diff options
| author | eichlan <eichlan@jotunn.office.penny-arcade.com> | 2026-06-23 15:31:51 -0700 |
|---|---|---|
| committer | eichlan <eichlan@jotunn.office.penny-arcade.com> | 2026-06-23 15:31:51 -0700 |
| commit | 062048ef6e3e060bcf841ea2ec92e026f09d8da0 (patch) | |
| tree | 22f7dad062b16f25fcfa1a28af13ac2937348ef0 /crimtag/src/lexer.rs | |
| parent | 7c299c86bbac76ec2ac199d68eb4dda093a64e3a (diff) | |
| download | crimtag-062048ef6e3e060bcf841ea2ec92e026f09d8da0.tar.gz crimtag-062048ef6e3e060bcf841ea2ec92e026f09d8da0.tar.bz2 crimtag-062048ef6e3e060bcf841ea2ec92e026f09d8da0.tar.xz crimtag-062048ef6e3e060bcf841ea2ec92e026f09d8da0.zip | |
Roorganized into workspace and packages.
This is to accomidate a new proc_macro package, which can't have
anything else in it.
Diffstat (limited to 'crimtag/src/lexer.rs')
| -rw-r--r-- | crimtag/src/lexer.rs | 368 |
1 files changed, 368 insertions, 0 deletions
diff --git a/crimtag/src/lexer.rs b/crimtag/src/lexer.rs new file mode 100644 index 0000000..d4d39e9 --- /dev/null +++ b/crimtag/src/lexer.rs | |||
| @@ -0,0 +1,368 @@ | |||
| 1 | use std::iter::Iterator; | ||
| 2 | //use core::error::Error; | ||
| 3 | use std::str::CharIndices; | ||
| 4 | use std::fmt; | ||
| 5 | |||
| 6 | use crate::Position; | ||
| 7 | |||
| 8 | #[derive(Debug,Copy,Clone,PartialEq,Eq)] | ||
| 9 | pub enum ErrorType { | ||
| 10 | UnexpectedChar(char), | ||
| 11 | } | ||
| 12 | |||
| 13 | #[derive(Debug,Copy,Clone,PartialEq,Eq)] | ||
| 14 | pub enum SymbolType<'a> { | ||
| 15 | StartFlat, | ||
| 16 | StartPoint, | ||
| 17 | EndFlat, | ||
| 18 | EndPoint, | ||
| 19 | |||
| 20 | View, | ||
| 21 | Output, | ||
| 22 | Show, | ||
| 23 | Loop, | ||
| 24 | If, | ||
| 25 | ElIf, | ||
| 26 | Else, | ||
| 27 | |||
| 28 | Sharp, | ||
| 29 | Equals, | ||
| 30 | Period, | ||
| 31 | Token(&'a str), | ||
| 32 | Literal(&'a str), | ||
| 33 | Text(&'a str), | ||
| 34 | Error{ what: ErrorType }, | ||
| 35 | } | ||
| 36 | |||
| 37 | #[derive(Copy,Clone)] | ||
| 38 | pub struct Symbol<'a> { | ||
| 39 | symbol: SymbolType<'a>, | ||
| 40 | start: Position, | ||
| 41 | end: Position, | ||
| 42 | } | ||
| 43 | |||
| 44 | impl<'a> fmt::Debug for Symbol<'a> { | ||
| 45 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { | ||
| 46 | write!(f, "{:?} @ {:?}-{:?}", self.symbol, self.start, self.end ) | ||
| 47 | } | ||
| 48 | } | ||
| 49 | |||
| 50 | impl<'a> Symbol<'a> { | ||
| 51 | pub fn new( symbol: SymbolType<'a>, start: Position, end: Position ) -> Self { | ||
| 52 | Self { | ||
| 53 | symbol, start, end, | ||
| 54 | } | ||
| 55 | } | ||
| 56 | |||
| 57 | pub fn check_type<T: Fn( &SymbolType ) -> bool>(&self, f: T ) -> bool { | ||
| 58 | f( &self.symbol ) | ||
| 59 | } | ||
| 60 | |||
| 61 | pub fn symbol(&self) -> &SymbolType<'a> { | ||
| 62 | &self.symbol | ||
| 63 | } | ||
| 64 | |||
| 65 | pub fn start(&self) -> &Position { | ||
| 66 | &self.start | ||
| 67 | } | ||
| 68 | |||
| 69 | /* | ||
| 70 | pub fn end(&self) -> &Position { | ||
| 71 | &self.end | ||
| 72 | } | ||
| 73 | */ | ||
| 74 | } | ||
| 75 | |||
| 76 | #[derive(Debug)] | ||
