1use gpui::SharedString;
2use linkify::LinkFinder;
3pub use pulldown_cmark::TagEnd as MarkdownTagEnd;
4use pulldown_cmark::{Alignment, HeadingLevel, LinkType, MetadataBlockKind, Options, Parser};
5use std::ops::Range;
6
7pub fn parse_markdown(text: &str) -> Vec<(Range<usize>, MarkdownEvent)> {
8 let mut events = Vec::new();
9 let mut within_link = false;
10 let mut within_metadata = false;
11 for (pulldown_event, mut range) in Parser::new_ext(text, Options::all()).into_offset_iter() {
12 if within_metadata {
13 if let pulldown_cmark::Event::End(pulldown_cmark::TagEnd::MetadataBlock { .. }) =
14 pulldown_event
15 {
16 within_metadata = false;
17 }
18 continue;
19 }
20 match pulldown_event {
21 pulldown_cmark::Event::Start(tag) => {
22 match tag {
23 pulldown_cmark::Tag::Link { .. } => within_link = true,
24 pulldown_cmark::Tag::MetadataBlock { .. } => within_metadata = true,
25 _ => {}
26 }
27 events.push((range, MarkdownEvent::Start(tag.into())))
28 }
29 pulldown_cmark::Event::End(tag) => {
30 if let pulldown_cmark::TagEnd::Link = tag {
31 within_link = false;
32 }
33 events.push((range, MarkdownEvent::End(tag)));
34 }
35 pulldown_cmark::Event::Text(_) => {
36 // Automatically detect links in text if we're not already within a markdown
37 // link.
38 if !within_link {
39 let mut finder = LinkFinder::new();
40 finder.kinds(&[linkify::LinkKind::Url]);
41 let text_range = range.clone();
42 for link in finder.links(&text[text_range.clone()]) {
43 let link_range =
44 text_range.start + link.start()..text_range.start + link.end();
45
46 if link_range.start > range.start {
47 events.push((range.start..link_range.start, MarkdownEvent::Text));
48 }
49
50 events.push((
51 link_range.clone(),
52 MarkdownEvent::Start(MarkdownTag::Link {
53 link_type: LinkType::Autolink,
54 dest_url: SharedString::from(link.as_str().to_string()),
55 title: SharedString::default(),
56 id: SharedString::default(),
57 }),
58 ));
59 events.push((link_range.clone(), MarkdownEvent::Text));
60 events.push((link_range.clone(), MarkdownEvent::End(MarkdownTagEnd::Link)));
61
62 range.start = link_range.end;
63 }
64 }
65
66 if range.start < range.end {
67 events.push((range, MarkdownEvent::Text));
68 }
69 }
70 pulldown_cmark::Event::Code(_) => {
71 range.start += 1;
72 range.end -= 1;
73 events.push((range, MarkdownEvent::Code))
74 }
75 pulldown_cmark::Event::Html(_) => events.push((range, MarkdownEvent::Html)),
76 pulldown_cmark::Event::InlineHtml(_) => events.push((range, MarkdownEvent::InlineHtml)),
77 pulldown_cmark::Event::FootnoteReference(_) => {
78 events.push((range, MarkdownEvent::FootnoteReference))
79 }
80 pulldown_cmark::Event::SoftBreak => events.push((range, MarkdownEvent::SoftBreak)),
81 pulldown_cmark::Event::HardBreak => events.push((range, MarkdownEvent::HardBreak)),
82 pulldown_cmark::Event::Rule => events.push((range, MarkdownEvent::Rule)),
83 pulldown_cmark::Event::TaskListMarker(checked) => {
84 events.push((range, MarkdownEvent::TaskListMarker(checked)))
85 }
86 }
87 }
88 events
89}
90
91pub fn parse_links_only(text: &str) -> Vec<(Range<usize>, MarkdownEvent)> {
92 let mut events = Vec::new();
93 let mut finder = LinkFinder::new();
94 finder.kinds(&[linkify::LinkKind::Url]);
95 let mut text_range = Range {
96 start: 0,
97 end: text.len(),
98 };
99 for link in finder.links(&text[text_range.clone()]) {
100 let link_range = text_range.start + link.start()..text_range.start + link.end();
101
102 if link_range.start > text_range.start {
103 events.push((text_range.start..link_range.start, MarkdownEvent::Text));
104 }
105
106 events.push((
107 link_range.clone(),
108 MarkdownEvent::Start(MarkdownTag::Link {
109 link_type: LinkType::Autolink,
110 dest_url: SharedString::from(link.as_str().to_string()),
111 title: SharedString::default(),
112 id: SharedString::default(),
113 }),
114 ));
115 events.push((link_range.clone(), MarkdownEvent::Text));
116 events.push((link_range.clone(), MarkdownEvent::End(MarkdownTagEnd::Link)));
117
118 text_range.start = link_range.end;
119 }
120
121 events.push((text_range, MarkdownEvent::Text));
122
123 events
124}
125
126/// A static-lifetime equivalent of pulldown_cmark::Event so we can cache the
127/// parse result for rendering without resorting to unsafe lifetime coercion.
