| 1 | // Take a look at the license at the top of the repository in the LICENSE file. |
| 2 | |
| 3 | use std::convert::TryFrom; |
| 4 | use std::fmt; |
| 5 | use std::iter::Peekable; |
| 6 | use std::str::CharIndices; |
| 7 | |
| 8 | #[derive (Debug, PartialEq, Eq, Clone, Copy)] |
| 9 | pub enum ReservedChar { |
| 10 | Comma, |
| 11 | SuperiorThan, |
| 12 | OpenParenthese, |
| 13 | CloseParenthese, |
| 14 | OpenCurlyBrace, |
| 15 | CloseCurlyBrace, |
| 16 | OpenBracket, |
| 17 | CloseBracket, |
| 18 | Colon, |
| 19 | SemiColon, |
| 20 | Slash, |
| 21 | Plus, |
| 22 | EqualSign, |
| 23 | Space, |
| 24 | Tab, |
| 25 | Backline, |
| 26 | Star, |
| 27 | Quote, |
| 28 | DoubleQuote, |
| 29 | Pipe, |
| 30 | Tilde, |
| 31 | Dollar, |
| 32 | Circumflex, |
| 33 | Backslash, |
| 34 | } |
| 35 | |
| 36 | impl fmt::Display for ReservedChar { |
| 37 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 38 | write!( |
| 39 | f, |
| 40 | " {}" , |
| 41 | match *self { |
| 42 | ReservedChar::Comma => ',' , |
| 43 | ReservedChar::OpenParenthese => '(' , |
| 44 | ReservedChar::CloseParenthese => ')' , |
| 45 | ReservedChar::OpenCurlyBrace => '{' , |
| 46 | ReservedChar::CloseCurlyBrace => '}' , |
| 47 | ReservedChar::OpenBracket => '[' , |
| 48 | ReservedChar::CloseBracket => ']' , |
| 49 | ReservedChar::Colon => ':' , |
| 50 | ReservedChar::SemiColon => ';' , |
| 51 | ReservedChar::Slash => '/' , |
| 52 | ReservedChar::Star => '*' , |
| 53 | ReservedChar::Plus => '+' , |
| 54 | ReservedChar::EqualSign => '=' , |
| 55 | ReservedChar::Space => ' ' , |
| 56 | ReservedChar::Tab => ' \t' , |
| 57 | ReservedChar::Backline => ' \n' , |
| 58 | ReservedChar::SuperiorThan => '>' , |
| 59 | ReservedChar::Quote => ' \'' , |
| 60 | ReservedChar::DoubleQuote => '"' , |
| 61 | ReservedChar::Pipe => '|' , |
| 62 | ReservedChar::Tilde => '~' , |
| 63 | ReservedChar::Dollar => '$' , |
| 64 | ReservedChar::Circumflex => '^' , |
| 65 | ReservedChar::Backslash => ' \\' , |
| 66 | } |
| 67 | ) |
| 68 | } |
| 69 | } |
| 70 | |
| 71 | impl TryFrom<char> for ReservedChar { |
| 72 | type Error = &'static str; |
| 73 | |
| 74 | fn try_from(value: char) -> Result<ReservedChar, Self::Error> { |
| 75 | match value { |
| 76 | ' \'' => Ok(ReservedChar::Quote), |
| 77 | '"' => Ok(ReservedChar::DoubleQuote), |
| 78 | ',' => Ok(ReservedChar::Comma), |
| 79 | '(' => Ok(ReservedChar::OpenParenthese), |
| 80 | ')' => Ok(ReservedChar::CloseParenthese), |
| 81 | '{' => Ok(ReservedChar::OpenCurlyBrace), |
| 82 | '}' => Ok(ReservedChar::CloseCurlyBrace), |
| 83 | '[' => Ok(ReservedChar::OpenBracket), |
| 84 | ']' => Ok(ReservedChar::CloseBracket), |
| 85 | ':' => Ok(ReservedChar::Colon), |
| 86 | ';' => Ok(ReservedChar::SemiColon), |
| 87 | '/' => Ok(ReservedChar::Slash), |
| 88 | '*' => Ok(ReservedChar::Star), |
| 89 | '+' => Ok(ReservedChar::Plus), |
| 90 | '=' => Ok(ReservedChar::EqualSign), |
| 91 | ' ' => Ok(ReservedChar::Space), |
| 92 | ' \t' => Ok(ReservedChar::Tab), |
| 93 | ' \n' | ' \r' => Ok(ReservedChar::Backline), |
| 94 | '>' => Ok(ReservedChar::SuperiorThan), |
| 95 | '|' => Ok(ReservedChar::Pipe), |
| 96 | '~' => Ok(ReservedChar::Tilde), |
| 97 | '$' => Ok(ReservedChar::Dollar), |
| 98 | '^' => Ok(ReservedChar::Circumflex), |
| 99 | ' \\' => Ok(ReservedChar::Backslash), |
| 100 | _ => Err("Unknown reserved char" ), |
