html: Allow to navigate through the HTML tree

crates/ruma-html/CHANGELOG.md

@@ -8,6 +8,8 @@ Breaking Changes:
 Improvements:
 
 - Add support for deprecated HTML tags, according to Matrix 1.10
+- Allow to navigate through the HTML tree with `Html::first_child()`,
+  `Html::last_child()` or `Html::children()`
 
 # 0.1.0
 

crates/ruma-html/src/html.rs

@@ -1,4 +1,4 @@
-use std::{collections::BTreeSet, fmt, io};
+use std::{collections::BTreeSet, fmt, io, iter::FusedIterator};
 
 use as_variant::as_variant;
 use html5ever::{
@@ -122,6 +122,30 @@ impl Html {
     pub(crate) fn root(&self) -> &Node {
         &self.nodes[self.root_id()]
     }
+
+    /// Whether the root node of the HTML has children.
+    pub fn has_children(&self) -> bool {
+        self.root().first_child.is_some()
+    }
+
+    /// The first child node of the root node of the HTML.
+    ///
+    /// Returns `None` if the root node has no children.
+    pub fn first_child(&self) -> Option<NodeRef<'_>> {
+        self.root().first_child.map(|id| NodeRef::new(self, id))
+    }
+
+    /// The last child node of the root node of the HTML .
+    ///
+    /// Returns `None` if the root node has no children.
+    pub fn last_child(&self) -> Option<NodeRef<'_>> {
+        self.root().last_child.map(|id| NodeRef::new(self, id))
+    }
+
+    /// Iterate through the children of the root node of the HTML.
+    pub fn children(&self) -> Children<'_> {
+        Children::new(self.first_child())
+    }
 }
 
 impl Default for Html {
@@ -329,6 +353,11 @@ impl Node {
         as_variant!(&mut self.data, NodeData::Element)
     }
 
+    /// Returns the text content of this `Node`, if it is a `NodeData::Text`.
+    fn as_text(&self) -> Option<&StrTendril> {
+        as_variant!(&self.data, NodeData::Text)
+    }
+
     /// Returns the mutable text content of this `Node`, if it is a `NodeData::Text`.
     fn as_text_mut(&mut self) -> Option<&mut StrTendril> {
         as_variant!(&mut self.data, NodeData::Text)
@@ -375,9 +404,9 @@ impl Node {
 }
 
 /// The data of a `Node`.
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 #[allow(clippy::exhaustive_enums)]
-pub(crate) enum NodeData {
+pub enum NodeData {
     /// The root node of the `Html`.
     Document,
 
@@ -392,7 +421,7 @@ pub(crate) enum NodeData {
 }
 
 /// The data of an HTML element.
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 #[allow(clippy::exhaustive_structs)]
 pub struct ElementData {
     /// The qualified name of the element.
@@ -402,6 +431,123 @@ pub struct ElementData {
     pub attrs: BTreeSet<Attribute>,
 }
 
