0.7.0

• Breaking: ruby wrapper shorthand now requires a paired wrapper in both renderer and analyzer.

  • Wrapper form is treated as shorthand only when both tags are present: <ruby>base《reading》</ruby>.
  • Unclosed wrapper input (for example <RUBY>漢字《かんじ》) is no longer treated as wrapped shorthand.
  • Bare shorthand (漢字《かんじ》) behavior is unchanged.
  • I thought I was fixing it, but it seems it was broken.

• Breaking: wrapper tag handling is now case-insensitive and consistent.

  • Closed uppercase wrappers such as <RUBY>...</RUBY> now follow the same paired-wrapper semantics as lowercase tags.
  • Legacy ambiguous optional-wrapper interpretation was removed.

• Added/Expanded (analyzer): editor-side APIs were expanded and aligned with plugin runtime behavior.

  • Added/solidified helpers: normalizeLineWindow, expandToParagraphBoundaries, shouldFullAnalyze, analyzeLineWindow.
  • Added analyzer parity regression tests for wrapped/unwrapped ruby and edge cases.

• Changed: analyzer packaging/layout was cleaned up (public subpath unchanged).

  • Public import path remains: @peaceroad/markdown-it-renderer-inline-text/analyzer.
  • Export target now points to ./src/analyzer.js (internal file layout change).

• Performance: analyzer hot paths were optimized without API changes.

  • Reduced temporary allocations in ruby scanning.
  • Added ruby-only fast path in inline scanning.
  • Reduced repeated trim() work in paragraph propagation by precomputing blank-line flags.
  • Removed redundant helpers and default-arg allocations in hot code paths.

• Tests/Maintenance: test suite structure was cleaned up.

  • test/option-assertions.js and test/analyzer-assertions.js were split/extended for option and analyzer regressions.