static void getFastaOffsets(struct brokenRefPepTbl *brpTbl,
struct sqlConnection *conn,
struct extFileTbl* extFileTbl,
char *faPath)
/* parse fasta file to get offsets of proteins */
{
struct gbFa *fa = gbFaOpen(faPath, "r");
char acc[GB_ACC_BUFSZ];
struct brokenRefPep *brp;
HGID extId = extFileTblGet(extFileTbl, conn, faPath);
gbVerbMsg(5, "scanning fasta: %s", faPath);
while (gbFaReadNext(fa))
{
gbVerbMsg(5, " %s: %lld", fa->id, (long long)fa->recOff);
/* save only if same acecss, version, and file (to match mrna fa) */
short ver = gbSplitAccVer(fa->id, acc);
brp = hashFindVal(brpTbl->protAccHash, acc);
if ((brp != NULL) && (ver == brp->protVer) && sameString(faPath, brp->newFaPath))
{
gbFaGetSeq(fa); /* force read of sequence data */
brp->newFaId = extId;
brp->newFaOff = fa->recOff;
brp->newSeqSize = fa->seqLen;
brp->newRecSize = fa->off-fa->recOff;
gbVerbMsg(5, " save: %s %lld for %lld\n", fa->id, (long long)fa->recOff, (long long)fa->off);
}
}
gbFaClose(&fa);
}
static HGID getExtFileId(struct sqlConnection *conn, char* relPath)
/* get the extFile id for a file, prepending the gbdb root dir */
{
char path[PATH_LEN];
path[0] = '\0';
if (gGbdbGenBank[0] != '\0')
{
strcpy(path, gGbdbGenBank);
strcat(path, "/");
}
strcat(path, relPath);
if (extFiles == NULL)
{
gbVerbMsg(4, "loading extFile table");
extFiles = extFileTblLoad(conn);
}
return extFileTblGet(extFiles, conn, path);
}
