Skip to main content

abyss_core/parser/
tokens.rs

1use chumsky::{error::Rich, extra, prelude::*, span::SimpleSpan as ChumskySpan};
2use ordered_float::OrderedFloat;
3use std::fmt;
4
5use crate::ast::Type;
6
7use super::SimpleSpan;
8
9#[derive(Debug, Clone, PartialEq, Eq, Hash)]
10pub enum Token {
11    Forge,
12    Morph,
13    Core,
14    Oracle,
15    Orbit,
16    Resume,
17    Eject,
18    Engrave,
19    Reveal,
20    Artifact,
21    Identifier(String),
22    Type(Type),
23    OmenLiteral(bool),
24    Arcana(i64),
25    Aether(OrderedFloat<f64>),
26    Rune(String),
27    Semicolon,
28    Colon,
29    Comma,
30    Arrow,
31    FatArrow,
32    DoubleColon,
33    Assign,
34    AddAssign,
35    SubAssign,
36    MulAssign,
37    DivAssign,
38    ModAssign,
39    PowArcanaAssign,
40    PowAetherAssign,
41    Equal,
42    NotEqual,
43    LessThan,
44    LessThanOrEqual,
45    GreaterThan,
46    GreaterThanOrEqual,
47    Plus,
48    Minus,
49    Star,
50    Slash,
51    Percent,
52    Caret,
53    DoubleStar,
54    DoublePipe,
55    DoubleAmpersand,
56    Bang,
57    OpenParen,
58    CloseParen,
59    OpenBrace,
60    CloseBrace,
61    OpenBracket,
62    CloseBracket,
63    RangeInclusive,
64    RangeExclusive,
65    Dot,
66}
67
68pub type SpannedToken = (Token, SimpleSpan<usize>);
69
70impl fmt::Display for Token {
71    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
72        match self {
73            Token::Forge => write!(f, "forge"),
74            Token::Morph => write!(f, "morph"),
75            Token::Core => write!(f, "core"),
76            Token::Oracle => write!(f, "oracle"),
77            Token::Orbit => write!(f, "orbit"),
78            Token::Resume => write!(f, "resume"),
79            Token::Eject => write!(f, "eject"),
80            Token::Engrave => write!(f, "engrave"),
81            Token::Reveal => write!(f, "reveal"),
82            Token::Artifact => write!(f, "artifact"),
83            Token::Identifier(name) => write!(f, "identifier `{name}`"),
84            Token::Type(ty) => write!(f, "type `{ty:?}`"),
85            Token::OmenLiteral(true) => write!(f, "boon"),
86            Token::OmenLiteral(false) => write!(f, "hex"),
87            Token::Arcana(value) => write!(f, "arcana literal {value}"),
88            Token::Aether(value) => write!(f, "aether literal {value}"),
89            Token::Rune(value) => write!(f, "rune literal \"{value}\""),
90            Token::Semicolon => write!(f, ";"),
91            Token::Colon => write!(f, ":"),
92            Token::Comma => write!(f, ","),
93            Token::Arrow => write!(f, "->"),
94            Token::FatArrow => write!(f, "=>"),
95            Token::DoubleColon => write!(f, "::"),
96            Token::Assign => write!(f, "="),
97            Token::AddAssign => write!(f, "+="),
98            Token::SubAssign => write!(f, "-="),
99            Token::MulAssign => write!(f, "*="),
100            Token::DivAssign => write!(f, "/="),
101            Token::ModAssign => write!(f, "%="),
102            Token::PowArcanaAssign => write!(f, "^="),
103            Token::PowAetherAssign => write!(f, "**="),
104            Token::Equal => write!(f, "=="),
105            Token::NotEqual => write!(f, "!="),
106            Token::LessThan => write!(f, "<"),
107            Token::LessThanOrEqual => write!(f, "<="),
108            Token::GreaterThan => write!(f, ">"),
109            Token::GreaterThanOrEqual => write!(f, ">="),
110            Token::Plus => write!(f, "+"),
111            Token::Minus => write!(f, "-"),
112            Token::Star => write!(f, "*"),
113            Token::Slash => write!(f, "/"),
114            Token::Percent => write!(f, "%"),
115            Token::Caret => write!(f, "^"),
116            Token::DoubleStar => write!(f, "**"),
