I think when prefetching is enabled (which is probably a setting not relevant to ParquetExec anyway we end up with a DataSourceRDD in the scan instead of a FileScanRDD. This extracts the partitions from that RDD type.
Total number of tests run: 754
Suites: completed 32, aborted 0
Tests: succeeded 609, failed 145, canceled 1, ignored 46, pending 0
I think when prefetching is enabled (which is probably a setting not relevant to ParquetExec anyway we end up with a DataSourceRDD in the scan instead of a FileScanRDD. This extracts the partitions from that RDD type.