path: root/src/token.rs

//! Provides the [`Token`] type.

use std::collections::{btree_map, BTreeMap};
use std::iter::FromIterator;
use std::ops::Index;

/// A type for the tokens emitted by a WHATWG-compliant HTML tokenizer.
#[derive(Debug, Eq, PartialEq)]
pub enum Token {
    /// A literal character, a resolved character reference,
    /// or part of a resolved character reference (since some
    /// character references resolve to two `char`s).
    Char(char),
    /// An HTML start tag.
    StartTag(StartTag),
    /// An HTML end tag.
    EndTag(EndTag),
    /// An HTML comment.
    Comment(String),
    /// An HTML doctype declaration.
    Doctype(Doctype),
    /// An end-of-file token.
    EndOfFile,
}

/// An HTML start tag, such as `<p>` or `<a>`.
#[derive(Debug, Eq, PartialEq)]
pub struct StartTag {
    /// Whether this tag is self-closing. If it is self-closing, no following [`EndTag`] should be
    /// expected.
    pub self_closing: bool,

    /// The tag name.
    /// Uppercase ASCII characters (A-Z) have been converted to lowercase.
    pub name: String,

    /// A mapping for any HTML attributes this start tag may have.
    ///
    /// Duplicate attributes are ignored after the first one as per WHATWG spec.
    pub attributes: AttributeMap,
}

/// An HTML end/close tag, such as `</p>` or `</a>`.
#[derive(Debug, Eq, PartialEq)]
pub struct EndTag {
    /// The tag name.
    /// Uppercase ASCII characters (A-Z) have been converted to lowercase.
    pub name: String,
}

/// A doctype. Some examples:
///
/// * `<!DOCTYPE {name}>`
/// * `<!DOCTYPE {name} PUBLIC '{public_id}'>`
/// * `<!DOCTYPE {name} SYSTEM '{system_id}'>`
/// * `<!DOCTYPE {name} PUBLIC '{public_id}' '{system_id}'>`
#[derive(Debug, Eq, PartialEq)]
pub struct Doctype {
    /// The [force-quirks flag].
    ///
    /// [force-quirks flag]: https://html.spec.whatwg.org/multipage/parsing.html#force-quirks-flag
    pub force_quirks: bool,

    /// The doctype's name. Uppercase ASCII characters (A-Z) have been
    /// converted to lowercase. For HTML documents this should be "html".
    pub name: Option<String>,

    /// The doctype's public identifier.
    pub public_id: Option<String>,

    /// The doctype's system identifier.
    pub system_id: Option<String>,
}

/// A map of HTML attributes.
///
/// Does not preserve the order of attributes.
/// Iterating always yields attributes in order by name.
///
/// # Example
///
/// ```
/// # use html5tokenizer::attr::AttributeMap;
/// let attrs: AttributeMap = vec![("href".into(), "http://example.com".into())]
///     .into_iter()
///     .collect();
/// assert_eq!(&attrs["href"], "http://example.com");
/// ```
#[derive(Debug, Default, PartialEq, Eq)]
pub struct AttributeMap {
    pub(crate) inner: BTreeMap<String, AttrInternal>,
}

/// The value type internally used by the [`AttributeMap`].
/// Not part of the public API.
#[derive(Default, Debug, Eq)]
pub(crate) struct AttrInternal {
    pub value: String,
    pub trace_idx: Option<AttributeTraceIdx>,
}

/// The index of an [`AttributeTrace`] within an [`AttributeTraceList`].
///
/// [`AttributeTrace`]: crate::trace::AttributeTrace
/// [`AttributeTraceList`]: crate::trace::AttributeTraceList
#[derive(Clone, Copy, Eq, PartialEq, Debug)]
pub struct AttributeTraceIdx(
    // Using NonZeroUsize so that `Option<AttributeTraceIdx>`
    // has the same size as `AttributeTraceIdx`.
    pub std::num::NonZeroUsize,
);

impl PartialEq for AttrInternal {
    fn eq(&self, other: &Self) -> bool {
        // We intentionally don't include the trace_idx,
        // so that PartialEq of Token only compares semantics.
        self.value == other.value
    }
}

/// An HTML attribute borrowed from an [`AttributeMap`].
#[derive(Debug, Eq, PartialEq)]
pub struct Attribute<'a> {
    name: &'a str,
    map_val: &'a AttrInternal,
}

