jotdown/src/inline.rs

use crate::lex;
use crate::Span;

use lex::Delimiter;
use lex::Symbol;

use Atom::*;
use Container::*;

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Atom {
    Softbreak,
    Hardbreak,
    Escape,
    Nbsp,
    Ellipsis,
    EnDash,
    EmDash,
}

#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub enum Container {
    Span,
    // typesetting
    Subscript,
    Superscript,
    Insert,
    Delete,
    Emphasis,
    Strong,
    Mark,
    // smart quoting
    SingleQuoted,
    DoubleQuoted,
    // Verbatim
    Verbatim,
    /// Span is the format.
    RawFormat,
    InlineMath,
    DisplayMath,
    /// Span is the reference link tag.
    ReferenceLink,
    /// Span is the reference link tag.
    ReferenceImage,
    /// Span is the URL.
    InlineLink,
    /// Span is the URL.
    InlineImage,

    AutoLink,
}

#[derive(Debug, PartialEq, Eq)]
pub enum EventKind {
    Enter(Container),
    Exit(Container),
    Atom(Atom),
    Str,
    Attributes,
}

#[derive(Debug, PartialEq, Eq)]
pub struct Event {
    pub kind: EventKind,
    pub span: Span,
}

pub struct Parser<I> {
    /// Lexer, hosting upcoming source.
    lexer: lex::Lexer<I>,
    /// Span of current event.
    span: Span,
    /// Stack with kind and index of _potential_ openers for typesetting containers.
    typesets: Vec<(Container, usize)>,
    /// Stack with index of _potential_ span/link openers.
    spans: Vec<(usize, bool)>,
    //attributes: Vec<(Span, usize)>,
    /// Buffer queue for next events. Events are buffered until no modifications due to future
    /// characters are needed.
    events: std::collections::VecDeque<Event>,
}

impl<I: Iterator<Item = char> + Clone> Parser<I> {
    pub fn new(chars: I) -> Self {
        Self {
            lexer: lex::Lexer::new(chars),
            span: Span::new(0, 0),
            typesets: Vec::new(),
            spans: Vec::new(),
            events: std::collections::VecDeque::new(),
        }
    }

    fn eat(&mut self) -> Option<lex::Token> {
        let tok = self.lexer.next();
        if let Some(t) = &tok {
            self.span = self.span.extend(t.len);
        }
        tok
    }

    fn peek(&mut self) -> Option<&lex::Token> {
        self.lexer.peek()
    }

    fn reset_span(&mut self) {
        self.span = Span::empty_at(self.span.end());
    }

    fn parse_event(&mut self) -> Option<Event> {
        self.reset_span();
        self.eat().map(|first| {
            self.parse_verbatim(&first)
                .or_else(|| self.parse_span(&first))
                .or_else(|| self.parse_typeset(&first))
                .or_else(|| self.parse_atom(&first))
                .unwrap_or(Event {
                    kind: EventKind::Str,
                    span: self.span,
                })
        })
    }

    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()
                        {
                            Some((
                                if first.len == 2 {
                                    DisplayMath
                                } else {
                                    InlineMath
                                },
                                *len,
                            ))
                        } else {
                            None
                        }
                    })
                    .flatten();
                if math_opt.is_some() {
                    self.eat(); // backticks
                }
                math_opt
            }
            lex::Kind::Seq(lex::Sequence::Backtick) => Some((Verbatim, first.len)),
            _ => None,
        }
        .map(|(mut kind, opener_len)| {
            let opener_event = self.events.len();
            self.events.push_back(Event {
                kind: EventKind::Enter(kind),
                span: self.span,
            });

            let mut span_inner = Span::empty_at(self.span.end());
            let mut span_outer = None;

