Misaka_Company 3b0374aa31 perf: enable fast_executemany and eliminate duplicate Excel reads ...

- Set cursor.fast_executemany = True (pyodbc 5.x moved it to cursor)
- merge_sources() now returns source row counts, avoiding re-reading files
- Single commit at end of batch_insert instead of per-batch commits
- Pre-build all rows before SQL execution
- Result: PG_2022 migration improved from ~5min to ~36s (7.6x faster)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-06-01 14:33:05 +08:00

17 KiB

Raw Blame History

View Raw