static ajint dbiblast_ncblreadhdr(AjPStr* hdrline, PBlastDb db, ajint start,
ajint end)
{
ajint size;
ajint llen;
PMemFile hfp;
size = ajStrGetRes(*hdrline);
hfp = db->HFile;
if(end)
{
llen = end - start;
if(db->IsBlast2)
llen += 1;
if(llen > size)
llen = size;
}
else
llen = size;
/*ajDebug("ncblreadhdr start %d end %d llen %d\n", start, end, llen);*/
if(db->IsBlast2)
{
dbiblast_memfseek(hfp,start,0);
dbiblast_memfreadS(hdrline,(size_t)1,(size_t)(llen-1),hfp);
}
else
{
dbiblast_memfseek(hfp,start+1,0); /* skip the '>' character */
dbiblast_memfreadS(hdrline,(size_t)1,(size_t)(llen-1),hfp);
}
ajStrSetValidLen(hdrline, (llen-1));
return llen;
}
ajint ajUserGet(AjPStr* pthis, const char* fmt, ...)
{
AjPStr thys;
const char *cp;
char *buff;
va_list args;
ajint ipos;
ajint isize;
ajint ilen;
ajint jlen;
ajint fileBuffSize = ajFileValueBuffsize();
va_start(args, fmt);
ajFmtVError(fmt, args);
va_end(args);
if(ajFileValueRedirectStdin())
{
ajUser("(Standard input in use: using default)");
ajStrAssignC(pthis, "");
return ajStrGetLen(*pthis);
}
ajStrSetRes(pthis, fileBuffSize);
buff = ajStrGetuniquePtr(pthis);
thys = *pthis;
isize = ajStrGetRes(thys);
ilen = 0;
ipos = 0;
/*ajDebug("ajUserGet buffer len: %d res: %d ptr: %x\n",
ajStrGetLen(thys), ajStrGetRes(thys), thys->Ptr);*/
while(buff)
{
#ifndef __ppc__
cp = fgets(&buff[ipos], isize, stdin);
#else
cp = ajSysFuncFgets(&buff[ipos], isize, stdin);
#endif
if(!cp && !ipos)
{
if(feof(stdin))
{
ajErr("Unable to get reply from user - end of standard input");
ajExitBad();
}
else
ajFatal("Error reading from user: '******'\n",
strerror(errno));
}
jlen = strlen(&buff[ipos]);
ilen += jlen;
/*
** We need to read again if:
** We have read the entire buffer
** and we don't have a newline at the end
** (must be careful about that - we may just have read enough)
*/
ajStrSetValidLen(pthis, ilen);
thys = *pthis;
if((jlen == (isize-1)) &&
(ajStrGetCharLast(thys) != '\n'))
{
ajStrSetRes(pthis, ajStrGetRes(thys)+fileBuffSize);
thys = *pthis;
/*ajDebug("more to do: jlen: %d ipos: %d isize: %d ilen: %d "
"Size: %d\n",
jlen, ipos, isize, ilen, ajStrGetRes(thys));*/
ipos += jlen;
buff = ajStrGetuniquePtr(pthis);
isize = ajStrGetRes(thys) - ipos;
/* ajDebug("expand to: ipos: %d isize: %d Size: %d\n",
ipos, isize, ajStrGetRes(thys)); */
}
else
buff = NULL;
}
ajStrSetValidLen(pthis, ilen);
if(ajStrGetCharLast(*pthis) == '\n')
ajStrCutEnd(pthis, 1);
/* PC files have \r\n Macintosh files have just \r : this fixes both */
if(ajStrGetCharLast(*pthis) == '\r')
{
/*ajDebug("Remove carriage-return characters from PC-style files\n");*/
ajStrCutEnd(pthis, 1);
}
ajStrTrimWhite(pthis);
return ajStrGetLen(*pthis);
}
static AjBool dbiblast_blastopenlib(const AjPStr name, AjBool usesrc,
ajint blastv, char dbtype,
PBlastDb* pdb)
{
AjPStr hname = NULL;
AjPStr sname = NULL;
AjPStr tname = NULL;
static AjPStr dbname = NULL;
ajint rdtmp = 0;
ajint rdtmp2 = 0;
ajint itype;
ajint ttop;
PMemFile TFile = NULL;
PBlastDb ret;
for(itype=0; blasttypes[itype].ExtT; itype++)
{
if((blastv == 1) && blasttypes[itype].IsBlast2)
continue;
if((blastv == 2) && !blasttypes[itype].IsBlast2)
continue;
if((dbtype == 'P') && !blasttypes[itype].IsProtein)
continue;
if((dbtype == 'N') && blasttypes[itype].IsProtein)
continue;
if(dbiblast_wrongtype(name, blasttypes[itype].ExtT))
continue;
dbiblast_dbname(&dbname,name,blasttypes[itype].ExtT);
dbiblast_newname(&tname,dbname,blasttypes[itype].ExtT);
TFile = dbiblast_memfopenfile(tname);
if(TFile)
break;
}
if(!TFile)
return ajFalse;
AJNEW0(*pdb);
ret = *pdb;
ret->TFile = TFile;
ajStrAssignS(&ret->Name, dbname);
ajDebug("Name '%S'\n", ret->Name);
/* find and open the 'table' file(s) */
if(!ret->TFile)
ajFatal(" cannot open %S table file %S\n", dbname, tname);
ajDebug("Successfully opened table file for type %d\n", itype);
ret->IsProtein = blasttypes[itype].IsProtein;
ret->IsBlast2 = blasttypes[itype].IsBlast2;
/* read the type and format - all databases */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->DbType);
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->DbFormat);
ret->HeaderLen += 8;
ajDebug("dbtype: %x dbformat: %x\n", ret->DbType, ret->DbFormat);
/* Open the header and (compressed) sequence files */
/* for DNA, also look for the FASTA file */
dbiblast_newname(&hname,dbname,blasttypes[itype].ExtH);
if((ret->HFile = dbiblast_memfopenfile(hname))==NULL)
ajFatal(" cannot open %S header file\n",hname);
dbiblast_newname(&sname,dbname,blasttypes[itype].ExtS);
if((ret->SFile = dbiblast_memfopenfile(sname))==NULL)
ajFatal(" cannot open %S sequence file\n",sname);
if(!ret->IsBlast2 && !ret->IsProtein && usesrc)
/* this can fail */
if((ret->FFile = dbiblast_memfopenfile(dbname))==NULL)
ajDebug(" cannot open %S source file\n",dbname);
/* read the title - all formats */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->TitleLen);
/* blast2 does not align after the title */
if(ret->IsBlast2)
rdtmp = ret->TitleLen;
else
rdtmp = ret->TitleLen + ((ret->TitleLen%4 !=0 ) ?
