1 // Copyright 2013-2014 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! Markdown formatting for rustdoc
13 //! This module implements markdown formatting through the hoedown C-library
14 //! (bundled into the rust runtime). This module self-contains the C bindings
15 //! and necessary legwork to render markdown, and exposes all of the
16 //! functionality through a unit-struct, `Markdown`, which has an implementation
17 //! of `fmt::Display`. Example usage:
20 //! use rustdoc::html::markdown::Markdown;
22 //! let s = "My *markdown* _text_";
23 //! let html = format!("{}", Markdown(s));
24 //! // ... something using html
27 #![allow(non_camel_case_types)]
30 use std::ascii::AsciiExt;
31 use std::cell::RefCell;
32 use std::default::Default;
33 //use std::ffi::CString;
34 use std::fmt::{self, Write};
37 use syntax::feature_gate::UnstableFeatures;
38 use syntax::codemap::Span;
40 use html::render::derive_id;
41 use html::toc::TocBuilder;
43 use html::escape::Escape;
46 use pulldown_cmark::{self, Event, Parser, Tag};
48 /// A unit struct which has the `fmt::Display` trait implemented. When
49 /// formatted, this struct will emit the HTML corresponding to the rendered
50 /// version of the contained markdown string.
51 // The second parameter is whether we need a shorter version or not.
52 pub struct Markdown<'a>(pub &'a str, pub bool);
53 /// A unit struct like `Markdown`, that renders the markdown with a
54 /// table of contents.
55 pub struct MarkdownWithToc<'a>(pub &'a str);
56 /// A unit struct like `Markdown`, that renders the markdown escaping HTML tags.
57 pub struct MarkdownHtml<'a>(pub &'a str);
59 /// Returns Some(code) if `s` is a line that should be stripped from
60 /// documentation but used in example code. `code` is the portion of
61 /// `s` that should be used in tests. (None for lines that should be
63 fn stripped_filtered_line<'a>(s: &'a str) -> Option<&'a str> {
64 let trimmed = s.trim();
67 } else if trimmed.starts_with("# ") {
74 /// Returns a new string with all consecutive whitespace collapsed into
77 /// Any leading or trailing whitespace will be trimmed.
78 fn collapse_whitespace(s: &str) -> String {
79 s.split_whitespace().collect::<Vec<_>>().join(" ")
82 // Information about the playground if a URL has been specified, containing an
83 // optional crate name and the URL.
84 thread_local!(pub static PLAYGROUND: RefCell<Option<(Option<String>, String)>> = {
89 pub fn render(w: &mut fmt::Formatter,
92 shorter: bool) -> fmt::Result {
93 fn block(parser: &mut Parser, buffer: &mut String, lang: &str) {
94 let mut origtext = String::new();
96 let event = parser.next();
97 if let Some(event) = event {
99 Event::End(Tag::CodeBlock(_)) => break,
100 Event::Text(ref s) => {
101 origtext.push_str(s);
109 let origtext = origtext.trim_left();
110 debug!("docblock: ==============\n{:?}\n=======", origtext);
112 let lines = origtext.lines().filter(|l| {
113 stripped_filtered_line(*l).is_none()
115 let text = lines.collect::<Vec<&str>>().join("\n");
116 let block_info = if lang.is_empty() {
117 LangString::all_false()
119 LangString::parse(lang)
121 if !block_info.rust {
122 buffer.push_str(&format!("<pre><code class=\"language-{}\">{}</code></pre>",
126 PLAYGROUND.with(|play| {
127 // insert newline to clearly separate it from the
128 // previous block so we can shorten the html output
130 let playground_button = play.borrow().as_ref().and_then(|&(ref krate, ref url)| {
134 let test = origtext.lines().map(|l| {
135 stripped_filtered_line(l).unwrap_or(l)
136 }).collect::<Vec<&str>>().join("\n");
137 let krate = krate.as_ref().map(|s| &**s);
138 let test = test::maketest(&test, krate, false,
139 &Default::default());
140 let channel = if test.contains("#![feature(") {
141 "&version=nightly"
145 // These characters don't need to be escaped in a URI.
146 // FIXME: use a library function for percent encoding.
