PCSAN LIBCALL LoadCSAN(CharPtr pcFile)
{
FILE *f;
Char fullpath [PATH_MAX];
CharPtr ptr;
PCSAN pcsanHead = NULL;
PCSAN pcsanTemp = NULL;
Char pcBuf[250];
CharPtr ppcParse[125];
CharPtr pcTemp = NULL;
CharPtr pcTest = NULL;
CharPtr pcSeqPart = NULL;
ValNodePtr pvn = NULL;
long int val = 0;
Int4 i = 0;
Int4 iString = 0;
Int4 iLen =0;
Int4 iWC =0;
Int4 iWL = 0;
Int4 iGi = 0;
if (!pcFile) return(NULL);
ProgramPath (fullpath, sizeof (fullpath) - 1);
ptr = StringRChr (fullpath, DIRDELIMCHR);
if (ptr != NULL) {
*ptr = '\0';
}
/* printf("FULLPATH[%s]\n",IsNullStr(fullpath)); */
FileBuildPath (fullpath, NULL, pcFile);
/* printf("FULLPATH[%s]\n",IsNullStr(fullpath)); */
if ((f = FileOpen (fullpath, "r")) == NULL)
{
if (FindPath ("NCBI", "NCBI", "DATA", fullpath, sizeof (fullpath) - 1))
{
FileBuildPath (fullpath, NULL, pcFile);
f = FileOpen (fullpath, "r");
if (f == NULL)
{
ErrPost(CTX_NCBIOBJ, 1, "Couldn't open [%s] in LoadCSA", IsNullStr(fullpath));
return (NULL);
}
}
else
{
ErrPost(CTX_NCBIOBJ, 1, "FindPath failed in LoadNeighbors");
return (NULL);
}
}
do /* get each line, tokenize into ppcParse */
{
pcBuf[0] = '\0';
pcTest = fgets(pcBuf, (size_t)250, f);
/* printf("[%s]\n",IsNullStr(pcTest)); */
if (pcTest)
{
iLen = StringLen(pcTest);
/* Break into words delimited by whitespace or "|" */
/* gi|123456| */
/* printf("LineLength %d\n",(int) iLen); */
for (i=0; i < 125; i++) ppcParse[i] = NULL;
pcTemp = pcTest;
ppcParse[0] = pcTest;
iWC = 1;
while ((*pcTemp != '\0') && (iWC < 125))
{
if ((*pcTemp == '|') || (*pcTemp == ' ') || (*pcTemp == '\t') || (*pcTemp == '\n'))
{
*pcTemp = '\0'; /* terminate current word */
do {
pcTemp++; /* skip whitespace to next word */
} while ((*pcTemp == '|') || (*pcTemp == ' ') || (*pcTemp == '\t') || (*pcTemp == '\n'));
ppcParse[iWC] = pcTemp; /* start next word */
iWC++;
}
else
pcTemp++;
}
/* DEBUG TEST */
/* for (i=0; i<125; i++)
if (ppcParse[i] != NULL)
{
printf("%ld/%ld [%s]\n",i,iWC, IsNullStr(ppcParse[i]));
} */
/* If ppcParse[0] is "gi" then this is a VALID sequence line and the last
word parsed is the sequence */
if (ppcParse[0][0] != '\0')
{
iWL = StringLen(ppcParse[0]);
if (iWL == 2)
if (((ppcParse[0][0] == 'g') ||
(ppcParse[0][0] == 'G')) &&
((ppcParse[0][1] == 'i') ||
(ppcParse[0][1] == 'I')))
{
/* next word should be GI */
iGi = 0;
if (StrToInt4 (ppcParse[1], &iGi))
{
/* printf("GI is (%ld)\n",iGi); */
if (iGi == 0) goto parserr;
/* find the 2nd last string = sequence */
/* last string is stripped off line terminator, by the way... */
iString=2; /* start after GI */
while (ppcParse[iString+1] != NULL)
{ /* iString is the LAST sting - line terminator */
