|
|
|
@ -70,7 +70,9 @@ where
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
|
|
|
|
|
pub enum Error {}
|
|
|
|
|
pub enum Error {
|
|
|
|
|
UnclosedExtension,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
|
|
|
|
|
pub struct Header<S> {
|
|
|
|
@ -79,13 +81,20 @@ pub struct Header<S> {
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
|
|
|
|
|
pub enum Inline<S> {
|
|
|
|
|
Text(S),
|
|
|
|
|
pub struct Text<S>(pub S);
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
|
|
|
|
|
pub struct Extension<S>(pub S);
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
|
|
|
|
|
pub enum ParagraphPiece<S> {
|
|
|
|
|
Text(Text<S>),
|
|
|
|
|
Extension(Extension<S>),
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Clone, Eq, PartialEq)]
|
|
|
|
|
pub struct Paragraph<S> {
|
|
|
|
|
pieces: Vec<Spanned<Inline<S>>>,
|
|
|
|
|
pieces: Vec<Spanned<ParagraphPiece<S>>>,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Clone, Eq, PartialEq)]
|
|
|
|
@ -96,6 +105,7 @@ pub enum Block<S> {
|
|
|
|
|
|
|
|
|
|
pub type Result<T> = core::result::Result<T, Spanned<Error>>;
|
|
|
|
|
|
|
|
|
|
#[derive(Copy, Clone)]
|
|
|
|
|
struct State<'a> {
|
|
|
|
|
corpus: &'a str,
|
|
|
|
|
current_offset: ByteOffset,
|
|
|
|
@ -119,20 +129,27 @@ impl<'a> State<'a> {
|
|
|
|
|
))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn peek_n(&self, buf: &mut [Option<(char, ByteOffset)>]) {
|
|
|
|
|
let mut self_ = *self;
|
|
|
|
|
for slot in buf {
|
|
|
|
|
*slot = self_.next();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn next(&mut self) -> Option<(char, ByteOffset)> {
|
|
|
|
|
if self.is_eof() {
|
|
|
|
|
return None;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
let out = Some((
|
|
|
|
|
self.corpus
|
|
|
|
|
.char_at_byte_offset(self.current_offset)
|
|
|
|
|
.unwrap()
|
|
|
|
|
.unwrap(),
|
|
|
|
|
self.current_offset,
|
|
|
|
|
));
|
|
|
|
|
let char = self
|
|
|
|
|
.corpus
|
|
|
|
|
.char_at_byte_offset(self.current_offset)
|
|
|
|
|
.unwrap()
|
|
|
|
|
.unwrap();
|
|
|
|
|
|
|
|
|
|
let out = Some((char, self.current_offset));
|
|
|
|
|
|
|
|
|
|
self.current_offset.0 += 1;
|
|
|
|
|
self.current_offset.0 += char.len_utf8();
|
|
|
|
|
|
|
|
|
|
out
|
|
|
|
|
}
|
|
|
|
@ -141,6 +158,12 @@ impl<'a> State<'a> {
|
|
|
|
|
self.next();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn forward_n(&mut self, n: usize) {
|
|
|
|
|
for _ in 0..n {
|
|
|
|
|
self.forward();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn skip_whitespace(&mut self) {
|
|
|
|
|
while let Some((ch, _)) = self.peek() {
|
|
|
|
|
if !ch.is_whitespace() {
|
|
|
|
@ -156,7 +179,6 @@ fn header<'a, S>(state: &mut State<'a>) -> Result<Header<S>>
|
|
|
|
|
where
|
|
|
|
|
S: From<&'a str>,
|
|
|
|
|
{
|
|
|
|
|
let start = state.current_offset;
|
|
|
|
|
let mut level = 1;
|
|
|
|
|
|
|
|
|
|
state.forward();
|
|
|
|
@ -190,45 +212,72 @@ where
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn inline_text<'a, S>(state: &mut State<'a>) -> Result<(S, bool)>