| 77 | enum Mode { | ||
| 78 | Text, | ||
| 79 | InTag, | ||
| 80 | } | ||
| 81 | |||
| 82 | pub struct Lexer<'a> { | ||
| 83 | data: &'a str, | ||
| 84 | chars: CharIndices<'a>, | ||
| 85 | cur: [Option<(usize, char)>;2], | ||
| 86 | icur: usize, | ||
| 87 | mode: Mode, | ||
| 88 | pos: Position, | ||
| 89 | } | ||
| 90 | |||
| 91 | fn is_valid_token_char( c: char, first: bool ) -> bool { | ||
| 92 | if c.is_whitespace() { | ||
| 93 | return false; | ||
| 94 | } | ||
| 95 | if first { | ||
| 96 | match c { | ||
| 97 | 'a'..'z' | 'A'..'Z' | '_' => true, | ||
| 98 | _ => false | ||
| 99 | } | ||
| 100 | } else { | ||
| 101 | match c { | ||
| 102 | 'a'..'z' | 'A'..'Z' | '0'..'9' | '_' => true, | ||
| 103 | _ => false | ||
| 104 | } | ||
| 105 | } | ||
| 106 | } | ||
| 107 | |||
| 108 | impl<'a> Lexer<'a> { | ||
| 109 | pub fn new( data: &'a str ) -> Lexer<'a> { | ||
| 110 | let mut chars = data.char_indices(); | ||
| 111 | let cur = chars.next(); | ||
| 112 | let cur2 = chars.next(); | ||
| 113 | //println!(" - cur: {:?}, peek: {:?}", cur, cur2 ); | ||
| 114 | Lexer { | ||
| 115 | data, | ||
| 116 | chars, | ||
| 117 | cur: [cur, cur2], | ||
| 118 | icur: 0, | ||
| 119 | mode: Mode::Text, | ||
| 120 | pos: Position::new(1,1), | ||
| 121 | } | ||
| 122 | } | ||
| 123 | |||
| 124 | fn next(&mut self) -> Option<char> { | ||
| 125 | self.cur[self.icur] = self.chars.next(); | ||
| 126 | self.icur = (self.icur+1)%2; | ||
| 127 | //println!(" - cur: {:?}, peek: {:?}", self.cur(), self.peek() ); | ||
| 128 | if let Some((_,chr)) = self.cur[self.icur] { | ||
| 129 | if chr == '\n' { | ||
| 130 | self.pos.column = 1; | ||
| 131 | self.pos.line += 1; | ||
| 132 | } else { | ||
| 133 | self.pos.column += 1; | ||
| 134 | } | ||
| 135 | Some(chr) | ||
| 136 | } else { | ||
| 137 | None | ||
| 138 | } | ||
| 139 | } | ||
| 140 | |||
| 141 | fn cur(&self) -> Option<char> { | ||
| 142 | if let Some((_, chr)) = self.cur[self.icur] { | ||
| 143 | Some(chr) | ||
| 144 | } else { | ||
| 145 | None | ||
| 146 | } | ||
| 147 | } | ||
| 148 | |||
| 149 | fn cur_index(&self) -> usize { | ||
| 150 | if let Some((idx, _)) = self.cur[self.icur] { | ||
| 151 | idx | ||
| 152 | } else { | ||
| 153 | self.chars.offset() | ||
| 154 | } | ||
| 155 | } | ||
| 156 | |||
| 157 | fn peek(&self) -> Option<char> { | ||
| 158 | if let Some((_,chr)) = self.cur[(self.icur+1)%2] { | ||
| 159 | Some(chr) | ||
| 160 | } else { | ||
| 161 | None | ||
| 162 | } | ||
| 163 | } | ||
| 164 | |||
| 165 | fn peek_index(&self) -> usize { | ||
| 166 | if let Some((idx, _)) = self.cur[(self.icur+1)%2] { | ||
| 167 | idx | ||
| 168 | } else { | ||
| 169 | self.chars.offset() | ||
| 170 | } | ||
| 171 | } | ||
| 172 | |||
| 173 | fn error( &self, what: ErrorType ) -> Option<Symbol<'a>> { | ||
| 174 | Some(Symbol::new( SymbolType::Error { | ||
| 175 | what: what | ||
| 176 | }, self.pos, self.pos )) | ||
| 177 | } | ||
| 178 | |||
| 179 | fn skip_ws( &mut self ) { | ||
| 180 | while self.cur().is_some_and(|x|x.is_whitespace()) { | ||
| 181 | self.next(); | ||
| 182 | } | ||
| 183 | } | ||
