1 // ***************************************************************************
\r
2 // BGZF.h (c) 2009 Derek Barnett, Michael Strömberg
\r
3 // Marth Lab, Department of Biology, Boston College
\r
4 // All rights reserved.
\r
5 // ---------------------------------------------------------------------------
\r
6 // Last modified: 11 January 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
28 // Platform-specific type definitions
\r
29 #ifndef BAMTOOLS_TYPES
\r
30 #define BAMTOOLS_TYPES
\r
32 typedef char int8_t;
\r
33 typedef unsigned char uint8_t;
\r
34 typedef short int16_t;
\r
35 typedef unsigned short uint16_t;
\r
36 typedef int int32_t;
\r
37 typedef unsigned int uint32_t;
\r
38 typedef long long int64_t;
\r
39 typedef unsigned long long uint64_t;
\r
43 #endif // BAMTOOLS_TYPES
\r
45 namespace BamTools {
\r
48 const int GZIP_ID1 = 31;
\r
49 const int GZIP_ID2 = 139;
\r
50 const int CM_DEFLATE = 8;
\r
51 const int FLG_FEXTRA = 4;
\r
52 const int OS_UNKNOWN = 255;
\r
53 const int BGZF_XLEN = 6;
\r
54 const int BGZF_ID1 = 66;
\r
55 const int BGZF_ID2 = 67;
\r
56 const int BGZF_LEN = 2;
\r
57 const int GZIP_WINDOW_BITS = -15;
\r
58 const int Z_DEFAULT_MEM_LEVEL = 8;
\r
61 const int BLOCK_HEADER_LENGTH = 18;
\r
62 const int BLOCK_FOOTER_LENGTH = 8;
\r
63 const int MAX_BLOCK_SIZE = 65536;
\r
64 const int DEFAULT_BLOCK_SIZE = 65536;
\r
69 unsigned int UncompressedBlockSize;
\r
70 unsigned int CompressedBlockSize;
\r
71 unsigned int BlockLength;
\r
72 unsigned int BlockOffset;
\r
73 uint64_t BlockAddress;
\r
77 char* UncompressedBlock;
\r
78 char* CompressedBlock;
\r
80 // constructor & destructor
\r
86 // compresses the current block
\r
87 int DeflateBlock(void);
\r
88 // flushes the data in the BGZF block
\r
89 void FlushBlock(void);
\r
90 // de-compresses the current block
\r
91 int InflateBlock(const int& blockLength);
\r
92 // opens the BGZF file for reading (mode is either "rb" for reading, or "wb" for writing
\r
93 void Open(const std::string& filename, const char* mode);
\r
94 // reads BGZF data into a byte buffer
\r
95 int Read(char* data, const unsigned int dataLength);
\r
97 int ReadBlock(void);
\r
98 // seek to position in BAM file
\r
99 bool Seek(int64_t position);
\r
100 // get file position in BAM file
\r
101 int64_t Tell(void);
\r
102 // writes the supplied data into the BGZF buffer
\r
103 unsigned int Write(const char* data, const unsigned int dataLen);
\r
105 // checks BGZF block header
\r
106 static inline bool CheckBlockHeader(char* header);
\r
107 // packs an unsigned integer into the specified buffer
\r
108 static inline void PackUnsignedInt(char* buffer, unsigned int value);
\r
109 // packs an unsigned short into the specified buffer
\r
110 static inline void PackUnsignedShort(char* buffer, unsigned short value);
\r
111 // unpacks a buffer into a signed int
\r
112 static inline signed int UnpackSignedInt(char* buffer);
\r
113 // unpacks a buffer into a unsigned int
\r
114 static inline unsigned int UnpackUnsignedInt(char* buffer);
\r
115 // unpacks a buffer into a unsigned short
\r
116 static inline unsigned short UnpackUnsignedShort(char* buffer);
\r
119 // -------------------------------------------------------------
\r
122 bool BgzfData::CheckBlockHeader(char* header) {
\r
123 return (header[0] == GZIP_ID1 &&
\r
124 header[1] == (char)GZIP_ID2 &&
\r
125 header[2] == Z_DEFLATED &&
\r
126 (header[3] & FLG_FEXTRA) != 0 &&
\r
127 BgzfData::UnpackUnsignedShort(&header[10]) == BGZF_XLEN &&
\r
128 header[12] == BGZF_ID1 &&
\r
129 header[13] == BGZF_ID2 &&
\r
130 BgzfData::UnpackUnsignedShort(&header[14]) == BGZF_LEN );
\r
133 // packs an unsigned integer into the specified buffer
\r
135 void BgzfData::PackUnsignedInt(char* buffer, unsigned int value) {
\r
136 buffer[0] = (char)value;
\r
137 buffer[1] = (char)(value >> 8);
\r
138 buffer[2] = (char)(value >> 16);
\r
139 buffer[3] = (char)(value >> 24);
\r
142 // packs an unsigned short into the specified buffer
\r
144 void BgzfData::PackUnsignedShort(char* buffer, unsigned short value) {
\r
145 buffer[0] = (char)value;
\r
146 buffer[1] = (char)(value >> 8);
\r
149 // unpacks a buffer into a signed int
\r
151 signed int BgzfData::UnpackSignedInt(char* buffer) {
\r
152 union { signed int value; unsigned char valueBuffer[sizeof(signed int)]; } un;
\r
154 un.valueBuffer[0] = buffer[0];
\r
155 un.valueBuffer[1] = buffer[1];
\r
156 un.valueBuffer[2] = buffer[2];
\r
157 un.valueBuffer[3] = buffer[3];
\r
161 // unpacks a buffer into an unsigned int
\r
163 unsigned int BgzfData::UnpackUnsignedInt(char* buffer) {
\r
164 union { unsigned int value; unsigned char valueBuffer[sizeof(unsigned int)]; } un;
\r
166 un.valueBuffer[0] = buffer[0];
\r
167 un.valueBuffer[1] = buffer[1];
\r
168 un.valueBuffer[2] = buffer[2];
\r
169 un.valueBuffer[3] = buffer[3];
\r
173 // unpacks a buffer into an unsigned short
\r
175 unsigned short BgzfData::UnpackUnsignedShort(char* buffer) {
\r
176 union { unsigned short value; unsigned char valueBuffer[sizeof(unsigned short)]; } un;
\r
178 un.valueBuffer[0] = buffer[0];
\r
179 un.valueBuffer[1] = buffer[1];
\r
183 } // namespace BamTools
\r