@ruleprint/collector-tests

Derives derived rules from vitest/jest test files: every it/test becomes one rule whose title is the chain of enclosing describe titles plus the test title, joined by >.

import { testsCollector } from '@ruleprint/collector-tests';

const candidates = await testsCollector.collect(
  { path: 'test/shipping.spec.ts', content },
  { warn: console.warn },
);

What it recognises

| Shape | Result | | --------------------------------------------------------------- | ---------------------------------------- | | describe / suite / context | a level in the title chain | | it / test | one rule | | .skip, .only, .concurrent, .sequential, .fails | same as the plain call | | .each(...)('title %s'), .for, .runIf(...), .skipIf(...) | one rule, title kept raw (%s, $name) | | .todo('...') | ignored: there is no body to derive from | | template-literal titles | kept raw, e.g. exactly ${LIMIT} items | | empty or non-literal titles | ignored |

Files are matched by name: *.test.*, *.spec.* and anything under __tests__/, with .js, .jsx, .ts, .tsx and their .cjs/.mjs variants. .tsx/.jsx files use the TSX grammar; everything else uses the TypeScript grammar.

Limitations

• Detection is by identifier name. import { it as spec } is not followed (see ADR-0003).
• On a syntax error the file is still processed and ctx.warn is called once, but tree-sitter's recovery may drop the enclosing describe, leaving only the leaf title.
• lastRunStatus is not filled in; wiring the vitest reporter is a later milestone.

Dogfooding

pnpm --filter @ruleprint/collector-tests dogfood <dir> [--json]

Walks <dir> (skipping node_modules, dist, coverage, .git) and prints every candidate, grouped by file. --json prints the raw RuleCandidate[].