yeya24 commented on PR #8315:
URL: https://github.com/apache/arrow-rs/pull/8315#issuecomment-3282706933

   Thanks @alamb, I have addressed your comments by exposing predicates via 
methods.
   
   For our usecase, we basically need a `ParquetRecordBatchStreamReader` like 
reader. Instead of reading and materializing the final results, it only filters 
the rows based on Row Filter and Row selection and returns the final rows that 
match the predicates in the Parquet file. The final result can be either 
`RowSelection` or a boolean array. Then our own customized reader can read and 
materialize the rows based on the filter results.
   
   If we are open to adding a reader for this usecase, then we are happy. If 
not, then we have to reimplement a reader which requires a lot of other structs 
and methods to be exposed as public. I can list some:
   - 
[ReadPlanBuilder](https://github.com/apache/arrow-rs/blob/main/parquet/src/arrow/arrow_reader/read_plan.rs#L32)
   - 
[InMemoryRowGroup](https://github.com/apache/arrow-rs/blob/aa626e12de8bc0d0f56b5349239cae1be8d1a195/parquet/src/arrow/async_reader/mod.rs#L967)
 and relevant fields like 
[ColumnChunkData](https://github.com/apache/arrow-rs/blob/aa626e12de8bc0d0f56b5349239cae1be8d1a195/parquet/src/arrow/async_reader/mod.rs#L1133)
   
   Would you be open to expose those specific fields or we are open to add a 
new reader that only filters and returns a boolean array? Thanks


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to