markdown.rs

  1//! Provides Markdown-related constructs.
  2
  3use std::sync::Arc;
  4use std::{ops::Range, path::PathBuf};
  5
  6use crate::{HighlightId, Language, LanguageRegistry};
  7use gpui::{px, FontStyle, FontWeight, HighlightStyle, StrikethroughStyle, UnderlineStyle};
  8use pulldown_cmark::{CodeBlockKind, Event, Parser, Tag, TagEnd};
  9
 10/// Parsed Markdown content.
 11#[derive(Debug, Clone, Default)]
 12pub struct ParsedMarkdown {
 13    /// The Markdown text.
 14    pub text: String,
 15    /// The list of highlights contained in the Markdown document.
 16    pub highlights: Vec<(Range<usize>, MarkdownHighlight)>,
 17    /// The regions of the various ranges in the Markdown document.
 18    pub region_ranges: Vec<Range<usize>>,
 19    /// The regions of the Markdown document.
 20    pub regions: Vec<ParsedRegion>,
 21}
 22
 23/// A run of highlighted Markdown text.
 24#[derive(Debug, Clone, PartialEq, Eq)]
 25pub enum MarkdownHighlight {
 26    /// A styled Markdown highlight.
 27    Style(MarkdownHighlightStyle),
 28    /// A highlighted code block.
 29    Code(HighlightId),
 30}
 31
 32impl MarkdownHighlight {
 33    /// Converts this [`MarkdownHighlight`] to a [`HighlightStyle`].
 34    pub fn to_highlight_style(&self, theme: &theme::SyntaxTheme) -> Option<HighlightStyle> {
 35        match self {
 36            MarkdownHighlight::Style(style) => {
 37                let mut highlight = HighlightStyle::default();
 38
 39                if style.italic {
 40                    highlight.font_style = Some(FontStyle::Italic);
 41                }
 42
 43                if style.underline {
 44                    highlight.underline = Some(UnderlineStyle {
 45                        thickness: px(1.),
 46                        ..Default::default()
 47                    });
 48                }
 49
 50                if style.strikethrough {
 51                    highlight.strikethrough = Some(StrikethroughStyle {
 52                        thickness: px(1.),
 53                        ..Default::default()
 54                    });
 55                }
 56
 57                if style.weight != FontWeight::default() {
 58                    highlight.font_weight = Some(style.weight);
 59                }
 60
 61                Some(highlight)
 62            }
 63
 64            MarkdownHighlight::Code(id) => id.style(theme),
 65        }
 66    }
 67}
 68
 69/// The style for a Markdown highlight.
 70#[derive(Debug, Clone, Default, PartialEq, Eq)]
 71pub struct MarkdownHighlightStyle {
 72    /// Whether the text should be italicized.
 73    pub italic: bool,
 74    /// Whether the text should be underlined.
 75    pub underline: bool,
 76    /// Whether the text should be struck through.
 77    pub strikethrough: bool,
 78    /// The weight of the text.
 79    pub weight: FontWeight,
 80}
 81
 82/// A parsed region in a Markdown document.
 83#[derive(Debug, Clone)]
 84pub struct ParsedRegion {
 85    /// Whether the region is a code block.
 86    pub code: bool,
 87    /// The link contained in this region, if it has one.
 88    pub link: Option<Link>,
 89}
 90
 91/// A Markdown link.
 92#[derive(Debug, Clone)]
 93pub enum Link {
 94    /// A link to a webpage.
 95    Web {
 96        /// The URL of the webpage.
 97        url: String,
 98    },
 99    /// A link to a path on the filesystem.
100    Path {
101        /// The path to the item.
102        path: PathBuf,
103    },
104}
105
106impl Link {
107    fn identify(text: String) -> Option<Link> {
108        if text.starts_with("http") {
109            return Some(Link::Web { url: text });
110        }
111
112        let path = PathBuf::from(text);
113        if path.is_absolute() {
114            return Some(Link::Path { path });
115        }
116
117        None
118    }
119}
120
121/// Parses a string of Markdown.
