feat: recover affected tests and extension mapping

[lzehrung]

Summary

• recover the unmerged commits from P1: add affected tests command #146 and its stacked extension-mapping work on top of current main
• add codegraph affected for deterministic changed-file-to-test selection, including deleted-source traversal
• add configurable literal suffix-to-language mappings through CLI, sessions, graph building, caches, and docs
• integrate both features with the current lazy CLI dispatcher and warm incremental index

Verification

• npm run check (231 files, 2,707 tests passed, 19 skipped)
• focused affected/config/session/cache/graph suites: 198 tests passed
• smoke: node ./dist/cli.js affected src/cli.ts --root . --quiet --cache memory returned five candidate test paths

Draft for review; do not merge automatically.

[Copilot]

Pull request overview

Adds a deterministic “affected tests” workflow and restores/extends extension-to-language mapping support so custom literal suffixes participate consistently in discovery, indexing, graph building, cache keys, and graph delta behavior.

Changes:

• Introduces codegraph affected (CLI plumbing + help/docs) with JSON/quiet output, depth-limited reverse-dependency traversal, filtering, stdin input, and git base/head diff input (including deleted-path traversal).
• Adds languages.extensions in codegraph.config.json plus normalization/validation, and threads languageExtensions through sessions/agent sessions/indexer/graph builder/SQL edge collection.
• Updates cache/manifest compatibility and graph-delta computation so mapping changes invalidate/recompute the right artifacts, with extensive new/updated tests.

Reviewed changes

Copilot reviewed 38 out of 38 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tests/session.test.ts	Verifies session build-option normalization and config-mapping application/reuse behavior.
tests/project-file-discovery.test.ts	Adds POSIX-only coverage for case-sensitive ignore globs (including symlink crawling).
tests/graph-delta.test.ts	Ensures graph delta reports edge changes when mappings change without file edits.
tests/codegraph-config.test.ts	Adds validation/normalization tests for languages.extensions and cache-option comparisons.
tests/cli-command-modules.test.ts	Ensures command module handlers forward languageExtensions correctly.
tests/cache-invalidation.test.ts	Covers incremental discovery and graph/cache recomputation when mappings change.
tests/agent-session.test.ts	Verifies agent session discovery/indexing honors config/programmatic mappings.
tests/affected.test.ts	New end-to-end coverage for codegraph affected behaviors and outputs.
src/sql/sourceGraph.ts	Makes SQL corpus/edge collection respect languageExtensions mapping.
src/session.ts	Normalizes session identity to treat language-extension mappings consistently.
src/review/deleted.ts	Generalizes deleted-import importer discovery and re-derives test importers via filtering.
src/languages/filePrep.ts	Threads languageExtensions into file preparation so remapped suffixes parse correctly.
src/languages.ts	Adds mapping normalization/validation helpers and extends supportForFile/languageForFile.
src/indexer/types.ts	Extends BuildOptions with languageExtensions and re-exports relevant types.
src/indexer/parse-context.ts	Passes languageExtensions into file preparation for indexing.
src/indexer/imports.ts	Passes languageExtensions through import collection path.
src/indexer/build-workers.ts	Threads languageExtensions into worker-side file preparation.
src/indexer/build-index.ts	Extends discovery patterns, cache signatures, SQL handling, and graph delta for mappings.
src/indexer/build-cache/options.ts	Normalizes/compares language-extension mappings for manifest compatibility.
src/indexer/build-cache/module-cache.ts	Ensures bloom filter build uses mapped language support.
src/indexer/build-cache.ts	Re-exports normalizeLanguageExtensions from build-cache surface.
src/graph-edge-collector.ts	Uses mapped language support for SQL edge routing and file prep.
src/graph-builder.ts	Ensures SQL fact cache and edge collection honor languageExtensions.
src/config.ts	Adds languages.extensions config parsing + validation + normalization.
src/cli/options.ts	Registers affected schema and CLI value parsing for --filter.
src/cli/inspect.ts	Threads languageExtensions into inspect/hotspots and language counts.
src/cli/index.ts	Threads languageExtensions into index command build options.
src/cli/help.ts	Adds affected help text and updates suggested commands/examples.
src/cli/graphDelta.ts	Forwards languageExtensions into graph-delta handler options.
src/cli/commandCatalog.ts	Registers affected in command catalog.
src/cli/affected.ts	Implements the affected command (inputs, traversal, filtering, outputs).
src/cli.ts	Wires affected into dispatcher; propagates config mappings into CLI build options and discovery patterns.
src/agent/session.ts	Extends agent session file planning/build options with languageExtensions + discovery patterns.
README.md	Documents affected as the recommended “which tests should I run?” workflow.
docs/plans/2026-07-03-16-config-extension-mapping.md	Updates plan notes for literal suffix rules and non-remappable SFC suffixes.
docs/library-api.md	Documents passing languageExtensions from config into programmatic builds.
docs/cli.md	Documents languages.extensions config and the new affected command usage/semantics.
codegraph-skill/codegraph/SKILL.md	Updates skill surface/docs to include affected and mapping behavior.

---

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

[Copilot]

Pull request overview

Copilot reviewed 38 out of 38 changed files in this pull request and generated no new comments.

Suppressed comments (2)

src/indexer/build-index.ts:596

• Filtering SQL files now calls supportForFile() for every project file, which is significantly more expensive than the prior extname check (it normalizes/sorts extension maps and may read samples for .h files). When no languageExtensions are configured (common case), this is avoidable overhead during indexing.

  const sqlFiles = normalizedFiles
    .filter((file) => supportForFile(file, opts?.languageExtensions)?.id === "sql")
    .sort((left, right) => left.localeCompare(right));

src/languages.ts:133

• supportForFile() always runs mappedSupportForFile(), which normalizes/sorts the extension map even when no extension map is provided. Given how frequently supportForFile() is called during indexing, short-circuiting when extensionMap is undefined avoids unnecessary work in the default configuration.

export function supportForFile(
  filename: string,
  extensionMap?: LanguageExtensionMap | undefined,
): LanguageSupport | undefined {
  const mapped = mappedSupportForFile(filename, extensionMap);
  if (mapped) return mapped;
  const ext = path.extname(filename).toLowerCase();