jotdown/src/inline.rs

401 lines
11 KiB
Rust
Raw Normal View History

2022-11-21 13:44:59 -05:00
use crate::lex;
2022-11-21 16:40:11 -05:00
use crate::Span;
2022-11-16 16:11:55 -05:00
2022-11-21 13:44:59 -05:00
use lex::Delimiter;
use lex::Symbol;
2022-11-16 16:11:55 -05:00
use Atom::*;
use Container::*;
2022-11-27 18:10:28 -05:00
use Node::*;
2022-11-16 16:11:55 -05:00
#[derive(Debug, Clone, PartialEq, Eq)]
2022-11-21 13:44:59 -05:00
pub enum Atom {
2022-11-16 16:11:55 -05:00
Softbreak,
Hardbreak,
Escape,
2022-11-21 13:44:59 -05:00
Nbsp,
OpenMarker, // ??
2022-11-28 18:33:43 -05:00
Ellipsis,
2022-11-20 13:13:48 -05:00
ImageMarker, // ??
2022-11-21 13:44:59 -05:00
EmDash,
EnDash,
2022-11-26 19:12:56 -05:00
Lt,
Gt,
Ampersand,
Quote,
}
#[derive(Debug, Clone, PartialEq, Eq)]
2022-11-27 18:10:28 -05:00
pub enum Node {
2022-11-26 19:12:56 -05:00
Str,
// link
2022-11-27 15:59:54 -05:00
Url,
ImageSource,
LinkReference,
2022-11-21 13:44:59 -05:00
FootnoteReference,
2022-11-26 19:12:56 -05:00
// verbatim
Verbatim,
2022-11-21 13:44:59 -05:00
RawFormat,
InlineMath,
2022-11-27 15:59:54 -05:00
DisplayMath,
2022-11-16 16:11:55 -05:00
}
2022-11-20 13:13:48 -05:00
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
2022-11-16 16:11:55 -05:00
pub enum Container {
Span,
2022-11-27 15:59:54 -05:00
Attributes,
2022-11-16 16:11:55 -05:00
// typesetting
Subscript,
Superscript,
Insert,
Delete,
2022-11-21 13:44:59 -05:00
Emphasis,
2022-11-16 16:11:55 -05:00
Strong,
2022-11-27 17:56:19 -05:00
Mark,
2022-11-16 16:11:55 -05:00
// smart quoting
SingleQuoted,
DoubleQuoted,
}
2022-11-21 16:40:11 -05:00
#[derive(Debug, PartialEq, Eq)]
2022-11-27 18:10:28 -05:00
pub enum EventKind {
2022-11-28 13:08:49 -05:00
Enter(Container),
2022-11-22 13:19:21 -05:00
Exit(Container),
2022-11-21 13:44:59 -05:00
Atom(Atom),
2022-11-26 19:12:56 -05:00
Node(Node),
2022-11-20 13:13:48 -05:00
}
2022-11-27 18:10:28 -05:00
#[derive(Debug, PartialEq, Eq)]
pub struct Event {
pub kind: EventKind,
pub span: Span,
}
2022-11-22 13:19:21 -05:00
#[derive(Debug, Clone, Copy)]
2022-11-21 13:44:59 -05:00
pub enum Dir {
Open,
Close,
2022-11-20 13:13:48 -05:00
Both,
}
2022-11-16 16:11:55 -05:00
2022-11-22 13:19:21 -05:00
pub struct Parser<'s> {
2022-11-27 17:56:19 -05:00
openers: Vec<(Container, usize)>,
events: std::collections::VecDeque<Event>,
2022-11-26 19:12:56 -05:00
span: Span,
lexer: std::iter::Peekable<lex::Lexer<'s>>,
2022-11-16 16:11:55 -05:00
}
2022-11-22 13:19:21 -05:00
impl<'s> Parser<'s> {
2022-11-21 13:56:11 -05:00
pub fn new() -> Self {
2022-11-16 16:11:55 -05:00
Self {
openers: Vec::new(),
2022-11-27 17:56:19 -05:00
events: std::collections::VecDeque::new(),
2022-11-26 19:12:56 -05:00
span: Span::new(0, 0),
lexer: lex::Lexer::new("").peekable(),
2022-11-20 13:13:48 -05:00
}
}
2022-11-22 13:19:21 -05:00
pub fn parse(&mut self, src: &'s str) {
2022-11-26 19:12:56 -05:00
self.lexer = lex::Lexer::new(src).peekable();
2022-11-22 13:19:21 -05:00
}
2022-11-26 19:12:56 -05:00
fn eat(&mut self) -> Option<lex::Token> {
let tok = self.lexer.next();
if let Some(t) = &tok {
