[dchaudhari7177]

Fixes #1165.

Cause

MockIncludeDirective.run calls:

self.renderer.nested_render_text(file_content, startline + 1, ...)

nested_render_text adds that lineno to the 0-based token maps, and _render_tokens then applies its own 0-based → 1-based + 1. startline is already the 0-based offset of the first included line, so adding one here reports every warning inside an included file one line too low. Exactly as diagnosed in the issue — a one-token change.

The existing fixture encoded the bug

tests/test_renderers/fixtures/mock_include_errors.md expected:

tmpdir/bad.md:2: (WARNING/2) Unknown interpreted text role "a".

while bad.md is written as "{a}b" — a single line. That expectation is now :1.

Regression cases

The issue asked for these before touching the line, so the fixture gains five more, each with the offending role in a paragraph of its own (MyST attributes an inline warning to its enclosing block's start line, which is worth being deliberate about here):

case	file	role on	expected
plain include	bad_line3.md	3	:3
include after content in the parent	bad_line3.md	3	:3
:start-line: 2	bad_skipped.md	4	:4
front matter stripped	bad_frontmatter.md	5	:5
nested include	bad_inner.md via bad_outer.md	3	:3

All six cases fail on master and pass with the change — I checked by reverting only mocking.py and leaving the tests in place.

The nested case is worth calling out: before this change it appeared correct for a role sharing a paragraph with the preceding line, because the two errors cancelled. Putting the role in its own paragraph shows it was off by one like the rest.

:heading-offset: was also probed and is unaffected (heading offsetting shifts levels, not maps), as is front-matter stripping — the front_matter token is popped before the maps are adjusted, so it shifts nothing.

Test run

1240 passed, 18 skipped. Three failures are pre-existing on an unmodified checkout in this environment (test_cmdline[40-linkify], test_extended_syntaxes, test_extended_syntaxes_text) — I diffed the FAILED sets before and after, and those three are identical on both sides.

🤖 Generated with Claude Code

[chrisjsewell]

Thank you @dchaudhari7177. Your fixture cases were the best regression set among the PRs for #1165, and your note that the nested case only appeared correct because two errors cancelled was spot on. We've landed your commit via #1169 (1915654), which also fixes :start-after: adding a character index to the line (#1216) and negative :start-line: in the same lines, so the separate PRs would otherwise have conflicted. You're credited as a co-author on the merge commit. Closing as superseded. Your #1197 has also landed for #1167 (253427e), with follow-up commits from us. Thanks again!