iString++;
}
pcsanTemp = pcsanHead;
while (pcsanTemp)
{ /* look for existing pcsan */
if (iGi == pcsanTemp->iGi)
{
break;
}
pcsanTemp = pcsanTemp->next;
}
if (!pcsanTemp) /* otherwise add a new one */
{
pcsanTemp = NewCSAN();
pcsanTemp->next = pcsanHead;
pcsanHead = pcsanTemp; /* put at head of list */
pcsanTemp->iGi = iGi;
}
/* Initialize fields */
pcSeqPart = NULL;
pcSeqPart = StringSave(ppcParse[iString-1]);
/* printf("SEQ [%ld]= [%s]\n",(long) iString-1, IsNullStr(pcSeqPart)); */
ValNodeAddStr(&pcsanTemp->pvnSeqAlnPart, 0, pcSeqPart);
} /* if GI */
} /* if line starts with GI */
} /* if line starts with a word */
} /* if pcTest */
} while (pcTest);
printf("REPORT\n");
pcsanTemp = pcsanHead;
while (pcsanTemp)
{
printf("GI = [%ld]\n", (long) pcsanTemp->iGi);
pvn = pcsanTemp->pvnSeqAlnPart;
while (pvn)
{
printf(">%s<\n",(char *) IsNullStr(pvn->data.ptrvalue));
pvn = pvn->next;
}
pcsanTemp = pcsanTemp->next;
}
FileClose(f);
return pcsanHead;
parserr:
FileClose(f);
ErrPostEx(SEV_FATAL,0,0, "Error in input format \n");
FreeCSAN(pcsanHead);
return NULL;
}
BiostrucPtr LIBCALL
MMDBBiostrucGet (DocUid uid, Int4 mdlLvl, Int4 maxModels)
{
AsnIoPtr aip = NULL;
BiostrucPtr pbs = NULL;
int index;
Int4 acc;
#ifdef MMDB_PUBSTRUCT
extern WWWInfoPtr info;
if(!file_db)
{
Int4 state = 0;
char *dbserver = (database[0]?database:NULL);
acc = uid;
if(WWWFindName(info, "ps_acc") >=0)
aip = PubStruct_readasn(dbserver, acc);
else
{
index = WWWFindName(info, "ps_state");
if (index >= 0)
{
char * item;
item = WWWGetValueByIndex(info,index);
if (!StrToInt4(item,&state))
ErrPostEx(SEV_ERROR,0,0, "cant convert Pubstruct_state<%s> into integer",item);
}
if (state == 0 )
aip = PubStruct_viewasn(dbserver, uid);
else
{
acc = PubStruct_lookup(dbserver,uid, state);
/*
fprintf(stderr,"%s:%d: acc=%d <<- %s(mmdb=%d,state=%d)\n",
__FILE__,__LINE__,acc,dbserver,uid,state);
*/
if (acc>0)
aip = PubStruct_readasn(dbserver, acc);
}
}
}
else
#endif
{
Char path[PATH_MAX];
sprintf(path, "%s%ld.val", database, (long) uid);
#ifdef MMDB_UNIXCOMPRESSED
{
FILE *pipe;
Char compath[PATH_MAX];
sprintf(compath,"%s -c %s.gz ", gunzip, path);
pipe=popen(compath,"rb");
if (pipe == 0)
{
ErrPostEx(SEV_FATAL,0,0, "MMDBBiostrucGet failed: Can't find gunzip in path.\n");
return NULL;
}
aip = AsnIoNew(ASNIO_BIN_IN, pipe , NULL, NULL, NULL);
}
#else
if (FileLength(path) == 0)
return NULL;
aip = AsnIoOpen(path, "rb");
#endif
}
if (!aip)
return NULL;
pbs = BiostrucAsnGet(aip, NULL, mdlLvl, maxModels);
#ifdef MMDB_PUBSTRUCT
if(!file_db)
PubStruct_closeasn (aip,1);
else
#endif
AsnIoClose (aip);
return pbs;
}