clairemcginty commented on code in PR #3098:
URL: https://github.com/apache/parquet-java/pull/3098#discussion_r1922778199
##########
parquet-hadoop/src/main/java/org/apache/parquet/filter2/statisticslevel/StatisticsFilter.java:
##########
@@ -217,6 +219,70 @@ public <T extends Comparable<T>> Boolean visit(Contains<T>
contains) {
return contains.filter(this, (l, r) -> l || r, (l, r) -> l && r, v ->
BLOCK_MIGHT_MATCH);
}
+ @Override
+ public Boolean visit(Size size) {
+ final ColumnChunkMetaData metadata =
getColumnChunk(size.getColumn().getColumnPath());
+ if (metadata == null) {
+ // the column isn't in this file, so fail eq/gt/gte targeting size > 0
+ final boolean blockCannotMatch =
+ size.filter((eq) -> eq > 0, (lt) -> false, (lte) -> false, (gt) ->
gt >= 0, (gte) -> gte > 0);
+ return blockCannotMatch ? BLOCK_CANNOT_MATCH : BLOCK_MIGHT_MATCH;
+ }
+
+ final SizeStatistics stats = metadata.getSizeStatistics();
+ final List<Long> repetitionLevelHistogram =
stats.getRepetitionLevelHistogram();
+ final List<Long> definitionLevelHistogram =
stats.getDefinitionLevelHistogram();
+
+ if (repetitionLevelHistogram.isEmpty() ||
definitionLevelHistogram.isEmpty()) {
+ return BLOCK_MIGHT_MATCH;
+ }
+
+ // If all values have repetition level 0, then no array has more than 1
element
+ if (repetitionLevelHistogram.size() == 1
+ || repetitionLevelHistogram.subList(1,
repetitionLevelHistogram.size()).stream()
+ .allMatch(l -> l == 0)) {
+
+ // Null list fields are treated as having size 0
+ if (( // all lists are nulls
+ definitionLevelHistogram.subList(1,
definitionLevelHistogram.size()).stream()
+ .allMatch(l -> l == 0))
+ || // all lists are size 0
+ (definitionLevelHistogram.get(0) == 0
+ && definitionLevelHistogram.subList(2,
definitionLevelHistogram.size()).stream()
+ .allMatch(l -> l == 0))) {
+
+ final boolean blockCannotMatch =
+ size.filter((eq) -> eq > 0, (lt) -> false, (lte) -> false, (gt) ->
gt >= 0, (gte) -> gte > 0);
+ return blockCannotMatch ? BLOCK_CANNOT_MATCH : BLOCK_MIGHT_MATCH;
+ }
+
+ long maxDefinitionLevel =
definitionLevelHistogram.get(definitionLevelHistogram.size() - 1);
+
+ // If all repetition levels are zero and all definitions level are >
MAX_DEFINITION_LEVEL - 1, all lists
+ // are of size 1
+ if (definitionLevelHistogram.stream().allMatch(l -> l >
maxDefinitionLevel - 1)) {
+ final boolean blockCannotMatch = size.filter(
+ (eq) -> eq != 1, (lt) -> lt <= 1, (lte) -> lte < 1, (gt) -> gt >=
1, (gte) -> gte > 1);
+
+ return blockCannotMatch ? BLOCK_CANNOT_MATCH : BLOCK_MIGHT_MATCH;
+ }
+ }
+ long nonNullElementCount =
+ repetitionLevelHistogram.stream().mapToLong(l -> l).sum() -
definitionLevelHistogram.get(0);
+ long numNonNullRecords = repetitionLevelHistogram.get(0) -
definitionLevelHistogram.get(0);
+
+ // Given the total number of elements and non-null fields, we can compute
the max size of any array field
+ long maxArrayElementCount = 1 + (nonNullElementCount - numNonNullRecords);
+ final boolean blockCannotMatch = size.filter(
+ (eq) -> eq > maxArrayElementCount,
+ (lt) -> false,
+ (lte) -> false,
+ (gt) -> gt >= maxArrayElementCount,
+ (gte) -> gte > maxArrayElementCount);
+
+ return blockCannotMatch ? BLOCK_CANNOT_MATCH : BLOCK_MIGHT_MATCH;
Review Comment:
good question... I updated the unit tests to test out various nesting levels
and they all passed:
https://github.com/apache/parquet-java/pull/3098/commits/39981e9f8eeec2f8492fbfcb35e997aa299af18d
--
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: issues-unsubscr...@parquet.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@parquet.apache.org
For additional commands, e-mail: issues-h...@parquet.apache.org