147 fn dont_escape(c: u8) -> bool {
148 (b'a' <= c && c <= b'z') ||
149 (b'A' <= c && c <= b'Z') ||
150 (b'0' <= c && c <= b'9') ||
151 c == b'-' || c == b'_' || c == b'.' ||
152 c == b'~' || c == b'!' || c == b'\'' ||
153 c == b'(' || c == b')' || c == b'*'
155 let mut test_escaped = String::new();
156 for b in test.bytes() {
158 test_escaped.push(char::from(b));
160 write!(test_escaped, "%{:02X}", b).unwrap();
164 r#"<a class="test-arrow" target="_blank" href="{}?code={}{}">Run</a>"#,
165 url, test_escaped, channel
168 buffer.push_str(&highlight::render_with_highlighting(
170 Some("rust-example-rendered"),
172 playground_button.as_ref().map(String::as_str)));
176 fn header(parser: &mut Parser, level: i32, toc_builder: &mut Option<TocBuilder>,
177 buffer: &mut String) {
178 let mut ret = String::new();
180 let event = parser.next();
181 if let Some(event) = event {
183 Event::End(Tag::Header(_)) => break,
184 Event::Text(ref s) => {
194 let id = ret.clone();
195 // Discard '<em>', '<code>' tags and some escaped characters,
196 // transform the contents of the header into a hyphenated string
197 // without non-alphanumeric characters other than '-' and '_'.
199 // This is a terrible hack working around how hoedown gives us rendered
200 // html for text rather than the raw text.
201 let id = id.chars().filter_map(|c| {
202 if c.is_alphanumeric() || c == '-' || c == '_' {
204 Some(c.to_ascii_lowercase())
208 } else if c.is_whitespace() && c.is_ascii() {
213 }).collect::<String>();
215 let id = derive_id(id);
217 let sec = toc_builder.as_mut().map_or("".to_owned(), |builder| {
218 format!("{} ", builder.push(level as u32, ret.clone(), id.clone()))
222 buffer.push_str(&format!("<h{lvl} id=\"{id}\" class=\"section-header\">\
223 <a href=\"#{id}\">{sec}{}</a></h{lvl}>",
224 ret, lvl = level, id = id, sec = sec));
227 fn codespan(parser: &mut Parser, buffer: &mut String) {
228 let mut content = String::new();
230 let event = parser.next();
231 if let Some(event) = event {
233 Event::End(Tag::Code) => break,
234 Event::Text(ref s) => {
243 buffer.push_str(&format!("<code>{}</code>", Escape(&collapse_whitespace(&content))));
246 fn link(parser: &mut Parser, buffer: &mut String, url: &str, mut title: String) {
248 let event = parser.next();
249 if let Some(event) = event {
251 Event::End(Tag::Link(_, _)) => break,
252 Event::Text(ref s) => {
261 buffer.push_str(&format!("<a href=\"{}\">{}</a>", url, title));
264 fn paragraph(parser: &mut Parser, buffer: &mut String, toc_builder: &mut Option<TocBuilder>,
266 let mut content = String::new();
268 let event = parser.next();
269 if let Some(event) = event {
271 Event::End(Tag::Paragraph) => break,
272 Event::Text(ref s) => {
276 looper(parser, &mut content, Some(x), toc_builder, shorter);
283 buffer.push_str(&format!("<p>{}</p>", content));
286 fn cell(parser: &mut Parser, buffer: &mut String, toc_builder: &mut Option<TocBuilder>,
288 let mut content = String::new();
290 let event = parser.next();
291 if let Some(event) = event {
293 Event::End(Tag::TableHead) |
294 Event::End(Tag::Table(_)) |
295 Event::End(Tag::TableRow) |
296 Event::End(Tag::TableCell) => break,
297 Event::Text(ref s) => {
301 looper(parser, &mut content, Some(x), toc_builder, shorter);
308 buffer.push_str(&format!("<td>{}</td>", content.trim()));
311 fn row(parser: &mut Parser, buffer: &mut String, toc_builder: &mut Option<TocBuilder>,
313 let mut content = String::new();
315 let event = parser.next();
316 if let Some(event) = event {
318 Event::End(Tag::TableHead) |
319 Event::End(Tag::Table(_)) |
320 Event::End(Tag::TableRow) => break,
321 Event::Start(Tag::TableCell) => {