117            Token::DoublePipe => write!(f, "||"),
118            Token::DoubleAmpersand => write!(f, "&&"),
119            Token::Bang => write!(f, "!"),
120            Token::OpenParen => write!(f, "("),
121            Token::CloseParen => write!(f, ")"),
122            Token::OpenBrace => write!(f, "{{"),
123            Token::CloseBrace => write!(f, "}}"),
124            Token::OpenBracket => write!(f, "["),
125            Token::CloseBracket => write!(f, "]"),
126            Token::RangeInclusive => write!(f, "..="),
127            Token::RangeExclusive => write!(f, ".."),
128            Token::Dot => write!(f, "."),
129        }
130    }
131}
132
133type LexerExtra<'src> = extra::Err<Rich<'src, char, ChumskySpan<usize>>>;
134
135pub fn lexer<'src>() -> impl Parser<'src, &'src str, Vec<SpannedToken>, LexerExtra<'src>> {
136    use chumsky::text;
137
138    let sign = just::<&str, _, LexerExtra<'src>>("-")
139        .to(String::from("-"))
140        .or_not();
141
142    let digits = |radix| text::digits::<_, LexerExtra<'src>>(radix).collect::<String>();
143
144    let aether = sign
145        .clone()
146        .then(digits(10))
147        .then_ignore(just::<&str, _, LexerExtra<'src>>("."))
148        .then(digits(10))
149        .map(|((sign, int_part), frac_part)| {
150            let mut number = String::new();
151            if let Some(sign) = sign {
152                number.push_str(&sign);
153            }
154            number.push_str(&int_part);
155            number.push('.');
156            number.push_str(&frac_part);
157            let value = number
158                .parse::<f64>()
159                .expect("parser should only construct valid f64 literals");
160            Token::Aether(OrderedFloat(value))
161        });
162
163    let arcana = sign.then(digits(10)).map(|(sign, value)| {
164        let mut number = String::new();
165        if let Some(sign) = sign {
166            number.push_str(&sign);
167        }
168        number.push_str(&value);
169        Token::Arcana(
170            number
171                .parse::<i64>()
172                .expect("parser should only construct valid i64 literals"),
173        )
174    });
175
176    let escape = just::<char, _, LexerExtra<'src>>('\\').ignore_then(
177        one_of::<_, _, LexerExtra<'src>>(r#""ntr\"#).map(|c| match c {
178            '"' => '"',
179            'n' => '\n',
180            't' => '\t',
181            'r' => '\r',
182            '\\' => '\\',
183            other => other, // fallback: just use the char as is
184        }),
185    );
186    let rune_char =
187        escape.or(any::<_, LexerExtra<'src>>().filter(|c: &char| *c != '"' && *c != '\\'));
188    let rune = just::<char, _, LexerExtra<'src>>('"')
189        .ignore_then(rune_char.repeated().collect::<String>())
190        .then_ignore(just::<char, _, LexerExtra<'src>>('"'))
191        .map(Token::Rune);
192
193    let ident = text::ident::<_, LexerExtra<'src>>().map(|ident: &'src str| match ident {
194        "forge" => Token::Forge,
195        "morph" => Token::Morph,
196        "core" => Token::Core,
197        "oracle" => Token::Oracle,
198        "orbit" => Token::Orbit,
199        "resume" => Token::Resume,
200        "eject" => Token::Eject,
201        "engrave" => Token::Engrave,
202        "reveal" => Token::Reveal,
203        "trans" => Token::Identifier("trans".to_string()),
204        "as" => Token::Identifier("as".to_string()),
205        "artifact" => Token::Artifact,
206        "arcana" => Token::Type(Type::Arcana),
207        "aether" => Token::Type(Type::Aether),
208        "rune" => Token::Type(Type::Rune),
209        "omen" => Token::Type(Type::Omen),
210        "abyss" => Token::Type(Type::Abyss),
