1 //! Markdown formatting for rustdoc.
3 //! This module implements markdown formatting through the pulldown-cmark library.
6 //! #![feature(rustc_private)]
8 //! extern crate rustc_span;
10 //! use rustc_span::edition::Edition;
11 //! use rustdoc::html::markdown::{HeadingOffset, IdMap, Markdown, ErrorCodes};
13 //! let s = "My *markdown* _text_";
14 //! let mut id_map = IdMap::new();
15 //! let md = Markdown {
19 //! error_codes: ErrorCodes::Yes,
20 //! edition: Edition::Edition2015,
21 //! playground: &None,
22 //! heading_offset: HeadingOffset::H2,
24 //! let html = md.into_string();
25 //! // ... something using html
28 use rustc_data_structures::fx::FxHashMap;
29 use rustc_hir::def_id::DefId;
31 use rustc_middle::ty::TyCtxt;
32 use rustc_span::edition::Edition;
35 use once_cell::sync::Lazy;
37 use std::cell::RefCell;
38 use std::collections::VecDeque;
39 use std::default::Default;
41 use std::ops::{ControlFlow, Range};
44 use crate::clean::RenderedLink;
46 use crate::html::escape::Escape;
47 use crate::html::format::Buffer;
48 use crate::html::highlight;
49 use crate::html::length_limit::HtmlWithLimit;
50 use crate::html::toc::TocBuilder;
53 html, BrokenLink, CodeBlockKind, CowStr, Event, LinkType, Options, Parser, Tag,
59 const MAX_HEADER_LEVEL: u32 = 6;
61 /// Options for rendering Markdown in the main body of documentation.
62 pub(crate) fn main_body_opts() -> Options {
63 Options::ENABLE_TABLES
64 | Options::ENABLE_FOOTNOTES
65 | Options::ENABLE_STRIKETHROUGH
66 | Options::ENABLE_TASKLISTS
67 | Options::ENABLE_SMART_PUNCTUATION
70 /// Options for rendering Markdown in summaries (e.g., in search results).
71 pub(crate) fn summary_opts() -> Options {
72 Options::ENABLE_TABLES
73 | Options::ENABLE_FOOTNOTES
74 | Options::ENABLE_STRIKETHROUGH
75 | Options::ENABLE_TASKLISTS
76 | Options::ENABLE_SMART_PUNCTUATION
79 #[derive(Debug, Clone, Copy)]
80 pub enum HeadingOffset {
89 /// When `to_string` is called, this struct will emit the HTML corresponding to
90 /// the rendered version of the contained markdown string.
91 pub struct Markdown<'a> {
93 /// A list of link replacements.
94 pub links: &'a [RenderedLink],
95 /// The current list of used header IDs.
96 pub ids: &'a mut IdMap,
97 /// Whether to allow the use of explicit error codes in doctest lang strings.
98 pub error_codes: ErrorCodes,
99 /// Default edition to use when parsing doctests (to add a `fn main`).
100 pub edition: Edition,
101 pub playground: &'a Option<Playground>,
102 /// Offset at which we render headings.
103 /// E.g. if `heading_offset: HeadingOffset::H2`, then `# something` renders an `<h2>`.
104 pub heading_offset: HeadingOffset,
106 /// A tuple struct like `Markdown` that renders the markdown with a table of contents.
107 pub(crate) struct MarkdownWithToc<'a>(
109 pub(crate) &'a mut IdMap,
110 pub(crate) ErrorCodes,
112 pub(crate) &'a Option<Playground>,
114 /// A tuple struct like `Markdown` that renders the markdown escaping HTML tags.
115 pub(crate) struct MarkdownHtml<'a>(
117 pub(crate) &'a mut IdMap,
118 pub(crate) ErrorCodes,
120 pub(crate) &'a Option<Playground>,
122 /// A tuple struct like `Markdown` that renders only the first paragraph.
123 pub(crate) struct MarkdownSummaryLine<'a>(pub &'a str, pub &'a [RenderedLink]);
125 #[derive(Copy, Clone, PartialEq, Debug)]
126 pub enum ErrorCodes {
132 pub(crate) fn from(b: bool) -> Self {
134 true => ErrorCodes::Yes,
135 false => ErrorCodes::No,
139 pub(crate) fn as_bool(self) -> bool {
141 ErrorCodes::Yes => true,
142 ErrorCodes::No => false,
147 /// Controls whether a line will be hidden or shown in HTML output.
149 /// All lines are used in documentation tests.
156 fn for_html(self) -> Option<Cow<'a, str>> {
158 Line::Shown(l) => Some(l),
159 Line::Hidden(_) => None,
163 fn for_code(self) -> Cow<'a, str> {
166 Line::Hidden(l) => Cow::Borrowed(l),
171 // FIXME: There is a minor inconsistency here. For lines that start with ##, we
172 // have no easy way of removing a potential single space after the hashes, which
173 // is done in the single # case. This inconsistency seems okay, if non-ideal. In
174 // order to fix it we'd have to iterate to find the first non-# character, and
175 // then reallocate to remove it; which would make us return a String.
176 fn map_line(s: &str) -> Line<'_> {
177 let trimmed = s.trim();
178 if trimmed.starts_with("##") {
179 Line::Shown(Cow::Owned(s.replacen("##", "#", 1)))
180 } else if let Some(stripped) = trimmed.strip_prefix("# ") {
182 Line::Hidden(stripped)
183 } else if trimmed == "#" {
184 // We cannot handle '#text' because it could be #[attr].