            while let Some(t) = self.eat() {
                if matches!(t.kind, lex::Kind::Seq(lex::Sequence::Backtick)) && t.len == opener_len
                {
                    if matches!(kind, Verbatim)
                        && matches!(
                            self.lexer.peek().map(|t| &t.kind),
                            Some(lex::Kind::Open(Delimiter::BraceEqual))
                        )
                    {
                        let mut ahead = self.lexer.inner().clone();
                        let mut end = false;
                        let len = (&mut ahead)
                            .take_while(|c| {
                                if *c == '{' {
                                    return false;
                                }
                                if *c == '}' {
                                    end = true;
                                };
                                !end && !c.is_whitespace()
                            })
                            .count();
                        if len > 0 && end {
                            self.lexer = lex::Lexer::new(ahead);
                            let span_format = Span::by_len(self.span.end() + "{=".len(), len);
                            kind = RawFormat;
                            self.events[opener_event].kind = EventKind::Enter(kind);
                            self.events[opener_event].span = span_format;
                            self.span = span_format.translate(1); // }
                            span_outer = Some(span_format);
                        }
                    }
                    break;
                }
                span_inner = span_inner.extend(t.len);
                self.reset_span();
            }

            self.events.push_back(Event {
                kind: EventKind::Str,
                span: span_inner,
            });

            Event {
                kind: EventKind::Exit(kind),
                span: span_outer.unwrap_or(self.span),
            }
        })
    }

    fn parse_span(&mut self, first: &lex::Token) -> Option<Event> {
        match first.kind {
            lex::Kind::Sym(Symbol::ExclaimBracket) => Some((true, true)),
            lex::Kind::Open(Delimiter::Bracket) => Some((true, false)),
            lex::Kind::Close(Delimiter::Bracket) => Some((false, false)),
            _ => None,
        }
        .and_then(|(open, img)| {
            if open {
                self.spans.push((self.events.len(), img));
                // use str for now, replace if closed later
                Some(Event {
                    kind: EventKind::Str,
                    span: self.span,
                })
            } else if !self.spans.is_empty() {
                let mut ahead = self.lexer.inner().clone();
                let img = self.spans.last().unwrap().1;
                match ahead.next() {
                    Some(opener @ ('[' | '(')) => {
                        let (closer, kind) = match opener {
                            '[' if img => (']', ReferenceImage),
                            '[' => (']', ReferenceLink),
                            '(' if img => (')', InlineImage),
                            '(' => (')', InlineLink),
                            _ => unreachable!(),
                        };
                        let mut end = false;
                        let len = (&mut ahead)
                            .take_while(|c| {
                                if *c == closer {
                                    end = true;
                                };
                                !end && *c != opener
                            })
                            .count();
                        end.then(|| {
                            let span = Span::by_len(self.span.end() + 1, len);
                            (kind, span)
                        })
                    }
                    Some('{') => todo!(),
                    _ => None,
                }
                .map(|(kind, span)| {
                    self.lexer = lex::Lexer::new(ahead);
                    let (opener_event, _) = self.spans.pop().unwrap();
                    self.events[opener_event].kind = EventKind::Enter(kind);
                    self.events[opener_event].span = span;
                    self.span = span.translate(1);
                    Event {
                        kind: EventKind::Exit(kind),
                        span,
                    }
                })
            } else {
                None
            }
        })
    }

    fn parse_typeset(&mut self, first: &lex::Token) -> Option<Event> {
        enum Dir {
            Open,
            Close,
            Both,
        }

        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)),
            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)),
            lex::Kind::Open(Delimiter::BraceEqual) => Some((Mark, Dir::Open)),
            lex::Kind::Close(Delimiter::BraceEqual) => Some((Mark, Dir::Close)),
            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,
        }
        .map(|(cont, dir)| {
            self.typesets
                .iter()
                .rposition(|(c, _)| *c == cont)
                .and_then(|o| {
                    matches!(dir, Dir::Close | Dir::Both).then(|| {
                        let (_, e) = &mut self.typesets[o];
                        self.events[*e].kind = EventKind::Enter(cont);
                        self.typesets.drain(o..);
                        EventKind::Exit(cont)
                    })
                })
                .unwrap_or_else(|| {
                    self.typesets.push((cont, self.events.len()));
                    // use str for now, replace if closed later
                    EventKind::Str
                })
        })
        .map(|kind| Event {
            kind,
            span: self.span,
        })
    }

    fn parse_atom(&mut self, first: &lex::Token) -> Option<Event> {
        let atom = match first.kind {
            lex::Kind::Newline => Softbreak,
            lex::Kind::Hardbreak => Hardbreak,
            lex::Kind::Escape => Escape,
            lex::Kind::Nbsp => Nbsp,
            lex::Kind::Seq(lex::Sequence::Period) if first.len == 3 => Ellipsis,
            lex::Kind::Seq(lex::Sequence::Hyphen) if first.len == 2 => EnDash,
            lex::Kind::Seq(lex::Sequence::Hyphen) if first.len == 3 => EmDash,
            _ => return None,
        };