128#[derive(Clone, Debug, PartialEq)]
129pub enum MarkdownEvent {
130 /// Start of a tagged element. Events that are yielded after this event
131 /// and before its corresponding `End` event are inside this element.
132 /// Start and end events are guaranteed to be balanced.
133 Start(MarkdownTag),
134 /// End of a tagged element.
135 End(MarkdownTagEnd),
136 /// A text node.
137 Text,
138 /// An inline code node.
139 Code,
140 /// An HTML node.
141 Html,
142 /// An inline HTML node.
143 InlineHtml,
144 /// A reference to a footnote with given label, which may or may not be defined
145 /// by an event with a `Tag::FootnoteDefinition` tag. Definitions and references to them may
146 /// occur in any order.
147 FootnoteReference,
148 /// A soft line break.
149 SoftBreak,
150 /// A hard line break.
151 HardBreak,
152 /// A horizontal ruler.
153 Rule,
154 /// A task list marker, rendered as a checkbox in HTML. Contains a true when it is checked.
155 TaskListMarker(bool),
156}
157
158/// Tags for elements that can contain other elements.
159#[derive(Clone, Debug, PartialEq)]
160pub enum MarkdownTag {
161 /// A paragraph of text and other inline elements.
162 Paragraph,
163
164 /// A heading, with optional identifier, classes and custom attributes.
165 /// The identifier is prefixed with `#` and the last one in the attributes
166 /// list is chosen, classes are prefixed with `.` and custom attributes
167 /// have no prefix and can optionally have a value (`myattr` o `myattr=myvalue`).
168 Heading {
169 level: HeadingLevel,
170 id: Option<SharedString>,
171 classes: Vec<SharedString>,
172 /// The first item of the tuple is the attr and second one the value.
173 attrs: Vec<(SharedString, Option<SharedString>)>,
174 },
175
176 BlockQuote,
177
178 /// A code block.
179 CodeBlock(CodeBlockKind),
180
181 /// A HTML block.
182 HtmlBlock,
183
184 /// A list. If the list is ordered the field indicates the number of the first item.
185 /// Contains only list items.
186 List(Option<u64>), // TODO: add delim and tight for ast (not needed for html)
187
188 /// A list item.
189 Item,
190
191 /// A footnote definition. The value contained is the footnote's label by which it can
192 /// be referred to.
193 FootnoteDefinition(SharedString),
194
195 /// A table. Contains a vector describing the text-alignment for each of its columns.
196 Table(Vec<Alignment>),
197
198 /// A table header. Contains only `TableCell`s. Note that the table body starts immediately
199 /// after the closure of the `TableHead` tag. There is no `TableBody` tag.
200 TableHead,
201
202 /// A table row. Is used both for header rows as body rows. Contains only `TableCell`s.
203 TableRow,
204 TableCell,
205
206 // span-level tags
207 Emphasis,
208 Strong,
209 Strikethrough,
210
211 /// A link.
212 Link {
213 link_type: LinkType,
214 dest_url: SharedString,
215 title: SharedString,
216 /// Identifier of reference links, e.g. `world` in the link `[hello][world]`.
