aboutsummaryrefslogtreecommitdiff
path: root/src/emitter.rs
diff options
context:
space:
mode:
Diffstat (limited to 'src/emitter.rs')
-rw-r--r--src/emitter.rs81
1 files changed, 42 insertions, 39 deletions
diff --git a/src/emitter.rs b/src/emitter.rs
index 18b2539..b3fdb99 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -4,9 +4,11 @@ use std::collections::BTreeSet;
use std::collections::VecDeque;
use std::marker::PhantomData;
use std::mem;
+use std::ops::Range;
-use crate::spans::Position;
-use crate::spans::Span;
+use crate::offset::NoopOffset;
+use crate::offset::Offset;
+use crate::offset::Position;
use crate::Error;
/// An emitter is an object providing methods to the tokenizer to produce ("emit") tokens.
@@ -160,17 +162,17 @@ pub trait Emitter<R> {
}
/// The default implementation of [`Emitter`], used to produce tokens.
-pub struct DefaultEmitter<R, S = ()> {
+pub struct DefaultEmitter<R, O = NoopOffset> {
current_characters: String,
- current_token: Option<Token<S>>,
- current_attribute: Option<(String, Attribute<S>)>,
+ current_token: Option<Token<O>>,
+ current_attribute: Option<(String, Attribute<O>)>,
seen_attributes: BTreeSet<String>,
- emitted_tokens: VecDeque<Token<S>>,
+ emitted_tokens: VecDeque<Token<O>>,
reader: PhantomData<R>,
- attr_in_end_tag_span: Option<S>,
+ attr_in_end_tag_span: Option<Range<O>>,
}
-impl<R, S> Default for DefaultEmitter<R, S> {
+impl<R, O> Default for DefaultEmitter<R, O> {
fn default() -> Self {
DefaultEmitter {
current_characters: String::new(),
@@ -184,13 +186,16 @@ impl<R, S> Default for DefaultEmitter<R, S> {
}
}
-impl<R, S: Span> DefaultEmitter<R, S> {
- fn emit_token(&mut self, token: Token<S>) {
+impl<R, O> DefaultEmitter<R, O> {
+ fn emit_token(&mut self, token: Token<O>) {
self.flush_current_characters();
self.emitted_tokens.push_front(token);
}
- fn flush_current_attribute(&mut self) {
+ fn flush_current_attribute(&mut self)
+ where
+ O: Clone,
+ {
if let Some((k, v)) = self.current_attribute.take() {
match self.current_token {
Some(Token::StartTag(ref mut tag)) => match tag.attributes.entry(k) {
@@ -223,22 +228,22 @@ impl<R, S: Span> DefaultEmitter<R, S> {
self.emit_token(Token::String(s));
}
- fn push_error(&mut self, error: Error, span: S) {
+ fn push_error(&mut self, error: Error, span: Range<O>) {
// bypass character flushing in self.emit_token: we don't need the error location to be
// that exact
self.emitted_tokens.push_front(Token::Error { error, span });
}
}
-impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
- type Token = Token<S>;
+impl<O: Offset, R: Position<O>> Emitter<R> for DefaultEmitter<R, O> {
+ type Token = Token<O>;
fn emit_eof(&mut self) {
self.flush_current_characters();
}
fn emit_error(&mut self, error: Error, reader: &R) {
- self.push_error(error, S::new(reader.position(), reader.position()));
+ self.push_error(error, reader.position()..reader.position());
}
fn pop_token(&mut self) -> Option<Self::Token> {
@@ -251,7 +256,7 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
fn init_start_tag(&mut self, reader: &R) {
self.current_token = Some(Token::StartTag(StartTag {
- name_span: S::new(reader.position(), reader.position()),
+ name_span: reader.position()..reader.position(),
self_closing: false,
name: String::new(),
attributes: Default::default(),
@@ -259,7 +264,7 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
}
fn init_end_tag(&mut self, reader: &R) {
self.current_token = Some(Token::EndTag(EndTag {
- name_span: S::new(reader.position(), reader.position()),
+ name_span: reader.position()..reader.position(),
name: String::new(),
}));
self.seen_attributes.clear();
@@ -327,7 +332,7 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
..