        Some(Event {
            kind: EventKind::Atom(atom),
            span: self.span,
        })
    }
}

impl<I: Iterator<Item = char> + Clone> Iterator for Parser<I> {
    type Item = Event;

    fn next(&mut self) -> Option<Self::Item> {
        while self.events.is_empty()
            || !self.typesets.is_empty()
            || !self.spans.is_empty()
            || self // for merge
                .events
                .back()
                .map_or(false, |ev| matches!(ev.kind, EventKind::Str))
        {
            if let Some(ev) = self.parse_event() {
                self.events.push_back(ev);
            } else {
                break;
            }
        }

        self.events.pop_front().map(|e| {
            if matches!(e.kind, EventKind::Str) {
                // merge str events
                let mut span = e.span;
                while self
                    .events
                    .front()
                    .map_or(false, |ev| matches!(ev.kind, EventKind::Str))
                {
                    let ev = self.events.pop_front().unwrap();
                    assert_eq!(span.end(), ev.span.start());
                    span = span.union(ev.span);
                }
                Event {
                    kind: EventKind::Str,
                    span,
                }
            } else {
                e
            }
        })
    }
}

#[cfg(test)]
mod test {
    use crate::Span;

    use super::Atom::*;
    use super::Container::*;
    use super::EventKind::*;
    use super::Verbatim;

    macro_rules! test_parse {
        ($($st:ident,)? $src:expr $(,$($token:expr),* $(,)?)?) => {
            #[allow(unused)]
            let mut p = super::Parser::new($src.chars());
            let actual = p.map(|ev| (ev.kind, ev.span.of($src))).collect::<Vec<_>>();
            let expected = &[$($($token),*,)?];
            assert_eq!(actual, expected, "\n\n{}\n\n", $src);
        };
    }

    #[test]
    fn str() {
        test_parse!("abc", (Str, "abc"));
        test_parse!("abc def", (Str, "abc def"));
    }

    #[test]
    fn verbatim() {
        test_parse!(
            "`abc`",
            (Enter(Verbatim), "`"),
            (Str, "abc"),
            (Exit(Verbatim), "`"),
        );
        test_parse!(
            "`abc\ndef`",
            (Enter(Verbatim), "`"),
            (Str, "abc\ndef"),
            (Exit(Verbatim), "`"),
        );
        test_parse!(
            "`abc&def`",
            (Enter(Verbatim), "`"),
            (Str, "abc&def"),
            (Exit(Verbatim), "`"),
        );
        test_parse!(
            "`abc",
            (Enter(Verbatim), "`"),
            (Str, "abc"),
            (Exit(Verbatim), ""),
        );
        test_parse!(
            "``abc``",
            (Enter(Verbatim), "``"),
            (Str, "abc"),
            (Exit(Verbatim), "``"),
        );
        test_parse!(
            "abc `def`",
            (Str, "abc "),
            (Enter(Verbatim), "`"),
            (Str, "def"),
            (Exit(Verbatim), "`"),
        );
        test_parse!(
            "abc`def`",
            (Str, "abc"),
            (Enter(Verbatim), "`"),
            (Str, "def"),
            (Exit(Verbatim), "`"),
        );
    }

    #[test]
    fn math() {
        test_parse!(
            "$`abc`",
            (Enter(InlineMath), "$`"),
            (Str, "abc"),
            (Exit(InlineMath), "`"),
        );
        test_parse!(
            "$`abc` str",
            (Enter(InlineMath), "$`"),
            (Str, "abc"),
            (Exit(InlineMath), "`"),
            (Str, " str"),
        );
        test_parse!(
            "$$`abc`",
            (Enter(DisplayMath), "$$`"),
            (Str, "abc"),
            (Exit(DisplayMath), "`"),
        );
        test_parse!(
            "$`abc",
            (Enter(InlineMath), "$`"),
            (Str, "abc"),
            (Exit(InlineMath), ""),
        );
        test_parse!(
            "$```abc```",
            (Enter(InlineMath), "$```"),
            (Str, "abc"),
            (Exit(InlineMath), "```"),
        );
    }