4-(ret->TitleLen%4) : 0);
ajStrAssignResC(&ret->Title, rdtmp+1, "");
ajDebug("IsBlast2: %B title_len: %d rdtmp: %d title_str: '%S'\n",
ret->IsBlast2, ret->TitleLen, rdtmp, ret->Title);
ajStrTrace(ret->Title);
dbiblast_memfreadS(&ret->Title,(size_t)1,(size_t)rdtmp,ret->TFile);
if(ret->IsBlast2)
ajStrSetValidLen(&ret->Title, ret->TitleLen);
else
ajStrSetValidLen(&ret->Title, ret->TitleLen-1);
ajDebug("title_len: %d rdtmp: %d title_str: '%S'\n",
ret->TitleLen, rdtmp, ret->Title);
ret->HeaderLen += 4 + rdtmp;
/* read the date - blast2 */
if(ret->IsBlast2)
{
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->DateLen);
rdtmp2 = ret->DateLen;
ajStrAssignResC(&ret->Date, rdtmp2+1, "");
dbiblast_memfreadS(&ret->Date,(size_t)1,(size_t)rdtmp2,ret->TFile);
ajStrSetValid(&ret->Date);
ret->DateLen = ajStrGetLen(ret->Date);
ajDebug("datelen: %d rdtmp: %d date: '%S'\n",
ret->DateLen, rdtmp2, ret->Date);
ret->HeaderLen += 4 + rdtmp2;
}
/* read the rest of the header (different for protein and DNA) */
if(!ret->IsBlast2 && !ret->IsProtein)
{
/* length of source lines */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->LineLen);
ret->HeaderLen += 4;
}
/* all formats have the next 3 */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->Size);
if(ret->IsProtein)
{ /* mad, but they are the other way for DNA */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->TotLen);
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->MaxSeqLen);
}
else
{
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->MaxSeqLen);
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->TotLen);
}
ret->HeaderLen += 12;
if(!ret->IsBlast2 && !ret->IsProtein)
{ /* Blast 1.4 DNA only */
/* compressed db length */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->CompLen);
/* count of nt's cleaned */
dbiblast_memreadUInt4(ret->TFile,(ajuint*)&ret->CleanCount);
ret->HeaderLen += 8;
}
ajDebug(" size: %u, totlen: %d maxseqlen: %u\n",
ret->Size, ret->TotLen, ret->MaxSeqLen);
ajDebug(" linelen: %u, complen: %d cleancount: %d\n",
ret->LineLen, ret->CompLen, ret->CleanCount);
/* Now for the tables of offsets. Again maddeningly different in each */
if(ret->IsBlast2)
{
ttop = ret->TopHdr = ret->HeaderLen; /* header first */
ttop = ret->TopCmp = ttop + (ret->Size+1) * 4; /* then sequence */
if(!ret->IsProtein) /* Blast 2 DNA only */
ttop = ret->TopAmb = ttop + (ret->Size+1) * 4;
}
else
{
ttop = ret->TopCmp = ret->HeaderLen + ret->CleanCount*4; /* comp seq */
if(!ret->IsProtein) /* Blast 1.4 DNA only */
ttop = ret->TopSrc = ttop + (ret->Size+1) * 4;
ttop = ret->TopHdr = ttop + (ret->Size+1) * 4; /* headers for all */
if(!ret->IsProtein) /* Blast 1.4 DNA only */
ttop = ret->TopAmb = ttop + (ret->Size+1) * 4;
}
ajDebug("table file index starts at %d\n", ret->HeaderLen);
ajDebug("table file csq starts at %d\n", ret->TopCmp);
ajDebug("table file src starts at %d\n", ret->TopSrc);
ajDebug("table file hdr starts at %d\n", ret->TopHdr);
ajDebug("table file amb starts at %d\n", ret->TopAmb);
ajStrDel(&hname);
ajStrDel(&sname);
ajStrDel(&tname);
ajStrDel(&dbname);
return ajTrue;
}