domtree/
token.rs

1//! The token type produced by the streaming [`Tokenizer`](crate::Tokenizer).
2//!
3//! Tokens borrow directly from the source string — tokenizing allocates nothing
4//! except the per-tag attribute list. Text and attribute values are returned
5//! **raw** (HTML entities are *not* decoded here); decoding is deferred until a
6//! tree node is built or [`text`](crate::NodeRef::text) is requested, which
7//! keeps the token stream zero-copy.
8
9use alloc::vec::Vec;
10
11/// A single lexical token of HTML.
12#[derive(Clone, Debug, PartialEq)]
13pub enum Token<'a> {
14    /// A start tag such as `<div class="x">` or a self-closing `<br/>`.
15    StartTag {
16        /// Tag name, exactly as written (not lower-cased).
17        name: &'a str,
18        /// Attributes in source order. Duplicate names are kept as-is.
19        attrs: Vec<(&'a str, AttrValue<'a>)>,
20        /// Whether the tag used XML-style self-closing syntax (`/>`).
21        self_closing: bool,
22    },
23    /// An end tag such as `</div>`.
24    EndTag {
25        /// Tag name, exactly as written.
26        name: &'a str,
27    },
28    /// A run of character data. Raw (undecoded) slice of the source.
29    Text(&'a str),
30    /// The inner text of an `<!-- ... -->` comment (or a bogus comment).
31    Comment(&'a str),
32    /// The inner text of a `<!doctype ...>` declaration.
33    Doctype(&'a str),
34    /// The inner text of a `<![CDATA[ ... ]]>` section.
35    Cdata(&'a str),
36}
37
38/// The value side of an attribute.
39#[derive(Clone, Copy, Debug, PartialEq, Eq)]
40pub enum AttrValue<'a> {
41    /// `name="value"` or `name='value'` — slice excludes the quotes.
42    Quoted(&'a str),
43    /// `name=value` with no quotes.
44    Unquoted(&'a str),
45    /// A boolean attribute with no value, e.g. `disabled`.
46    Empty,
47}
48
49impl<'a> AttrValue<'a> {
50    /// The raw value as a string slice (`""` for [`Empty`](AttrValue::Empty)).
51    pub fn as_str(&self) -> &'a str {
52        match self {
53            AttrValue::Quoted(s) | AttrValue::Unquoted(s) => s,
54            AttrValue::Empty => "",
55        }
56    }
57}
domtree/token.rs

domtree/
token.rs