217 id: SharedString,
218 },
219
220 /// An image. The first field is the link type, the second the destination URL and the third is a title,
221 /// the fourth is the link identifier.
222 Image {
223 link_type: LinkType,
224 dest_url: SharedString,
225 title: SharedString,
226 /// Identifier of reference links, e.g. `world` in the link `[hello][world]`.
227 id: SharedString,
228 },
229
230 /// A metadata block.
231 MetadataBlock(MetadataBlockKind),
232}
233
234#[derive(Clone, Debug, PartialEq)]
235pub enum CodeBlockKind {
236 Indented,
237 /// The value contained in the tag describes the language of the code, which may be empty.
238 Fenced(SharedString),
239}
240
241impl From<pulldown_cmark::Tag<'_>> for MarkdownTag {
242 fn from(tag: pulldown_cmark::Tag) -> Self {
243 match tag {
244 pulldown_cmark::Tag::Paragraph => MarkdownTag::Paragraph,
245 pulldown_cmark::Tag::Heading {
246 level,
247 id,
248 classes,
249 attrs,
250 } => {
251 let id = id.map(|id| SharedString::from(id.into_string()));
252 let classes = classes
253 .into_iter()
254 .map(|c| SharedString::from(c.into_string()))
255 .collect();
256 let attrs = attrs
257 .into_iter()
258 .map(|(key, value)| {
259 (
260 SharedString::from(key.into_string()),
261 value.map(|v| SharedString::from(v.into_string())),
262 )
263 })
264 .collect();
265 MarkdownTag::Heading {
266 level,
267 id,
268 classes,
269 attrs,
270 }
271 }
272 pulldown_cmark::Tag::BlockQuote => MarkdownTag::BlockQuote,
273 pulldown_cmark::Tag::CodeBlock(kind) => match kind {
274 pulldown_cmark::CodeBlockKind::Indented => {
275 MarkdownTag::CodeBlock(CodeBlockKind::Indented)
276 }
277 pulldown_cmark::CodeBlockKind::Fenced(info) => MarkdownTag::CodeBlock(
278 CodeBlockKind::Fenced(SharedString::from(info.into_string())),
279 ),
280 },
281 pulldown_cmark::Tag::List(start_number) => MarkdownTag::List(start_number),
282 pulldown_cmark::Tag::Item => MarkdownTag::Item,
283 pulldown_cmark::Tag::FootnoteDefinition(label) => {
284 MarkdownTag::FootnoteDefinition(SharedString::from(label.to_string()))
285 }
286 pulldown_cmark::Tag::Table(alignments) => MarkdownTag::Table(alignments),
287 pulldown_cmark::Tag::TableHead => MarkdownTag::TableHead,
288 pulldown_cmark::Tag::TableRow => MarkdownTag::TableRow,
289 pulldown_cmark::Tag::TableCell => MarkdownTag::TableCell,
290 pulldown_cmark::Tag::Emphasis => MarkdownTag::Emphasis,
291 pulldown_cmark::Tag::Strong => MarkdownTag::Strong,
292 pulldown_cmark::Tag::Strikethrough => MarkdownTag::Strikethrough,
293 pulldown_cmark::Tag::Link {
294 link_type,
295 dest_url,
296 title,
297 id,
298 } => MarkdownTag::Link {
299 link_type,
300 dest_url: SharedString::from(dest_url.into_string()),
301 title: SharedString::from(title.into_string()),
302 id: SharedString::from(id.into_string()),
303 },
304 pulldown_cmark::Tag::Image {
305 link_type,
306 dest_url,
307 title,
308 id,
309 } => MarkdownTag::Image {
310 link_type,
311 dest_url: SharedString::from(dest_url.into_string()),
312 title: SharedString::from(title.into_string()),
313 id: SharedString::from(id.into_string()),
314 },
315 pulldown_cmark::Tag::HtmlBlock => MarkdownTag::HtmlBlock,
316 pulldown_cmark::Tag::MetadataBlock(kind) => MarkdownTag::MetadataBlock(kind),
317 }
318 }
319}