211        "scroll" => Token::Type(Type::Scroll),
212        "lexicon" => Token::Type(Type::Lexicon),
213        "materia" => Token::Type(Type::Materia),
214        "glyph" => Token::Type(Type::Glyph),
215        "boon" => Token::OmenLiteral(true),
216        "hex" => Token::OmenLiteral(false),
217        _ => Token::Identifier(ident.to_string()),
218    });
219
220    let multi_char_symbols = choice((
221        just::<&str, _, LexerExtra<'src>>("**=").to(Token::PowAetherAssign),
222        just::<&str, _, LexerExtra<'src>>("**").to(Token::DoubleStar),
223        just::<&str, _, LexerExtra<'src>>("^=").to(Token::PowArcanaAssign),
224        just::<&str, _, LexerExtra<'src>>("+=").to(Token::AddAssign),
225        just::<&str, _, LexerExtra<'src>>("-=").to(Token::SubAssign),
226        just::<&str, _, LexerExtra<'src>>("*=").to(Token::MulAssign),
227        just::<&str, _, LexerExtra<'src>>("/=").to(Token::DivAssign),
228        just::<&str, _, LexerExtra<'src>>("%=").to(Token::ModAssign),
229        just::<&str, _, LexerExtra<'src>>("=>").to(Token::FatArrow),
230        just::<&str, _, LexerExtra<'src>>("::").to(Token::DoubleColon),
231        just::<&str, _, LexerExtra<'src>>("->").to(Token::Arrow),
232        just::<&str, _, LexerExtra<'src>>("||").to(Token::DoublePipe),
233        just::<&str, _, LexerExtra<'src>>("&&").to(Token::DoubleAmpersand),
234        just::<&str, _, LexerExtra<'src>>("==").to(Token::Equal),
235        just::<&str, _, LexerExtra<'src>>("!=").to(Token::NotEqual),
236        just::<&str, _, LexerExtra<'src>>("<=").to(Token::LessThanOrEqual),
237        just::<&str, _, LexerExtra<'src>>(">=").to(Token::GreaterThanOrEqual),
238        just::<&str, _, LexerExtra<'src>>("..=").to(Token::RangeInclusive),
239        just::<&str, _, LexerExtra<'src>>("..").to(Token::RangeExclusive),
240    ));
241
242    let single_char_symbols = choice((
243        just::<char, _, LexerExtra<'src>>('=').to(Token::Assign),
244        just::<char, _, LexerExtra<'src>>('+').to(Token::Plus),
245        just::<char, _, LexerExtra<'src>>('-').to(Token::Minus),
246        just::<char, _, LexerExtra<'src>>('*').to(Token::Star),
247        just::<char, _, LexerExtra<'src>>('/').to(Token::Slash),
248        just::<char, _, LexerExtra<'src>>('%').to(Token::Percent),
249        just::<char, _, LexerExtra<'src>>('^').to(Token::Caret),
250        just::<char, _, LexerExtra<'src>>('<').to(Token::LessThan),
251        just::<char, _, LexerExtra<'src>>('>').to(Token::GreaterThan),
252        just::<char, _, LexerExtra<'src>>('!').to(Token::Bang),
253        just::<char, _, LexerExtra<'src>>(';').to(Token::Semicolon),
254        just::<char, _, LexerExtra<'src>>(':').to(Token::Colon),
255        just::<char, _, LexerExtra<'src>>(',').to(Token::Comma),
256        just::<char, _, LexerExtra<'src>>('(').to(Token::OpenParen),
257        just::<char, _, LexerExtra<'src>>(')').to(Token::CloseParen),
258        just::<char, _, LexerExtra<'src>>('{').to(Token::OpenBrace),
259        just::<char, _, LexerExtra<'src>>('}').to(Token::CloseBrace),
260        just::<char, _, LexerExtra<'src>>('[').to(Token::OpenBracket),
261        just::<char, _, LexerExtra<'src>>(']').to(Token::CloseBracket),
262        just::<char, _, LexerExtra<'src>>('.').to(Token::Dot),
263    ));
264
265    let token = choice((
266        aether,
267        arcana,
268        rune,
269        ident,
270        multi_char_symbols,
271        single_char_symbols,
272    ))
273    .map_with(|tok, extra| {
274        let span: ChumskySpan<usize> = extra.span();
275        (tok, SimpleSpan::new(span.start(), span.end()))
276    });
277
278    token
279        .padded_by(crate::parser::helpers::abyss_whitespace())
280        .repeated()
281        .collect()
282        .then_ignore(end())
283}