187 Line::Shown(Cow::Borrowed(s))
191 /// Convert chars from a title for an id.
193 /// "Hello, world!" -> "hello-world"
194 fn slugify(c: char) -> Option<char> {
195 if c.is_alphanumeric() || c == '-' || c == '_' {
196 if c.is_ascii() { Some(c.to_ascii_lowercase()) } else { Some(c) }
197 } else if c.is_whitespace() && c.is_ascii() {
204 #[derive(Clone, Debug)]
205 pub struct Playground {
206 pub crate_name: Option<String>,
210 /// Adds syntax highlighting and playground Run buttons to Rust code blocks.
211 struct CodeBlocks<'p, 'a, I: Iterator<Item = Event<'a>>> {
213 check_error_codes: ErrorCodes,
215 // Information about the playground if a URL has been specified, containing an
216 // optional crate name and the URL.
217 playground: &'p Option<Playground>,
220 impl<'p, 'a, I: Iterator<Item = Event<'a>>> CodeBlocks<'p, 'a, I> {
223 error_codes: ErrorCodes,
225 playground: &'p Option<Playground>,
227 CodeBlocks { inner: iter, check_error_codes: error_codes, edition, playground }
231 impl<'a, I: Iterator<Item = Event<'a>>> Iterator for CodeBlocks<'_, 'a, I> {
232 type Item = Event<'a>;
234 fn next(&mut self) -> Option<Self::Item> {
235 let event = self.inner.next();
240 let Some(Event::Start(Tag::CodeBlock(kind))) = event else {
244 let mut origtext = String::new();
245 for event in &mut self.inner {
247 Event::End(Tag::CodeBlock(..)) => break,
248 Event::Text(ref s) => {
249 origtext.push_str(s);
254 let lines = origtext.lines().filter_map(|l| map_line(l).for_html());
255 let text = lines.intersperse("\n".into()).collect::<String>();
257 let parse_result = match kind {
258 CodeBlockKind::Fenced(ref lang) => {
260 LangString::parse_without_check(lang, self.check_error_codes, false);
261 if !parse_result.rust {
262 return Some(Event::Html(
264 "<div class=\"example-wrap\">\
265 <pre class=\"language-{}\"><code>{}</code></pre>\
275 CodeBlockKind::Indented => Default::default(),
278 compile_fail = parse_result.compile_fail;
279 should_panic = parse_result.should_panic;
280 ignore = parse_result.ignore;
281 edition = parse_result.edition;
283 let explicit_edition = edition.is_some();
284 let edition = edition.unwrap_or(self.edition);
286 let playground_button = self.playground.as_ref().and_then(|playground| {
287 let krate = &playground.crate_name;
288 let url = &playground.url;
294 .map(|l| map_line(l).for_code())
295 .intersperse("\n".into())
296 .collect::<String>();
297 let krate = krate.as_ref().map(|s| &**s);
299 doctest::make_test(&test, krate, false, &Default::default(), edition, None);
300 let channel = if test.contains("#![feature(") { "&version=nightly" } else { "" };
302 // These characters don't need to be escaped in a URI.
303 // FIXME: use a library function for percent encoding.
304 fn dont_escape(c: u8) -> bool {
305 (b'a' <= c && c <= b'z')
306 || (b'A' <= c && c <= b'Z')
307 || (b'0' <= c && c <= b'9')
318 let mut test_escaped = String::new();
319 for b in test.bytes() {
321 test_escaped.push(char::from(b));
323 write!(test_escaped, "%{:02X}", b).unwrap();
327 r#"<a class="test-arrow" target="_blank" href="{}?code={}{}&edition={}">Run</a>"#,
328 url, test_escaped, channel, edition,
332 let tooltip = if ignore != Ignore::None {
333 highlight::Tooltip::Ignore
334 } else if compile_fail {
335 highlight::Tooltip::CompileFail
336 } else if should_panic {
337 highlight::Tooltip::ShouldPanic
338 } else if explicit_edition {
339 highlight::Tooltip::Edition(edition)
341 highlight::Tooltip::None
344 // insert newline to clearly separate it from the
345 // previous block so we can shorten the html output
346 let mut s = Buffer::new();
349 highlight::render_example_with_highlighting(
353 playground_button.as_deref(),
355 Some(Event::Html(s.into_inner().into()))
359 /// Make headings links with anchor IDs and build up TOC.
360 struct LinkReplacer<'a, I: Iterator<Item = Event<'a>>> {
362 links: &'a [RenderedLink],
363 shortcut_link: Option<&'a RenderedLink>,
366 impl<'a, I: Iterator<Item = Event<'a>>> LinkReplacer<'a, I> {
367 fn new(iter: I, links: &'a [RenderedLink]) -> Self {
368 LinkReplacer { inner: iter, links, shortcut_link: None }
372 impl<'a, I: Iterator<Item = Event<'a>>> Iterator for LinkReplacer<'a, I> {
373 type Item = Event<'a>;
375 fn next(&mut self) -> Option<Self::Item> {
376 let mut event = self.inner.next();
378 // Replace intra-doc links and remove disambiguators from shortcut links (`[fn@f]`).
380 // This is a shortcut link that was resolved by the broken_link_callback: `[fn@f]`
381 // Remove any disambiguator.
382 Some(Event::Start(Tag::Link(
383 // [fn@f] or [fn@f][]
384 LinkType::ShortcutUnknown | LinkType::CollapsedUnknown,
388 debug!("saw start of shortcut link to {} with title {}", dest, title);
389 // If this is a shortcut link, it was resolved by the broken_link_callback.