| 184 | |||
| 185 | fn next_symbol(&mut self) -> Option<Symbol<'a>> { | ||
| 186 | // If we hit the end then we're already done. | ||
| 187 | if self.cur().is_none() { | ||
| 188 | return None; | ||
| 189 | } | ||
| 190 | |||
| 191 | match self.mode { | ||
| 192 | Mode::Text => { | ||
| 193 | if let Some(sym) = self.parse_start_tag() { | ||
| 194 | Some(sym) | ||
| 195 | } else { | ||
| 196 | self.parse_text() | ||
| 197 | } | ||
| 198 | } | ||
| 199 | Mode::InTag => { | ||
| 200 | if let Some(sym) = self.parse_end_tag() { | ||
| 201 | Some(sym) | ||
| 202 | } else { | ||
| 203 | self.parse_token() | ||
| 204 | } | ||
| 205 | } | ||
| 206 | } | ||
| 207 | } | ||
| 208 | |||
| 209 | fn parse_text(&mut self) -> Option<Symbol<'a>> { | ||
| 210 | let start = self.cur_index(); | ||
| 211 | let start_pos = self.pos.clone(); | ||
| 212 | while self.next().is_some() && !self.is_start_tag() { } | ||
| 213 | let end = self.cur_index(); | ||
| 214 | let end_pos = self.pos.clone(); | ||
| 215 | let s = &self.data[start..end]; | ||
| 216 | //println!(" text: >>>{}<<<", s); | ||
| 217 | if start == end { | ||
| 218 | None | ||
| 219 | } else { | ||
| 220 | Some(Symbol::new( SymbolType::Text(s), start_pos, end_pos )) | ||
| 221 | } | ||
| 222 | } | ||
| 223 | |||
| 224 | fn parse_token(&mut self) -> Option<Symbol<'a>> { | ||
| 225 | self.skip_ws(); | ||
| 226 | match self.cur() { | ||
| 227 | Some('"') => { | ||
| 228 | return self.parse_literal_str(); | ||
| 229 | } | ||
| 230 | Some('#') => { | ||
| 231 | self.next(); | ||
| 232 | return Some(Symbol::new(SymbolType::Sharp, self.pos, self.pos)); | ||
| 233 | } | ||
| 234 | Some('=') => { | ||
| 235 | self.next(); | ||
| 236 | return Some(Symbol::new(SymbolType::Equals, self.pos, self.pos)); | ||
| 237 | } | ||
| 238 | Some('.') => { | ||
| 239 | self.next(); | ||
| 240 | return Some(Symbol::new(SymbolType::Period, self.pos, self.pos)); | ||
| 241 | } | ||
| 242 | _ => {} | ||
| 243 | } | ||
| 244 | let start = self.cur_index(); | ||
| 245 | let start_pos = self.pos.clone(); | ||
| 246 | |||
| 247 | let mut first = true; | ||
| 248 | while self.next().is_some_and(|ch| is_valid_token_char(ch, first) ) && | ||
| 249 | !self.is_end_tag() { first = false; } | ||
| 250 | let end = self.cur_index(); | ||
| 251 | let end_pos = self.pos.clone(); | ||
| 252 | let s = &self.data[start..end]; | ||
| 253 | if start == end { | ||
| 254 | None | ||
| 255 | } else { | ||
| 256 | Some(Symbol::new( match s { | ||
| 257 | "view" => SymbolType::View, | ||
| 258 | "show" => SymbolType::Show, | ||
| 259 | "output" => SymbolType::Output, | ||
| 260 | "loop" => SymbolType::Loop, | ||
| 261 | "if" => SymbolType::If, | ||
| 262 | "elif" => SymbolType::ElIf, | ||
| 263 | "else" => SymbolType::Else, | ||
| 264 | _ => SymbolType::Token(s) | ||
| 265 | }, start_pos, end_pos )) | ||
| 266 | } | ||
| 267 | } | ||
| 268 | |||
| 269 | fn parse_literal_str(&mut self) -> Option<Symbol<'a>> { | ||
| 270 | if let Some(chr) = self.cur() && chr != '"' { | ||
| 271 | return self.error( ErrorType::UnexpectedChar(chr) ); | ||
| 272 | } | ||
