[Lohith248]

Summary

Make curate(..., output=None) and CLI --dry-run accept saved SQL ending in ;, a final line comment, or both. These queries already write manifests successfully; the dry-run count wrapper introduces their syntax errors.

 dry_run(sql)
   _reject_tenant_sql(sql)
-  execute("SELECT count(*) FROM (" + sql + ")")
+  connection.sql(sql).aggregate("count(*)").fetchone()

Count the parsed DuckDB relation after the unchanged statement gate. Extend existing SQL-surface and CLI tests, including --sql-file.

Evidence

• Before: five regression cases fail on main 43380ac; both CLI input modes exit 2. Four selected controls pass, including manifest writes.
  After: 183 catalog/curation tests pass, including existing non-single-SELECT refusal cases.

• A real synthetic MCAP → App.test(record=True) → catalog → saved SQL-file run previously failed dry-run but wrote a one-row manifest. Afterward, both CLI modes exit 0 with one row and joints: 1/1 (100%). Dry-run creates no manifest.

• Exact validation from the repository root:

  uv run --locked pytest tests/test_catalog_curation.py -q
  UV_PROJECT_ENVIRONMENT="$HOME/.cache/hflow-onboarding/venv311" uv run --python 3.11 --locked pytest tests/test_catalog_curation.py -q -k 'sql_surface or dry_run or reports_duckdb_execution_errors'
  uv run --locked pytest -q -n 4 -rs
  uv run --locked ruff check --fix
  uv run --locked ruff format
  uv run --locked ty check
  git diff --check

  Python 3.14 full suite: 2,564 passed, 8 skipped. Python 3.11 focused suite: 42 passed, 141 deselected in a separate environment. Skips require network opt-in, model assets, Docker, or a live bucket. Quality checks passed; full-suite prerequisites included FFmpeg 8.1 and a C compiler.

Merge Danger

Door: two-way

Revertible without data migration.

Blast Radius: curation

Only report-only counts change. Statement validation and constrained connection settings remain in place. No dependencies, stored formats, canonical bytes, or manifest-publication changes.

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

[greptile-apps]

Confidence Score: 5/5

[Medium impact] The PR appears safe to merge; no actionable issues were found.

Summary

Changes dry-run counting to use the parsed DuckDB relation instead of wrapping SQL text.

• Dry-run curation counts queries with trailing semicolons or comments.

No actionable issues found. Tests were inspected, not run; DuckDB is unavailable in the review environment.

_{Reviews (1) · Last reviewed commit: "fix(curation): count parsed queries duri..." · Reviewed by Greptile}

[kstonekuan]

Thanks for the fix. Ruff, formatting, ty, and the full suite passed (2564 passed, 8 skipped). LGTM.