static BioseqPtr FetchOnlyBioseqFromID (CharPtr str)
{
BioseqPtr bsp = NULL;
Uint2 entityID;
SeqEntryPtr sep;
SeqIdPtr sip = NULL;
Int4 uid;
time_t t1, t2;
t1 = time (NULL);
uid = 0;
TrimSpacesAroundString (str);
if (IsAllDigits (str)) {
if (PvtStrToLong (str, &uid)) {
sip = ValNodeNew(NULL);
sip->choice = SEQID_GI;
sip->data.intvalue = uid;
} else {
uid = 0;
}
} else {
sip = SeqIdFromAccessionDotVersion (str);
uid = AccessionToGi (str);
}
sep = NULL;
if (uid > 0) {
sep = PubSeqSynchronousQuery (uid, 3, 0); /* retcode was 0 */
if (sep != NULL) {
bsp = BioseqFindInSeqEntry (sip, sep);
entityID = ObjMgrGetEntityIDForChoice (sep);
}
}
sip = SeqIdFree (sip);
if (debug_mode) {
t2 = time (NULL);
if (t2 - t1 > 1) {
printf ("Time to download %s from ID:%d\n", str, (int) (t2 - t1));
}
}
return bsp;
}
static Int2 LIBCALLBACK TPASmartBioseqFetchFunc (Pointer data)
{
BioseqPtr bsp;
Char cmmd [256];
Pointer dataptr;
Uint2 datatype;
Uint2 entityID;
FILE* fp;
OMProcControlPtr ompcp;
ObjMgrProcPtr ompp;
Char path [PATH_MAX];
Char err_path [PATH_MAX];
SeqEntryPtr sep = NULL;
SeqIdPtr sip;
TextSeqIdPtr tsip;
ompcp = (OMProcControlPtr) data;
if (ompcp == NULL) return OM_MSG_RET_ERROR;
ompp = ompcp->proc;
if (ompp == NULL) return OM_MSG_RET_ERROR;
sip = (SeqIdPtr) ompcp->input_data;
if (sip == NULL) return OM_MSG_RET_ERROR;
if (sip->choice != SEQID_TPG) return OM_MSG_RET_ERROR;
tsip = (TextSeqIdPtr) sip->data.ptrvalue;
if (tsip == NULL || StringHasNoText (tsip->accession)) return OM_MSG_RET_ERROR;
if (tpasmartfetchcmd == NULL) {
if (GetAppParam ("SEQUIN", "TPASMART", "FETCHSCRIPT", NULL, cmmd, sizeof (cmmd))) {
tpasmartfetchcmd = StringSaveNoNull (cmmd);
}
}
if (tpasmartfetchcmd == NULL) return OM_MSG_RET_ERROR;
TmpNam (path);
#ifdef OS_UNIX
sprintf (err_path, "%s.err", path);
sprintf (cmmd, "csh %s %s > %s 2>%s", tpasmartfetchcmd, tsip->accession, path, err_path);
system (cmmd);
#endif
#ifdef OS_MSWIN
sprintf (cmmd, "%s %s -o %s", tpasmartfetchcmd, tsip->accession, path);
system (cmmd);
#endif
fp = FileOpen (path, "r");
if (fp == NULL) {
FileRemove (path);
#ifdef OS_UNIX
FileRemove (err_path);
#endif
return OM_MSG_RET_ERROR;
}
dataptr = ReadAsnFastaOrFlatFile (fp, &datatype, &entityID, FALSE, FALSE, TRUE, FALSE);
FileClose (fp);
FileRemove (path);
#ifdef OS_UNIX
FileRemove (err_path);
#endif
if (dataptr == NULL) return OM_MSG_RET_OK;
sep = GetTopSeqEntryForEntityID (entityID);
if (sep == NULL) return OM_MSG_RET_ERROR;
bsp = BioseqFindInSeqEntry (sip, sep);
ompcp->output_data = (Pointer) bsp;
ompcp->output_entityID = ObjMgrGetEntityIDForChoice (sep);
return OM_MSG_RET_DONE;
}
static void ProcessMultipleRecord (
CharPtr filename,
CSpeedFlagPtr cfp
)
{
AsnIoPtr aip;
AsnTypePtr atp;
BioseqPtr bsp;
Char buf [41];
Uint2 entityID;
FILE *fp;
SeqEntryPtr fsep;
Char longest [41];
Int4 numrecords, x;
SeqEntryPtr sep;
