@pretextbook/latex-style-pretext

Editor intelligence for LaTeX-style PreTeXt — LaTeX source that converts to PreTeXt via @pretextbook/unified-latex-to-pretext.

This package is platform-agnostic and returns LSP-shaped values (vscode-languageserver-types), so it can be consumed both by the VS Code extension's LSP server (Node) and by the Monaco editor in pretext-plus-editor (browser). It ships raw TypeScript with no build step, following the @pretextbook/completions precedent.

What's here

| Module | Purpose | | -------------- | ------------------------------------------------------------------------------- | | data/ | Curated tables of the environments and macros the PreTeXt conversion supports | | scan/ | Lightweight linear document scanner (math regions, env stack, labels, comments) | | completions/ | Completion engine (Phase 2) | | lint/ | Diagnostics engine (Phase 3) | | clean/ | Legacy-LaTeX cleanup rules, shared with @pretextbook/import |

Keeping the tables in sync with the converter

The curated tables in data/ mirror what @pretextbook/unified-latex-to-pretext will actually convert. Two guards keep them honest, in opposite directions:

• converter-drift.spec.ts converts a usage of every curated entry and fails if the converter no longer recognizes it — this catches support being removed or renamed.
• converter-coverage.spec.ts compares the tables against data/converter-support.json and fails if the converter supports something the tables lack — this catches support being added.

converter-support.json is generated, never hand-edited:

npm run refresh:latex-support     # regenerate from the installed converter
npm run check:latex-support       # fail if it is stale (CI runs this)

The snapshot has two halves, because the converter only exports half its surface:

• environments / macros / plusMacros / plusTypes come from the records the converter exports, complete with argument signatures.
• probedMacros / probedEnvironments are discovered by converting \name{x} and \begin{name}…\end{name} for every PreTeXt element name in the schema, because ordinary LaTeX and PreTeXt inline markup (\alert, \q, …) are handled through internal tables the converter does not export.

To update after bumping the converter: run npm run refresh:latex-support, then npm run test -w @pretextbook/latex-style-pretext. The coverage spec names every entry that needs adding; the drift spec names every entry that needs removing or promoting out of KNOWN_UNCONVERTED.