+/// A reference to an HTML node.
+#[derive(Debug, Clone, Copy)]
+#[non_exhaustive]
+pub struct NodeRef<'a> {
+    /// The `Html` struct containing the nodes.
+    pub(crate) html: &'a Html,
+    /// The referenced node.
+    pub(crate) node: &'a Node,
+}
+
+impl<'a> NodeRef<'a> {
+    /// Construct a new `NodeRef` for the given HTML and node ID.
+    fn new(html: &'a Html, id: usize) -> Self {
+        Self { html, node: &html.nodes[id] }
+    }
+
+    /// Construct a new `NodeRef` from the same HTML as this node with the given node ID.
+    fn with_id(&self, id: usize) -> Self {
+        let html = self.html;
+        Self::new(html, id)
+    }
+
+    /// The data of the node.
+    pub fn data(&self) -> &'a NodeData {
+        &self.node.data
+    }
+
+    /// Returns the data of this node if it is a `NodeData::Element`.
+    pub fn as_element(&self) -> Option<&'a ElementData> {
+        self.node.as_element()
+    }
+
+    /// Returns the text content of this node, if it is a `NodeData::Text`.
+    pub fn as_text(&self) -> Option<&'a StrTendril> {
+        self.node.as_text()
+    }
+
+    /// The parent node of this node.
+    ///
+    /// Returns `None` if the parent is the root node.
+    pub fn parent(&self) -> Option<NodeRef<'a>> {
+        let parent_id = self.node.parent?;
+
+        // We don't want users to be able to navigate to the root.
+        if parent_id == self.html.root_id() {
+            return None;
+        }
+
+        Some(self.with_id(parent_id))
+    }
+
+    /// The next sibling node of this node.
+    ///
+    /// Returns `None` if this is the last of its siblings.
+    pub fn next_sibling(&self) -> Option<NodeRef<'a>> {
+        Some(self.with_id(self.node.next_sibling?))
+    }
+
+    /// The previous sibling node of this node.
+    ///
+    /// Returns `None` if this is the first of its siblings.
+    pub fn prev_sibling(&self) -> Option<NodeRef<'a>> {
+        Some(self.with_id(self.node.prev_sibling?))
+    }
+
+    /// Whether this node has children.
+    pub fn has_children(&self) -> bool {
+        self.node.first_child.is_some()
+    }
+
+    /// The first child node of this node.
+    ///
+    /// Returns `None` if this node has no children.
+    pub fn first_child(&self) -> Option<NodeRef<'a>> {
+        Some(self.with_id(self.node.first_child?))
+    }
+
+    /// The last child node of this node.
+    ///
+    /// Returns `None` if this node has no children.
+    pub fn last_child(&self) -> Option<NodeRef<'a>> {
+        Some(self.with_id(self.node.last_child?))
+    }
+
+    /// Get an iterator through the children of this node.
+    pub fn children(&self) -> Children<'a> {
+        Children::new(self.first_child())
+    }
+}
+
+/// An iterator through the children of a node.
+///
+/// Can be constructed with [`Html::children()`] or [`NodeRef::children()`].
+#[derive(Debug, Clone, Copy)]
+pub struct Children<'a> {
+    next: Option<NodeRef<'a>>,
+}
+
+impl<'a> Children<'a> {
+    /// Construct a `Children` starting from the given node.
+    fn new(start_node: Option<NodeRef<'a>>) -> Self {
+        Self { next: start_node }
+    }
+}
+
+impl<'a> Iterator for Children<'a> {
+    type Item = NodeRef<'a>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        let next = self.next?;
+        self.next = next.next_sibling();
+        Some(next)
+    }
+}
+
+impl<'a> FusedIterator for Children<'a> {}
+
 #[cfg(test)]
 mod tests {
     use super::Html;

crates/ruma-html/src/lib.rs

@@ -14,8 +14,4 @@ mod helpers;
 mod html;
 mod sanitizer_config;
 
-pub use self::{
-    helpers::*,
-    html::{ElementData, Html},
-    sanitizer_config::SanitizerConfig,
-};
+pub use self::{helpers::*, html::*, sanitizer_config::SanitizerConfig};

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

@@ -1 +1,2 @@
+mod navigate;
 mod sanitize;