time_t starttime, stoptime, worsttime;
#ifdef OS_UNIX
Char cmmd [256];
CharPtr gzcatprog;
int ret;
Boolean usedPopen = FALSE;
#endif
if (cfp == NULL) return;
if (StringHasNoText (filename)) return;
#ifndef OS_UNIX
if (cfp->compressed) {
Message (MSG_POSTERR, "Can only decompress on-the-fly on UNIX machines");
return;
}
#endif
#ifdef OS_UNIX
if (cfp->compressed) {
gzcatprog = getenv ("NCBI_UNCOMPRESS_BINARY");
if (gzcatprog != NULL) {
sprintf (cmmd, "%s %s", gzcatprog, filename);
} else {
ret = system ("gzcat -h >/dev/null 2>&1");
if (ret == 0) {
sprintf (cmmd, "gzcat %s", filename);
} else if (ret == -1) {
Message (MSG_POSTERR, "Unable to fork or exec gzcat in ScanBioseqSetRelease");
return;
} else {
ret = system ("zcat -h >/dev/null 2>&1");
if (ret == 0) {
sprintf (cmmd, "zcat %s", filename);
} else if (ret == -1) {
Message (MSG_POSTERR, "Unable to fork or exec zcat in ScanBioseqSetRelease");
return;
} else {
Message (MSG_POSTERR, "Unable to find zcat or gzcat in ScanBioseqSetRelease - please edit your PATH environment variable");
return;
}
}
}
fp = popen (cmmd, /* cfp->binary? "rb" : */ "r");
usedPopen = TRUE;
} else {
fp = FileOpen (filename, cfp->binary? "rb" : "r");
}
#else
fp = FileOpen (filename, cfp->binary? "rb" : "r");
#endif
if (fp == NULL) {
Message (MSG_POSTERR, "FileOpen failed for input file '%s'", filename);
return;
}
aip = AsnIoNew (cfp->binary? ASNIO_BIN_IN : ASNIO_TEXT_IN, fp, NULL, NULL, NULL);
if (aip == NULL) {
Message (MSG_ERROR, "AsnIoNew failed for input file '%s'", filename);
return;
}
if (cfp->logfp != NULL) {
fprintf (cfp->logfp, "%s\n\n", filename);
fflush (cfp->logfp);
}
longest [0] = '\0';
worsttime = 0;
numrecords = 0;
atp = cfp->atp_bss;
while ((atp = AsnReadId (aip, cfp->amp, atp)) != NULL) {
if (atp == cfp->atp_se) {
sep = SeqEntryAsnRead (aip, atp);
if (sep != NULL) {
entityID = ObjMgrGetEntityIDForChoice (sep);
fsep = FindNthBioseq (sep, 1);
if (fsep != NULL && fsep->choice == 1) {
bsp = (BioseqPtr) fsep->data.ptrvalue;
if (bsp != NULL) {
SeqIdWrite (bsp->id, buf, PRINTID_FASTA_LONG, sizeof (buf));
if (cfp->logfp != NULL) {
fprintf (cfp->logfp, "%s\n", buf);
fflush (cfp->logfp);
}
}
}
starttime = GetSecs ();
for (x = 0; x < cfp->maxcount; x++) {
DoProcess (sep, entityID, cfp);
}
stoptime = GetSecs ();
if (stoptime - starttime > worsttime) {
worsttime = stoptime - starttime;
StringCpy (longest, buf);
}
numrecords++;
ObjMgrFreeByEntityID (entityID);
}
} else {
AsnReadVal (aip, atp, NULL);
}
}
AsnIoFree (aip, FALSE);
#ifdef OS_UNIX
if (usedPopen) {
pclose (fp);
} else {
FileClose (fp);
}
#else
FileClose (fp);
#endif
if (cfp->logfp != NULL && (! StringHasNoText (longest))) {
fprintf (cfp->logfp, "Longest processing time %ld seconds on %s\n",
(long) worsttime, longest);
fprintf (cfp->logfp, "Total number of records %ld\n", (long) numrecords);
fflush (cfp->logfp);
}
}
