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, Options, 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    for event in Parser::new_ext(markdown, Options::all()) {
169        let prev_len = text.len();
170        match event {
171            Event::Text(t) => {
172                if let Some(language) = &current_language {
173                    highlight_code(text, highlights, t.as_ref(), language);
174                } else {
175                    text.push_str(t.as_ref());
176
177                    let mut style = MarkdownHighlightStyle::default();
178
179                    if bold_depth > 0 {
180                        style.weight = FontWeight::BOLD;
181                    }
182
183                    if italic_depth > 0 {
184                        style.italic = true;
185                    }
186
187                    if strikethrough_depth > 0 {
188                        style.strikethrough = true;
189                    }
190
191                    if let Some(link) = link_url.clone().and_then(|u| Link::identify(u)) {
192                        region_ranges.push(prev_len..text.len());
193                        regions.push(ParsedRegion {
194                            code: false,
195                            link: Some(link),
196                        });
197                        style.underline = true;
198                    }
199
200                    if style != MarkdownHighlightStyle::default() {
201                        let mut new_highlight = true;
202                        if let Some((last_range, MarkdownHighlight::Style(last_style))) =
203                            highlights.last_mut()
204                        {
205                            if last_range.end == prev_len && last_style == &style {
206                                last_range.end = text.len();
207                                new_highlight = false;
208                            }
209                        }
210                        if new_highlight {
211                            let range = prev_len..text.len();
212                            highlights.push((range, MarkdownHighlight::Style(style)));
213                        }
214                    }
215                }
216            }
217
218            Event::Code(t) => {
219                text.push_str(t.as_ref());
220                region_ranges.push(prev_len..text.len());
221
222                let link = link_url.clone().and_then(|u| Link::identify(u));
223                if link.is_some() {
224                    highlights.push((
225                        prev_len..text.len(),
226                        MarkdownHighlight::Style(MarkdownHighlightStyle {
227                            underline: true,
228                            ..Default::default()
229                        }),
230                    ));
231                }
232                regions.push(ParsedRegion { code: true, link });
233            }
234
235            Event::Start(tag) => match tag {
236                Tag::Paragraph => new_paragraph(text, &mut list_stack),
237
238                Tag::Heading {
239                    level: _,
240                    id: _,
241                    classes: _,
242                    attrs: _,
243                } => {
244                    new_paragraph(text, &mut list_stack);
245                    bold_depth += 1;
246                }
247
248                Tag::CodeBlock(kind) => {
249                    new_paragraph(text, &mut list_stack);
250                    current_language = if let CodeBlockKind::Fenced(language) = kind {
251                        language_registry
252                            .language_for_name(language.as_ref())
253                            .await
254                            .ok()
255                    } else {
256                        language.clone()
257                    }
258                }
259
260                Tag::Emphasis => italic_depth += 1,
261
262                Tag::Strong => bold_depth += 1,
263
264                Tag::Strikethrough => strikethrough_depth += 1,
265
266                Tag::Link {
267                    link_type: _,
268                    dest_url,
269                    title: _,
270                    id: _,
271                } => link_url = Some(dest_url.to_string()),
272
273                Tag::List(number) => {
274                    list_stack.push((number, false));
275                }
276
277                Tag::Item => {
278                    let len = list_stack.len();
279                    if let Some((list_number, has_content)) = list_stack.last_mut() {
280                        *has_content = false;
281                        if !text.is_empty() && !text.ends_with('\n') {
282                            text.push('\n');
283                        }
284                        for _ in 0..len - 1 {
285                            text.push_str("  ");
286                        }
287                        if let Some(number) = list_number {
288                            text.push_str(&format!("{}. ", number));
289                            *number += 1;
290                            *has_content = false;
291                        } else {
292                            text.push_str("- ");
293                        }
294                    }
295                }
296
297                _ => {}
298            },
299
300            Event::End(tag) => match tag {
301                TagEnd::Heading(_) => bold_depth -= 1,
302                TagEnd::CodeBlock => current_language = None,
303                TagEnd::Emphasis => italic_depth -= 1,
304                TagEnd::Strong => bold_depth -= 1,
305                TagEnd::Strikethrough => strikethrough_depth -= 1,
306                TagEnd::Link => link_url = None,
307                TagEnd::List(_) => drop(list_stack.pop()),
308                _ => {}
309            },
310
311            Event::HardBreak => text.push('\n'),
312
313            Event::SoftBreak => text.push(' '),
314
315            _ => {}
316        }
317    }
318}
319
320/// Appends a highlighted run of text to the provided `text` buffer.
321pub fn highlight_code(
322    text: &mut String,
323    highlights: &mut Vec<(Range<usize>, MarkdownHighlight)>,
324    content: &str,
325    language: &Arc<Language>,
326) {
327    let prev_len = text.len();
328    text.push_str(content);
329    for (range, highlight_id) in language.highlight_text(&content.into(), 0..content.len()) {
330        let highlight = MarkdownHighlight::Code(highlight_id);
331        highlights.push((prev_len + range.start..prev_len + range.end, highlight));
332    }
333}
334
335/// Appends a new paragraph to the provided `text` buffer.
336pub fn new_paragraph(text: &mut String, list_stack: &mut Vec<(Option<u64>, bool)>) {
337    let mut is_subsequent_paragraph_of_list = false;
338    if let Some((_, has_content)) = list_stack.last_mut() {
339        if *has_content {
340            is_subsequent_paragraph_of_list = true;
341        } else {
342            *has_content = true;
343            return;
344        }
345    }
346
347    if !text.is_empty() {
348        if !text.ends_with('\n') {
349            text.push('\n');
350        }
351        text.push('\n');
352    }
353    for _ in 0..list_stack.len().saturating_sub(1) {
354        text.push_str("  ");
355    }
356    if is_subsequent_paragraph_of_list {
357        text.push_str("  ");
358    }
359}