| 101 | } |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | impl ReservedChar { |
| 106 | fn is_useless(&self) -> bool { |
| 107 | *self == ReservedChar::Space |
| 108 | || *self == ReservedChar::Tab |
| 109 | || *self == ReservedChar::Backline |
| 110 | } |
| 111 | |
| 112 | fn is_operator(&self) -> bool { |
| 113 | Operator::try_from(*self).is_ok() |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | #[derive (Debug, PartialEq, Eq, Clone, Copy)] |
| 118 | pub enum Operator { |
| 119 | Plus, |
| 120 | Multiply, |
| 121 | Minus, |
| 122 | Modulo, |
| 123 | Divide, |
| 124 | } |
| 125 | |
| 126 | impl fmt::Display for Operator { |
| 127 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 128 | write!( |
| 129 | f, |
| 130 | " {}" , |
| 131 | match *self { |
| 132 | Operator::Plus => '+' , |
| 133 | Operator::Multiply => '*' , |
| 134 | Operator::Minus => '-' , |
| 135 | Operator::Modulo => '%' , |
| 136 | Operator::Divide => '/' , |
| 137 | } |
| 138 | ) |
| 139 | } |
| 140 | } |
| 141 | |
| 142 | impl TryFrom<char> for Operator { |
| 143 | type Error = &'static str; |
| 144 | |
| 145 | fn try_from(value: char) -> Result<Operator, Self::Error> { |
| 146 | match value { |
| 147 | '+' => Ok(Operator::Plus), |
| 148 | '*' => Ok(Operator::Multiply), |
| 149 | '-' => Ok(Operator::Minus), |
| 150 | '%' => Ok(Operator::Modulo), |
| 151 | '/' => Ok(Operator::Divide), |
| 152 | _ => Err("Unknown operator" ), |
| 153 | } |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | impl TryFrom<ReservedChar> for Operator { |
| 158 | type Error = &'static str; |
| 159 | |
| 160 | fn try_from(value: ReservedChar) -> Result<Operator, Self::Error> { |
| 161 | match value { |
| 162 | ReservedChar::Slash => Ok(Operator::Divide), |
| 163 | ReservedChar::Star => Ok(Operator::Multiply), |
| 164 | ReservedChar::Plus => Ok(Operator::Plus), |
| 165 | _ => Err("Unknown operator" ), |
| 166 | } |
| 167 | } |
| 168 | } |
| 169 | |
| 170 | #[derive (Eq, PartialEq, Clone, Debug)] |
| 171 | pub enum SelectorElement<'a> { |
| 172 | PseudoClass(&'a str), |
| 173 | Class(&'a str), |
| 174 | Id(&'a str), |
| 175 | Tag(&'a str), |
| 176 | Media(&'a str), |
| 177 | } |
| 178 | |
| 179 | impl<'a> TryFrom<&'a str> for SelectorElement<'a> { |
| 180 | type Error = &'static str; |
| 181 | |
| 182 | fn try_from(value: &'a str) -> Result<SelectorElement<'a>, Self::Error> { |
| 183 | if let Some(value) = value.strip_prefix('.' ) { |
| 184 | if value.is_empty() { |
| 185 | Err("cannot determine selector" ) |
| 186 | } else { |
| 187 | Ok(SelectorElement::Class(value)) |
| 188 | } |
| 189 | } else if let Some(value) = value.strip_prefix('#' ) { |
| 190 | if value.is_empty() { |
| 191 | Err("cannot determine selector" ) |
| 192 | } else { |
| 193 | Ok(SelectorElement::Id(value)) |
| 194 | } |
| 195 | } else if let Some(value) = value.strip_prefix('@' ) { |
| 196 | if value.is_empty() { |
| 197 | Err("cannot determine selector" ) |
| 198 | } else { |
| 199 | Ok(SelectorElement::Media(value)) |
| 200 | } |
| 201 | } else if let Some(value) = value.strip_prefix(':' ) { |
| 202 | if value.is_empty() { |
| 203 | Err("cannot determine selector" ) |
| 204 | } else { |
| 205 | Ok(SelectorElement::PseudoClass(value)) |
| 206 | } |
| 207 | } else if value.chars().next().unwrap_or(' ' ).is_alphabetic() { |