322 cell(parser, &mut content, toc_builder, shorter);
325 looper(parser, &mut content, Some(x), toc_builder, shorter);
332 buffer.push_str(&format!("<tr>{}</tr>", content));
335 fn head(parser: &mut Parser, buffer: &mut String, toc_builder: &mut Option<TocBuilder>,
337 let mut content = String::new();
339 let event = parser.next();
340 if let Some(event) = event {
342 Event::End(Tag::TableHead) | Event::End(Tag::Table(_)) => break,
343 Event::Start(Tag::TableCell) => {
344 cell(parser, &mut content, toc_builder, shorter);
347 looper(parser, &mut content, Some(x), toc_builder, shorter);
354 if content.is_empty() {
357 buffer.push_str(&format!("<thead><tr>{}</tr></thead>", content.replace("td>", "th>")));
360 fn table(parser: &mut Parser, buffer: &mut String, toc_builder: &mut Option<TocBuilder>,
362 let mut content = String::new();
363 let mut rows = String::new();
365 let event = parser.next();
366 if let Some(event) = event {
368 Event::End(Tag::Table(_)) => break,
369 Event::Start(Tag::TableHead) => {
370 head(parser, &mut content, toc_builder, shorter);
372 Event::Start(Tag::TableRow) => {
373 row(parser, &mut rows, toc_builder, shorter);
381 buffer.push_str(&format!("<table>{}{}</table>",
383 if shorter || rows.is_empty() {
386 format!("<tbody>{}</tbody>", rows)
390 fn looper<'a>(parser: &'a mut Parser, buffer: &mut String, next_event: Option<Event<'a>>,
391 toc_builder: &mut Option<TocBuilder>, shorter: bool) -> bool {
392 if let Some(event) = next_event {
394 Event::Start(Tag::CodeBlock(lang)) => {
395 block(parser, buffer, &*lang);
397 Event::Start(Tag::Header(level)) => {
398 header(parser, level, toc_builder, buffer);
400 Event::Start(Tag::Code) => {
401 codespan(parser, buffer);
403 Event::Start(Tag::Paragraph) => {
404 paragraph(parser, buffer, toc_builder, shorter);
406 Event::Start(Tag::Link(ref url, ref t)) => {
407 link(parser, buffer, url, t.as_ref().to_owned());
409 Event::Start(Tag::Table(_)) => {
410 table(parser, buffer, toc_builder, shorter);
420 let mut toc_builder = if print_toc {
421 Some(TocBuilder::new())
425 let mut buffer = String::new();
426 let mut parser = Parser::new_ext(s, pulldown_cmark::OPTION_ENABLE_TABLES);
428 let next_event = parser.next();
429 if !looper(&mut parser, &mut buffer, next_event, &mut toc_builder, shorter) {
433 let mut ret = toc_builder.map_or(Ok(()), |builder| {
434 write!(w, "<nav id=\"TOC\">{}</nav>", builder.into_toc())
438 ret = w.write_str(&buffer);
443 pub fn find_testable_code(doc: &str, tests: &mut ::test::Collector, position: Span) {
444 tests.set_position(position);
446 let mut parser = Parser::new(doc);
447 let mut prev_offset = 0;
448 let mut nb_lines = 0;
449 let mut register_header = None;
450 'main: while let Some(event) = parser.next() {
452 Event::Start(Tag::CodeBlock(s)) => {
453 let block_info = if s.is_empty() {
454 LangString::all_false()
456 LangString::parse(&*s)
458 if !block_info.rust {
461 let mut test_s = String::new();
462 let mut offset = None;
464 let event = parser.next();
465 if let Some(event) = event {
467 Event::End(Tag::CodeBlock(_)) => break,
468 Event::Text(ref s) => {
470 if offset.is_none() {
471 offset = Some(parser.get_offset());
480 let offset = offset.unwrap_or(0);
481 let lines = test_s.lines().map(|l| {
482 stripped_filtered_line(l).unwrap_or(l)
484 let text = lines.collect::<Vec<&str>>().join("\n");
485 nb_lines += doc[prev_offset..offset].lines().count();
486 let line = tests.get_line() + (nb_lines - 1);
487 let filename = tests.get_filename();
488 tests.add_test(text.to_owned(),
489 block_info.should_panic, block_info.no_run,
490 block_info.ignore, block_info.test_harness,
