Files
iptvnator/apps
4grayandClaude Fable 5 7a86485b4c fix(epg): deduplicate program rows in SQL before applying row limits
Codex follow-up: the JS-level slot dedup ran after the SQL LIMIT, so
duplicate cross-source rows consumed the cap and truncated real data.
Moved the dedup into SQL with GROUP BY, applied before the limits:

- selectChannelPrograms / selectLegacyChannelPrograms: GROUP BY
  (channel_id, start, title) before ORDER BY start LIMIT 500, so the
  timeline cap counts distinct programmes rather than duplicate rows
- selectCurrentProgramsForChannelIds: GROUP BY channel_id before
  LIMIT channelIds.length, so duplicate cross-source current slots can't
  starve other channels of their current-programme preview

The JS toEpgPrograms() dedup stays as a safety net (e.g. legacy NULL-source
rows the unique index treats as distinct). Test query-chain mocks updated
for the new groupBy link.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-19 19:45:59 +02:00
..