| 208 | Ok(SelectorElement::Tag(value)) |
| 209 | } else { |
| 210 | Err("unknown selector" ) |
| 211 | } |
| 212 | } |
| 213 | } |
| 214 | |
| 215 | impl fmt::Display for SelectorElement<'_> { |
| 216 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 217 | match *self { |
| 218 | SelectorElement::Class(c: &str) => write!(f, ". {}" , c), |
| 219 | SelectorElement::Id(i: &str) => write!(f, "# {}" , i), |
| 220 | SelectorElement::Tag(t: &str) => write!(f, " {}" , t), |
| 221 | SelectorElement::Media(m: &str) => write!(f, "@ {} " , m), |
| 222 | SelectorElement::PseudoClass(pc: &str) => write!(f, ": {}" , pc), |
| 223 | } |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | #[derive (Eq, PartialEq, Clone, Debug, Copy)] |
| 228 | pub enum SelectorOperator { |
| 229 | /// `~=` |
| 230 | OneAttributeEquals, |
| 231 | /// `|=` |
| 232 | EqualsOrStartsWithFollowedByDash, |
| 233 | /// `$=` |
| 234 | EndsWith, |
| 235 | /// `^=` |
| 236 | FirstStartsWith, |
| 237 | /// `*=` |
| 238 | Contains, |
| 239 | } |
| 240 | |
| 241 | impl fmt::Display for SelectorOperator { |
| 242 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 243 | match *self { |
| 244 | SelectorOperator::OneAttributeEquals => write!(f, "~=" ), |
| 245 | SelectorOperator::EqualsOrStartsWithFollowedByDash => write!(f, "|=" ), |
| 246 | SelectorOperator::EndsWith => write!(f, "$=" ), |
| 247 | SelectorOperator::FirstStartsWith => write!(f, "^=" ), |
| 248 | SelectorOperator::Contains => write!(f, "*=" ), |
| 249 | } |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | #[derive (Eq, PartialEq, Clone, Debug)] |
| 254 | pub enum Token<'a> { |
| 255 | /// Comment. |
| 256 | Comment(&'a str), |
| 257 | /// Comment starting with `/**`. |
| 258 | License(&'a str), |
| 259 | Char(ReservedChar), |
| 260 | Other(&'a str), |
| 261 | SelectorElement(SelectorElement<'a>), |
| 262 | String(&'a str), |
| 263 | SelectorOperator(SelectorOperator), |
| 264 | Operator(Operator), |
| 265 | } |
| 266 | |
| 267 | impl fmt::Display for Token<'_> { |
| 268 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 269 | match *self { |
| 270 | // Token::AtRule(at_rule) => write!(f, "{}", at_rule, content), |
| 271 | // Token::ElementRule(selectors) => write!(f, "{}", x), |
| 272 | Token::Comment(c: &str) => write!(f, " {}" , c), |
| 273 | Token::License(l: &str) => writeln!(f, "/*! {}*/" , l), |
| 274 | Token::Char(c: ReservedChar) => write!(f, " {}" , c), |
| 275 | Token::Other(s: &str) => write!(f, " {}" , s), |
| 276 | Token::SelectorElement(ref se: &SelectorElement<'_>) => write!(f, " {}" , se), |
| 277 | Token::String(s: &str) => write!(f, " {}" , s), |
| 278 | Token::SelectorOperator(so: SelectorOperator) => write!(f, " {}" , so), |
| 279 | Token::Operator(op: Operator) => write!(f, " {}" , op), |
| 280 | } |
| 281 | } |
| 282 | } |
| 283 | |
| 284 | impl Token<'_> { |
| 285 | fn is_comment(&self) -> bool { |
| 286 | matches!(*self, Token::Comment(_)) |
| 287 | } |
| 288 | |
| 289 | fn is_char(&self) -> bool { |
| 290 | matches!(*self, Token::Char(_)) |
| 291 | } |
| 292 | |
| 293 | fn get_char(&self) -> Option<ReservedChar> { |
| 294 | match *self { |
| 295 | Token::Char(c) => Some(c), |
| 296 | _ => None, |
| 297 | } |
| 298 | } |
| 299 | |
| 300 | fn is_useless(&self) -> bool { |
| 301 | match *self { |
| 302 | Token::Char(c) => c.is_useless(), |
