WORK-144
ID:WORK-144Status:done

Extract pure scanner functions into scanner-core module

Split scanner.ts so that pure functions (parseFileContent, scanPlanSources, and all supporting AST helpers) live in a new scanner-core.ts with zero Node.js imports. The existing scanner.ts re-imports from scanner-core.ts and adds filesystem-dependent functions.

Priority:highComplexity:moderateSource:SPEC-040
claude/file-naming-convention-LJdwR View source

Criteria completion

No incremental history — criteria tracking started on Apr 21.

Relationships

Implements 1
Related 1
Branches 1
claude/file-naming-convention-LJdwR current done
main done
History 2
  1. bbbb94c
    Created (done)by bjornolofandersson
  2. f2b3512
    Content editedby Claude
    Add {ID}-{slug}.md filename convention + migrate filenames subcommand

Acceptance Criteria

  • New file runes/plan/src/scanner-core.ts exists and contains all pure scanner functions
  • scanner-core.ts has zero imports from node:fs, node:path, node:child_process, or @refrakt-md/content
  • parseFileContent(source, relPath) is exported from scanner-core.ts
  • scanPlanSources(sources) is exported from scanner-core.ts
  • All supporting pure functions (extractCriteria, extractResolution, extractRefs, extractScopedRefs, walkNodes, extractHeadingText, matchKnownSection, extractTitle) move to scanner-core.ts
  • scanner.ts imports from scanner-core.ts and re-exports the pure functions alongside its filesystem functions
  • All existing imports from ./scanner and the main entry point continue to work (no breaking changes)
  • index.ts re-exports parseFileContent and scanPlanSources unchanged
  • Existing tests pass without modification

Approach

  1. Create scanner-core.ts and move all pure functions into it. Keep the Markdoc import (pure JS, no Node.js APIs) and any type imports from ./types.ts.
  2. In scanner.ts, replace the moved function bodies with import { parseFileContent, scanPlanSources, ... } from './scanner-core.js' and export { parseFileContent, scanPlanSources } from './scanner-core.js'.
  3. Filesystem-dependent functions (parseFile, scanPlanFiles, collectMdFiles, readCache, writeCache) stay in scanner.ts.
  4. Verify that scanner.ts is the only file that imports node:fs, node:path, and @refrakt-md/content.
  5. Run existing tests to confirm no regressions.

Dependencies

None — this is the first extraction step.

References

  • SPEC-040 — Edge Runtime Compatibility for Plan Package

Resolution

Completed: 2026-04-14

Branch: claude/edge-runtime-refactor-HOg8v

What was done

  • Created runes/plan/src/scanner-core.ts with all pure scanner functions (parseFileContent, scanPlanSources, extractCriteria, extractResolution, extractRefs, extractScopedRefs, walkNodes, extractHeadingText, matchKnownSection, extractTitle, and supporting constants)
  • Rewrote runes/plan/src/scanner.ts to import from scanner-core and re-export pure functions; kept only filesystem-dependent code (parseFile, scanPlanFiles, collectMdFiles, readCache, writeCache, getGitMtimes)
  • index.ts re-exports unchanged — no breaking changes to public API

Notes

  • scanner-core.ts depends only on @markdoc/markdoc (pure JS) and @refrakt-md/runes (escapeFenceTags only)
  • All 321 plan tests pass without modification