markdown.rs 94.5%

52 of 55 lines run, 15 of 17 functions · run #504 on main · run missed

//! Markdown to sanitised HTML: pulldown-cmark, then ammonia with an
//! allow-list (hub spec §15.2). Relative links and images are rebased on
//! `base`, so a README's `docs/x.md` opens in the hub.

use pulldown_cmark::{html, Event, Options, Parser, Tag};

/// Rendered Markdown by (text, link bases): 32 MiB.
fn cache() -> &'static super::cache::Lru<String> {
    static C: std::sync::OnceLock<super::cache::Lru<String>> = std::sync::OnceLock::new();
    C.get_or_init(|| super::cache::Lru::new(32 << 20))
}

/// (hits, misses, bytes) of the rendered Markdown cache.
pub fn cache_stats() -> (u64, u64, usize) {
    cache().stats()
}

/// `base_file` and `base_raw` are URL prefixes (ending in `/`) for links to
/// files and for images. Cached by the text and both bases.
pub fn render(text: &str, base_file: &str, base_raw: &str) -> String {
    let key = super::cache::key(&[b"md1", text.as_bytes(), base_file.as_bytes(), base_raw.as_bytes()]);
    cache().get_or(key, || render_uncached(text, base_file, base_raw)).as_ref().clone()
}

fn render_uncached(text: &str, base_file: &str, base_raw: &str) -> String {
    let mut opts = Options::empty();
    opts.insert(Options::ENABLE_TABLES);
    opts.insert(Options::ENABLE_STRIKETHROUGH);
    opts.insert(Options::ENABLE_TASKLISTS);
    opts.insert(Options::ENABLE_FOOTNOTES);
    opts.insert(Options::ENABLE_HEADING_ATTRIBUTES);
    let parser = Parser::new_ext(text, opts).map(|ev| match ev {
        Event::Start(Tag::Link { link_type, dest_url, title, id }) => {
            Event::Start(Tag::Link { link_type, dest_url: rebase(&dest_url, base_file).into(), title, id })
        }
        Event::Start(Tag::Image { link_type, dest_url, title, id }) => {
            Event::Start(Tag::Image { link_type, dest_url: rebase(&dest_url, base_raw).into(), title, id })
        }
        other => other,
    });
    let mut out = String::with_capacity(text.len() * 2);
    html::push_html(&mut out, parser);
    let mut builder = ammonia::Builder::default();
    builder.add_generic_attributes(["id", "class"]);
    builder.add_tag_attributes("input", ["type", "checked", "disabled"]);
    builder.add_tags(["input"]);
    builder.link_rel(Some("noopener noreferrer"));
    builder.clean(&out).to_string()
}

/// Leave absolute URLs, fragments and mail links alone; rebase the rest.
fn rebase(url: &str, base: &str) -> String {
    if url.starts_with('#') || url.contains("://") || url.starts_with("mailto:") || url.starts_with('/') || url.starts_with("data:") {
        return url.to_string();
    }
    let url = url.strip_prefix("./").unwrap_or(url);
    format!("{base}{url}")
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn renders_and_sanitises() {
        let html = render(
            "# Hi\n\nSee [docs](docs/x.md) and ![i](img.png) <script>alert(1)</script>\n\n- [ ] task\n",
            "/a/b/file/main/",
            "/a/b/file/main/",
        );
        assert!(html.contains("<h1>Hi</h1>"));
        assert!(html.contains("href=\"/a/b/file/main/docs/x.md\""));
        assert!(html.contains("src=\"/a/b/file/main/img.png\""));
        assert!(!html.contains("<script"));
        assert!(html.contains("type=\"checkbox\""));
        assert!(render("[x](https://e.com) [y](#frag)", "/p/", "/p/").contains("href=\"https://e.com\""));
    }
}

