fix(export): prefix CSV data keys that collide with id/timestamp/duration

[TimeToBuildBob]

Follow-up to the Greptile note on #178: the bucket CSV export (#179) writes id,timestamp,duration followed by the union of event data keys, so a data key named id, timestamp or duration produced a duplicate header name.

A data key that would repeat an earlier header name now gets a data. prefix, repeated until unique (duration → data.duration; id next to a literal data.id → data.data.id). Only the header changes; values are still read by the original key. Non-colliding keys and the >32-key single data column are unchanged. Same scheme as ActivityWatch/aw-server-rust#795, so both servers keep emitting the same header.

Tests: test_export_bucket_csv_prefixes_colliding_data_keys (API round-trip) and test_csv_header_prefix_repeats_until_unique; tests/test_server.py 15 passed, ruff and mypy clean.

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Changes how CSV export column names handle collisions.

This PR appears safe to merge; no actionable issues were found.

What we checked:

• Renamed columns keep their values: _header_names returns a separate list. Rows still use the original data_keys, so each renamed column keeps its matching value.
• Wide exports keep their format: Wide exports pass only data to _header_names. That name does not collide with the fixed columns, and rows still contain one JSON value.

Summary

CSV exports now prefix colliding event-data headers with data., repeating the prefix until the name is unique.

• Values still come from the original data keys.
• The single JSON column for exports with more than 32 keys stays unchanged.
• New tests cover reserved names and repeated prefixes.

_{Reviews (1) · Last reviewed commit: "fix(export): prefix CSV data keys that c..."}

[TimeToBuildBob]

Status check: Greptile reviewed at 5/5 (no actionable issues). CI: ubuntu, windows, lint, format, and typecheck all pass. macOS runner has been queued/running for ~1.5h — appears slow on GitHub's side, not a test failure. Ready for maintainer review and merge when macOS finishes.

[TimeToBuildBob]

CI-green and mergeable (Greptile 5/5) — waiting only on a maintainer click.

This PR is ready to merge, but the bot has pull-only access to this repo and can't self-merge — surfacing it here so it isn't lost. The monitoring loop will stop re-flagging it now that this note is posted.

[TimeToBuildBob]

🤖 AI code review

This PR changes the CSV export header generation in aw_server/csv_export.py by adding a _header_names() function that prefixes data keys colliding with the fixed id/timestamp/duration columns (or with earlier data keys) with a repeated 'data.' prefix until unique. The events_to_csv() function now uses this for the header row, while values are still read by the original data key. Two tests are added: an API round-trip test for colliding keys and a unit test for the prefix-repeats-until-unique behavior.

Safe to merge — no P0/P1 findings

Confidence 5/5

✅ No findings. The diff looks correct to me on this pass.

Files changed (2) — the diff as I read it

• aw_server/csv_export.py — Adds _header_names() to generate unique CSV header names by prefixing colliding data keys with 'data.' repeatedly, and uses it in events_to_csv().
• tests/test_server.py — Adds test_export_bucket_csv_prefixes_colliding_data_keys and test_csv_header_prefix_repeats_until_unique to cover the new header naming.

Previous review passes

commit	score	findings	engine	when
12213131270e	5/5	0	llm	2026-10-07 09:43 UTC
1d2e92699b28	5/5	0	subscription	2026-10-10 12:08 UTC

_{Reviewed 1d2e92699b28 · openrouter/deepseek/deepseek-v4-flash-0731 · llm engine · 107s · about this reviewer}

_{Maintainer commands}

_{@TimeToBuildBob review (own line) — fresh review · @TimeToBuildBob fix — a worker acts on the findings. Once per comment; 👀 = received.}

[TimeToBuildBob]

The conflict is already resolved in 1d2e926: the earlier rebase preserved both the heartbeat-cache tests from master and the CSV collision tests. Verified the current head is mergeable and all six CI checks pass. No further rebase or force-push needed; awaiting maintainer merge.

[TimeToBuildBob]

Rebased onto current master and pushed 4f031cd. Resolved the overlap in tests/test_server.py by retaining both master's bulk-delete/heartbeat-cache regressions and this PR's CSV collision tests; CSV behavior is unchanged.

Verification: the five test modules used by make test pass (94 tests), Ruff 0.15.22 (matching CI) and Black pass on both PR files, and the range-diff shows only the test insertion context changed. GitHub now reports the head mergeable; fresh CI is pending.