Nithurshen opened a new pull request, #18901: URL: https://github.com/apache/datafusion/pull/18901
## Which issue does this PR close? - Closes #18795 ## Rationale for this change Currently, scanning CSV files does not correctly report the `elapsed_compute` metric in `BaselineMetrics`. When running `EXPLAIN ANALYZE` on a CSV table, the `elapsed_compute` time is reported as negligible (e.g., `1ns` or `42ns`), even for large files, because the time spent parsing the CSV data is not being measured. This PR ensures that the time spent reading and parsing CSV batches is correctly accounted for in the execution metrics. ## What changes are included in this PR? - Modified `datafusion/datasource-csv/src/source.rs`. - Updated `CsvOpener` to store the `partition_index`. - Initialized `BaselineMetrics` within C`svOpener::open`. - Wrapped the underlying CSV reader iterator using `std::iter::from_fn`. This allows us to start a timer before calling `reader.next()` and stop it immediately after, ensuring the CPU time spent parsing the CSV is captured in `elapsed_compute`. ## Are these changes tested? <!-- We typically require tests for all PRs in order to: 1. Prevent the code from being accidentally broken by subsequent changes 2. Serve as another way to document the expected behavior of the code If tests are not included in your PR, please explain why (for example, are they covered by existing tests)? --> Yes. Existing unit tests passed (`cargo test -p datafusion-datasource-csv`). Additionally, I verified the fix manually using `datafusion-cli` and `EXPLAIN ANALYZE`. **Before the fix:** elapsed_compute was reported as 85ns (effectively zero). ``` DataSourceExec: ... metrics=[output_rows=3, elapsed_compute=85ns, ... time_elapsed_processing=1.54ms ...] ``` **After the fix**: elapsed_compute is reported as 1.11ms, accurately reflecting the parsing time. ``` DataSourceExec: ... metrics=[output_rows=3, elapsed_compute=1.11ms, ... time_elapsed_processing=1.26ms ...] ``` ## Are there any user-facing changes? Users viewing `EXPLAIN ANALYZE` output for queries involving CSV files will now see accurate `elapsed_compute` values instead of near-zero values. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