self.span = self.span.extend(t.len);
2022-11-22 13:19:21 -05:00
}
2022-11-26 19:12:56 -05:00
tok
2022-11-21 13:56:11 -05:00
}
2022-11-26 19:12:56 -05:00
fn peek(&mut self) -> Option<&lex::Token> {
self.lexer.peek()
}
2022-11-21 13:56:11 -05:00
2022-11-26 19:12:56 -05:00
fn reset_span(&mut self) {
self.span = Span::empty_at(self.span.end());
2022-11-21 13:56:11 -05:00
}
2022-11-27 18:10:28 -05:00
fn node(&self, kind: Node) -> Event {
Event {
kind: EventKind::Node(kind),
2022-11-26 19:12:56 -05:00
span: self.span,
2022-11-27 18:10:28 -05:00
}
2022-11-21 17:32:28 -05:00
}
2022-11-21 16:40:11 -05:00
2022-11-26 19:12:56 -05:00
fn parse_event(&mut self) -> Option<Event> {
self.reset_span();
self.eat().map(|first| {
self.parse_verbatim(&first)
.or_else(|| self.parse_container(&first))
.or_else(|| self.parse_atom(&first))
.unwrap_or_else(|| self.node(Str))
})
}
2022-11-21 16:40:11 -05:00
2022-11-26 19:12:56 -05:00
fn parse_atom(&mut self, first: &lex::Token) -> Option<Event> {
2022-11-27 18:10:28 -05:00
let atom = match first.kind {
lex::Kind::Escape => Escape,
lex::Kind::Nbsp => Nbsp,
lex::Kind::Sym(lex::Symbol::Lt) => Lt,
lex::Kind::Sym(lex::Symbol::Gt) => Gt,
lex::Kind::Sym(lex::Symbol::Quote2) => Quote,
_ => return None,
};
Some(Event {
kind: EventKind::Atom(atom),
span: self.span,
})
2022-11-26 19:12:56 -05:00
}
2022-11-20 13:13:48 -05:00
2022-11-26 19:12:56 -05:00
fn parse_verbatim(&mut self, first: &lex::Token) -> Option<Event> {
match first.kind {
lex::Kind::Seq(lex::Sequence::Dollar) => {
let math_opt = (first.len <= 2)
.then(|| {
if let Some(lex::Token {
kind: lex::Kind::Seq(lex::Sequence::Backtick),
len,
}) = self.peek()
{
2022-11-27 15:59:54 -05:00
Some((
if first.len == 2 {
DisplayMath
} else {
InlineMath
},
*len,
))
2022-11-26 19:12:56 -05:00
} else {
None
2022-11-21 13:44:59 -05:00
}
2022-11-26 19:12:56 -05:00
})
.flatten();
if math_opt.is_some() {
self.eat(); // backticks
2022-11-21 13:44:59 -05:00
}
2022-11-26 19:12:56 -05:00
math_opt
2022-11-21 13:44:59 -05:00
}
2022-11-26 19:12:56 -05:00
lex::Kind::Seq(lex::Sequence::Backtick) => Some((Verbatim, first.len)),
_ => None,
2022-11-21 16:40:11 -05:00
}
2022-11-26 19:12:56 -05:00
.map(|(kind, opener_len)| {
let mut span = Span::empty_at(self.span.end());
while let Some(tok) = self.eat() {
if matches!(tok.kind, lex::Kind::Seq(lex::Sequence::Backtick))
&& tok.len == opener_len
{
break;
}
span = span.extend(tok.len);
2022-11-21 13:44:59 -05:00
}
2022-11-27 18:10:28 -05:00
Event {
kind: EventKind::Node(kind),
span,
}
2022-11-26 19:12:56 -05:00
})
}
2022-11-21 13:44:59 -05:00
2022-11-26 19:12:56 -05:00
fn parse_container(&mut self, first: &lex::Token) -> Option<Event> {
match first.kind {
lex::Kind::Sym(Symbol::Asterisk) => Some((Strong, Dir::Both)),
lex::Kind::Sym(Symbol::Underscore) => Some((Emphasis, Dir::Both)),
lex::Kind::Sym(Symbol::Caret) => Some((Superscript, Dir::Both)),
lex::Kind::Sym(Symbol::Tilde) => Some((Subscript, Dir::Both)),
lex::Kind::Sym(Symbol::Quote1) => Some((SingleQuoted, Dir::Both)),
lex::Kind::Sym(Symbol::Quote2) => Some((DoubleQuoted, Dir::Both)),