| 303 | _ => false, |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | fn is_a_media(&self) -> bool { |
| 308 | matches!(*self, Token::SelectorElement(SelectorElement::Media(_))) |
| 309 | } |
| 310 | |
| 311 | fn is_a_selector_element(&self) -> bool { |
| 312 | matches!(*self, Token::SelectorElement(_)) |
| 313 | } |
| 314 | |
| 315 | fn is_a_license(&self) -> bool { |
| 316 | matches!(*self, Token::License(_)) |
| 317 | } |
| 318 | |
| 319 | fn is_operator(&self) -> bool { |
| 320 | match *self { |
| 321 | Token::Operator(_) => true, |
| 322 | Token::Char(c) => c.is_operator(), |
| 323 | _ => false, |
| 324 | } |
| 325 | } |
| 326 | } |
| 327 | |
| 328 | impl PartialEq<ReservedChar> for Token<'_> { |
| 329 | fn eq(&self, other: &ReservedChar) -> bool { |
| 330 | match *self { |
| 331 | Token::Char(c: ReservedChar) => c == *other, |
| 332 | _ => false, |
| 333 | } |
| 334 | } |
| 335 | } |
| 336 | |
| 337 | fn get_comment<'a>( |
| 338 | source: &'a str, |
| 339 | iterator: &mut Peekable<CharIndices<'_>>, |
| 340 | start_pos: &mut usize, |
| 341 | ) -> Option<Token<'a>> { |
| 342 | let mut prev = ReservedChar::Quote; |
| 343 | // eat the forward slash |
| 344 | let mut content_start_pos = *start_pos + 1; |
| 345 | let builder = if let Some((_, c)) = iterator.next() { |
| 346 | if c == '!' || (c == '*' && iterator.peek().map(|(_, c)| c) != Some(&'/' )) { |
| 347 | content_start_pos += 1; |
| 348 | Token::License |
| 349 | } else { |
| 350 | if let Ok(c) = ReservedChar::try_from(c) { |
| 351 | prev = c; |
| 352 | } |
| 353 | Token::Comment |
| 354 | } |
| 355 | } else { |
| 356 | Token::Comment |
| 357 | }; |
| 358 | |
| 359 | for (pos, c) in iterator { |
| 360 | if let Ok(c) = ReservedChar::try_from(c) { |
| 361 | if c == ReservedChar::Slash && prev == ReservedChar::Star { |
| 362 | let ret = Some(builder(&source[content_start_pos..pos - 1])); |
| 363 | *start_pos = pos; |
| 364 | return ret; |
| 365 | } |
| 366 | prev = c; |
| 367 | } else { |
| 368 | prev = ReservedChar::Space; |
| 369 | } |
| 370 | } |
| 371 | None |
| 372 | } |
| 373 | |
| 374 | fn get_string<'a>( |
| 375 | source: &'a str, |
| 376 | iterator: &mut Peekable<CharIndices<'_>>, |
| 377 | start_pos: &mut usize, |
| 378 | start: ReservedChar, |
| 379 | ) -> Option<Token<'a>> { |
| 380 | while let Some((pos: usize, c: char)) = iterator.next() { |
| 381 | if c == ' \\' { |
| 382 | // we skip next character |
| 383 | iterator.next(); |
| 384 | continue; |
| 385 | } |
| 386 | if let Ok(c: ReservedChar) = ReservedChar::try_from(c) { |
| 387 | if c == start { |
| 388 | let ret: Option> = Some(Token::String(&source[*start_pos..pos + 1])); |
| 389 | *start_pos = pos; |
| 390 | return ret; |
| 391 | } |
| 392 | } |
| 393 | } |
| 394 | None |
| 395 | } |
| 396 | |
| 397 | fn fill_other<'a>( |
| 398 | source: &'a str, |
| 399 | v: &mut Vec<Token<'a>>, |
| 400 | start: usize, |
| 401 | pos: usize, |
| 402 | is_in_block: isize, |
| 403 | is_in_media: bool, |
| 404 | is_in_attribute_selector: bool, |
| 405 | ) { |
| 406 | if start < pos { |
| 407 | if !is_in_attribute_selector |
| 408 | && ((is_in_block == 0 && !is_in_media) || (is_in_media && is_in_block == 1)) |
| 409 | { |
| 410 | let mut is_pseudo_class = false; |
| 411 | let mut add = 0; |
| 412 | if let Some(&Token::Char(ReservedChar::Colon)) = v.last() { |
