Markdown code span regex crosses blank-line/paragraph boundaries, swallowing a subsequent header's highlighting

[denysshnurenko]

What happened

An inline code span (single backtick) whose closing backtick appears after
a header, separated by a blank line, gets matched as one greedy token that
swallows the header text — so the header loses its heading-color styling.

Repro:

here is `an open span

## Swallowed Header

and it closes` here.

Expected: "## Swallowed Header" renders with heading styling.
Actual: it's consumed into the code-span token and rendered as plain
backtick-string text instead.

Root cause

lexers/markdown.go:102:

{"`[^`]+`", LiteralStringBacktick, nil},

[^]` matches newlines by default in Go's RE2 engine, so this rule greedily
matches across blank lines / paragraph and block boundaries. Per CommonMark,
a code span should not cross a blank line — this rule has no such bound.

Related

#1267 reports a similar symptom (list-item code blocks losing highlighting)
and was closed as "completed", but the exact same regex is still present at
HEAD (af5ff891, 2026-06-17, is the only recent commit touching
markdown.go, and it's an unrelated Iterator→iter.Seq refactor) — so whatever
closed #1267 doesn't look like it touched this rule. Might be worth
re-checking whether #1267 is actually resolved, since it and this report
look like they share the same underlying cause.

Context

Found via revdiff (a diff-review TUI that uses chroma for markdown
rendering) — see umputun/revdiff#260 for the original report and the
maintainer's diagnosis that traced it back to this rule.

Environment

• chroma: current master (HEAD as of 2026-07-15)