/// An owned HTML attribute.
#[derive(Debug, PartialEq, Eq)]
pub struct AttributeOwned {
    /// The attribute name.
    /// Uppercase ASCII characters (A-Z) have been converted to lowercase.
    pub name: String,
    /// The attribute value. Character references have been resolved.
    pub value: String,
    /// The index of the corresponding [`AttributeTrace`] in the
    /// `attribute_traces` field of [`StartTagTrace`], in case this attribute
    /// was present in the source and the [`Emitter`] has tracked this.
    ///
    /// [`AttributeTrace`]: super::trace::AttributeTrace
    /// [`StartTagTrace`]: super::trace::AttributeTrace
    /// [`Emitter`]: super::Emitter
    pub trace_idx: Option<AttributeTraceIdx>,
}

impl AttributeMap {
    /// Returns the attribute with the given name.
    ///
    /// The name must not contain any uppercase ASCII character (A-Z)
    /// or the method will always return `None`.
    pub fn get(&self, name: &str) -> Option<Attribute> {
        self.inner
            .get_key_value(name)
            .map(|(name, map_val)| Attribute { name, map_val })
    }
}

impl<'a> Attribute<'a> {
    /// Returns the attribute name.
    /// Uppercase ASCII characters (A-Z) have been converted to lowercase.
    pub fn name(&self) -> &'a str {
        self.name
    }

    /// Returns the attribute value. Character references have been resolved.
    pub fn value(&self) -> &'a str {
        &self.map_val.value
    }

    /// Returns the index of the corresponding [`AttributeTrace`] in the
    /// `attribute_traces` field of [`StartTagTrace`], in case this attribute
    /// was present in the source and the [`Emitter`] has tracked that.
    ///
    /// [`AttributeTrace`]: super::trace::AttributeTrace
    /// [`StartTagTrace`]: super::trace::AttributeTrace
    /// [`Emitter`]: super::Emitter
    pub fn trace_idx(&self) -> Option<AttributeTraceIdx> {
        self.map_val.trace_idx
    }
}

// We cannot impl Index<Output=Attribute> because Index::index returns a reference of
// the Output type (and you cannot return a value referencing a temporary value).
impl Index<&str> for AttributeMap {
    type Output = str;

    /// Returns the attribute value with the given name.
    ///
    /// The name must not contain any uppercase ASCII character (A-Z)
    /// or the method will always panic.
    fn index(&self, name: &str) -> &Self::Output {
        &self.inner[name].value
    }
}

impl IntoIterator for AttributeMap {
    type Item = AttributeOwned;

    type IntoIter = AttrIntoIter;

    fn into_iter(self) -> Self::IntoIter {
        AttrIntoIter(self.inner.into_iter())
    }
}

/// A consuming iterator over the attributes of an [`AttributeMap`].
pub struct AttrIntoIter(btree_map::IntoIter<String, AttrInternal>);

impl Iterator for AttrIntoIter {
    type Item = AttributeOwned;

    fn next(&mut self) -> Option<Self::Item> {
        let (name, map_val) = self.0.next()?;
        Some(AttributeOwned {
            name,
            value: map_val.value,
            trace_idx: map_val.trace_idx,
        })
    }
}

impl<'a> IntoIterator for &'a AttributeMap {
    type Item = Attribute<'a>;

    type IntoIter = AttrIter<'a>;

    fn into_iter(self) -> Self::IntoIter {
        AttrIter(self.inner.iter())
    }
}

/// A borrowed iterator over the attributes of an [`AttributeMap`].
pub struct AttrIter<'a>(btree_map::Iter<'a, String, AttrInternal>);

impl<'a> Iterator for AttrIter<'a> {
    type Item = Attribute<'a>;

    fn next(&mut self) -> Option<Self::Item> {
        let (name, map_val) = self.0.next()?;
        Some(Attribute { name, map_val })
    }
}

impl FromIterator<(String, String)> for AttributeMap {
    fn from_iter<T: IntoIterator<Item = (String, String)>>(iter: T) -> Self {
        Self {
            inner: iter
                .into_iter()
                .map(|(name, value)| {
                    (
                        name,
                        AttrInternal {
                            value,
                            trace_idx: None,
                        },
                    )
                })
                .collect(),
        }
    }
}