|
|
|
|
|
fn extension<'a, S>(state: &mut State<'a>) -> Result<Extension<S>>
|
|
|
|
|
where
|
|
|
|
|
S: From<&'a str>,
|
|
|
|
|
{
|
|
|
|
|
debug_assert!({
|
|
|
|
|
let mut peek_buf = [None; 2];
|
|
|
|
|
state.peek_n(&mut peek_buf);
|
|
|
|
|
|
|
|
|
|
matches!(peek_buf, [Some(('@', _)), Some(('{', _))])
|
|
|
|
|
});
|
|
|
|
|
|
|
|
|
|
state.forward();
|
|
|
|
|
state.forward();
|
|
|
|
|
|
|
|
|
|
let start = state.current_offset;
|
|
|
|
|
|
|
|
|
|
let mut end: Option<ByteOffset> = None;
|
|
|
|
|
while let Some((ch, bo)) = state.next() {
|
|
|
|
|
match ch {
|
|
|
|
|
'\n' => match end {
|
|
|
|
|
Some(end) => {
|
|
|
|
|
return Ok((state.corpus[start.0..end.0].into(), true))
|
|
|
|
|
}
|
|
|
|
|
None => end = Some(bo),
|
|
|
|
|
},
|
|
|
|
|
loop {
|
|
|
|
|
let mut peek_buf = [None; 2];
|
|
|
|
|
state.peek_n(&mut peek_buf);
|
|
|
|
|
|
|
|
|
|
match peek_buf {
|
|
|
|
|
[None, _] => {
|
|
|
|
|
return Err(spanned(
|
|
|
|
|
start,
|
|
|
|
|
state.current_offset,
|
|
|
|
|
Error::UnclosedExtension,
|
|
|
|
|
))
|
|
|
|
|
}
|
|
|
|
|
[Some(('}', bo)), Some(('@', _))] => {
|
|
|
|
|
state.forward();
|
|
|
|
|
state.forward();
|
|
|
|
|
|
|
|
|
|
let content = &state.corpus[start.0..bo.0];
|
|
|
|
|
|
|
|
|
|
return Ok(Extension(content.into()));
|
|
|
|
|
}
|
|
|
|
|
_ => {
|
|
|
|
|
end = None;
|
|
|
|
|
state.forward();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Ok((
|
|
|
|
|
state.corpus
|
|
|
|
|
[start.0..end.map(|e| e.0).unwrap_or_else(|| state.corpus.len())]
|
|
|
|
|
.into(),
|
|
|
|
|
true,
|
|
|
|
|
))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn inline<'a, S>(state: &mut State<'a>) -> Result<(Inline<S>, bool)>
|
|
|
|
|
|
|
|
|
|
fn text<'a, S>(state: &mut State<'a>) -> Result<Text<S>>
|
|
|
|
|
where
|
|
|
|
|
S: From<&'a str>,
|
|
|
|
|
{
|
|
|
|
|
match state.peek() {
|
|
|
|
|
_ => {
|
|
|
|
|
let (inline, done) = inline_text(state)?;
|
|
|
|
|
Ok((Inline::Text(inline), done))
|
|
|
|
|
let start = state.current_offset;
|
|
|
|
|
|
|
|
|
|
loop {
|
|
|
|
|
let mut peek_buf = [None; 2];
|
|
|
|
|
state.peek_n(&mut peek_buf);
|
|
|
|
|
|
|
|
|
|
match peek_buf {
|
|
|
|
|
[None, _]
|
|
|
|
|
| [Some(('\n', _)), Some(('\n', _))]
|
|
|
|
|
| [Some(('\n', _)), None]
|
|
|
|
|
| [Some(('@', _)), Some(('{', _))] => break,
|
|
|
|
|
_ => state.forward(),
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
let end = state.current_offset;
|
|
|
|
|
|
|
|
|
|
Ok(Text(state.corpus[start.0..end.0].into()))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn paragraph<'a, S>(state: &mut State<'a>) -> Result<Paragraph<S>>
|
|
|
|
@ -239,14 +288,35 @@ where
|
|
|
|
|
|
|
|
|
|
loop {
|
|
|
|
|
let start = state.current_offset;
|
|
|
|
|
let (piece, done) = inline::<S>(state)?;
|
|
|
|
|
let end = state.current_offset;
|
|
|
|
|
|
|
|
|
|
pieces.push(spanned(start, end, piece));
|
|
|
|
|
// detect what type of thing we're about to parse.