390 // So the URL will already be updated properly.
391 let link = self.links.iter().find(|&link| *link.href == **dest);
392 // Since this is an external iterator, we can't replace the inner text just yet.
393 // Store that we saw a link so we know to replace it later.
394 if let Some(link) = link {
395 trace!("it matched");
396 assert!(self.shortcut_link.is_none(), "shortcut links cannot be nested");
397 self.shortcut_link = Some(link);
400 // Now that we're done with the shortcut link, don't replace any more text.
401 Some(Event::End(Tag::Link(
402 LinkType::ShortcutUnknown | LinkType::CollapsedUnknown,
406 debug!("saw end of shortcut link to {}", dest);
407 if self.links.iter().any(|link| *link.href == **dest) {
408 assert!(self.shortcut_link.is_some(), "saw closing link without opening tag");
409 self.shortcut_link = None;
412 // Handle backticks in inline code blocks, but only if we're in the middle of a shortcut link.
414 Some(Event::Code(text)) => {
415 trace!("saw code {}", text);
416 if let Some(link) = self.shortcut_link {
417 trace!("original text was {}", link.original_text);
418 // NOTE: this only replaces if the code block is the *entire* text.
419 // If only part of the link has code highlighting, the disambiguator will not be removed.
421 // This is a limitation from `collect_intra_doc_links`: it passes a full link,
422 // and does not distinguish at all between code blocks.
423 // So we could never be sure we weren't replacing too much:
424 // [fn@my_`f`unc] is treated the same as [my_func()] in that pass.
426 // NOTE: &[1..len() - 1] is to strip the backticks
427 if **text == link.original_text[1..link.original_text.len() - 1] {
428 debug!("replacing {} with {}", text, link.new_text);
429 *text = CowStr::Borrowed(&link.new_text);
433 // Replace plain text in links, but only in the middle of a shortcut link.
435 Some(Event::Text(text)) => {
436 trace!("saw text {}", text);
437 if let Some(link) = self.shortcut_link {
438 trace!("original text was {}", link.original_text);
439 // NOTE: same limitations as `Event::Code`
440 if **text == *link.original_text {
441 debug!("replacing {} with {}", text, link.new_text);
442 *text = CowStr::Borrowed(&link.new_text);
446 // If this is a link, but not a shortcut link,
447 // replace the URL, since the broken_link_callback was not called.
448 Some(Event::Start(Tag::Link(_, dest, _))) => {
449 if let Some(link) = self.links.iter().find(|&link| *link.original_text == **dest) {
450 *dest = CowStr::Borrowed(link.href.as_ref());
453 // Anything else couldn't have been a valid Rust path, so no need to replace the text.
457 // Yield the modified event
462 /// Wrap HTML tables into `<div>` to prevent having the doc blocks width being too big.
463 struct TableWrapper<'a, I: Iterator<Item = Event<'a>>> {
465 stored_events: VecDeque<Event<'a>>,
468 impl<'a, I: Iterator<Item = Event<'a>>> TableWrapper<'a, I> {
469 fn new(iter: I) -> Self {
470 Self { inner: iter, stored_events: VecDeque::new() }
474 impl<'a, I: Iterator<Item = Event<'a>>> Iterator for TableWrapper<'a, I> {
475 type Item = Event<'a>;
477 fn next(&mut self) -> Option<Self::Item> {
478 if let Some(first) = self.stored_events.pop_front() {
482 let event = self.inner.next()?;
485 Event::Start(Tag::Table(t)) => {
486 self.stored_events.push_back(Event::Start(Tag::Table(t)));
487 Event::Html(CowStr::Borrowed("<div>"))
489 Event::End(Tag::Table(t)) => {
490 self.stored_events.push_back(Event::Html(CowStr::Borrowed("</div>")));
491 Event::End(Tag::Table(t))
498 type SpannedEvent<'a> = (Event<'a>, Range<usize>);
500 /// Make headings links with anchor IDs and build up TOC.
501 struct HeadingLinks<'a, 'b, 'ids, I> {
503 toc: Option<&'b mut TocBuilder>,
504 buf: VecDeque<SpannedEvent<'a>>,
505 id_map: &'ids mut IdMap,
506 heading_offset: HeadingOffset,
509 impl<'a, 'b, 'ids, I> HeadingLinks<'a, 'b, 'ids, I> {
512 toc: Option<&'b mut TocBuilder>,
513 ids: &'ids mut IdMap,
514 heading_offset: HeadingOffset,
516 HeadingLinks { inner: iter, toc, buf: VecDeque::new(), id_map: ids, heading_offset }
520 impl<'a, 'b, 'ids, I: Iterator<Item = SpannedEvent<'a>>> Iterator
521 for HeadingLinks<'a, 'b, 'ids, I>
523 type Item = SpannedEvent<'a>;
525 fn next(&mut self) -> Option<Self::Item> {
526 if let Some(e) = self.buf.pop_front() {
530 let event = self.inner.next();
531 if let Some((Event::Start(Tag::Heading(level, _, _)), _)) = event {
532 let mut id = String::new();
533 for event in &mut self.inner {
535 Event::End(Tag::Heading(..)) => break,
536 Event::Start(Tag::Link(_, _, _)) | Event::End(Tag::Link(..)) => {}
537 Event::Text(text) | Event::Code(text) => {
538 id.extend(text.chars().filter_map(slugify));
539 self.buf.push_back(event);
541 _ => self.buf.push_back(event),
544 let id = self.id_map.derive(id);
546 if let Some(ref mut builder) = self.toc {
547 let mut html_header = String::new();
548 html::push_html(&mut html_header, self.buf.iter().map(|(ev, _)| ev.clone()));
549 let sec = builder.push(level as u32, html_header, id.clone());
550 self.buf.push_front((Event::Html(format!("{} ", sec).into()), 0..0));
554 std::cmp::min(level as u32 + (self.heading_offset as u32), MAX_HEADER_LEVEL);
555 self.buf.push_back((Event::Html(format!("</a></h{}>", level).into()), 0..0));
557 let start_tags = format!(
558 "<h{level} id=\"{id}\">\
563 return Some((Event::Html(start_tags.into()), 0..0));
569 /// Extracts just the first paragraph.