})) => {
name.push_str(s);
- name_span.push_str(s);
+ name_span.end += s.len();
}
Some(Token::EndTag(EndTag {
ref mut name,
@@ -335,7 +340,7 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
..
})) => {
name.push_str(s);
- name_span.push_str(s);
+ name_span.end += s.len();
}
_ => debug_assert!(false),
}
@@ -368,28 +373,26 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
self.current_attribute = Some((
String::new(),
Attribute {
- name_span: S::new(reader.position(), reader.position()),
+ name_span: reader.position()..reader.position(),
value: String::new(),
- value_span: S::default(),
+ value_span: Range::default(),
},
));
}
fn init_attribute_value(&mut self, reader: &R, quoted: bool) {
- self.current_attribute.as_mut().unwrap().1.value_span = S::new(
- reader.position() + quoted as usize,
- reader.position() + quoted as usize,
- );
+ self.current_attribute.as_mut().unwrap().1.value_span =
+ reader.position() + quoted as usize..reader.position() + quoted as usize;
}
fn push_attribute_name(&mut self, s: &str) {
let current_attr = self.current_attribute.as_mut().unwrap();
current_attr.0.push_str(s);
- current_attr.1.name_span.push_str(s);
+ current_attr.1.name_span.end += s.len();
}
fn push_attribute_value(&mut self, s: &str) {
let current_attr = self.current_attribute.as_mut().unwrap();
current_attr.1.value.push_str(s);
- current_attr.1.value_span.push_str(s);
+ current_attr.1.value_span.end += s.len();
}
fn set_doctype_public_identifier(&mut self, value: &str) {
if let Some(Token::Doctype(Doctype {
@@ -439,7 +442,7 @@ impl<R: Position<S::Offset>, S: Span> Emitter<R> for DefaultEmitter<R, S> {
/// An HTML start tag, such as `<p>` or `<a>`.
#[derive(Debug, Eq, PartialEq)]
-pub struct StartTag<S> {
+pub struct StartTag<O> {
/// Whether this tag is self-closing. If it is self-closing, no following [`EndTag`] should be
/// expected.
pub self_closing: bool,
@@ -451,33 +454,33 @@ pub struct StartTag<S> {
///
/// Duplicate attributes are ignored after the first one as per WHATWG spec. Implement your own
/// [`Emitter`] to tweak this behavior.
- pub attributes: BTreeMap<String, Attribute<S>>,
+ pub attributes: BTreeMap<String, Attribute<O>>,
/// The source code span of the tag name.
- pub name_span: S,
+ pub name_span: Range<O>,
}
/// A HTML attribute value (plus spans).
#[derive(Debug, Eq, PartialEq)]
-pub struct Attribute<S> {
+pub struct Attribute<O> {
/// The value of the attribute.
pub value: String,
/// The source code span of the attribute name.
- pub name_span: S,
+ pub name_span: Range<O>,
/// The source code span of the attribute value.
- pub value_span: S,
+ pub value_span: Range<O>,
}
/// A HTML end/close tag, such as `</p>` or `</a>`.
#[derive(Debug, Eq, PartialEq)]
-pub struct EndTag<S> {
+pub struct EndTag<O> {
/// The ending tag's name, such as `"p"` or `"a"`.
pub name: String,
/// The source code span of the tag name.
- pub name_span: S,
+ pub name_span: Range<O>,
}
/// A doctype. Some examples:
@@ -504,11 +507,11 @@ pub struct Doctype {
/// The token type used by default. You can define your own token type by implementing the
/// [`Emitter`] trait.
#[derive(Debug, Eq, PartialEq)]
-pub enum Token<S> {
+pub enum Token<O> {
/// A HTML start tag.
- StartTag(StartTag<S>),
+ StartTag(StartTag<O>),
/// A HTML end tag.
- EndTag(EndTag<S>),
+ EndTag(EndTag<O>),
/// A literal string.
String(String),
/// A HTML comment.
@@ -523,6 +526,6 @@ pub enum Token<S> {
/// What kind of error occured.
error: Error,
/// The source code span of the error.
- span: S,
+ span: Range<O>,
},
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-09-20 18:40:04 +0000