    #[test]
    fn raw_format() {
        test_parse!(
            "`raw`{=format}",
            (Enter(RawFormat), "format"),
            (Str, "raw"),
            (Exit(RawFormat), "format"),
        );
        test_parse!(
            "before `raw`{=format} after",
            (Str, "before "),
            (Enter(RawFormat), "format"),
            (Str, "raw"),
            (Exit(RawFormat), "format"),
            (Str, " after"),
        );
    }

    #[test]
    fn raw_attr() {
        test_parse!(
            "`raw`{=format #id}",
            (Enter(Verbatim), "`"),
            (Str, "raw"),
            (Exit(Verbatim), "`"),
            (Str, "{=format #id}"),
        );
    }

    #[test]
    fn span_tag() {
        test_parse!(
            "[text][tag]",
            (Enter(ReferenceLink), "tag"),
            (Str, "text"),
            (Exit(ReferenceLink), "tag"),
        );
        test_parse!(
            "![text][tag]",
            (Enter(ReferenceImage), "tag"),
            (Str, "text"),
            (Exit(ReferenceImage), "tag"),
        );
        test_parse!(
            "before [text][tag] after",
            (Str, "before "),
            (Enter(ReferenceLink), "tag"),
            (Str, "text"),
            (Exit(ReferenceLink), "tag"),
            (Str, " after"),
        );
        test_parse!(
            "[[inner][i]][o]",
            (Enter(ReferenceLink), "o"),
            (Enter(ReferenceLink), "i"),
            (Str, "inner"),
            (Exit(ReferenceLink), "i"),
            (Exit(ReferenceLink), "o"),
        );
    }

    #[test]
    fn span_url() {
        test_parse!(
            "before [text](url) after",
            (Str, "before "),
            (Enter(InlineLink), "url"),
            (Str, "text"),
            (Exit(InlineLink), "url"),
            (Str, " after"),
        );
        test_parse!(
            "[outer [inner](i)](o)",
            (Enter(InlineLink), "o"),
            (Str, "outer "),
            (Enter(InlineLink), "i"),
            (Str, "inner"),
            (Exit(InlineLink), "i"),
            (Exit(InlineLink), "o"),
        );
    }

    #[test]
    fn typeset_basic() {
        test_parse!(
            "_abc_",
            (Enter(Emphasis), "_"),
            (Str, "abc"),
            (Exit(Emphasis), "_"),
        );
        test_parse!(
            "{_abc_}",
            (Enter(Emphasis), "{_"),
            (Str, "abc"),
            (Exit(Emphasis), "_}"),
        );
    }

    #[test]
    fn typeset_nest() {
        test_parse!(
            "{_{_abc_}_}",
            (Enter(Emphasis), "{_"),
            (Enter(Emphasis), "{_"),
            (Str, "abc"),
            (Exit(Emphasis), "_}"),
            (Exit(Emphasis), "_}"),
        );
        test_parse!(
            "*_abc_*",
            (Enter(Strong), "*"),
            (Enter(Emphasis), "_"),
            (Str, "abc"),
            (Exit(Emphasis), "_"),
            (Exit(Strong), "*"),
        );
    }

    #[test]
    fn typeset_unopened() {
        test_parse!("*}abc", (Str, "*}abc"));
    }

    #[test]
    fn typeset_close_parent() {
        test_parse!(
            "{*{_abc*}",
            (Enter(Strong), "{*"),
            (Str, "{_abc"),
            (Exit(Strong), "*}"),
        );
    }

    #[test]
    fn typeset_close_block() {
        test_parse!("{_abc", (Str, "{_abc"));
        test_parse!("{_{*{_abc", (Str, "{_{*{_abc"));
    }
}