Improve suggestion in case a bare URL is surrounded by brackets

This commit is contained in:
Guillaume Gomez 2025-09-10 15:44:42 +02:00
parent be8de5d6a0
commit 5a36fe2387

View file

@ -17,7 +17,10 @@ use crate::core::DocContext;
use crate::html::markdown::main_body_opts;
pub(super) fn visit_item(cx: &DocContext<'_>, item: &Item, hir_id: HirId, dox: &str) {
let report_diag = |cx: &DocContext<'_>, msg: &'static str, range: Range<usize>| {
let report_diag = |cx: &DocContext<'_>,
msg: &'static str,
range: Range<usize>,
without_brackets: Option<&str>| {
let maybe_sp = source_span_for_markdown_range(cx.tcx, dox, &range, &item.attrs.doc_strings)
.map(|(sp, _)| sp);
let sp = maybe_sp.unwrap_or_else(|| item.attr_span(cx.tcx));
@ -27,14 +30,22 @@ pub(super) fn visit_item(cx: &DocContext<'_>, item: &Item, hir_id: HirId, dox: &
// The fallback of using the attribute span is suitable for
// highlighting where the error is, but not for placing the < and >
if let Some(sp) = maybe_sp {
lint.multipart_suggestion(
"use an automatic link instead",
vec![
(sp.shrink_to_lo(), "<".to_string()),
(sp.shrink_to_hi(), ">".to_string()),
],
Applicability::MachineApplicable,
);
if let Some(without_brackets) = without_brackets {
lint.multipart_suggestion(
"use an automatic link instead",
vec![(sp, format!("<{without_brackets}>"))],
Applicability::MachineApplicable,
);
} else {
lint.multipart_suggestion(
"use an automatic link instead",
vec![
(sp.shrink_to_lo(), "<".to_string()),
(sp.shrink_to_hi(), ">".to_string()),
],
Applicability::MachineApplicable,
);
}
}
});
};
@ -43,7 +54,7 @@ pub(super) fn visit_item(cx: &DocContext<'_>, item: &Item, hir_id: HirId, dox: &
while let Some((event, range)) = p.next() {
match event {
Event::Text(s) => find_raw_urls(cx, &s, range, &report_diag),
Event::Text(s) => find_raw_urls(cx, dox, &s, range, &report_diag),
// We don't want to check the text inside code blocks or links.
Event::Start(tag @ (Tag::CodeBlock(_) | Tag::Link { .. })) => {
for (event, _) in p.by_ref() {
@ -67,25 +78,41 @@ static URL_REGEX: LazyLock<Regex> = LazyLock::new(|| {
r"https?://", // url scheme
r"([-a-zA-Z0-9@:%._\+~#=]{2,256}\.)+", // one or more subdomains
r"[a-zA-Z]{2,63}", // root domain
r"\b([-a-zA-Z0-9@:%_\+.~#?&/=]*)" // optional query or url fragments
r"\b([-a-zA-Z0-9@:%_\+.~#?&/=]*)", // optional query or url fragments
))
.expect("failed to build regex")
});
fn find_raw_urls(
cx: &DocContext<'_>,
whole_text: &str,
text: &str,
range: Range<usize>,
f: &impl Fn(&DocContext<'_>, &'static str, Range<usize>),
f: &impl Fn(&DocContext<'_>, &'static str, Range<usize>, Option<&str>),
) {
trace!("looking for raw urls in {text}");
// For now, we only check "full" URLs (meaning, starting with "http://" or "https://").
for match_ in URL_REGEX.find_iter(text) {
let url_range = match_.range();
let mut without_brackets = None;
let mut extra_range = 0;
// If the whole text is contained inside `[]`, then we need to replace the brackets and
// not just add `<>`.
if whole_text[..range.start + url_range.start].ends_with('[')
&& range.start + url_range.end <= whole_text.len()
&& whole_text[range.start + url_range.end..].starts_with(']')
{
extra_range = 1;
without_brackets = Some(match_.as_str());
}
f(
cx,
"this URL is not a hyperlink",
Range { start: range.start + url_range.start, end: range.start + url_range.end },
Range {
start: range.start + url_range.start - extra_range,
end: range.start + url_range.end + extra_range,
},
without_brackets,
);
}
}