570 struct SummaryLine<'a, I: Iterator<Item = Event<'a>>> {
576 impl<'a, I: Iterator<Item = Event<'a>>> SummaryLine<'a, I> {
577 fn new(iter: I) -> Self {
578 SummaryLine { inner: iter, started: false, depth: 0 }
582 fn check_if_allowed_tag(t: &Tag<'_>) -> bool {
585 Tag::Paragraph | Tag::Item | Tag::Emphasis | Tag::Strong | Tag::Link(..) | Tag::BlockQuote
589 fn is_forbidden_tag(t: &Tag<'_>) -> bool {
590 matches!(t, Tag::CodeBlock(_) | Tag::Table(_) | Tag::TableHead | Tag::TableRow | Tag::TableCell)
593 impl<'a, I: Iterator<Item = Event<'a>>> Iterator for SummaryLine<'a, I> {
594 type Item = Event<'a>;
596 fn next(&mut self) -> Option<Self::Item> {
597 if self.started && self.depth == 0 {
603 if let Some(event) = self.inner.next() {
604 let mut is_start = true;
605 let is_allowed_tag = match event {
606 Event::Start(ref c) => {
607 if is_forbidden_tag(c) {
611 check_if_allowed_tag(c)
613 Event::End(ref c) => {
614 if is_forbidden_tag(c) {
619 check_if_allowed_tag(c)
623 return if !is_allowed_tag {
625 Some(Event::Start(Tag::Paragraph))
627 Some(Event::End(Tag::Paragraph))
637 /// Moves all footnote definitions to the end and add back links to the
639 struct Footnotes<'a, I> {
641 footnotes: FxHashMap<String, (Vec<Event<'a>>, u16)>,
644 impl<'a, I> Footnotes<'a, I> {
645 fn new(iter: I) -> Self {
646 Footnotes { inner: iter, footnotes: FxHashMap::default() }
649 fn get_entry(&mut self, key: &str) -> &mut (Vec<Event<'a>>, u16) {
650 let new_id = self.footnotes.len() + 1;
651 let key = key.to_owned();
652 self.footnotes.entry(key).or_insert((Vec::new(), new_id as u16))
656 impl<'a, I: Iterator<Item = SpannedEvent<'a>>> Iterator for Footnotes<'a, I> {
657 type Item = SpannedEvent<'a>;
659 fn next(&mut self) -> Option<Self::Item> {
661 match self.inner.next() {
662 Some((Event::FootnoteReference(ref reference), range)) => {
663 let entry = self.get_entry(reference);
664 let reference = format!(
665 "<sup id=\"fnref{0}\"><a href=\"#fn{0}\">{0}</a></sup>",
668 return Some((Event::Html(reference.into()), range));
670 Some((Event::Start(Tag::FootnoteDefinition(def)), _)) => {
671 let mut content = Vec::new();
672 for (event, _) in &mut self.inner {
673 if let Event::End(Tag::FootnoteDefinition(..)) = event {
678 let entry = self.get_entry(&def);
679 (*entry).0 = content;
681 Some(e) => return Some(e),
683 if !self.footnotes.is_empty() {
684 let mut v: Vec<_> = self.footnotes.drain().map(|(_, x)| x).collect();
685 v.sort_by(|a, b| a.1.cmp(&b.1));
686 let mut ret = String::from("<div class=\"footnotes\"><hr><ol>");
687 for (mut content, id) in v {
688 write!(ret, "<li id=\"fn{}\">", id).unwrap();
689 let mut is_paragraph = false;
690 if let Some(&Event::End(Tag::Paragraph)) = content.last() {
694 html::push_html(&mut ret, content.into_iter());
695 write!(ret, " <a href=\"#fnref{}\">↩</a>", id).unwrap();
697 ret.push_str("</p>");
699 ret.push_str("</li>");
701 ret.push_str("</ol></div>");
702 return Some((Event::Html(ret.into()), 0..0));
712 pub(crate) fn find_testable_code<T: doctest::Tester>(
715 error_codes: ErrorCodes,
716 enable_per_target_ignores: bool,
717 extra_info: Option<&ExtraInfo<'_>>,
719 let mut parser = Parser::new(doc).into_offset_iter();
720 let mut prev_offset = 0;
721 let mut nb_lines = 0;
722 let mut register_header = None;
723 while let Some((event, offset)) = parser.next() {
725 Event::Start(Tag::CodeBlock(kind)) => {
726 let block_info = match kind {
727 CodeBlockKind::Fenced(ref lang) => {
734 enable_per_target_ignores,
739 CodeBlockKind::Indented => Default::default(),
741 if !block_info.rust {
745 let mut test_s = String::new();
747 while let Some((Event::Text(s), _)) = parser.next() {
752 .map(|l| map_line(l).for_code())
753 .collect::<Vec<Cow<'_, str>>>()
756 nb_lines += doc[prev_offset..offset.start].lines().count();
757 // If there are characters between the preceding line ending and
758 // this code block, `str::lines` will return an additional line,
759 // which we subtract here.
