Add --until and --sort sweep flags to analyze/label

[ashleyzhang01]

Add --until flag to analyze/label for backfilling a specific date range (e.g. a missed day). Add --sort sweep mode that orders by assembled_at (analyze) or analyzed_at (label) DESC instead of bot_reviewed_at. Since we only analyze merged PRs, ones that merge long after discovery fall outside the --since window of the primary run; the sweep pass catches these stragglers in addition to the freshness-first processing.

[Copilot]

Pull request overview

This PR extends the DB-backed analyze and label CLI subcommands to support bounded backfills and an alternate “sweep” prioritization order, aimed at catching PRs that become eligible long after their original discovery/review timestamp.

Changes:

• Added --until (exclusive upper bound) to analyze/label, including a shared _parse_time_bound helper to normalize CLI time inputs (relative days, bare dates).
• Added --sort {reviewed,sweep} to analyze/label, and introduced new SQL query variants to support sweep ordering.
• Added/extended SQLite integration tests covering until bounds and sweep ordering behavior.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
online/etl/tests/test_repository.py	Adds integration tests for until bounds and analyze sweep ordering.
online/etl/tests/test_main.py	Adds unit tests for the new _parse_time_bound CLI helper.
online/etl/pipeline/label.py	Plumbs until/sort_by into the labeling pipeline call to the repository.
online/etl/pipeline/analyze.py	Plumbs until/sort_by into the analysis pipeline call to the repository.
online/etl/main.py	Adds --until/--sort flags to CLI and centralizes time-bound parsing.
online/etl/db/repository.py	Extends repository query APIs with until + sort_by branching.
online/etl/db/queries.py	Adds bounded and sweep-ordered SQL query variants for analyze/label selection.
online/etl/db/connection.py	Updates SQLite param translation to strip Postgres ::type casts used in new bounded queries.

Comments suppressed due to low confidence (1)

online/etl/db/repository.py:334

• sort_by == "sweep" bypasses the since/until bounds here too, meaning label --since/--until --sort sweep will label PRs outside the requested reviewed-at window. Align the sweep branch with the bounded logic (e.g., add bounded+sorted queries ordered by p.analyzed_at DESC while still filtering on p.bot_reviewed_at).

        if sort_by == "sweep":
            if chatbot_id is not None:
                return await self.db.fetchall(
                    q.GET_ANALYZED_NOT_LABELED_BY_ASSEMBLED, (chatbot_id, limit)
                )
            return await self.db.fetchall(
                q.GET_ALL_ANALYZED_NOT_LABELED_BY_ASSEMBLED, (limit,)
            )
        if until is not None:

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[christama]

Config