void bedSqlFieldsExceptForChrom(struct hTableInfo *hti,
int *retFieldCount, char **retFields)
/* Given tableInfo figure out what fields are needed to
* add to a database query to have information to create
* a bed. The chromosome is not one of these fields - we
* assume that is already known since we're processing one
* chromosome at a time. Return a comma separated (no last
* comma) list of fields that can be freeMem'd, and the count
* of fields (this *including* the chromosome). */
{
struct dyString *fields = dyStringNew(128);
int fieldCount = fieldCount = hTableInfoBedFieldCount(hti);
dyStringPrintf(fields, "%s,%s", hti->startField, hti->endField);
if (fieldCount >= 4)
{
if (hti->nameField[0] != 0)
dyStringPrintf(fields, ",%s", hti->nameField);
else /* Put in . as placeholder. */
dyStringPrintf(fields, ",'.'");
}
if (fieldCount >= 5)
{
if (hti->scoreField[0] != 0)
dyStringPrintf(fields, ",%s", hti->scoreField);
else
dyStringPrintf(fields, ",0", hti->startField);
}
if (fieldCount >= 6)
{
if (hti->strandField[0] != 0)
dyStringPrintf(fields, ",%s", hti->strandField);
else
dyStringPrintf(fields, ",'.'");
}
if (fieldCount >= 8)
{
if (hti->cdsStartField[0] != 0)
dyStringPrintf(fields, ",%s,%s", hti->cdsStartField, hti->cdsEndField);
else
dyStringPrintf(fields, ",%s,%s", hti->startField, hti->endField);
}
if (fieldCount >= 12)
{
dyStringPrintf(fields, ",%s,%s,%s", hti->countField,
hti->endsSizesField, hti->startsField);
}
if (htiIsPsl(hti))
{
/* For psl format we need to know chrom size as well
* to handle reverse strand case. */
dyStringPrintf(fields, ",tSize");
}
*retFieldCount = fieldCount;
*retFields = dyStringCannibalize(&fields);
}
static struct bed *getChromAsBed(
struct sqlConnection *conn,
char *db, char *table, /* Database and table. */
char *chrom,
struct lm *lm,
int *retFieldCount) /* Where to allocate memory. */
/* Return a bed list of all items in the given range in table.
* Cleanup result via lmCleanup(&lm) rather than bedFreeList. */
{
char *fields = NULL;
struct sqlResult *sr;
struct hTableInfo *hti;
struct bed *bedList=NULL, *bed;
char **row;
int fieldCount;
boolean isPsl, isGenePred, isBedWithBlocks;
boolean pslKnowIfProtein = FALSE, pslIsProtein = FALSE;
hti = hFindTableInfoDb(db, chrom, table);
if (hti == NULL)
errAbort("Could not find table info for table %s", table);
bedSqlFieldsExceptForChrom(hti, &fieldCount, &fields);
isPsl = htiIsPsl(hti);
isGenePred = sameString("exonEnds", hti->endsSizesField);
isBedWithBlocks = (
(sameString("chromStarts", hti->startsField) ||
sameString("blockStarts", hti->startsField))
&& sameString("blockSizes", hti->endsSizesField));
/* All beds have at least chrom,start,end. We omit the chrom
* from the query since we already know it. */
sr = hExtendedChromQuery(conn, table, chrom, NULL, FALSE, fields, NULL);
while (sr != NULL && (row = sqlNextRow(sr)) != NULL)
{
bed = bedFromRow(chrom, row, fieldCount, isPsl, isGenePred,
isBedWithBlocks, &pslKnowIfProtein, &pslIsProtein, lm);
slAddHead(&bedList, bed);
}
freez(&fields);
sqlFreeResult(&sr);
slReverse(&bedList);
*retFieldCount = fieldCount;
return(bedList);
}
struct bed *getRegionAsBed(
char *db, char *table, /* Database and table. */
struct region *region, /* Region to get data for. */
char *filter, /* Filter to add to SQL where clause if any. */
struct hash *idHash, /* Restrict to id's in this hash if non-NULL. */
struct lm *lm, /* Where to allocate memory. */
int *retFieldCount) /* Number of fields. */
/* Return a bed list of all items in the given range in table.
* Cleanup result via lmCleanup(&lm) rather than bedFreeList. */
{
char *fields = NULL;
struct sqlResult *sr;
struct hTableInfo *hti;
struct bed *bedList=NULL, *bed;
char **row;
int fieldCount;
boolean isPsl, isGenePred, isBedWithBlocks;
boolean pslKnowIfProtein = FALSE, pslIsProtein = FALSE;
struct sqlConnection *conn = NULL;
char *dbTable = NULL;
if (isCustomTrack(table))
{
struct customTrack *ct = ctLookupName(table);
dbTable = ct->dbTableName;
conn = hAllocConn(CUSTOM_TRASH);
hti = hFindTableInfo(CUSTOM_TRASH, region->chrom, dbTable);
}
else
{
dbTable = table;
struct trackDb *tdb;
if(sameWord(db, database))
tdb = tdbForTrack(db, table, &fullTrackList);
else
tdb = hTrackDbForTrack(db, table);
conn = (tdb ? hAllocConnTrack(db, tdb) : hAllocConn(db));
hti = hFindTableInfo(db, region->chrom, table);
}
if (hti == NULL)
errAbort("Could not find table info for table %s.%s", db,table);
if (isWiggle(db, table))
{
bedList = getWiggleAsBed(db, table, region, filter, idHash, lm, conn);
fieldCount = 4;
}
else
{
bedSqlFieldsExceptForChrom(hti, &fieldCount, &fields);
isPsl = htiIsPsl(hti);
isGenePred = sameString("exonEnds", hti->endsSizesField);
isBedWithBlocks = (
(sameString("chromStarts", hti->startsField) ||
sameString("blockStarts", hti->startsField))
&& sameString("blockSizes", hti->endsSizesField));
/* All beds have at least chrom,start,end. We omit the chrom
* from the query since we already know it. */
sr = regionQuery(conn, dbTable, fields, region, TRUE, filter);
while (sr != NULL && (row = sqlNextRow(sr)) != NULL)
{
/* If have a name field apply hash filter. */
if (fieldCount >= 4 && idHash != NULL)
if (!hashLookup(idHash, row[2]))
continue;
bed = bedFromRow(region->chrom, row, fieldCount, isPsl, isGenePred,
isBedWithBlocks, &pslKnowIfProtein, &pslIsProtein, lm);
slAddHead(&bedList, bed);
}
freez(&fields);
sqlFreeResult(&sr);
slReverse(&bedList);
}
hFreeConn(&conn);
if (retFieldCount)
*retFieldCount = fieldCount;
return(bedList);
}