491 block_info.compile_fail, block_info.error_codes,
493 prev_offset = offset;
495 Event::Start(Tag::Header(level)) => {
496 register_header = Some(level as u32);
498 Event::Text(ref s) if register_header.is_some() => {
499 let level = register_header.unwrap();
501 tests.register_header("", level);
503 tests.register_header(s, level);
505 register_header = None;
512 #[derive(Eq, PartialEq, Clone, Debug)]
521 error_codes: Vec<String>,
525 fn all_false() -> LangString {
527 original: String::new(),
531 rust: true, // NB This used to be `notrust = false`
534 error_codes: Vec::new(),
538 fn parse(string: &str) -> LangString {
539 let mut seen_rust_tags = false;
540 let mut seen_other_tags = false;
541 let mut data = LangString::all_false();
542 let mut allow_compile_fail = false;
543 let mut allow_error_code_check = false;
544 if UnstableFeatures::from_environment().is_nightly_build() {
545 allow_compile_fail = true;
546 allow_error_code_check = true;
549 data.original = string.to_owned();
550 let tokens = string.split(|c: char|
551 !(c == '_' || c == '-' || c.is_alphanumeric())
554 for token in tokens {
557 "should_panic" => { data.should_panic = true; seen_rust_tags = true; },
558 "no_run" => { data.no_run = true; seen_rust_tags = true; },
559 "ignore" => { data.ignore = true; seen_rust_tags = true; },
560 "rust" => { data.rust = true; seen_rust_tags = true; },
561 "test_harness" => { data.test_harness = true; seen_rust_tags = true; },
562 "compile_fail" if allow_compile_fail => {
563 data.compile_fail = true;
564 seen_rust_tags = true;
567 x if allow_error_code_check && x.starts_with("E") && x.len() == 5 => {
568 if let Ok(_) = x[1..].parse::<u32>() {
569 data.error_codes.push(x.to_owned());
570 seen_rust_tags = true;
572 seen_other_tags = true;
575 _ => { seen_other_tags = true }
579 data.rust &= !seen_other_tags || seen_rust_tags;
585 impl<'a> fmt::Display for Markdown<'a> {
586 fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
587 let Markdown(md, shorter) = *self;
588 // This is actually common enough to special-case
589 if md.is_empty() { return Ok(()) }
590 render(fmt, md, false, shorter)
594 impl<'a> fmt::Display for MarkdownWithToc<'a> {
595 fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
596 let MarkdownWithToc(md) = *self;
597 render(fmt, md, true, false)
601 impl<'a> fmt::Display for MarkdownHtml<'a> {
602 fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
603 let MarkdownHtml(md) = *self;
604 // This is actually common enough to special-case
605 if md.is_empty() { return Ok(()) }
606 render(fmt, md, false, false)
610 pub fn plain_summary_line(md: &str) -> String {
611 struct ParserWrapper<'a> {
617 impl<'a> Iterator for ParserWrapper<'a> {
620 fn next(&mut self) -> Option<String> {
621 let next_event = self.inner.next();
622 if next_event.is_none() {
625 let next_event = next_event.unwrap();
626 let (ret, is_in) = match next_event {
627 Event::Start(Tag::Paragraph) => (None, 1),
628 Event::Start(Tag::Link(_, ref t)) if !self.is_first => {
629 (Some(t.as_ref().to_owned()), 1)
631 Event::Text(ref s) if self.is_in > 0 => (Some(s.as_ref().to_owned()), 0),
632 Event::End(Tag::Link(_, ref t)) => (Some(t.as_ref().to_owned()), -1),
633 Event::End(Tag::Paragraph) => (None, -1),
636 if is_in > 0 || (is_in < 0 && self.is_in > 0) {
640 self.is_first = false;
647 let mut s = String::with_capacity(md.len() * 3 / 2);
648 let mut p = ParserWrapper {
649 inner: Parser::new(md),
653 while let Some(t) = p.next() {
663 use super::{LangString, Markdown, MarkdownHtml};
664 use super::plain_summary_line;
665 use html::render::reset_ids;
668 fn test_lang_string_parse() {
