OXYGENHUB
# ca0c443c30cf oxygenhub / crates / hub / src / render / markdown.rs
3.2 KiB · Rust · last changed Cache rendered Markdown and highlighted files by content, in memory-bounded LRUs 1 month ago source blame history raw
1 
//! Markdown to sanitised HTML: pulldown-cmark, then ammonia with an
2 
//! allow-list (hub spec §15.2). Relative links and images are rebased on
3 
//! `base`, so a README's `docs/x.md` opens in the hub.
4
5 
use pulldown_cmark::{html, Event, Options, Parser, Tag};
6
7 
/// Rendered Markdown by (text, link bases): 32 MiB.
8 
fn cache() -> &'static super::cache::Lru<String> {
9 
    static C: std::sync::OnceLock<super::cache::Lru<String>> = std::sync::OnceLock::new();
10 
    C.get_or_init(|| super::cache::Lru::new(32 << 20))
11 
}
12
13 
/// (hits, misses, bytes) of the rendered Markdown cache.
14 
pub fn cache_stats() -> (u64, u64, usize) {
15 
    cache().stats()
16 
}
17
18 
/// `base_file` and `base_raw` are URL prefixes (ending in `/`) for links to
19 
/// files and for images. Cached by the text and both bases.
20 
pub fn render(text: &str, base_file: &str, base_raw: &str) -> String {
21 
    let key = super::cache::key(&[b"md1", text.as_bytes(), base_file.as_bytes(), base_raw.as_bytes()]);
22 
    cache().get_or(key, || render_uncached(text, base_file, base_raw)).as_ref().clone()
23 
}
24
25 
fn render_uncached(text: &str, base_file: &str, base_raw: &str) -> String {
26 
    let mut opts = Options::empty();
27 
    opts.insert(Options::ENABLE_TABLES);
28 
    opts.insert(Options::ENABLE_STRIKETHROUGH);
29 
    opts.insert(Options::ENABLE_TASKLISTS);
30 
    opts.insert(Options::ENABLE_FOOTNOTES);
31 
    opts.insert(Options::ENABLE_HEADING_ATTRIBUTES);
32 
    let parser = Parser::new_ext(text, opts).map(|ev| match ev {
33 
        Event::Start(Tag::Link { link_type, dest_url, title, id }) => {
34 
            Event::Start(Tag::Link { link_type, dest_url: rebase(&dest_url, base_file).into(), title, id })
35 
        }
36 
        Event::Start(Tag::Image { link_type, dest_url, title, id }) => {
37 
            Event::Start(Tag::Image { link_type, dest_url: rebase(&dest_url, base_raw).into(), title, id })
38 
        }
39 
        other => other,
40 
    });
41 
    let mut out = String::with_capacity(text.len() * 2);
42 
    html::push_html(&mut out, parser);
43 
    let mut builder = ammonia::Builder::default();
44 
    builder.add_generic_attributes(["id", "class"]);
45 
    builder.add_tag_attributes("input", ["type", "checked", "disabled"]);
46 
    builder.add_tags(["input"]);
47 
    builder.link_rel(Some("noopener noreferrer"));
48 
    builder.clean(&out).to_string()
49 
}
50
51 
/// Leave absolute URLs, fragments and mail links alone; rebase the rest.
52 
fn rebase(url: &str, base: &str) -> String {
53 
    if url.starts_with('#') || url.contains("://") || url.starts_with("mailto:") || url.starts_with('/') || url.starts_with("data:") {
54 
        return url.to_string();
55 
    }
56 
    let url = url.strip_prefix("./").unwrap_or(url);
57 
    format!("{base}{url}")
58 
}
59
60 
#[cfg(test)]
61 
mod tests {
62 
    use super::*;
63
64 
    #[test]
65 
    fn renders_and_sanitises() {
66 
        let html = render(
67 
            "# Hi\n\nSee [docs](docs/x.md) and ![i](img.png) <script>alert(1)</script>\n\n- [ ] task\n",
68 
            "/a/b/file/main/",
69 
            "/a/b/file/main/",
70 
        );
71 
        assert!(html.contains("<h1>Hi</h1>"));
72 
        assert!(html.contains("href=\"/a/b/file/main/docs/x.md\""));
73 
        assert!(html.contains("src=\"/a/b/file/main/img.png\""));
74 
        assert!(!html.contains("<script"));
75 
        assert!(html.contains("type=\"checkbox\""));
76 
        assert!(render("[x](https://e.com) [y](#frag)", "/p/", "/p/").contains("href=\"https://e.com\""));
77 
    }
78 
}