2022-11-27 15:59:54 -05:00
lex::Kind::Open(Delimiter::Bracket) => Some((Span, Dir::Open)),
lex::Kind::Close(Delimiter::Bracket) => Some((Span, Dir::Close)),
2022-11-26 19:12:56 -05:00
lex::Kind::Open(Delimiter::BraceAsterisk) => Some((Strong, Dir::Open)),
lex::Kind::Close(Delimiter::BraceAsterisk) => Some((Strong, Dir::Close)),
lex::Kind::Open(Delimiter::BraceCaret) => Some((Superscript, Dir::Open)),
lex::Kind::Close(Delimiter::BraceCaret) => Some((Superscript, Dir::Close)),
2022-11-27 17:56:19 -05:00
lex::Kind::Open(Delimiter::BraceEqual) => Some((Mark, Dir::Open)),
lex::Kind::Close(Delimiter::BraceEqual) => Some((Mark, Dir::Close)),
2022-11-26 19:12:56 -05:00
lex::Kind::Open(Delimiter::BraceHyphen) => Some((Delete, Dir::Open)),
lex::Kind::Close(Delimiter::BraceHyphen) => Some((Delete, Dir::Close)),
lex::Kind::Open(Delimiter::BracePlus) => Some((Insert, Dir::Open)),
lex::Kind::Close(Delimiter::BracePlus) => Some((Insert, Dir::Close)),
lex::Kind::Open(Delimiter::BraceTilde) => Some((Subscript, Dir::Open)),
lex::Kind::Close(Delimiter::BraceTilde) => Some((Subscript, Dir::Close)),
lex::Kind::Open(Delimiter::BraceUnderscore) => Some((Emphasis, Dir::Open)),
lex::Kind::Close(Delimiter::BraceUnderscore) => Some((Emphasis, Dir::Close)),
_ => None,
2022-11-20 13:13:48 -05:00
}
2022-11-28 13:08:49 -05:00
.map(|(cont, dir)| {
2022-11-27 17:56:19 -05:00
self.openers
.iter()
2022-11-28 13:08:49 -05:00
.rposition(|(c, _)| *c == cont)
2022-11-27 17:56:19 -05:00
.and_then(|o| {
matches!(dir, Dir::Close | Dir::Both).then(|| {
2022-11-27 17:59:01 -05:00
let (_, e) = &mut self.openers[o];
2022-11-28 13:08:49 -05:00
self.events[*e].kind = EventKind::Enter(cont);
self.openers.drain(o..);
EventKind::Exit(cont)
2022-11-27 17:56:19 -05:00
})
})
.unwrap_or_else(|| {
2022-11-28 13:08:49 -05:00
self.openers.push((cont, self.events.len()));
// use str for now, replace if closed later
EventKind::Node(Str)
2022-11-27 17:56:19 -05:00
})
2022-11-26 19:12:56 -05:00
})
2022-11-27 18:10:28 -05:00
.map(|kind| Event {
kind,
span: self.span,
})
2022-11-26 19:12:56 -05:00
}
}
impl<'s> Iterator for Parser<'s> {
type Item = Event;
fn next(&mut self) -> Option<Self::Item> {
2022-11-27 18:34:30 -05:00
while self.events.is_empty()
|| !self.openers.is_empty()
2022-11-28 13:08:49 -05:00
|| self
.events
.back()
.map_or(false, |ev| matches!(ev.kind, EventKind::Node(Str)))
2022-11-27 18:34:30 -05:00
{
2022-11-27 17:56:19 -05:00
if let Some(ev) = self.parse_event() {
self.events.push_back(ev);
} else {
break;
}
}
2022-11-26 19:12:56 -05:00
2022-11-27 18:34:30 -05:00
self.events.pop_front().map(|e| {
2022-11-28 13:08:49 -05:00
if matches!(e.kind, EventKind::Node(Str)) {
2022-11-27 18:34:30 -05:00
// merge str events
let mut span = e.span;
2022-11-28 13:08:49 -05:00
while self
.events
.front()
.map_or(false, |ev| matches!(ev.kind, EventKind::Node(Str)))
{
2022-11-27 18:34:30 -05:00
span = span.union(self.events.pop_front().unwrap().span);
}
Event {
kind: EventKind::Node(Str),
span,
}
} else {
e
}
})
}
}
2022-11-21 16:40:11 -05:00
#[cfg(test)]
mod test {
