static int flagOrphans(struct sqlConnection *conn, struct gbSelect* select,
struct hash *seqHash, struct gbStatusTbl* statusTbl)
/* find orphans in gbSeq table that were not found in gbStatus and flag them
*in the status table */
{
struct hashCookie cookie;
struct hashEl* hel;
int numOrphans = 0;
/* Find orphans using seqHash. This was previously implemented using
* SELECT seq.acc FROM seq LEFT JOIN gbStatus ON (seq.acc = gbStatus.acc)
* WHERE (seq.type = '%s') AND (seq.srcDb = '%s')
* AND (gbStatus.seq IS NULL);
* however this proved many times slower than reading the sequences into
* a hash.
*/
cookie = hashFirst(seqHash);
while ((hel = hashNext(&cookie)) != NULL)
{
if (!hel->val)
{
struct gbStatus *status = gbStatusTblFind(statusTbl, hel->name);
if (status != NULL)
{
status->stateChg |= GB_ORPHAN;
if (!(status->stateChg & GB_NEW))
errAbort("oprhaned seq entry for %s is not flaged as new",
status->acc);
numOrphans++;
}
else
{
/* Orphan, but not in table. This could be an entry that was
* deleted from gbStatus table by hand. Leave versions empty */
fprintf(stderr, "Warning: %s is orphaned, but not a new entry, deleting\n",
hel->name);
status = gbStatusTblAdd(statusTbl, hel->name, 0, 0, select->type,
select->release->srcDb, 0, 0, 0,
"", "", 0);
slAddHead(&statusTbl->deleteList, status);
statusTbl->numDelete++;
status->stateChg = GB_DELETED;
}
}
}
return numOrphans;
}
static boolean processEntry(struct sqlConnection *conn,
struct gbStatusTbl* statusTbl,
HGID faFileId, HGID pepFaId)
/* Parse and process the next ra entry, check to see if it is selected
* for update. */
{
char* acc;
struct gbStatus* status;
if ((acc = gbMDParseEntry()) == NULL)
return FALSE;
status = gbStatusTblFind(statusTbl, acc);
if ((status != NULL) && (status->selectProc != NULL) && (!status->metaDone)
&& (status->selectProc->modDate == raModDate))
updateMetaData(conn, status, statusTbl, faFileId, pepFaId);
return TRUE;
}
static struct gbStatus *lookupStatus(char *accver, struct gbStatusTbl* statusTbl,
struct lineFile *lf, short *versionPtr)
/* get the status entry for an accession if it is selected, return NULL if not
* selected */
{
char acc[GB_ACC_BUFSZ];
short version = gbSplitAccVer(accver, acc);
if (version < 0)
errAbort("entry name \"%s\" is not a genbank accession with version: %s:%d",
accver, lf->fileName, lf->lineIx);
struct gbStatus* status = gbStatusTblFind(statusTbl, acc);
if ((status != NULL) && (status->selectAlign != NULL)
&& (status->selectAlign->version == version))
{
if (versionPtr != NULL)
*versionPtr = version;
return status;
}
else
return NULL;
}
