6;
\r
61 const int BGZF_ID1 = 66;
\r
62 const int BGZF_ID2 = 67;
\r
63 const int BGZF_LEN = 2;
\r
64 const int GZIP_WINDOW_BITS = -15;
\r
65 const int Z_DEFAULT_MEM_LEVEL = 8;
\r
68 const int BLOCK_HEADER_LENGTH = 18;
\r
69 const int BLOCK_FOOTER_LENGTH = 8;
\r
70 const int MAX_BLOCK_SIZE = 65536;
\r
71 const int DEFAULT_BLOCK_SIZE = 65536;
\r
77 unsigned int UncompressedBlockSize;
\r
78 unsigned int CompressedBlockSize;
\r
79 unsigned int BlockLength;
\r
80 unsigned int BlockOffset;
\r
81 uint64_t BlockAddress;
\r
84 bool IsWriteUncompressed;
\r
86 char* UncompressedBlock;
\r
87 char* CompressedBlock;
\r
89 // constructor & destructor
\r
94 // main interface methods
\r
98 // opens the BGZF file (mode is either "rb" for reading, or "wb" for writing)
\r
99 bool Open(const std::string& filename, const char* mode, bool isWriteUncompressed = false);
\r
100 // reads BGZF data into a byte buffer
\r
101 int Read(char* data, const unsigned int dataLength);
\r
102 // seek to position in BGZF file
\r
103 bool Seek(int64_t position);
\r
104 // get file position in BGZF file
\r
105 int64_t Tell(void);
\r
106 // writes the supplied data into the BGZF buffer
\r
107 unsigned int Write(const char* data, const unsigned int dataLen);
\r
109 // internal methods
\r
111 // compresses the current block
\r
112 int DeflateBlock(void);
\r
113 // flushes the data in the BGZF block
\r
114 void FlushBlock(void);
\r
115 // de-compresses the current block
\r
116 int InflateBlock(const int& blockLength);
\r
117 // reads a BGZF block
\r
118 bool ReadBlock(void);
\r
120 // static 'utility' methods
\r
122 // checks BGZF block header
\r
123 static inline bool CheckBlockHeader(char* header);
\r
124 // packs an unsigned integer into the specified buffer
\r
125 static inline void PackUnsignedInt(char* buffer, unsigned int value);
\r
126 // packs an unsigned short into the specified buffer
\r
127 static inline void PackUnsignedShort(char* buffer, unsigned short value);
\r
128 // unpacks a buffer into a double
\r
129 static inline double UnpackDouble(char* buffer);
\r
130 static inline double UnpackDouble(const char* buffer);
\r
131 // unpacks a buffer into a float
\r
132 static inline float UnpackFloat(char* buffer);
\r
133 static inline float UnpackFloat(const char* buffer);
\r
134 // unpacks a buffer into a signed int
\r
135 static inline signed int UnpackSignedInt(char* buffer);
\r
136 static inline signed int UnpackSignedInt(const char* buffer);
\r
137 // unpacks a buffer into a signed short
\r
138 static inline signed short UnpackSignedShort(char* buffer);
\r
139 static inline signed short UnpackSignedShort(const char* buffer);
\r
140 // unpacks a buffer into an unsigned int
\r
141 static inline unsigned int UnpackUnsignedInt(char* buffer);
\r
142 static inline unsigned int UnpackUnsignedInt(const char* buffer);
\r
143 // unpacks a buffer into an unsigned short
\r
144 static inline unsigned short UnpackUnsignedShort(char* buffer);
\r
145 static inline unsigned short UnpackUnsignedShort(const char* buffer);
\r
148 // -------------------------------------------------------------
\r
149 // static 'utility' method implementations
\r
151 // checks BGZF block header
\r
153 bool BgzfData::CheckBlockHeader(char* header) {
\r
154 return (header[0] == GZIP_ID1 &&
\r
155 header[1] == (char)GZIP_ID2 &&
\r
156 header[2] == Z_DEFLATED &&
\r
157 (header[3] & FLG_FEXTRA) != 0 &&
\r
158 BgzfData::UnpackUnsignedShort(&header[10]) == BGZF_XLEN &&
\r
159 header[12] == BGZF_ID1 &&
\r
160 header[13] == BGZF_ID2 &&
\r
161 BgzfData::UnpackUnsignedShort(&header[14]) == BGZF_LEN );
\r
164 // 'packs' an unsigned integer into the specified buffer
\r
166 void BgzfData::PackUnsignedInt(char* buffer, unsigned int value) {
\r
167 buffer[0] = (char)value;
\r
168 buffer[1] = (char)(value >> 8);
\r
169 buffer[2] = (char)(value >> 16);
\r
170 buffer[3] = (char)(value >> 24);
\r
173 // 'packs' an unsigned short into the specified buffer
\r
175 void BgzfData::PackUnsignedShort(char* buffer, unsigned short value) {
\r
176 buffer[0] = (char)value;
\r
177 buffer[1] = (char)(value >> 8);
\r
180 // 'unpacks' a buffer into a double (includes both non-const & const char* flavors)
