rumdl 0.3.1

v0.3.1

on Python PyPI

Added

• rumdl check-assets reports assets unreferenced in complete Markdown source
  scans, with explicit include/keep globs and text or versioned JSON output.
  Literal filename mentions are kept separately; incomplete scans emit no findings.
• MD022 lines-above-mode = "exact" removes surplus blank lines above headings
  while preserving comments and the existing minimum-spacing default (#838).
• Opt-in code-block tool version verification against uv.lock, declared
  pyproject.toml constraints, or pre-commit/prek hook revisions for Ruff,
  Black, djLint and SQLFluff. Supports automatic source selection and one
  configurable policy for mismatched or unresolved versions.
• MD013 link-text-punctuation = "trailing" optionally recognizes sentence
  endings in link text and image alt text while keeping each construct intact.
  When enabled, semantic reflow also preserves detected sentence boundaries
  when merging short lines.
• MD063 style = "consistent" infers heading capitalization per document while
  respecting ignored words, proper names, level filters, and inline disables.
• Independent code blocks can share the existing worker pool when error policies
  permit parallel execution, preserving tool order and applying formatted
  replacements together from their original source ranges.
• MD013 break-words adds project-specific or localized words to the semantic
  reflow defaults, with case-insensitive Unicode matching.
• custom-rules: prototype versioned checker interchange (602a8f7)
• code-block-tools: verify pre-commit and prek revisions (0cfe861)
• MD096: add opt-in table sorting by named column (43a7858)

Changed

• Code-block formatting warnings appear in source order, including single-worker
  runs.
• MD013 semantic reflow joins wrapped list and blockquote prose consistently
  with paragraphs, including items containing only one semantic sentence.
• Reduce silent-formatting work for nested lists and unsupported proof syntax,
  and avoid copying line text while checking Python-Markdown list structure.

Fixed

• Prevent stack overflows when releasing parsed MDX contexts with deeply
  nested JSX, including headings checked in parallel.
• MD013 sentence reflow uses surrounding context to distinguish technical labels
  (service A., plan B.) from name initials (#965).
  A lone initial after a lowercase word (edited by A. Smith) can now also split.
  Custom abbreviations preserve ambiguous initials globally, including when
  the same letter is used as a technical label.
• Silent formatting applies native bare-URL fixes in continued reference
  definitions and reference labels with empty destinations.
• Silent formatting applies native list-indentation fixes when display-math
  delimiters span blocks and close inside a code span.
• MD013 semantic break words keep offsets in the original Unicode text, avoiding
  incorrect splits after characters whose lowercase forms change byte length.
• MD013 semantic reflow keeps balanced parenthetical asides together during
  sentence detection and counts them as single units in paragraph diagnostics.
  Overlong asides prefer inner sentence boundaries before further width wrapping;
  explicit hard breaks remain separate reflow segments.
• MD007: honor start-indent at nested list levels (fbf3d10)
• MD005: keep adjacent ordered and bullet lists separate (#970) (2c1d543)
• MD033: guard HTML conversions with parser byte ranges (dea14d8)
• docs: retire AI search and retain keyword search (f9f4f2e)
• MD070: detect collisions in list items opened on a quoted item line (21c3149)
• MD070: judge nested fences relative to their list item or blockquote (68dbb3e)
• mdx: bundle maintained parser and avoid redundant image scans (c281ade)
• MD070: skip a fence that ends with its list item (#967) (3be9c0c)
• MD030: stop auto-fixing '1.Foo' lines that are paragraph text (abf05df)
• MD023: measure headings after quoted items from the quote they sit in (632c88d)
• parser: treat the lines of an HTML block inside a blockquote as raw HTML (751f025)
• MD044: preserve filename extension capitalization (25fef5b)
• code-block-tools: avoid parsing jq output as diagnostics (79a3abd)

Performance

• MD044: bound fallback code and link scans (3dedd68)
• MD044: index WikiLink destinations once (f1d5c0f)
• MD044: reuse URL ranges and character columns (a19906f)
• headings: retain anchor ranges across plain attributes (538ecb3)
• headings: bound backtick run rescans (08a9f4d)
• headings: bound image delimiter rescans (9cb0aee)
• headings: build anchor display text in linear time (8945a71)
• headings: reuse trailing anchor ranges (fb4af16)
• MD063: avoid repeated unmatched tag scans (1303d3a)
• MD030: reuse indented block lookups (7797130)
• headings: sweep opaque ranges during anchor extraction (80ec3bc)
• MD063: sweep heading regions in source order (51602e0)
• lint-context: reuse list continuation indentation (f0588d4)
• MD031: memoize textual list membership (9a6a4fd)
• avoid repeated fence scans in spacing rules (96f6873)
• inline-config: index directive code ranges (9415ab6)
• inline-config: reuse suppression and comment lookups (692e326)
• MD077: index native fence queries (e8e1749)
• MD070: index fence closure lookups (37106d6)
• workspace: reuse link searches and source columns (f52e95b)
• MD057: cache link diagnostic columns (12ab11b)
• MD093: cache heading warning columns (8a23e39)
• MD084: cache fix offsets and defer Unicode formatting (e4cdbce)
• parser: index heading link exclusions (e7a5ec6)
• MD089: index image links and warning columns (6c38dc2)
• MD095: reuse indexed source positions (93515db)
• parser: index comment exclusions and line containment (87f6930)
• MD049: cache markup exclusions and source columns (1fb5463)
• MD037: index warning exclusions and columns (e5984c1)
• emphasis: bound delimiter pairing searches (ef1e7c2)
• emphasis: index skip checks and source columns (f612dd0)
• MD054: reuse parsed diagnostic positions (2102f60)
• MD054: avoid quadratic fix preparation (09873a9)
• MD033: index allowed-inside regions (ebb66a5)
• MD033: avoid rescanning diagnostic line prefixes (a4f4cdd)
• parser: reuse column indexes for inline elements (a4abcc4)
• parser: index link columns on long lines (143b17d)
• MD034: avoid quadratic scans on URL-heavy lines (5513676)
• stdin-batch: avoid rebuilding declared link targets (e4b902d)

Downloads

Installation

Using uv (Recommended)

uv tool install rumdl

Using pip

pip install rumdl

Using pipx

pipx install rumdl

Direct Download

Download the appropriate binary for your platform from the table above, extract it, and add it to your PATH.