2022-11-26 19:12:56 -05:00
use crate::Span;
2022-11-21 16:40:11 -05:00
use super::Atom::*;
2022-11-22 13:19:21 -05:00
use super::Container::*;
2022-11-27 18:10:28 -05:00
use super::EventKind::*;
use super::Node::*;
2022-11-21 16:40:11 -05:00
2022-11-22 13:48:17 -05:00
macro_rules! test_parse {
($($st:ident,)? $src:expr $(,$($token:expr),* $(,)?)?) => {
#[allow(unused)]
let mut p = super::Parser::new();
p.parse($src);
2022-11-28 14:12:49 -05:00
let actual = p.map(|ev| (ev.kind, ev.span.of($src))).collect::<Vec<_>>();
2022-11-22 13:48:17 -05:00
let expected = &[$($($token),*,)?];
assert_eq!(actual, expected, "\n\n{}\n\n", $src);
};
}
2022-11-27 18:10:28 -05:00
impl super::EventKind {
pub fn span(self, start: usize, end: usize) -> super::Event {
super::Event {
2022-11-26 19:12:56 -05:00
span: Span::new(start, end),
kind: self,
}
}
}
2022-11-22 13:48:17 -05:00
#[test]
fn str() {
2022-11-28 14:12:49 -05:00
test_parse!("abc", (Node(Str), "abc"));
test_parse!("abc def", (Node(Str), "abc def"));
2022-11-26 19:12:56 -05:00
}
#[test]
fn verbatim() {
2022-11-28 14:12:49 -05:00
test_parse!("`abc`", (Node(Verbatim), "abc"));
test_parse!("`abc", (Node(Verbatim), "abc"));
test_parse!("``abc``", (Node(Verbatim), "abc"));
test_parse!("abc `def`", (Node(Str), "abc "), (Node(Verbatim), "def"));
2022-11-26 19:12:56 -05:00
}
2022-11-27 15:59:54 -05:00
#[test]
fn math() {
2022-11-28 14:12:49 -05:00
test_parse!("$`abc`", (Node(InlineMath), "abc"));
test_parse!("$`abc` str", (Node(InlineMath), "abc"), (Node(Str), " str"));
test_parse!("$$`abc`", (Node(DisplayMath), "abc"));
test_parse!("$`abc", (Node(InlineMath), "abc"));
test_parse!("$```abc```", (Node(InlineMath), "abc"),);
2022-11-27 15:59:54 -05:00
}
2022-11-26 19:12:56 -05:00
#[test]
fn container_basic() {
test_parse!(
"_abc_",
2022-11-28 14:12:49 -05:00
(Enter(Emphasis), "_"),
(Node(Str), "abc"),
(Exit(Emphasis), "_"),
2022-11-26 19:12:56 -05:00
);
test_parse!(
"{_abc_}",
2022-11-28 14:12:49 -05:00
(Enter(Emphasis), "{_"),
(Node(Str), "abc"),
(Exit(Emphasis), "_}"),
2022-11-26 19:12:56 -05:00
);
}
#[test]
fn container_nest() {
test_parse!(
"{_{_abc_}_}",
2022-11-28 14:12:49 -05:00
(Enter(Emphasis), "{_"),
(Enter(Emphasis), "{_"),
(Node(Str), "abc"),
(Exit(Emphasis), "_}"),
(Exit(Emphasis), "_}"),
2022-11-26 19:12:56 -05:00
);
test_parse!(
"*_abc_*",
2022-11-28 14:12:49 -05:00
(Enter(Strong), "*"),
(Enter(Emphasis), "_"),
(Node(Str), "abc"),
(Exit(Emphasis), "_"),
(Exit(Strong), "*"),
2022-11-26 19:12:56 -05:00
);
}
#[test]
fn container_unopened() {
2022-11-28 14:12:49 -05:00
test_parse!("*}abc", (Node(Str), "*}abc"));
2022-11-26 19:12:56 -05:00
}
#[test]
fn container_close_parent() {
test_parse!(
"{*{_abc*}",
2022-11-28 14:12:49 -05:00
(Enter(Strong), "{*"),
(Node(Str), "{_abc"),
(Exit(Strong), "*}"),
2022-11-26 19:12:56 -05:00
);
2022-11-22 13:48:17 -05:00
}
2022-11-21 16:40:11 -05:00
#[test]
2022-11-26 19:12:56 -05:00
fn container_close_block() {
2022-11-28 14:12:49 -05:00
test_parse!("{_abc", (Node(Str), "{_abc"));
test_parse!("{_{*{_abc", (Node(Str), "{_{*{_abc"));
2022-11-20 13:13:48 -05:00
}
}