rumdl_lib/utils/
parser_options.rs

1use pulldown_cmark::Options;
2
3/// Standard pulldown-cmark options for rumdl parsing.
4///
5/// Uses an explicit allowlist rather than `Options::all()` to prevent
6/// future pulldown-cmark releases from silently changing parse behavior.
7///
8/// Notably excludes `ENABLE_YAML_STYLE_METADATA_BLOCKS` and
9/// `ENABLE_PLUSES_DELIMITED_METADATA_BLOCKS` because rumdl handles
10/// front matter detection independently. These options cause pulldown-cmark
11/// to misinterpret `---` horizontal rules as metadata delimiters,
12/// corrupting code block detection across the entire document.
13pub fn rumdl_parser_options() -> Options {
14    let mut options = Options::empty();
15    options.insert(Options::ENABLE_TABLES);
16    options.insert(Options::ENABLE_FOOTNOTES);
17    options.insert(Options::ENABLE_STRIKETHROUGH);
18    options.insert(Options::ENABLE_TASKLISTS);
19    options.insert(Options::ENABLE_SMART_PUNCTUATION);
20    options.insert(Options::ENABLE_HEADING_ATTRIBUTES);
21    options.insert(Options::ENABLE_MATH);
22    options.insert(Options::ENABLE_GFM);
23    options.insert(Options::ENABLE_DEFINITION_LIST);
24    options.insert(Options::ENABLE_SUPERSCRIPT);
25    options.insert(Options::ENABLE_SUBSCRIPT);
26    options.insert(Options::ENABLE_WIKILINKS);
27    options
28}
29
30/// The byte ranges of the code spans in `content` as a reader without math
31/// finds them, sorted by start.
32///
33/// rumdl reads `$...$` and `$$...$$` as math, and a backtick inside math opens
34/// no code span. CommonMark has no math, and there that backtick opens a span
35/// closing at the next backtick run of its length, inside the math or past it.
36/// A rewrite that has to keep a code span's text as written needs both
37/// readings, so this one comes from the same parse with only math left out.
38pub fn code_span_ranges_without_math(content: &str) -> Vec<(usize, usize)> {
39    if !content.contains('`') {
40        return Vec::new();
41    }
42    let mut options = rumdl_parser_options();
43    options.remove(Options::ENABLE_MATH);
44    pulldown_cmark::Parser::new_ext(content, options)
45        .into_offset_iter()
46        .filter_map(|(event, range)| {
47            matches!(event, pulldown_cmark::Event::Code(_)).then_some((range.start, range.end))
48        })
49        .collect()
50}
rumdl_lib/utils/parser_options.rs

rumdl_lib/utils/
parser_options.rs
