ruma_html/
helpers.rs

1//! Convenience methods and types to sanitize HTML messages.
2
3use crate::{Html, HtmlSanitizerMode, SanitizerConfig};
4
5/// Sanitize the given HTML string.
6///
7/// This removes the [tags and attributes] that are not listed in the Matrix specification.
8///
9/// It can also optionally remove the [rich reply] fallback.
10///
11/// [tags and attributes]: https://spec.matrix.org/latest/client-server-api/#mroommessage-msgtypes
12/// [rich reply]: https://spec.matrix.org/latest/client-server-api/#rich-replies
13pub fn sanitize_html(
14    s: &str,
15    mode: HtmlSanitizerMode,
16    remove_reply_fallback: RemoveReplyFallback,
17) -> String {
18    let mut config = match mode {
19        HtmlSanitizerMode::Strict => SanitizerConfig::strict(),
20        HtmlSanitizerMode::Compat => SanitizerConfig::compat(),
21    };
22
23    if remove_reply_fallback == RemoveReplyFallback::Yes {
24        config = config.remove_reply_fallback();
25    }
26
27    sanitize_inner(s, &config)
28}
29
30/// Whether to remove the [rich reply] fallback while sanitizing.
31///
32/// [rich reply]: https://spec.matrix.org/latest/client-server-api/#rich-replies
33#[derive(Clone, Copy, Debug, PartialEq, Eq)]
34#[allow(clippy::exhaustive_enums)]
35pub enum RemoveReplyFallback {
36    /// Remove the rich reply fallback.
37    Yes,
38
39    /// Don't remove the rich reply fallback.
40    No,
41}
42
43/// Remove the [rich reply] fallback of the given HTML string.
44///
45/// Due to the fact that the HTML is parsed, note that malformed HTML and comments will be stripped
46/// from the output.
47///
48/// [rich reply]: https://spec.matrix.org/latest/client-server-api/#rich-replies
49pub fn remove_html_reply_fallback(s: &str) -> String {
50    let config = SanitizerConfig::new().remove_reply_fallback();
51    sanitize_inner(s, &config)
52}
53
54fn sanitize_inner(s: &str, config: &SanitizerConfig) -> String {
55    let html = Html::parse(s);
56    html.sanitize_with(config);
57    html.to_string()
58}