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

@@ -0,0 +1,140 @@
+use ruma_html::Html;
+
+#[test]
+fn navigate_tree() {
+    let raw_html = "\
+        <h1>Title</h1>\
+        <div class=\"text\">\
+            <p>This is some <em>text</em></p>\
+        </div>\
+    ";
+    let html = Html::parse(raw_html);
+
+    assert!(html.has_children());
+    assert!(html.first_child().is_some());
+    assert!(html.last_child().is_some());
+
+    let mut html_children = html.children();
+
+    // `<h1>` element.
+    let h1_node = html_children.next().unwrap();
+
+    let h1_element = h1_node.as_element().unwrap();
+    assert_eq!(&h1_element.name.local, "h1");
+    assert!(h1_element.attrs.is_empty());
+
+    assert!(h1_node.parent().is_none());
+    assert!(h1_node.next_sibling().is_some());
+    assert!(h1_node.prev_sibling().is_none());
+    assert!(h1_node.has_children());
+    assert!(h1_node.first_child().is_some());
+    assert!(h1_node.last_child().is_some());
+
+    let mut h1_children = h1_node.children();
+
+    // Text of `<h1>` element.
+    let h1_text_node = h1_children.next().unwrap();
+    let h1_text = h1_text_node.as_text().unwrap();
+    assert_eq!(h1_text.as_ref(), "Title");
+
+    assert!(h1_text_node.parent().is_some());
+    assert!(h1_text_node.next_sibling().is_none());
+    assert!(h1_text_node.prev_sibling().is_none());
+    assert!(!h1_text_node.has_children());
+    assert!(h1_text_node.first_child().is_none());
+    assert!(h1_text_node.last_child().is_none());
+
+    let mut h1_text_children = h1_text_node.children();
+    assert!(h1_text_children.next().is_none());
+
+    assert!(h1_children.next().is_none());
+
+    // `<div>` element.
+    let div_node = html_children.next().unwrap();
+
+    let div_element = div_node.as_element().unwrap();
+    assert_eq!(&div_element.name.local, "div");
+    assert_eq!(div_element.attrs.len(), 1);
+    let class_attr = div_element.attrs.first().unwrap();
+    assert_eq!(&class_attr.name.local, "class");
+    assert_eq!(class_attr.value.as_ref(), "text");
+
+    assert!(div_node.parent().is_none());
+    assert!(div_node.next_sibling().is_none());
+    assert!(div_node.prev_sibling().is_some());
+    assert!(div_node.has_children());
+    assert!(div_node.first_child().is_some());
+    assert!(div_node.last_child().is_some());
+
+    let mut div_children = div_node.children();
+
+    // `<p>` element.
+    let p_node = div_children.next().unwrap();
+
+    let p_element = p_node.as_element().unwrap();
+    assert_eq!(&p_element.name.local, "p");
+    assert!(p_element.attrs.is_empty());
+
+    assert!(p_node.parent().is_some());
+    assert!(p_node.next_sibling().is_none());
+    assert!(p_node.prev_sibling().is_none());
+    assert!(p_node.has_children());
+    assert!(p_node.first_child().is_some());
+    assert!(p_node.last_child().is_some());
+
+    let mut p_children = p_node.children();
+
+    // Text of `<p>` element.
+    let p_text_node = p_children.next().unwrap();
+    let p_text = p_text_node.as_text().unwrap();
+    assert_eq!(p_text.as_ref(), "This is some ");
+
+    assert!(p_text_node.parent().is_some());
+    assert!(p_text_node.next_sibling().is_some());
+    assert!(p_text_node.prev_sibling().is_none());
+    assert!(!p_text_node.has_children());
+    assert!(p_text_node.first_child().is_none());
+    assert!(p_text_node.last_child().is_none());
+
+    let mut p_text_children = p_text_node.children();
+    assert!(p_text_children.next().is_none());
+
+    // `<em>` element.
+    let em_node = p_children.next().unwrap();
+
+    let em_element = em_node.as_element().unwrap();
+    assert_eq!(&em_element.name.local, "em");
+    assert!(em_element.attrs.is_empty());
+
+    assert!(em_node.parent().is_some());
+    assert!(em_node.next_sibling().is_none());
+    assert!(em_node.prev_sibling().is_some());
+    assert!(em_node.has_children());
+    assert!(em_node.first_child().is_some());
+    assert!(em_node.last_child().is_some());
+
+    let mut em_children = em_node.children();
+
+    // Text of `<em>` element.
+    let em_text_node = em_children.next().unwrap();
+    let em_text = em_text_node.as_text().unwrap();
+    assert_eq!(em_text.as_ref(), "text");
+
+    assert!(em_text_node.parent().is_some());
+    assert!(em_text_node.next_sibling().is_none());
+    assert!(em_text_node.prev_sibling().is_none());
+    assert!(!em_text_node.has_children());
+    assert!(em_text_node.first_child().is_none());
+    assert!(em_text_node.last_child().is_none());
+
+    let mut em_text_children = em_text_node.children();
+    assert!(em_text_children.next().is_none());
+
+    assert!(em_children.next().is_none());
+
+    assert!(p_children.next().is_none());
+
+    assert!(div_children.next().is_none());
+
+    assert!(html_children.next().is_none());
+}