670 should_panic: bool, no_run: bool, ignore: bool, rust: bool, test_harness: bool,
671 compile_fail: bool, error_codes: Vec<String>) {
672 assert_eq!(LangString::parse(s), LangString {
673 should_panic: should_panic,
677 test_harness: test_harness,
678 compile_fail: compile_fail,
679 error_codes: error_codes,
680 original: s.to_owned(),
684 // marker | should_panic| no_run| ignore| rust | test_harness| compile_fail
686 t("", false, false, false, true, false, false, Vec::new());
687 t("rust", false, false, false, true, false, false, Vec::new());
688 t("sh", false, false, false, false, false, false, Vec::new());
689 t("ignore", false, false, true, true, false, false, Vec::new());
690 t("should_panic", true, false, false, true, false, false, Vec::new());
691 t("no_run", false, true, false, true, false, false, Vec::new());
692 t("test_harness", false, false, false, true, true, false, Vec::new());
693 t("compile_fail", false, true, false, true, false, true, Vec::new());
694 t("{.no_run .example}", false, true, false, true, false, false, Vec::new());
695 t("{.sh .should_panic}", true, false, false, true, false, false, Vec::new());
696 t("{.example .rust}", false, false, false, true, false, false, Vec::new());
697 t("{.test_harness .rust}", false, false, false, true, true, false, Vec::new());
702 let markdown = "# title";
703 format!("{}", Markdown(markdown));
709 fn t(input: &str, expect: &str) {
710 let output = format!("{}", Markdown(input));
711 assert_eq!(output, expect);
715 t("# Foo bar", "\n<h1 id='foo-bar' class='section-header'>\
716 <a href='#foo-bar'>Foo bar</a></h1>");
717 t("## Foo-bar_baz qux", "\n<h2 id='foo-bar_baz-qux' class=\'section-\
718 header'><a href='#foo-bar_baz-qux'>Foo-bar_baz qux</a></h2>");
719 t("### **Foo** *bar* baz!?!& -_qux_-%",
720 "\n<h3 id='foo-bar-baz--_qux_-' class='section-header'>\
721 <a href='#foo-bar-baz--_qux_-'><strong>Foo</strong> \
722 <em>bar</em> baz!?!& -_qux_-%</a></h3>");
723 t("####**Foo?** & \\*bar?!* _`baz`_ ❤ #qux",
724 "\n<h4 id='foo--bar--baz--qux' class='section-header'>\
725 <a href='#foo--bar--baz--qux'><strong>Foo?</strong> & *bar?!* \
726 <em><code>baz</code></em> ❤ #qux</a></h4>");
730 fn test_header_ids_multiple_blocks() {
731 fn t(input: &str, expect: &str) {
732 let output = format!("{}", Markdown(input));
733 assert_eq!(output, expect);
737 t("# Example", "\n<h1 id='example' class='section-header'>\
738 <a href='#example'>Example</a></h1>");
739 t("# Panics", "\n<h1 id='panics' class='section-header'>\
740 <a href='#panics'>Panics</a></h1>");
741 t("# Example", "\n<h1 id='example-1' class='section-header'>\
742 <a href='#example-1'>Example</a></h1>");
743 t("# Main", "\n<h1 id='main-1' class='section-header'>\
744 <a href='#main-1'>Main</a></h1>");
745 t("# Example", "\n<h1 id='example-2' class='section-header'>\
746 <a href='#example-2'>Example</a></h1>");
747 t("# Panics", "\n<h1 id='panics-1' class='section-header'>\
748 <a href='#panics-1'>Panics</a></h1>");
756 fn test_plain_summary_line() {
757 fn t(input: &str, expect: &str) {
758 let output = plain_summary_line(input);
759 assert_eq!(output, expect);
762 t("hello [Rust](https://www.rust-lang.org) :)", "hello Rust :)");
763 t("code `let x = i32;` ...", "code `let x = i32;` ...");
764 t("type `Type<'static>` ...", "type `Type<'static>` ...");
765 t("# top header", "top header");
766 t("## header", "header");
770 fn test_markdown_html_escape() {
771 fn t(input: &str, expect: &str) {
772 let output = format!("{}", MarkdownHtml(input));
773 assert_eq!(output, expect);
776 t("`Struct<'a, T>`", "<p><code>Struct<'a, T></code></p>\n");
777 t("Struct<'a, T>", "<p>Struct<'a, T></p>\n");