\r
182 double BgzfData::UnpackDouble(char* buffer) {
\r
183 union { double value; unsigned char valueBuffer[sizeof(double)]; } un;
\r
185 un.valueBuffer[0] = buffer[0];
\r
186 un.valueBuffer[1] = buffer[1];
\r
187 un.valueBuffer[2] = buffer[2];
\r
188 un.valueBuffer[3] = buffer[3];
\r
189 un.valueBuffer[4] = buffer[4];
\r
190 un.valueBuffer[5] = buffer[5];
\r
191 un.valueBuffer[6] = buffer[6];
\r
192 un.valueBuffer[7] = buffer[7];
\r
197 double BgzfData::UnpackDouble(const char* buffer) {
\r
198 union { double value; unsigned char valueBuffer[sizeof(double)]; } un;
\r
200 un.valueBuffer[0] = buffer[0];
\r
201 un.valueBuffer[1] = buffer[1];
\r
202 un.valueBuffer[2] = buffer[2];
\r
203 un.valueBuffer[3] = buffer[3];
\r
204 un.valueBuffer[4] = buffer[4];
\r
205 un.valueBuffer[5] = buffer[5];
\r
206 un.valueBuffer[6] = buffer[6];
\r
207 un.valueBuffer[7] = buffer[7];
\r
211 // 'unpacks' a buffer into a float (includes both non-const & const char* flavors)
\r
213 float BgzfData::UnpackFloat(char* buffer) {
\r
214 union { float value; unsigned char valueBuffer[sizeof(float)]; } un;
\r
216 un.valueBuffer[0] = buffer[0];
\r
217 un.valueBuffer[1] = buffer[1];
\r
218 un.valueBuffer[2] = buffer[2];
\r
219 un.valueBuffer[3] = buffer[3];
\r
224 float BgzfData::UnpackFloat(const char* buffer) {
\r
225 union { float value; unsigned char valueBuffer[sizeof(float)]; } un;
\r
227 un.valueBuffer[0] = buffer[0];
\r
228 un.valueBuffer[1] = buffer[1];
\r
229 un.valueBuffer[2] = buffer[2];
\r
230 un.valueBuffer[3] = buffer[3];
\r
234 // 'unpacks' a buffer into a signed int (includes both non-const & const char* flavors)
\r
236 signed int BgzfData::UnpackSignedInt(char* buffer) {
\r
237 union { signed int value; unsigned char valueBuffer[sizeof(signed int)]; } un;
\r
239 un.valueBuffer[0] = buffer[0];
\r
240 un.valueBuffer[1] = buffer[1];
\r
241 un.valueBuffer[2] = buffer[2];
\r
242 un.valueBuffer[3] = buffer[3];
\r
247 signed int BgzfData::UnpackSignedInt(const char* buffer) {
\r
248 union { signed int value; unsigned char valueBuffer[sizeof(signed int)]; } un;
\r
250 un.valueBuffer[0] = buffer[0];
\r
251 un.valueBuffer[1] = buffer[1];
\r
252 un.valueBuffer[2] = buffer[2];
\r
253 un.valueBuffer[3] = buffer[3];
\r
257 // 'unpacks' a buffer into a signed short (includes both non-const & const char* flavors)
\r
259 signed short BgzfData::UnpackSignedShort(char* buffer) {
\r
260 union { signed short value; unsigned char valueBuffer[sizeof(signed short)]; } un;
\r
262 un.valueBuffer[0] = buffer[0];
\r
263 un.valueBuffer[1] = buffer[1];
\r
268 signed short BgzfData::UnpackSignedShort(const char* buffer) {
\r
269 union { signed short value; unsigned char valueBuffer[sizeof(signed short)]; } un;
\r
271 un.valueBuffer[0] = buffer[0];
\r
272 un.valueBuffer[1] = buffer[1];
\r
276 // 'unpacks' a buffer into an unsigned int (includes both non-const & const char* flavors)
\r
278 unsigned int BgzfData::UnpackUnsignedInt(char* buffer) {
\r
279 union { unsigned int value; unsigned char valueBuffer[sizeof(unsigned int)]; } un;
\r
281 un.valueBuffer[0] = buffer[0];
\r
282 un.valueBuffer[1] = buffer[1];
\r
283 un.valueBuffer[2] = buffer[2];
\r
284 un.valueBuffer[3] = buffer[3];
\r
289 unsigned int BgzfData::UnpackUnsignedInt(const char* buffer) {
\r
290 union { unsigned int value; unsigned char valueBuffer[sizeof(unsigned int)]; } un;
\r
292 un.valueBuffer[0] = buffer[0];
\r
293 un.valueBuffer[1] = buffer[1];
\r
294 un.valueBuffer[2] = buffer[2];
\r
295 un.valueBuffer[3] = buffer[3];
\r
299 // 'unpacks' a buffer into an unsigned short (includes both non-const & const char* flavors)
\r
301 unsigned short BgzfData::UnpackUnsignedShort(char* buffer) {
\r
302 union { unsigned short value; unsigned char valueBuffer[sizeof(unsigned short)]; } un;
\r
304 un.valueBuffer[0] = buffer[0];
\r
305 un.valueBuffer[1] = buffer[1];
\r
310 unsigned short BgzfData::UnpackUnsignedShort(const char* buffer) {
\r
311 union { unsigned short value; unsigned char valueBuffer[sizeof(unsigned short)]; } un;
\r
313 un.valueBuffer[0] = buffer[0];
\r
314 un.valueBuffer[1] = buffer[1];
\r
318 } // namespace BamTools
\r