supabase / supabase / 29101184120

74%

push

github

test(sql-editor): characterization tests for SQLEditor (decompose 1/6) (#47820)

## Summary

Add some tests for the SQL editor so I can refactor it without
regressions. Tests are not best practice because they are intended to be
temporary and improving them would require refactoring first (currently
they are over-mocking and asserting on internal details).

Stacked on top of #47792 (`charislam/sql-editor-top-bar-controls`).

## What this adds

`apps/studio/tests/components/SQLEditor/SQLEditor.test.tsx` (11 tests):

- Run success → `addResult` + Results tab; EXPLAIN-shaped result
auto-switches to the explain tab; a non-EXPLAIN run switches back.
- Run error with `position` → error-highlight line math +
`deltaDecorations` + `revealLineInCenter`; the next run clears the
highlight.
- Run button refocuses the editor; disabled + short-circuits while a
diff is open.
- Diff request queued before mount drains exactly once (one-shot; no
re-apply on remount).
- Ask-AI widget renders only while the prompt is open (render-time
`editorRef.current` read).
- Destructive query → warning modal → confirm forces the re-run;
confirm-with-RLS appends enable-RLS statements.

## Test approach

Real Monaco / DiffEditor are replaced with lightweight fakes exposing a
controllable editor; child panels + orthogonal context hooks are
stubbed; the execute mutation runs for real against an MSW-mocked
`/platform/pg-meta/:ref/query`. Tests assert on public behavior so they
survive the internal refactor unchanged.

## Verification

- `pnpm --filter studio exec vitest run
tests/components/SQLEditor/SQLEditor.test.tsx` — 11/11 pass (stable
across repeated runs)
- `pnpm --filter studio typecheck` — clean
- `eslint` — 0 errors

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Tests**
* Added comprehensive coverage for SQL editor behavior, including query
execution, result and explain views, error highlighting, editor focus,
and diff ... (continued)

1197 of 1655 branches covered (72.33%)

Branch coverage included in aggregate %.

1771 of 2359 relevant lines covered (75.07%)

246.79 hits per line