| 1 |
|
#ifndef __GCDB_H |
| 2 |
|
#define __GCDB_H |
| 3 |
< |
#include <stdlib.h> |
| 3 |
> |
#include "GBase.h" |
| 4 |
|
#include <stddef.h> |
| 5 |
|
#include <fcntl.h> |
| 6 |
– |
#include <sys/stat.h> |
| 7 |
– |
#include "GBase.h" |
| 6 |
|
|
| 7 |
< |
#if defined(__WIN32__) || defined(WIN32) |
| 7 |
> |
#ifdef __WIN32__ |
| 8 |
|
#define PROT_READ 1 |
| 9 |
|
#define PROT_WRITE 2 |
| 10 |
|
#define PROT_READWRITE 3 |
| 15 |
|
#define W_OK 2 |
| 16 |
|
#define RW_OK 6 |
| 17 |
|
|
| 18 |
< |
#if !defined(MAP_FAILED) |
| 18 |
> |
#ifndef MAP_FAILED |
| 19 |
|
#define MAP_FAILED ((void *) -1) |
| 20 |
|
#endif |
| 21 |
|
void *mmap(char *,size_t,int,int,int,off_t); |
| 24 |
|
#include <sys/mman.h> |
| 25 |
|
#endif |
| 26 |
|
|
| 29 |
– |
#define MAX_UINT 0xFFFFFFFFUL |
| 30 |
– |
|
| 31 |
– |
|
| 27 |
|
//===================================================== |
| 28 |
|
//------------- buffer stuff ------------------- |
| 29 |
|
//===================================================== |
| 30 |
|
#define GCDBUFFER_INSIZE 8192 |
| 31 |
|
#define GCDBUFFER_OUTSIZE 8192 |
| 32 |
|
|
| 38 |
– |
|
| 33 |
|
typedef int (*opfunc)(int, char*, size_t); |
| 34 |
|
|
| 35 |
|
//typedef unsigned long gcdb_seek_pos; |
| 36 |
|
typedef off_t gcdb_seek_pos; |
| 37 |
< |
typedef unsigned int (*uint_conv)(void*); //uint conversion function pointer |
| 38 |
< |
typedef off_t (*offt_conv)(void*); //uint conversion function pointer |
| 37 |
> |
typedef unsigned int (*uint_conv_func)(void*); //uint conversion function pointer |
| 38 |
> |
typedef off_t (*offt_conv_func)(void*); //uint conversion function pointer |
| 39 |
> |
typedef int16_t (*int16_conv_func)(void*); //int16 conversion function pointer |
| 40 |
|
|
| 41 |
|
|
| 42 |
|
//conversion function --> to platform independent uint |
| 43 |
< |
extern uint_conv gcvt_uint; |
| 44 |
< |
extern offt_conv gcvt_offt; |
| 45 |
< |
|
| 46 |
< |
int endian_test(void); |
| 43 |
> |
extern uint_conv_func gcvt_uint; |
| 44 |
> |
extern offt_conv_func gcvt_offt; |
| 45 |
> |
extern int16_conv_func gcvt_int16; |
| 46 |
> |
/* |
| 47 |
|
unsigned int uint32_sun(void* x86int); |
| 48 |
|
unsigned int uint32_x86(void* x86int); |
| 49 |
|
//for file offsets: off_t runtime conversions: |
| 50 |
|
off_t offt_sun(void* offt); |
| 51 |
|
off_t offt_x86(void* offt); |
| 52 |
+ |
int16_t int16_sun(void* i16); |
| 53 |
+ |
int16_t int16_x86(void* i16); |
| 54 |
+ |
*/ |
| 55 |
|
|
| 56 |
+ |
void gcvt_endian_setup(); |
| 57 |
|
|
| 58 |
|
class GCDBuffer { |
| 59 |
|
public: |
| 68 |
|
fd=0; |
| 69 |
|
op=NULL; |
| 70 |
|
n=0; |
| 72 |
– |
//check endianness |
| 73 |
– |
gcvt_uint=(endian_test())? &uint32_sun : &uint32_x86; |
| 74 |
– |
gcvt_offt=(endian_test())? &offt_sun : &offt_x86; |
| 71 |
|
} |
| 72 |
|
GCDBuffer(opfunc aop,int afd,char *buf,unsigned int len) { |
| 73 |
|
//check endianness |
| 74 |
< |
gcvt_uint=(endian_test())? &uint32_sun : &uint32_x86; |
| 79 |
< |
gcvt_offt=(endian_test())? &offt_sun : &offt_x86; |
| 74 |
> |
gcvt_endian_setup(); |
| 75 |
|
init(aop, afd, buf, len); |
| 76 |
|
} |
| 77 |
|
void init(opfunc aop,int afd,char *buf,unsigned int len) { |
| 115 |
|
#define CDBMSK_OPT_C 0x00000002 |
| 116 |
|
#define CDBMSK_OPT_CADD 0x00000004 |
| 117 |
|
#define CDBMSK_OPT_COMPRESS 0x00000008 |