122pub async fn parse_markdown(
123    markdown: &str,
124    language_registry: &Arc<LanguageRegistry>,
125    language: Option<Arc<Language>>,
126) -> ParsedMarkdown {
127    let mut text = String::new();
128    let mut highlights = Vec::new();
129    let mut region_ranges = Vec::new();
130    let mut regions = Vec::new();
131
132    parse_markdown_block(
133        markdown,
134        language_registry,
135        language,
136        &mut text,
137        &mut highlights,
138        &mut region_ranges,
139        &mut regions,
140    )
141    .await;
142
143    ParsedMarkdown {
144        text,
145        highlights,
146        region_ranges,
147        regions,
148    }
149}
150
151/// Parses a Markdown block.
152pub async fn parse_markdown_block(
153    markdown: &str,
154    language_registry: &Arc<LanguageRegistry>,
155    language: Option<Arc<Language>>,
156    text: &mut String,
157    highlights: &mut Vec<(Range<usize>, MarkdownHighlight)>,
158    region_ranges: &mut Vec<Range<usize>>,
159    regions: &mut Vec<ParsedRegion>,
160) {
161    let mut bold_depth = 0;
162    let mut italic_depth = 0;
163    let mut strikethrough_depth = 0;
164    let mut link_url = None;
165    let mut current_language = None;
166    let mut list_stack = Vec::new();
167
168    let mut options = pulldown_cmark::Options::all();
169    options.remove(pulldown_cmark::Options::ENABLE_YAML_STYLE_METADATA_BLOCKS);
170
171    for event in Parser::new_ext(markdown, options) {
172        let prev_len = text.len();
173        match event {
174            Event::Text(t) => {
175                if let Some(language) = &current_language {
176                    highlight_code(text, highlights, t.as_ref(), language);
177                } else {
178                    text.push_str(t.as_ref());
179
180                    let mut style = MarkdownHighlightStyle::default();
181
182                    if bold_depth > 0 {
183                        style.weight = FontWeight::BOLD;
184                    }
185
186                    if italic_depth > 0 {
187                        style.italic = true;
188                    }
189
190                    if strikethrough_depth > 0 {
191                        style.strikethrough = true;
192                    }
193
194                    if let Some(link) = link_url.clone().and_then(|u| Link::identify(u)) {
195                        region_ranges.push(prev_len..text.len());
196                        regions.push(ParsedRegion {
197                            code: false,
198                            link: Some(link),
199                        });
200                        style.underline = true;
201                    }
202
203                    if style != MarkdownHighlightStyle::default() {
204                        let mut new_highlight = true;
205                        if let Some((last_range, MarkdownHighlight::Style(last_style))) =
206                            highlights.last_mut()
207                        {
208                            if last_range.end == prev_len && last_style == &style {
209                                last_range.end = text.len();
210                                new_highlight = false;
211                            }
212                        }
213                        if new_highlight {
214                            let range = prev_len..text.len();
215                            highlights.push((range, MarkdownHighlight::Style(style)));
216                        }
217                    }
218                }
219            }
220
221            Event::Code(t) => {
222                text.push_str(t.as_ref());
223                region_ranges.push(prev_len..text.len());
224
225                let link = link_url.clone().and_then(|u| Link::identify(u));
226                if link.is_some() {
227                    highlights.push((
228                        prev_len..text.len(),
229                        MarkdownHighlight::Style(MarkdownHighlightStyle {
230                            underline: true,
231                            ..Default::default()
232                        }),
233                    ));
234                }
235                regions.push(ParsedRegion { code: true, link });
236            }
237
238            Event::Start(tag) => match tag {
239                Tag::Paragraph => new_paragraph(text, &mut list_stack),
240
241                Tag::Heading {
242                    level: _,
243                    id: _,
244                    classes: _,
245                    attrs: _,
246                } => {
247                    new_paragraph(text, &mut list_stack);
248                    bold_depth += 1;
249                }
250
251                Tag::CodeBlock(kind) => {