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	253×	fn cache() -> &'static super::cache::Lru<String> {
9		static C: std::sync::OnceLock<super::cache::Lru<String>> = std::sync::OnceLock::new();
10	253×	C.get_or_init(\|\| super::cache::Lru::new(32 << 20))
11	253×	}
12
13		/// (hits, misses, bytes) of the rendered Markdown cache.
14	0×	pub fn cache_stats() -> (u64, u64, usize) {
15	0×	cache().stats()
16	0×	}
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	253×	pub fn render(text: &str, base_file: &str, base_raw: &str) -> String {
21	253×	let key = super::cache::key(&[b"md1", text.as_bytes(), base_file.as_bytes(), base_raw.as_bytes()]);
22	253×	cache().get_or(key, \|\| render_uncached(text, base_file, base_raw)).as_ref().clone()
23	253×	}
24
25	59×	fn render_uncached(text: &str, base_file: &str, base_raw: &str) -> String {
26	59×	let mut opts = Options::empty();
27	59×	opts.insert(Options::ENABLE_TABLES);
28	59×	opts.insert(Options::ENABLE_STRIKETHROUGH);
29	59×	opts.insert(Options::ENABLE_TASKLISTS);
30	59×	opts.insert(Options::ENABLE_FOOTNOTES);
31	59×	opts.insert(Options::ENABLE_HEADING_ATTRIBUTES);
32	91×	let parser = Parser::new_ext(text, opts).map(\|ev\| match ev {
33	16×	Event::Start(Tag::Link { link_type, dest_url, title, id }) => {
34	16×	Event::Start(Tag::Link { link_type, dest_url: rebase(&dest_url, base_file).into(), title, id })
35		}
36	4×	Event::Start(Tag::Image { link_type, dest_url, title, id }) => {
37	4×	Event::Start(Tag::Image { link_type, dest_url: rebase(&dest_url, base_raw).into(), title, id })
38		}
39	274×	other => other,
40	294×	});
41	59×	let mut out = String::with_capacity(text.len() * 2);
42	59×	html::push_html(&mut out, parser);
43	59×	let mut builder = ammonia::Builder::default();
44	59×	builder.add_generic_attributes(["id", "class"]);
45	59×	builder.add_tag_attributes("input", ["type", "checked", "disabled"]);
46	59×	builder.add_tags(["input"]);
47	59×	builder.link_rel(Some("noopener noreferrer"));
48	59×	builder.clean(&out).to_string()
49	59×	}
50
51		/// Leave absolute URLs, fragments and mail links alone; rebase the rest.
52	20×	fn rebase(url: &str, base: &str) -> String {
53	20×	if url.starts_with('#') \|\| url.contains("://") \|\| url.starts_with("mailto:") \|\| url.starts_with('/') \|\| url.starts_with("data:") {
54	8×	return url.to_string();
55	12×	}
56	12×	let url = url.strip_prefix("./").unwrap_or(url);
57	12×	format!("{base}{url}")
58	20×	}
59
60		#[cfg(test)]
61		mod tests {
62		use super::*;
63
64		#[test]
65	1×	fn renders_and_sanitises() {
66	1×	let html = render(
67	1×	"# Hi\n\nSee [docs](docs/x.md) and ![i](img.png) <script>alert(1)</script>\n\n- [ ] task\n",
68	1×	"/a/b/file/main/",
69	1×	"/a/b/file/main/",
70		);
71	1×	assert!(html.contains("<h1>Hi</h1>"));
72	1×	assert!(html.contains("href=\"/a/b/file/main/docs/x.md\""));
73	1×	assert!(html.contains("src=\"/a/b/file/main/img.png\""));
74	1×	assert!(!html.contains("<script"));
75	1×	assert!(html.contains("type=\"checkbox\""));
76	1×	assert!(render("[x](https://e.com) [y](#frag)", "/p/", "/p/").contains("href=\"https://e.com\""));
77	1×	}
78		}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

Coverage / crates/hub/src/render/markdown.rs 94.5%

Keyboard