| 273 | let start = self.peek_index(); | ||
| 274 | let start_pos = self.pos.clone(); | ||
| 275 | while self.next().is_some_and(|chr| chr != '"') { } | ||
| 276 | let end = self.cur_index(); | ||
| 277 | let end_pos = self.pos.clone(); | ||
| 278 | self.next(); | ||
| 279 | let s = &self.data[start..end]; | ||
| 280 | Some(Symbol::new(SymbolType::Literal(s), start_pos, end_pos )) | ||
| 281 | } | ||
| 282 | |||
| 283 | fn is_start_tag(&mut self) -> bool { | ||
| 284 | if let Some(cur) = self.cur() && (cur == '[' || cur == '<') && | ||
| 285 | let Some(peek) = self.peek() && peek == '|' { | ||
| 286 | true | ||
| 287 | } else { | ||
| 288 | false | ||
| 289 | } | ||
| 290 | } | ||
| 291 | |||
| 292 | fn parse_start_tag(&mut self) -> Option<Symbol<'a>> { | ||
| 293 | let start_pos = self.pos.clone(); | ||
| 294 | match self.cur() { | ||
| 295 | Some('[') => { | ||
| 296 | if let Some(p) = self.peek() && p == '|' { | ||
| 297 | self.next(); | ||
| 298 | let end_pos = self.pos.clone(); | ||
| 299 | self.next(); | ||
| 300 | self.mode = Mode::InTag; | ||
| 301 | Some(Symbol::new( SymbolType::StartFlat, start_pos, end_pos ) ) | ||
| 302 | } else { | ||
| 303 | None | ||
| 304 | } | ||
| 305 | } | ||
| 306 | Some('<') => { | ||
| 307 | if let Some(p) = self.peek() && p == '|' { | ||
| 308 | self.next(); | ||
| 309 | let end_pos = self.pos.clone(); | ||
| 310 | self.next(); | ||
| 311 | self.mode = Mode::InTag; | ||
| 312 | Some(Symbol::new(SymbolType::StartPoint, start_pos, end_pos ) ) | ||
| 313 | } else { | ||
| 314 | None | ||
| 315 | } | ||
| 316 | } | ||
| 317 | _ => { | ||
| 318 | None | ||
| 319 | } | ||
| 320 | } | ||
| 321 | } | ||
| 322 | |||
| 323 | fn is_end_tag(&mut self) -> bool { | ||
| 324 | if let Some(cur) = self.cur() && cur == '|' && | ||
| 325 | let Some(peek) = self.peek() && (peek == '>' || peek == ']') { | ||
| 326 | true | ||
| 327 | } else { | ||
| 328 | false | ||
| 329 | } | ||
| 330 | } | ||
| 331 | |||
| 332 | fn parse_end_tag(&mut self) -> Option<Symbol<'a>> { | ||
| 333 | self.skip_ws(); | ||
| 334 | let start_pos = self.pos.clone(); | ||
| 335 | if let Some(chr) = self.cur() && chr == '|' { | ||
| 336 | match self.peek() { | ||
| 337 | Some(']') => { | ||
| 338 | self.next(); | ||
| 339 | let end_pos = self.pos.clone(); | ||
| 340 | self.next(); | ||
| 341 | self.mode = Mode::Text; | ||
| 342 | Some(Symbol::new( SymbolType::EndFlat, start_pos, end_pos)) | ||
| 343 | } | ||
| 344 | Some('>') => { | ||
| 345 | self.next(); | ||
| 346 | let end_pos = self.pos.clone(); | ||
| 347 | self.next(); | ||
| 348 | self.mode = Mode::Text; | ||
| 349 | Some(Symbol::new( SymbolType::EndPoint, start_pos, end_pos)) | ||
| 350 | } | ||
| 351 | _ => { | ||
| 352 | None | ||
| 353 | } | ||
| 354 | } | ||
| 355 | } else { | ||
| 356 | None | ||
| 357 | } | ||
| 358 | } | ||
| 359 | } | ||
| 360 | |||
| 361 | impl<'a> Iterator for Lexer<'a> { | ||
| 362 | type Item = Symbol<'a>; | ||
| 363 | |||
| 364 | fn next(&mut self) -> Option<Self::Item> { | ||
| 365 | self.next_symbol() | ||
| 366 | } | ||
| 367 | } | ||
| 368 | |||
