richardstartin commented on code in PR #10254:
URL: https://github.com/apache/pinot/pull/10254#discussion_r1101868818
##########
pinot-core/src/main/java/org/apache/pinot/core/operator/filter/predicate/PredicateUtils.java:
##########
@@ -139,10 +143,40 @@ public static IntSet getDictIdSet(BaseInPredicate
inPredicate, Dictionary dictio
}
break;
case STRING:
- for (String value : values) {
- int dictId = dictionary.indexOf(value);
- if (dictId >= 0) {
- dictIdSet.add(dictId);
+ if (values.size() >
RequestContextUtils.THRESHOLD_OF_LARGE_IN_CLAUSE_VALUES) {
+ if (bloomFilter != null) {
+ for (String value : values) {
+ if (bloomFilter.mightContain(value)) {
+ int dictId = dictionary.indexOf(value);
+ if (dictId >= 0) {
+ dictIdSet.add(dictId);
+ }
+ }
+ }
+ } else {
+ int valueIdx = 0;
+ int dictIdx = 0;
+ while (valueIdx < values.size() && dictIdx < dictionary.length()) {
+ String value = values.get(valueIdx);
+ String dictValue = dictionary.getStringValue(dictIdx);
+ int compareResult = value.compareTo(dictValue);
Review Comment:
I think batching in the dictionary should give some nice opportunities to
optimise things, control thresholds etc.
```java
void getDictIds(List<String> strings, IntConsumer dictIds) {
if (strings.size() < SPARSE_THRESHOLD) {
for (String string : strings) {
int dictId = indexOf(string);
if (dictId >= 0) dictIds.acceptInt(dictId);
}
} else {
int i, j = 0;
byte[] utf8 = null;
boolean needNewUtf8 = true;
while (i < strings.size() && j < length()) {
if (needNewUtf8) {
utf8 = strings.get(i).getBytes(UTF_8);
needNewUtf8 = true;
}
int comparison = _valueReader.compareUtf8Bytes(j,
_numBytesPerValue, utf8);
if (comparison == 0) {
dictIds.acceptInt(j);
j++;
i++;
} else if (comparison < 0) {
i++;
} else {
j++;
needNewUtf8 = false;
}
}
}
}
...
BitSet dictIds = new BitSet();
dictionary.getDictIds(sortedValues, dictIds::add);
```
I think the bloom filter threshold should be managed outside the dictionary,
though.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org
