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);
}
void tjLoadSome(struct region *regionList,
struct joinedTables *joined, int fieldOffset, int keyOffset,
char *idField, struct hash *idHash,
struct slName *chopBefore, struct slName *chopAfter,
struct tableJoiner *tj, boolean isPositional, boolean isFirst)
/* Load up rows. */
{
struct region *region;
struct dyString *sqlFields = dyStringNew(0);
struct joinerDtf *dtf;
struct slRef *ref;
struct joinerPair *jp;
int fieldCount = 0, keyCount = 0;
int idFieldIx = -1;
struct sqlConnection *conn = hAllocConn(tj->database);
char *identifierFilter = NULL;
char *filter;
boolean needUpdateFilter = FALSE;
struct joinedRow *jr;
if (isFirst)
identifierFilter = identifierWhereClause(idField, idHash);
filter = filterClause(tj->database, tj->table, regionList->chrom, identifierFilter);
/* Record combined filter. */
// Show only the SQL filter built from filter page options, not identifierFilter,
// because identifierFilter can get enormous (like 126kB for 12,500 rsIDs).
char *filterNoIds = filterClause(tj->database, tj->table, regionList->chrom, NULL);
if (filterNoIds != NULL)
{
if (joined->filter == NULL)
joined->filter = dyStringNew(0);
else
dyStringAppend(joined->filter, " AND ");
dyStringAppend(joined->filter, filterNoIds);
if (!isFirst)
{
needUpdateFilter = TRUE;
for (jr = joined->rowList; jr != NULL; jr = jr->next)
jr->hitThisTable = FALSE;
}
}
/* Create field spec for sql - first fields user will see, and
* second keys if any. */
for (dtf = tj->fieldList; dtf != NULL; dtf = dtf->next)
{
struct joinerDtf *dupe = joinerDtfClone(dtf);
slAddTail(&joined->fieldList, dupe);
dyStringAddList(sqlFields, dtf->field);
++fieldCount;
}
for (ref = tj->keysOut; ref != NULL; ref = ref->next)
{
struct joinerDtf *dupe;
jp = ref->val;
dupe = joinerDtfClone(jp->a);
slAddTail(&joined->keyList, dupe);
dyStringAddList(sqlFields, jp->a->field);
++keyCount;
}
if (idHash != NULL)
{
if (idField == NULL)
internalErr();
idFieldIx = fieldCount + keyCount;
dyStringAddList(sqlFields, idField);
}
for (region = regionList; region != NULL; region = region->next)
{
char **row;
/* We free at end of loop so we get new one for each chromosome. */
char *filter = filterClause(tj->database, tj->table, region->chrom, identifierFilter);
struct sqlResult *sr = regionQuery(conn, tj->table,
sqlFields->string, region, isPositional, filter);
while (sr != NULL && (row = sqlNextRow(sr)) != NULL)
{
if (idFieldIx < 0)
{
if (jrRowAdd(joined, row, fieldCount, keyCount) == NULL)
break;
}
else
{
char *id = row[idFieldIx];
if (isFirst)
{
if (hashLookup(idHash, id))
{
if (jrRowAdd(joined, row, fieldCount, keyCount) == NULL)
break;
}
}
else
{
struct hashEl *bucket;
id = chopKey(chopBefore, chopAfter, id);
for (bucket = hashLookup(idHash, id); bucket != NULL;
bucket = hashLookupNext(bucket))
{
jr = bucket->val;
jr->hitThisTable = TRUE;
jrRowExpand(joined, jr, row,
fieldOffset, fieldCount, keyOffset, keyCount);
}
}
}
}
sqlFreeResult(&sr);
freez(&filter);
if (!isPositional)
break;
}
if (isFirst)
slReverse(&joined->rowList);
if (needUpdateFilter)
{
for (jr = joined->rowList; jr != NULL; jr = jr->next)
{
jr->passedFilter &= jr->hitThisTable;
}
}
tj->loaded = TRUE;
hFreeConn(&conn);
}