| 118 |
+ |
#define CDBMSK_OPT_GSEQ 0x00000010 |
| 119 |
|
//creates a compressed version of the database |
| 120 |
|
//uses plenty of unions for ensuring compatibility with |
| 121 |
|
// the old 'CIDX' info structure |
| 122 |
|
|
| 123 |
< |
//damn, sun and 64bit machines |
| 128 |
< |
// align this to 64bit -- so sizeof() is misled! |
| 123 |
> |
//trying to prevent [64bit] machines to align this to 64bit -- sizeof() gets it wrong! |
| 124 |
|
#pragma pack(4) |
| 125 |
< |
// I wish, but stupid gcc 2.95.3 alpha-decosf version does not |
| 126 |
< |
// recognize this pragma directive !!? |
| 127 |
< |
// |
| 125 |
> |
// eek, gcc 2.95.3 alpha-decosf version does not |
| 126 |
> |
// recognize this pragma directive |
| 127 |
> |
|
| 128 |
> |
|
| 129 |
|
struct cdbInfo { |
| 130 |
|
uint32 num_keys; |
| 131 |
|
union { |
| 151 |
|
uint32 old_dbnamelen; |
| 152 |
|
}; |
| 153 |
|
}; |
| 154 |
+ |
|
| 155 |
+ |
// for passing around index data: |
| 156 |
+ |
struct CIdxData32 { |
| 157 |
+ |
uint32 fpos; |
| 158 |
+ |
uint32 reclen; |
| 159 |
+ |
}; |
| 160 |
+ |
/* |
| 161 |
+ |
struct CIdxSeqData32 { //4+4+2+1 = 11 bytes |
| 162 |
+ |
uint32 fpos; |
| 163 |
+ |
uint32 reclen; |
| 164 |
+ |
uint16_t linelen; //line length for FASTA-formatted seq |
| 165 |
+ |
byte elen; //length of end-of-line delimiter: 1 (unix/mac) or 2 (Windows) |
| 166 |
+ |
}; |
| 167 |
+ |
*/ |
| 168 |
+ |
struct CIdxData { |
| 169 |
+ |
off_t fpos; //64bit value on Linux |
| 170 |
+ |
uint32 reclen; |
| 171 |
+ |
}; |
| 172 |
+ |
/* |
| 173 |
+ |
struct CIdxSeqData { //8+4+2+1 = 15 bytes |
| 174 |
+ |
off_t fpos; //64bit value on Linux |
| 175 |
+ |
uint32 reclen; |
| 176 |
+ |
uint16_t linelen; //line length for FASTA-formatted seq |
| 177 |
+ |
byte elen; //length of end-of-line delimiter: 1 (unix/mac) or 2 (Windows) |
| 178 |
+ |
}; |
| 179 |
+ |
*/ |
| 180 |
|
#pragma pack() |
| 181 |
|
|
| 182 |
|
extern int cdbInfoSIZE; |
| 183 |
+ |
extern int IdxDataSIZE; |
| 184 |
+ |
extern int IdxDataSIZE32; |
| 185 |
+ |
/* |
| 186 |
+ |
extern int IdxSeqDataSIZE; |
| 187 |
+ |
extern int IdxSeqDataSIZE32; |
| 188 |
+ |
*/ |
| 189 |
|
|
| 190 |
|
void uint32_pack(char *,uint32); |
| 191 |
|
void uint32_pack_big(char *,uint32); |
| 403 |
|
memcpy((void*)outbuf, (void*)(buf+bufpos), mlen); |
| 404 |
|
return mlen; |
| 405 |
|
} |
| 406 |
< |
|
| 406 |
> |
char peekChar() { |
| 407 |
> |
if (eob) return -1; |
| 408 |
> |
//if (eob || len>buflen) return -1; |
| 409 |
> |
if (1>bufused-bufpos) refill(true); |
| 410 |
> |
return *(buf+bufpos); |
| 411 |
> |
} |
| 412 |
|
uchar* peekStr(uchar* outbuf, int len) { |
| 413 |
|
int rd=peek(outbuf,len); |
| 414 |
|
if (rd>0) { outbuf[rd]='\0'; return outbuf; } |
| 415 |
|
else return NULL; |
| 416 |
|
} |
| 417 |
|
//looks ahead to check if what follows matches |
| 418 |
< |
int peekCmp(char* cmpstr, int cmplen=0) { |
| 418 |
> |
int peekCmp(char* cmpstr, int cmplen=-1) { |
| 419 |
> |
if (cmplen==0) return 0; |
| 420 |
|
if (eob) //GError("GReadBuf::peekcmp error: eob!\n"); |
| 421 |
|
return -2; |
| 422 |
< |
if (!cmplen) cmplen=strlen(cmpstr); |
| 422 |
> |
if (cmplen<0) cmplen=strlen(cmpstr); |
| 423 |
|
if (cmplen>bufused-bufpos) { |
| 424 |
|
refill(true); |
| 425 |
|
if (cmplen>bufused-bufpos) return -2; |