760 if nb_lines != 0 && !&doc[prev_offset..offset.start].ends_with('\n') {
763 let line = tests.get_line() + nb_lines + 1;
764 tests.add_test(text, block_info, line);
765 prev_offset = offset.start;
767 Event::Start(Tag::Heading(level, _, _)) => {
768 register_header = Some(level as u32);
770 Event::Text(ref s) if register_header.is_some() => {
771 let level = register_header.unwrap();
773 tests.register_header("", level);
775 tests.register_header(s, level);
777 register_header = None;
784 pub(crate) struct ExtraInfo<'tcx> {
795 impl<'tcx> ExtraInfo<'tcx> {
796 pub(crate) fn new(tcx: TyCtxt<'tcx>, hir_id: HirId, sp: Span) -> ExtraInfo<'tcx> {
797 ExtraInfo { id: ExtraInfoId::Hir(hir_id), sp, tcx }
800 pub(crate) fn new_did(tcx: TyCtxt<'tcx>, did: DefId, sp: Span) -> ExtraInfo<'tcx> {
801 ExtraInfo { id: ExtraInfoId::Def(did), sp, tcx }
804 fn error_invalid_codeblock_attr(&self, msg: &str, help: &str) {
805 let hir_id = match self.id {
806 ExtraInfoId::Hir(hir_id) => hir_id,
807 ExtraInfoId::Def(item_did) => {
808 match item_did.as_local() {
809 Some(item_did) => self.tcx.hir().local_def_id_to_hir_id(item_did),
811 // If non-local, no need to check anything.
817 self.tcx.struct_span_lint_hir(
818 crate::lint::INVALID_CODEBLOCK_ATTRIBUTES,
822 let mut diag = lint.build(msg);
830 #[derive(Eq, PartialEq, Clone, Debug)]
831 pub(crate) struct LangString {
833 pub(crate) should_panic: bool,
834 pub(crate) no_run: bool,
835 pub(crate) ignore: Ignore,
836 pub(crate) rust: bool,
837 pub(crate) test_harness: bool,
838 pub(crate) compile_fail: bool,
839 pub(crate) error_codes: Vec<String>,
840 pub(crate) edition: Option<Edition>,
843 #[derive(Eq, PartialEq, Clone, Debug)]
844 pub(crate) enum Ignore {
850 impl Default for LangString {
851 fn default() -> Self {
853 original: String::new(),
856 ignore: Ignore::None,
860 error_codes: Vec::new(),
867 fn parse_without_check(
869 allow_error_code_check: ErrorCodes,
870 enable_per_target_ignores: bool,
872 Self::parse(string, allow_error_code_check, enable_per_target_ignores, None)
875 fn tokens(string: &str) -> impl Iterator<Item = &str> {
876 // Pandoc, which Rust once used for generating documentation,
877 // expects lang strings to be surrounded by `{}` and for each token
878 // to be proceeded by a `.`. Since some of these lang strings are still
879 // loose in the wild, we strip a pair of surrounding `{}` from the lang
880 // string and a leading `.` from each token.
882 let string = string.trim();
884 let first = string.chars().next();
885 let last = string.chars().last();
887 let string = if first == Some('{') && last == Some('}') {
888 &string[1..string.len() - 1]
894 .split(|c| c == ',' || c == ' ' || c == '\t')
896 .map(|token| token.strip_prefix('.').unwrap_or(token))
897 .filter(|token| !token.is_empty())
902 allow_error_code_check: ErrorCodes,
903 enable_per_target_ignores: bool,
904 extra: Option<&ExtraInfo<'_>>,
906 let allow_error_code_check = allow_error_code_check.as_bool();
907 let mut seen_rust_tags = false;
908 let mut seen_other_tags = false;
909 let mut data = LangString::default();
910 let mut ignores = vec![];
912 data.original = string.to_owned();
914 for token in Self::tokens(string) {
917 data.should_panic = true;
918 seen_rust_tags = !seen_other_tags;
922 seen_rust_tags = !seen_other_tags;
925 data.ignore = Ignore::All;
926 seen_rust_tags = !seen_other_tags;
928 x if x.starts_with("ignore-") => {
929 if enable_per_target_ignores {
930 ignores.push(x.trim_start_matches("ignore-").to_owned());
931 seen_rust_tags = !seen_other_tags;
936 seen_rust_tags = true;
939 data.test_harness = true;
940 seen_rust_tags = !seen_other_tags || seen_rust_tags;
943 data.compile_fail = true;
944 seen_rust_tags = !seen_other_tags || seen_rust_tags;
947 x if x.starts_with("edition") => {
948 data.edition = x[7..].parse::<Edition>().ok();
950 x if allow_error_code_check && x.starts_with('E') && x.len() == 5 => {
951 if x[1..].parse::<u32>().is_ok() {
952 data.error_codes.push(x.to_owned());
953 seen_rust_tags = !seen_other_tags || seen_rust_tags;
955 seen_other_tags = true;
958 x if extra.is_some() => {
959 let s = x.to_lowercase();
960 if let Some((flag, help)) = if s == "compile-fail"
961 || s == "compile_fail"
962 || s == "compilefail"
966 "the code block will either not be tested if not marked as a rust one \