|
|
|
|
|
let mut peek_buf = [None; 2];
|
|
|
|
|
state.peek_n(&mut peek_buf);
|
|
|
|
|
|
|
|
|
|
if done {
|
|
|
|
|
break;
|
|
|
|
|
let piece = match peek_buf {
|
|
|
|
|
// eof
|
|
|
|
|
[None, _] => break,
|
|
|
|
|
// double newline
|
|
|
|
|
[Some(('\n', _)), Some(('\n', _))] => {
|
|
|
|
|
state.forward_n(2);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
// newline eof
|
|
|
|
|
[Some(('\n', _)), None] => {
|
|
|
|
|
state.forward();
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
// extension
|
|
|
|
|
[Some(('@', _)), Some(('{', _))] => {
|
|
|
|
|
ParagraphPiece::Extension(extension(state)?)
|
|
|
|
|
}
|
|
|
|
|
// regular text
|
|
|
|
|
_ => ParagraphPiece::Text(text(state)?),
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
let end = state.current_offset;
|
|
|
|
|
|
|
|
|
|
pieces.push(spanned(start, end, piece));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Ok(Paragraph { pieces })
|
|
|
|
@ -353,7 +423,7 @@ mod test {
|
|
|
|
|
pieces: vec![spanned(
|
|
|
|
|
0,
|
|
|
|
|
corpus.len(),
|
|
|
|
|
Inline::Text("Hello, world")
|
|
|
|
|
ParagraphPiece::Text(Text("Hello, world"))
|
|
|
|
|
)]
|
|
|
|
|
})
|
|
|
|
|
)],
|
|
|
|
@ -374,8 +444,8 @@ mod test {
|
|
|
|
|
Block::Paragraph(Paragraph {
|
|
|
|
|
pieces: vec![spanned(
|
|
|
|
|
0,
|
|
|
|
|
14,
|
|
|
|
|
Inline::Text("Hello, world")
|
|
|
|
|
12,
|
|
|
|
|
ParagraphPiece::Text(Text("Hello, world"))
|
|
|
|
|
)]
|
|
|
|
|
})
|
|
|
|
|
),
|
|
|
|
@ -385,12 +455,36 @@ mod test {
|
|
|
|
|
Block::Paragraph(Paragraph {
|
|
|
|
|
pieces: vec![spanned(
|
|
|
|
|
14,
|
|
|
|
|
corpus.len(),
|
|
|
|
|
Inline::Text("Goodbye, world")
|
|
|
|
|
corpus.len() - 1,
|
|
|
|
|
ParagraphPiece::Text(Text("Goodbye, world"))
|
|
|
|
|
)]
|
|
|
|
|
})
|
|
|
|
|
)
|
|
|
|
|
],
|
|
|
|
|
)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
fn paragraph_extension_1() {
|
|
|
|
|
let corpus = "Hello @{world}@";
|
|
|
|
|
|
|
|
|
|
let output = parse(corpus).unwrap();
|
|
|
|
|
assert_eq!(
|
|
|
|
|
&output,
|
|
|
|
|
&[spanned(
|
|
|
|
|
0,
|
|
|
|
|
corpus.len(),
|
|
|
|
|
Block::Paragraph(Paragraph {
|
|
|
|
|
pieces: vec![
|
|
|
|
|
spanned(0, 6, ParagraphPiece::Text(Text("Hello "))),
|
|
|
|
|
spanned(
|
|
|
|
|
6,
|
|
|
|
|
corpus.len(),
|
|
|
|
|
ParagraphPiece::Extension(Extension("world"))
|
|
|
|
|
)
|
|
|
|
|
]
|
|
|
|
|
})
|
|
|
|
|
)]
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|