252                    new_paragraph(text, &mut list_stack);
253                    current_language = if let CodeBlockKind::Fenced(language) = kind {
254                        language_registry
255                            .language_for_name_or_extension(language.as_ref())
256                            .await
257                            .ok()
258                    } else {
259                        language.clone()
260                    }
261                }
262
263                Tag::Emphasis => italic_depth += 1,
264
265                Tag::Strong => bold_depth += 1,
266
267                Tag::Strikethrough => strikethrough_depth += 1,
268
269                Tag::Link {
270                    link_type: _,
271                    dest_url,
272                    title: _,
273                    id: _,
274                } => link_url = Some(dest_url.to_string()),
275
276                Tag::List(number) => {
277                    list_stack.push((number, false));
278                }
279
280                Tag::Item => {
281                    let len = list_stack.len();
282                    if let Some((list_number, has_content)) = list_stack.last_mut() {
283                        *has_content = false;
284                        if !text.is_empty() && !text.ends_with('\n') {
285                            text.push('\n');
286                        }
287                        for _ in 0..len - 1 {
288                            text.push_str("  ");
289                        }
290                        if let Some(number) = list_number {
291                            text.push_str(&format!("{}. ", number));
292                            *number += 1;
293                            *has_content = false;
294                        } else {
295                            text.push_str("- ");
296                        }
297                    }
298                }
299
300                _ => {}
301            },
302
303            Event::End(tag) => match tag {
304                TagEnd::Heading(_) => bold_depth -= 1,
305                TagEnd::CodeBlock => current_language = None,
306                TagEnd::Emphasis => italic_depth -= 1,
307                TagEnd::Strong => bold_depth -= 1,
308                TagEnd::Strikethrough => strikethrough_depth -= 1,
309                TagEnd::Link => link_url = None,
310                TagEnd::List(_) => drop(list_stack.pop()),
311                _ => {}
312            },
313
314            Event::HardBreak => text.push('\n'),
315
316            Event::SoftBreak => text.push(' '),
317
318            _ => {}
319        }
320    }
321}
322
323/// Appends a highlighted run of text to the provided `text` buffer.
324pub fn highlight_code(
325    text: &mut String,
326    highlights: &mut Vec<(Range<usize>, MarkdownHighlight)>,
327    content: &str,
328    language: &Arc<Language>,
329) {
330    let prev_len = text.len();
331    text.push_str(content);
332    for (range, highlight_id) in language.highlight_text(&content.into(), 0..content.len()) {
333        let highlight = MarkdownHighlight::Code(highlight_id);
334        highlights.push((prev_len + range.start..prev_len + range.end, highlight));
335    }
336}
337
338/// Appends a new paragraph to the provided `text` buffer.
339pub fn new_paragraph(text: &mut String, list_stack: &mut Vec<(Option<u64>, bool)>) {
340    let mut is_subsequent_paragraph_of_list = false;
341    if let Some((_, has_content)) = list_stack.last_mut() {
342        if *has_content {
343            is_subsequent_paragraph_of_list = true;
344        } else {
345            *has_content = true;
346            return;
347        }
348    }
349
350    if !text.is_empty() {
351        if !text.ends_with('\n') {
352            text.push('\n');
353        }
354        text.push('\n');
355    }
356    for _ in 0..list_stack.len().saturating_sub(1) {
357        text.push_str("  ");
358    }
359    if is_subsequent_paragraph_of_list {
360        text.push_str("  ");
361    }
362}
363
364#[cfg(test)]
365mod tests {
366
367    #[test]
368    fn test_dividers() {
369        let input = r#"
370### instance-method `format`
371
372---
373→ `void`
374Parameters:
375- `const int &`
376- `const std::tm &`
377- `int & dest`
378
379---
380```cpp
381// In my_formatter_flag
382public: void format(const int &, const std::tm &, int &dest)
383```
384"#;
385
386        let mut options = pulldown_cmark::Options::all();
387        options.remove(pulldown_cmark::Options::ENABLE_YAML_STYLE_METADATA_BLOCKS);
388
389        let parser = pulldown_cmark::Parser::new_ext(input, options);
390        for event in parser.into_iter() {
391            println!("{:?}", event);
392        }
393    }
394}