967 or won't fail if it compiles successfully",
969 } else if s == "should-panic" || s == "should_panic" || s == "shouldpanic" {
972 "the code block will either not be tested if not marked as a rust one \
973 or won't fail if it doesn't panic when running",
975 } else if s == "no-run" || s == "no_run" || s == "norun" {
978 "the code block will either not be tested if not marked as a rust one \
979 or will be run (which you might not want)",
981 } else if s == "test-harness" || s == "test_harness" || s == "testharness" {
984 "the code block will either not be tested if not marked as a rust one \
985 or the code will be wrapped inside a main function",
990 if let Some(extra) = extra {
991 extra.error_invalid_codeblock_attr(
992 &format!("unknown attribute `{}`. Did you mean `{}`?", x, flag),
997 seen_other_tags = true;
999 _ => seen_other_tags = true,
1003 // ignore-foo overrides ignore
1004 if !ignores.is_empty() {
1005 data.ignore = Ignore::Some(ignores);
1008 data.rust &= !seen_other_tags || seen_rust_tags;
1015 pub fn into_string(self) -> String {
1026 // This is actually common enough to special-case
1028 return String::new();
1030 let mut replacer = |broken_link: BrokenLink<'_>| {
1033 .find(|link| link.original_text.as_str() == &*broken_link.reference)
1034 .map(|link| (link.href.as_str().into(), link.new_text.as_str().into()))
1037 let p = Parser::new_with_broken_link_callback(md, main_body_opts(), Some(&mut replacer));
1038 let p = p.into_offset_iter();
1040 let mut s = String::with_capacity(md.len() * 3 / 2);
1042 let p = HeadingLinks::new(p, None, ids, heading_offset);
1043 let p = Footnotes::new(p);
1044 let p = LinkReplacer::new(p.map(|(ev, _)| ev), links);
1045 let p = TableWrapper::new(p);
1046 let p = CodeBlocks::new(p, codes, edition, playground);
1047 html::push_html(&mut s, p);
1053 impl MarkdownWithToc<'_> {
1054 pub(crate) fn into_string(self) -> String {
1055 let MarkdownWithToc(md, ids, codes, edition, playground) = self;
1057 let p = Parser::new_ext(md, main_body_opts()).into_offset_iter();
1059 let mut s = String::with_capacity(md.len() * 3 / 2);
1061 let mut toc = TocBuilder::new();
1064 let p = HeadingLinks::new(p, Some(&mut toc), ids, HeadingOffset::H1);
1065 let p = Footnotes::new(p);
1066 let p = TableWrapper::new(p.map(|(ev, _)| ev));
1067 let p = CodeBlocks::new(p, codes, edition, playground);
1068 html::push_html(&mut s, p);
1071 format!("<nav id=\"TOC\">{}</nav>{}", toc.into_toc().print(), s)
1075 impl MarkdownHtml<'_> {
1076 pub(crate) fn into_string(self) -> String {
1077 let MarkdownHtml(md, ids, codes, edition, playground) = self;
1079 // This is actually common enough to special-case
1081 return String::new();
1083 let p = Parser::new_ext(md, main_body_opts()).into_offset_iter();
1085 // Treat inline HTML as plain text.
1086 let p = p.map(|event| match event.0 {
1087 Event::Html(text) => (Event::Text(text), event.1),
1091 let mut s = String::with_capacity(md.len() * 3 / 2);
1093 let p = HeadingLinks::new(p, None, ids, HeadingOffset::H1);
1094 let p = Footnotes::new(p);
1095 let p = TableWrapper::new(p.map(|(ev, _)| ev));
1096 let p = CodeBlocks::new(p, codes, edition, playground);
1097 html::push_html(&mut s, p);
1103 impl MarkdownSummaryLine<'_> {
1104 pub(crate) fn into_string(self) -> String {
1105 let MarkdownSummaryLine(md, links) = self;
1106 // This is actually common enough to special-case
1108 return String::new();
1111 let mut replacer = |broken_link: BrokenLink<'_>| {
1114 .find(|link| link.original_text.as_str() == &*broken_link.reference)
1115 .map(|link| (link.href.as_str().into(), link.new_text.as_str().into()))
1118 let p = Parser::new_with_broken_link_callback(md, summary_opts(), Some(&mut replacer));
1120 let mut s = String::new();
1122 html::push_html(&mut s, LinkReplacer::new(SummaryLine::new(p), links));
1128 /// Renders a subset of Markdown in the first paragraph of the provided Markdown.
1130 /// - *Italics*, **bold**, and `inline code` styles **are** rendered.
1131 /// - Headings and links are stripped (though the text *is* rendered).
1132 /// - HTML, code blocks, and everything else are ignored.
1134 /// Returns a tuple of the rendered HTML string and whether the output was shortened
1135 /// due to the provided `length_limit`.
