Strong with nested emphasis parses as "**" #657
Description
The string "****foo*bar*baz****" is parsed as follows:
"****foo*bar*baz****" -> [
Start(Paragraph)
Text(Borrowed("*"))
Text(Borrowed("*"))
Start(Emphasis)
Start(Emphasis)
Text(Borrowed("foo"))
End(Emphasis)
Text(Borrowed("bar"))
End(Emphasis)
Text(Borrowed("baz"))
Text(Borrowed("*"))
Text(Borrowed("*"))
Text(Borrowed("*"))
Text(Borrowed("*"))
End(Paragraph)
]This differs from the commonmark.js reference implementation which gives the following AST:
<paragraph>
<strong>
<emph>
<emph>
<text>foo</text>
</emph>
<text>bar</text>
</emph>
<text>baz</text>
</strong>
<text>**</text>
</paragraph>I found this while trying to make pulldown-cmark-to-cmark round-trip the Markdown text with a fuzzer. I'm trying to make a single paragraph of text work and the fuzzer looks like this:
#![no_main]
use libfuzzer_sys::fuzz_target;
use pulldown_cmark::Parser;
use pulldown_cmark::{Event, Tag};
use pulldown_cmark_to_cmark::cmark;
fn round_trip(text: &str) -> String {
let mut result = String::with_capacity(text.len());
let events = Parser::new(&text);
let events = events.collect::<Vec<_>>();
if std::env::var("EVENTS").is_ok() {
let mut width = 0;
eprintln!("{text:?} -> [");
for event in &events {
if let Event::End(_) = event {
width -= 2;
}
eprintln!(" {:width$}{event:?}", "");
if let Event::Start(_) = event {
width += 2;
}
}
eprintln!("]");
}
cmark(events.into_iter(), &mut result).unwrap();
result
}
fuzz_target!(|text: String| {
if text.contains(&['\n', '\r', '`']) {
return;
}
let round_trip_1 = round_trip(&text);
match Parser::new(&round_trip_1).collect::<Vec<_>>().as_slice() {
[Event::Start(Tag::Paragraph), .., Event::End(Tag::Paragraph)] => {}
_ => return,
}
let round_trip_2 = round_trip(&round_trip_1);
let round_trip_3 = round_trip(&round_trip_2);
assert_eq!(round_trip_2, round_trip_3);
});