coderaiser / putout-editor

100%

master: 100%

push

github

feature: char: syntax highlighting

* feature: chat: lezer syntax highlighting in SourceBlock, TransformDiff, AstCodePreview

highlight.ts turns JavaScript into HTML spans via @lezer/javascript and
@classHighlighter, escaping every text run first so a string literal holding
<script> stays inert. The tok-* classes are mapped to the existing colour
tokens in highlight.css, scoped under .tok-scope so they cannot bleed into the
AST tree.

The column marker in AstCodePreview now splices into the highlighted string
rather than the plain line: insertMarker counts only non-tag characters, so
the | lands at the same column without ever ending up inside a span tag.

412 tests, 100% branches/lines/functions/statements in packages/chat.

* test: chat: e2e: source: syntax

* test: chat: e2e

* test: chat: e2e for syntax highlighting

Three desktop tests pin what step 1 built, plus the two fixes the e2e
forced back into the source.

- AstCodePreview now renders its line through dangerouslySetInnerHTML.
  Before, markedLine's HTML arrived as a React text child and the page
  showed literal `<span class="tok-keyword">` text — the tok spans were
  strings, not elements. Unit specs read innerHTML, so they passed while
  the browser showed markup as text. Unselected lines go through escHtml
  so a `<div>` in source stays inert.
- AstCodePreview's container is `ast-code` again, not `ast-code-preview`.
  AstTree.css styles and hides `.ast-code`; the renamed class had no rules
  and no coarse-pointer hiding.
- `/transform` answers "needs the putout server and does not run in the
  browser", so no diff block ever renders in e2e. The TransformDiff test
  asserts the empty state and says so, rather than asserting spans that
  cannot appear.
- The AstCodePreview test selects VariableDeclarator first: the preview
  highlights only the selected line, so with no selection there is nothing
  to assert.

65 e2e pass (desktop + mobile); 412 unit, 100% across the board.

* chore: cha... (continued)

1081 of 1081 branches covered (100.0%)

Branch coverage included in aggregate %.

6702 of 6702 relevant lines covered (100.0%)

61.2 hits per line