1136 fn markdown_summary_with_limit(
1138 link_names: &[RenderedLink],
1139 length_limit: usize,
1140 ) -> (String, bool) {
1142 return (String::new(), false);
1145 let mut replacer = |broken_link: BrokenLink<'_>| {
1148 .find(|link| link.original_text.as_str() == &*broken_link.reference)
1149 .map(|link| (link.href.as_str().into(), link.new_text.as_str().into()))
1152 let p = Parser::new_with_broken_link_callback(md, summary_opts(), Some(&mut replacer));
1153 let mut p = LinkReplacer::new(p, link_names);
1155 let mut buf = HtmlWithLimit::new(length_limit);
1156 let mut stopped_early = false;
1157 p.try_for_each(|event| {
1159 Event::Text(text) => {
1161 text.split_inclusive(char::is_whitespace).try_for_each(|word| buf.push(word));
1163 stopped_early = true;
1167 Event::Code(code) => {
1168 buf.open_tag("code");
1169 let r = buf.push(code);
1171 stopped_early = true;
1177 Event::Start(tag) => match tag {
1178 Tag::Emphasis => buf.open_tag("em"),
1179 Tag::Strong => buf.open_tag("strong"),
1180 Tag::CodeBlock(..) => return ControlFlow::BREAK,
1183 Event::End(tag) => match tag {
1184 Tag::Emphasis | Tag::Strong => buf.close_tag(),
1185 Tag::Paragraph | Tag::Heading(..) => return ControlFlow::BREAK,
1188 Event::HardBreak | Event::SoftBreak => buf.push(" ")?,
1191 ControlFlow::CONTINUE
1194 (buf.finish(), stopped_early)
1197 /// Renders a shortened first paragraph of the given Markdown as a subset of Markdown,
1198 /// making it suitable for contexts like the search index.
1200 /// Will shorten to 59 or 60 characters, including an ellipsis (…) if it was shortened.
1202 /// See [`markdown_summary_with_limit`] for details about what is rendered and what is not.
1203 pub(crate) fn short_markdown_summary(markdown: &str, link_names: &[RenderedLink]) -> String {
1204 let (mut s, was_shortened) = markdown_summary_with_limit(markdown, link_names, 59);
1213 /// Renders the first paragraph of the provided markdown as plain text.
1214 /// Useful for alt-text.
1216 /// - Headings, links, and formatting are stripped.
1217 /// - Inline code is rendered as-is, surrounded by backticks.
1218 /// - HTML and code blocks are ignored.
1219 pub(crate) fn plain_text_summary(md: &str) -> String {
1221 return String::new();
1224 let mut s = String::with_capacity(md.len() * 3 / 2);
1226 for event in Parser::new_ext(md, summary_opts()) {
1228 Event::Text(text) => s.push_str(text),
1229 Event::Code(code) => {
1234 Event::HardBreak | Event::SoftBreak => s.push(' '),
1235 Event::Start(Tag::CodeBlock(..)) => break,
1236 Event::End(Tag::Paragraph) => break,
1237 Event::End(Tag::Heading(..)) => break,
1246 pub(crate) struct MarkdownLink {
1249 pub range: Range<usize>,
1252 pub(crate) fn markdown_links<R>(
1254 filter_map: impl Fn(MarkdownLink) -> Option<R>,
1260 let links = RefCell::new(vec![]);
1262 // FIXME: remove this function once pulldown_cmark can provide spans for link definitions.
1263 let locate = |s: &str, fallback: Range<usize>| unsafe {
1264 let s_start = s.as_ptr();
1265 let s_end = s_start.add(s.len());
1266 let md_start = md.as_ptr();
1267 let md_end = md_start.add(md.len());
1268 if md_start <= s_start && s_end <= md_end {
1269 let start = s_start.offset_from(md_start) as usize;
1270 let end = s_end.offset_from(md_start) as usize;
1277 let span_for_link = |link: &CowStr<'_>, span: Range<usize>| {
1278 // For diagnostics, we want to underline the link's definition but `span` will point at
1279 // where the link is used. This is a problem for reference-style links, where the definition
1280 // is separate from the usage.
1282 // `Borrowed` variant means the string (the link's destination) may come directly from
1283 // the markdown text and we can locate the original link destination.
1284 // NOTE: LinkReplacer also provides `Borrowed` but possibly from other sources,
1285 // so `locate()` can fall back to use `span`.
1286 CowStr::Borrowed(s) => locate(s, span),
1288 // For anything else, we can only use the provided range.
1289 CowStr::Boxed(_) | CowStr::Inlined(_) => span,
1293 let mut push = |link: BrokenLink<'_>| {
1294 let span = span_for_link(&link.reference, link.span);
1295 filter_map(MarkdownLink {
1296 kind: LinkType::ShortcutUnknown,
1297 link: link.reference.to_string(),
1300 .map(|link| links.borrow_mut().push(link));
1303 let p = Parser::new_with_broken_link_callback(md, main_body_opts(), Some(&mut push))
1304 .into_offset_iter();
1306 // There's no need to thread an IdMap through to here because
1307 // the IDs generated aren't going to be emitted anywhere.
1308 let mut ids = IdMap::new();
1309 let iter = Footnotes::new(HeadingLinks::new(p, None, &mut ids, HeadingOffset::H1));
1312 if let Event::Start(Tag::Link(
1313 // `<>` links cannot be intra-doc links so we skip them.
1314 kind @ (LinkType::Inline
1315 | LinkType::Reference
1316 | LinkType::ReferenceUnknown
1317 | LinkType::Collapsed
1318 | LinkType::CollapsedUnknown
1319 | LinkType::Shortcut
1320 | LinkType::ShortcutUnknown),
1325 debug!("found link: {dest}");
1326 let span = span_for_link(&dest, ev.1);
1327 filter_map(MarkdownLink { kind, link: dest.into_string(), range: span })
1328 .map(|link| links.borrow_mut().push(link));
1336 pub(crate) struct RustCodeBlock {
1337 /// The range in the markdown that the code block occupies. Note that this includes the fences
1338 /// for fenced code blocks.
1339 pub(crate) range: Range<usize>,
1340 /// The range in the markdown that the code within the code block occupies.
