ContextQMD
Back to Plate

Coverage Priority Map Post GTE 5

docs/plans/2026-03-24-coverage-priority-map-post-gte-5.md

53.0.56.6 KB
Original Source

Coverage Priority Map Post GTE 5

Inputs

  • Coverage source: lcov.info
  • Constraints:
    • exclude /react
    • no browser or e2e
    • no coverage vanity
    • score only files still worth real unit or editor-contract tests
    • sync against the March 17 pass plus the March 22-24 completed sweeps

Coverage Run

  • Command: bun test --coverage --coverage-reporter=lcov --coverage-dir=.coverage-repo-2026-03-24g --reporter=dots
  • Result: Fresh coverage: 2869 pass, 0 fail, 575 files, 3.59s.

Scoring Rules

  • Scope is packages/**/src/**.
  • Exclude /react, test files, barrels, declaration files, obvious type-only files, and React-hook/component files living outside /react.
  • Reward deterministic transforms, queries, merge helpers, parser/serializer helpers, plugin contracts, and small utilities with real uncovered behavior.
  • Penalize DOM-heavy leftovers, giant sludge files, thin wrapper plugins, and already-swept packages so crumbs do not crowd out better seams.
  • package_score is the sum of the top 5 remaining file scores in that package, not every leftover crumb.

Strong Take

Do not do another blind package sweep.

The honest next work is a small core-heavy batch, then a second ring of list/table/suggestion follow-ups.

If you want the strict next batch, use the raw score >= 6 set:

  1. core: pluginDeserializeHtml.ts — score 7
  2. diff: transformDiffTexts.ts — score 7
  3. core: getSelectedDomFragment.tsx — score 6
  4. markdown: deserializeMd.ts — score 6
  5. core: resolvePlugin.ts — score 6

My actual recommendation is slightly stricter than the raw score:

  • start with the non-DOM part of core
  • then markdown
  • then diff
  • only then touch the DOM-ish core leftover if you still care

Threshold Counts

  • score >= 7: 2 files
  • score >= 6: 5 files
  • score >= 5: 15 files
  • score >= 4: 39 files
  • score >= 3: 62 files
  • score >= 2: 195 files
  • score >= 1: 211 files

Packages By Value

  1. core — package score 29, top files pluginDeserializeHtml.ts:7; getSelectedDomFragment.tsx:6; resolvePlugin.ts:6; resolvePlugins.ts:5; htmlElementToLeaf.ts:5
  2. table — package score 21, top files withApplyTable.ts:5; getSelectedCellsBorders.ts:4; deleteColumn.ts:4; insertTableColumn.ts:4; insertTableRow.ts:4
  3. slate — package score 20, top files hasDOMNode.ts:4; hasEditableTarget.ts:4; hasSelectableTarget.ts:4; hasTarget.ts:4; hasRange.ts:4
  4. list-classic — package score 18, top files BaseTodoListPlugin.ts:5; withList.ts:5; withInsertFragmentList.ts:4; withDeleteForwardList.ts:2; withNormalizeList.ts:2
  5. suggestion — package score 18, top files deleteSuggestion.ts:5; BaseSuggestionPlugin.ts:4; withSuggestion.ts:4; rejectSuggestion.ts:3; acceptSuggestion.ts:2
  6. markdown — package score 15, top files deserializeMd.ts:6; fontRules.ts:3; columnRules.ts:2; convertNodesSerialize.ts:2; customMdxDeserialize.ts:2
  7. diff — package score 12, top files transformDiffTexts.ts:7; get-properties.ts:2; unused-char-generator.ts:2; transformDiffDescendants.ts:1
  8. docx — package score 11, top files docxListToList.ts:3; cleanDocxListElementsToList.ts:2; getVShapeSpid.ts:2; DocxPlugin.ts:2; getDocxIndent.ts:2
  9. dnd — package score 9, top files onDropNode.ts:5; getHoverDirection.ts:2; getNewDirection.ts:2
  10. code-block — package score 10, top files BaseCodeBlockPlugin.ts:4; insertCodeBlock.ts:2; htmlDeserializerCodeBlock.ts:2; formatter.ts:1; withCodeBlock.ts:1

Best Files By Value

  1. core — pluginDeserializeHtml.ts — score 7, coverage 88.8%, uncovered 18
  2. diff — transformDiffTexts.ts — score 7, coverage 96.2%, uncovered 8
  3. core — getSelectedDomFragment.tsx — score 6, coverage 13.9%, uncovered 31
  4. markdown — deserializeMd.ts — score 6, coverage 79.2%, uncovered 15
  5. core — resolvePlugin.ts — score 6, coverage 88.1%, uncovered 8
  6. autoformat — AutoformatPlugin.ts — score 5, coverage 37.0%, uncovered 46
  7. list-classic — BaseTodoListPlugin.ts — score 5, coverage 37.1%, uncovered 22
  8. table — withApplyTable.ts — score 5, coverage 83.8%, uncovered 16
  9. code-drawing — renderers.ts — score 5, coverage 85.4%, uncovered 15
  10. core — resolvePlugins.ts — score 5, coverage 95.8%, uncovered 15
  11. dnd — onDropNode.ts — score 5, coverage 87.5%, uncovered 14
  12. core — htmlElementToLeaf.ts — score 5, coverage 69.4%, uncovered 11
  13. list-classic — withList.ts — score 5, coverage 84.7%, uncovered 11
  14. suggestion — deleteSuggestion.ts — score 5, coverage 93.6%, uncovered 10
  15. core — cleanHtmlFontElements.ts — score 5, coverage 66.7%, uncovered 4

Caveats

  • Raw package totals still inflate core because it has a pile of medium leftovers. Trust the file ranking more than the package ranking.
  • A few non-React but still DOM-ish files remain in core and slate. They are valid candidates, but lower priority than the parser/plugin/value seams above.
  • Thin wrappers and giant serializer sludge were intentionally capped or filtered so they do not cosplay as the next frontier.

Full Data