html: Add support for mathematical messages

According to MSC2191 / Matrix 1.11

crates/ruma-html/CHANGELOG.md

@@ -1,5 +1,13 @@
 # [unreleased]
 
+Breaking Changes:
+
+- `MatrixElement::Div` is now a newtype variant.
+
+Improvements:
+
+- Add support for mathematical messages, according to MSC2191 / Matrix 1.11
+
 # 0.2.0
 
 Breaking Changes:

crates/ruma-html/src/html/matrix.rs

@@ -152,7 +152,7 @@ pub enum MatrixElement {
     /// [`<div>`], a content division element.
     ///
     /// [`<div>`]: https://developer.mozilla.org/en-US/docs/Web/HTML/Element/div
-    Div,
+    Div(DivData),
 
     /// [`<table>`], a table element.
     ///
@@ -268,7 +268,10 @@ impl MatrixElement {
             }
             b"hr" => (Self::Hr, attrs.clone()),
             b"br" => (Self::Br, attrs.clone()),
-            b"div" => (Self::Div, attrs.clone()),
+            b"div" => {
+                let (data, attrs) = DivData::parse(attrs);
+                (Self::Div(data), attrs)
+            }
             b"table" => (Self::Table, attrs.clone()),
             b"thead" => (Self::Thead, attrs.clone()),
             b"tbody" => (Self::Tbody, attrs.clone()),
@@ -599,12 +602,23 @@ pub struct SpanData {
     ///
     /// [spoiler message]: https://spec.matrix.org/latest/client-server-api/#spoiler-messages
     pub spoiler: Option<StrTendril>,
+
+    /// `data-mx-maths`, an inline Matrix [mathematical message].
+    ///
+    /// The value is the mathematical notation in [LaTeX] format.
+    ///
+    /// If this attribute is present, the content of the span is the fallback representation of the
+    /// mathematical notation.
+    ///
+    /// [mathematical message]: https://spec.matrix.org/latest/client-server-api/#mathematical-messages
+    /// [LaTeX]: https://www.latex-project.org/
+    pub maths: Option<StrTendril>,
 }
 
 impl SpanData {
     /// Construct an empty `SpanData`.
     fn new() -> Self {
-        Self { bg_color: None, color: None, spoiler: None }
+        Self { bg_color: None, color: None, spoiler: None, maths: None }
     }
 
     /// Parse the given attributes to construct a new `SpanData`.
@@ -629,6 +643,9 @@ impl SpanData {
                 b"data-mx-spoiler" => {
                     data.spoiler = Some(attr.value.clone());
                 }
+                b"data-mx-maths" => {
+                    data.maths = Some(attr.value.clone());
+                }
                 _ => {
                     remaining_attrs.insert(attr.clone());
                 }
@@ -722,3 +739,53 @@ impl ImageData {
         (data, remaining_attrs)
     }
 }
+
+/// The supported data of a `<div>` HTML element.
+#[derive(Debug, Clone)]
+#[non_exhaustive]
+pub struct DivData {
+    /// `data-mx-maths`, a Matrix [mathematical message] block.
+    ///
+    /// The value is the mathematical notation in [LaTeX] format.
+    ///
+    /// If this attribute is present, the content of the div is the fallback representation of the
+    /// mathematical notation.
+    ///
+    /// [mathematical message]: https://spec.matrix.org/latest/client-server-api/#mathematical-messages
+    /// [LaTeX]: https://www.latex-project.org/
+    pub maths: Option<StrTendril>,
+}
+
+impl DivData {
+    /// Construct an empty `DivData`.
+    fn new() -> Self {
+        Self { maths: None }
+    }
+
+    /// Parse the given attributes to construct a new `SpanData`.
+    ///
+    /// Returns a tuple containing the constructed data and the remaining unsupported attributes.
+    #[allow(clippy::mutable_key_type)]
+    fn parse(attrs: &BTreeSet<Attribute>) -> (Self, BTreeSet<Attribute>) {
+        let mut data = Self::new();
+        let mut remaining_attrs = BTreeSet::new();
+
+        for attr in attrs {
+            if attr.name.ns != ns!() {
+                remaining_attrs.insert(attr.clone());
+                continue;
+            }
+
+            match attr.name.local.as_bytes() {
+                b"data-mx-maths" => {
+                    data.maths = Some(attr.value.clone());
+                }
+                _ => {
+                    remaining_attrs.insert(attr.clone());
+                }
+            }
+        }
+
+        (data, remaining_attrs)
+    }
+}

crates/ruma-html/src/sanitizer_config/clean.rs

@@ -28,14 +28,16 @@ static ALLOWED_ATTRIBUTES_STRICT: Map<&str, &Set<&str>> = phf_map! {
     "img" => &ALLOWED_ATTRIBUTES_IMG_STRICT,
     "ol" => &ALLOWED_ATTRIBUTES_OL_STRICT,
     "code" => &ALLOWED_ATTRIBUTES_CODE_STRICT,
+    "div" => &ALLOWED_ATTRIBUTES_DIV_STRICT,
 };
 static ALLOWED_ATTRIBUTES_SPAN_STRICT: Set<&str> =
-    phf_set! { "data-mx-bg-color", "data-mx-color", "data-mx-spoiler" };
+    phf_set! { "data-mx-bg-color", "data-mx-color", "data-mx-spoiler", "data-mx-maths" };
 static ALLOWED_ATTRIBUTES_A_STRICT: Set<&str> = phf_set! { "name", "target", "href" };
 static ALLOWED_ATTRIBUTES_IMG_STRICT: Set<&str> =
     phf_set! { "width", "height", "alt", "title", "src" };
 static ALLOWED_ATTRIBUTES_OL_STRICT: Set<&str> = phf_set! { "start" };
 static ALLOWED_ATTRIBUTES_CODE_STRICT: Set<&str> = phf_set! { "class" };
+static ALLOWED_ATTRIBUTES_DIV_STRICT: Set<&str> = phf_set! { "data-mx-maths" };
 
 /// Attributes that were previously allowed on HTML elements according to the Matrix specification,
 /// with their replacement.

crates/ruma-html/tests/it/html/matrix.rs

@@ -26,7 +26,8 @@ fn elements() {
     // `<div>` element.
     let div_node = html_children.next().unwrap();
     let div_element = div_node.as_element().unwrap().to_matrix();
-    assert_matches!(div_element.element, MatrixElement::Div);
+    assert_matches!(div_element.element, MatrixElement::Div(div));
+    assert_eq!(div.maths, None);
     // The `class` attribute is not supported.
     assert_eq!(div_element.attrs.len(), 1);