1341 pub(crate) code: Range<usize>,
1342 pub(crate) is_fenced: bool,
1343 pub(crate) lang_string: LangString,
1346 /// Returns a range of bytes for each code block in the markdown that is tagged as `rust` or
1347 /// untagged (and assumed to be rust).
1348 pub(crate) fn rust_code_blocks(md: &str, extra_info: &ExtraInfo<'_>) -> Vec<RustCodeBlock> {
1349 let mut code_blocks = vec![];
1355 let mut p = Parser::new_ext(md, main_body_opts()).into_offset_iter();
1357 while let Some((event, offset)) = p.next() {
1358 if let Event::Start(Tag::CodeBlock(syntax)) = event {
1359 let (lang_string, code_start, code_end, range, is_fenced) = match syntax {
1360 CodeBlockKind::Fenced(syntax) => {
1361 let syntax = syntax.as_ref();
1362 let lang_string = if syntax.is_empty() {
1365 LangString::parse(&*syntax, ErrorCodes::Yes, false, Some(extra_info))
1367 if !lang_string.rust {
1370 let (code_start, mut code_end) = match p.next() {
1371 Some((Event::Text(_), offset)) => (offset.start, offset.end),
1372 Some((_, sub_offset)) => {
1373 let code = Range { start: sub_offset.start, end: sub_offset.start };
1374 code_blocks.push(RustCodeBlock {
1383 let code = Range { start: offset.end, end: offset.end };
1384 code_blocks.push(RustCodeBlock {
1393 while let Some((Event::Text(_), offset)) = p.next() {
1394 code_end = offset.end;
1396 (lang_string, code_start, code_end, offset, true)
1398 CodeBlockKind::Indented => {
1399 // The ending of the offset goes too far sometime so we reduce it by one in
1401 if offset.end > offset.start && md.get(offset.end..=offset.end) == Some("\n") {
1403 LangString::default(),
1406 Range { start: offset.start, end: offset.end - 1 },
1410 (LangString::default(), offset.start, offset.end, offset, false)
1415 code_blocks.push(RustCodeBlock {
1418 code: Range { start: code_start, end: code_end },
1427 #[derive(Clone, Default, Debug)]
1429 map: FxHashMap<Cow<'static, str>, usize>,
1432 // The map is pre-initialized and cloned each time to avoid reinitializing it repeatedly.
1433 static DEFAULT_ID_MAP: Lazy<FxHashMap<Cow<'static, str>, usize>> = Lazy::new(|| init_id_map());
1435 fn init_id_map() -> FxHashMap<Cow<'static, str>, usize> {
1436 let mut map = FxHashMap::default();
1437 // This is the list of IDs used in Javascript.
1438 map.insert("settings".into(), 1);
1439 map.insert("not-displayed".into(), 1);
1440 map.insert("alternative-display".into(), 1);
1441 map.insert("search".into(), 1);
1442 map.insert("crate-search".into(), 1);
1443 map.insert("crate-search-div".into(), 1);
1444 // This is the list of IDs used in HTML generated in Rust (including the ones
1445 // used in tera template files).
1446 map.insert("mainThemeStyle".into(), 1);
1447 map.insert("themeStyle".into(), 1);
1448 map.insert("settings-menu".into(), 1);
1449 map.insert("help-button".into(), 1);
1450 map.insert("main-content".into(), 1);
1451 map.insert("toggle-all-docs".into(), 1);
1452 map.insert("all-types".into(), 1);
1453 map.insert("default-settings".into(), 1);
1454 map.insert("rustdoc-vars".into(), 1);
1455 map.insert("sidebar-vars".into(), 1);
1456 map.insert("copy-path".into(), 1);
1457 map.insert("TOC".into(), 1);
1458 // This is the list of IDs used by rustdoc sections (but still generated by
1460 map.insert("fields".into(), 1);
1461 map.insert("variants".into(), 1);
1462 map.insert("implementors-list".into(), 1);
1463 map.insert("synthetic-implementors-list".into(), 1);
1464 map.insert("foreign-impls".into(), 1);
1465 map.insert("implementations".into(), 1);
1466 map.insert("trait-implementations".into(), 1);
1467 map.insert("synthetic-implementations".into(), 1);
1468 map.insert("blanket-implementations".into(), 1);
1469 map.insert("required-associated-types".into(), 1);
1470 map.insert("provided-associated-types".into(), 1);
1471 map.insert("provided-associated-consts".into(), 1);
1472 map.insert("required-associated-consts".into(), 1);
1473 map.insert("required-methods".into(), 1);
1474 map.insert("provided-methods".into(), 1);
1475 map.insert("implementors".into(), 1);
1476 map.insert("synthetic-implementors".into(), 1);
1477 map.insert("implementations-list".into(), 1);
1478 map.insert("trait-implementations-list".into(), 1);
1479 map.insert("synthetic-implementations-list".into(), 1);
1480 map.insert("blanket-implementations-list".into(), 1);
1481 map.insert("deref-methods".into(), 1);
1482 map.insert("layout".into(), 1);
1487 pub fn new() -> Self {
1488 IdMap { map: DEFAULT_ID_MAP.clone() }
1491 pub(crate) fn derive<S: AsRef<str> + ToString>(&mut self, candidate: S) -> String {
1492 let id = match self.map.get_mut(candidate.as_ref()) {
1493 None => candidate.to_string(),
1495 let id = format!("{}-{}", candidate.as_ref(), *a);
1501 self.map.insert(id.clone().into(), 1);