Python ETLEasyofficialBenchmark eligiblefree
Overlapping page dedup
The extractor retried page 3 and wrote duplicates. Destination must be unique on source_id; rerunning the job must not increase counts.
| Field | Detail |
|---|---|
| Creator | PipelineArena |
| Version | 2.0.0 |
| Time | 35 min |
| Scored checks | 11 |
Objective
Unique source_id; retry is a no-op.
| # | Success condition |
|---|---|
| 1 | UNIQUE(source_id) |
| 2 | Overlap pages produce one row |
| 3 | Second job run: same checksum |
| # | Visible instruction |
|---|---|
| 1 | Load the fixture pages as-is (they overlap). |
| 2 | Upsert on source_id. |
| 3 | Do not assume pages are disjoint. |
Scoring checks the leftover tables or files, replay safety, and seeded variations — not the exact syntax you used. Check implementation is never shown to participants.
Signed-in workspace: Open in app