@@ -94,6 +94,34 @@ fn extract_path_backwards(text: &str, end_pos: usize) -> Option<usize> {
9494 if current_pos == end_pos { None } else { Some ( current_pos) }
9595}
9696
97+ fn extract_path_forward ( text : & str , start_pos : usize ) -> Option < usize > {
98+ use rustc_lexer:: { is_id_continue, is_id_start} ;
99+ let mut current_pos = start_pos;
100+ loop {
101+ if current_pos < text. len ( ) && text[ current_pos..] . starts_with ( "::" ) {
102+ current_pos += 2 ;
103+ } else {
104+ break ;
105+ }
106+ let mut chars = text[ current_pos..] . chars ( ) ;
107+ if let Some ( c) = chars. next ( ) {
108+ if is_id_start ( c) {
109+ current_pos += c. len_utf8 ( ) ;
110+ } else {
111+ break ;
112+ }
113+ }
114+ while let Some ( c) = chars. next ( ) {
115+ if is_id_continue ( c) {
116+ current_pos += c. len_utf8 ( ) ;
117+ } else {
118+ break ;
119+ }
120+ }
121+ }
122+ if current_pos == start_pos { None } else { Some ( current_pos) }
123+ }
124+
97125fn is_valid_for_html_tag_name ( c : char , is_empty : bool ) -> bool {
98126 // https://spec.commonmark.org/0.30/#raw-html
99127 //
@@ -218,19 +246,68 @@ impl<'a, 'tcx> DocVisitor for InvalidHtmlTagsLinter<'a, 'tcx> {
218246 // If a tag looks like `<this>`, it might actually be a generic.
219247 // We don't try to detect stuff `<like, this>` because that's not valid HTML,
220248 // and we don't try to detect stuff `<like this>` because that's not valid Rust.
221- if let Some ( Some ( generics_start) ) = ( is_open_tag
222- && dox[ ..range. end ] . ends_with ( '>' ) )
249+ let mut generics_end = range. end ;
250+ if let Some ( Some ( mut generics_start) ) = ( is_open_tag
251+ && dox[ ..generics_end] . ends_with ( '>' ) )
223252 . then ( || extract_path_backwards ( & dox, range. start ) )
224253 {
254+ while generics_start != 0
255+ && generics_end < dox. len ( )
256+ && dox. as_bytes ( ) [ generics_start - 1 ] == b'<'
257+ && dox. as_bytes ( ) [ generics_end] == b'>'
258+ {
259+ generics_end += 1 ;
260+ generics_start -= 1 ;
261+ if let Some ( new_start) = extract_path_backwards ( & dox, generics_start) {
262+ generics_start = new_start;
263+ }
264+ if let Some ( new_end) = extract_path_forward ( & dox, generics_end) {
265+ generics_end = new_end;
266+ }
267+ }
268+ if let Some ( new_end) = extract_path_forward ( & dox, generics_end) {
269+ generics_end = new_end;
270+ }
225271 let generics_sp = match super :: source_span_for_markdown_range (
226272 tcx,
227273 & dox,
228- & ( generics_start..range . end ) ,
274+ & ( generics_start..generics_end ) ,
229275 & item. attrs ,
230276 ) {
231277 Some ( sp) => sp,
232278 None => item. attr_span ( tcx) ,
233279 } ;
280+ // Sometimes, we only extract part of a path. For example, consider this:
281+ //
282+ // <[u32] as IntoIter<u32>>::Item
283+ // ^^^^^ unclosed HTML tag `u32`
284+ //
285+ // We don't have any code for parsing fully-qualified trait paths.
286+ // In theory, we could add it, but doing it correctly would require
287+ // parsing the entire path grammar, which is problematic because of
288+ // overlap between the path grammar and Markdown.
289+ //
290+ // The example above shows that ambiguity. Is `[u32]` intended to be an
291+ // intra-doc link to the u32 primitive, or is it intended to be a slice?
292+ //
293+ // If the below conditional were removed, we would suggest this, which is
294+ // not what the user probably wants.
295+ //
296+ // <[u32] as `IntoIter<u32>`>::Item
297+ //
298+ // We know that the user actually wants to wrap the whole thing in a code
299+ // block, but the only reason we know that is because `u32` does not, in
300+ // fact, implement IntoIter. If the example looks like this:
301+ //
302+ // <[Vec<i32>] as IntoIter<i32>::Item
303+ //
304+ // The ideal fix would be significantly different.
305+ if ( generics_start > 0 && dox. as_bytes ( ) [ generics_start - 1 ] == b'<' )
306+ || ( generics_end < dox. len ( ) && dox. as_bytes ( ) [ generics_end] == b'>' )
307+ {
308+ diag. emit ( ) ;
309+ return ;
310+ }
234311 // multipart form is chosen here because ``Vec<i32>`` would be confusing.
235312 diag. multipart_suggestion (
236313 "try marking as source code" ,
@@ -278,7 +355,7 @@ impl<'a, 'tcx> DocVisitor for InvalidHtmlTagsLinter<'a, 'tcx> {
278355 for ( event, range) in p {
279356 match event {
280357 Event :: Start ( Tag :: CodeBlock ( _) ) => in_code_block = true ,
281- Event :: Html ( text) | Event :: Text ( text ) if !in_code_block => {
358+ Event :: Html ( text) if !in_code_block => {
282359 extract_tags ( & mut tags, & text, range, & mut is_in_comment, & report_diag)
283360 }
284361 Event :: End ( Tag :: CodeBlock ( _) ) => in_code_block = false ,
0 commit comments