razajafri commented on a change in pull request #31284:
URL: https://github.com/apache/spark/pull/31284#discussion_r572286331
##########
File path:
sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/VectorizedColumnReader.java
##########
@@ -147,7 +150,24 @@ public VectorizedColumnReader(
DictionaryPage dictionaryPage = pageReader.readDictionaryPage();
if (dictionaryPage != null) {
try {
- this.dictionary =
dictionaryPage.getEncoding().initDictionary(descriptor, dictionaryPage);
+ PrimitiveType primitiveType = descriptor.getPrimitiveType();
+ if (primitiveType.getOriginalType() == OriginalType.DECIMAL &&
+ primitiveType.getDecimalMetadata().getPrecision() <=
Decimal.MAX_INT_DIGITS() &&
+ primitiveType.getPrimitiveTypeName() == INT64) {
+ // We need to make sure that we initialize the right type for the
dictionary otherwise
+ // Encoding#initDictionary will initialize it to
PrimitiveTypeName.INT64 and
+ // WritableColumnVector will throw an exception when trying to
decode to an Int when the
+ // dictionary is in fact initialized as Long
+ PrimitiveType adjustedType = new
PrimitiveType(primitiveType.getRepetition(),
+ INT32, primitiveType.getTypeLength(),
+ primitiveType.getName(), primitiveType.getOriginalType(),
+ primitiveType.getDecimalMetadata(), primitiveType.getId());
+ ColumnDescriptor desc = new ColumnDescriptor(descriptor.getPath(),
+ adjustedType, descriptor.getMaxRepetitionLevel(),
descriptor.getMaxDefinitionLevel());
+ this.dictionary = dictionaryPage.getEncoding().initDictionary(desc,
dictionaryPage);
Review comment:
Thanks @cloud-fan. You have helped me uncover a bug in my code.
The dictionary is being written as INT64 so just by reading it as an INT32
is not the right solution. I will have to improve my test so this is tested
properly. The solution will be to read the dictionary as Long and cast it
instead of decoding it directly to Int.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