| 413 | is_pseudo_class = true; |
| 414 | add = 1; |
| 415 | } |
| 416 | if let Ok(s) = SelectorElement::try_from(&source[start - add..pos]) { |
| 417 | if is_pseudo_class { |
| 418 | v.pop(); |
| 419 | } |
| 420 | v.push(Token::SelectorElement(s)); |
| 421 | } else { |
| 422 | let s = &source[start..pos]; |
| 423 | v.push(Token::Other(s)); |
| 424 | } |
| 425 | } else { |
| 426 | v.push(Token::Other(&source[start..pos])); |
| 427 | } |
| 428 | } |
| 429 | } |
| 430 | |
| 431 | #[allow (clippy::comparison_chain)] |
| 432 | pub(super) fn tokenize(source: &str) -> Result<Tokens<'_>, &'static str> { |
| 433 | let mut v = Vec::with_capacity(1000); |
| 434 | let mut iterator = source.char_indices().peekable(); |
| 435 | let mut start = 0; |
| 436 | let mut is_in_block: isize = 0; |
| 437 | let mut is_in_media = false; |
| 438 | let mut is_in_attribute_selector = false; |
| 439 | |
| 440 | loop { |
| 441 | let (mut pos, c) = match iterator.next() { |
| 442 | Some(x) => x, |
| 443 | None => { |
| 444 | fill_other( |
| 445 | source, |
| 446 | &mut v, |
| 447 | start, |
| 448 | source.len(), |
| 449 | is_in_block, |
| 450 | is_in_media, |
| 451 | is_in_attribute_selector, |
| 452 | ); |
| 453 | break; |
| 454 | } |
| 455 | }; |
| 456 | if let Ok(c) = ReservedChar::try_from(c) { |
| 457 | fill_other( |
| 458 | source, |
| 459 | &mut v, |
| 460 | start, |
| 461 | pos, |
| 462 | is_in_block, |
| 463 | is_in_media, |
| 464 | is_in_attribute_selector, |
| 465 | ); |
| 466 | is_in_media = is_in_media |
| 467 | || v.last() |
| 468 | .unwrap_or(&Token::Char(ReservedChar::Space)) |
| 469 | .is_a_media(); |
| 470 | |
| 471 | match c { |
| 472 | ReservedChar::Backslash => { |
| 473 | v.push(Token::Char(ReservedChar::Backslash)); |
| 474 | |
| 475 | if let Some((idx, c)) = iterator.next() { |
| 476 | pos += c.len_utf8(); |
| 477 | v.push(Token::Other(&source[idx..idx + c.len_utf8()])); |
| 478 | } |
| 479 | } |
| 480 | ReservedChar::Quote | ReservedChar::DoubleQuote => { |
| 481 | if let Some(s) = get_string(source, &mut iterator, &mut pos, c) { |
| 482 | v.push(s); |
| 483 | } else { |
| 484 | return Err("Unclosed string" ); |
| 485 | } |
| 486 | } |
| 487 | ReservedChar::Slash if matches!(iterator.peek(), Some(&(_, '*' ))) => { |
| 488 | // This is a comment. |
| 489 | let _ = iterator.next(); |
| 490 | pos += 1; |
| 491 | if let Some(s) = get_comment(source, &mut iterator, &mut pos) { |
| 492 | v.push(s); |
| 493 | } else { |
| 494 | return Err("Unclosed comment" ); |
| 495 | } |
| 496 | } |
| 497 | ReservedChar::OpenBracket => { |
| 498 | if is_in_attribute_selector { |
| 499 | return Err("Already in attribute selector" ); |
| 500 | } |
| 501 | is_in_attribute_selector = true; |
| 502 | v.push(Token::Char(c)); |
| 503 | } |
| 504 | ReservedChar::CloseBracket => { |
| 505 | if !is_in_attribute_selector { |
| 506 | return Err("Unexpected ']'" ); |
| 507 | } |
| 508 | is_in_attribute_selector = false; |
| 509 | v.push(Token::Char(c)); |
| 510 | } |
| 511 | ReservedChar::OpenCurlyBrace => { |
| 512 | is_in_block += 1; |
| 513 | v.push(Token::Char(c)); |
| 514 | } |
| 515 | ReservedChar::CloseCurlyBrace => { |
| 516 | is_in_block -= 1; |
| 517 | if is_in_block < 0 { |
| 518 | return Err("Too much '}'" ); |
| 519 | } else if is_in_block == 0 { |
| 520 | is_in_media = false; |
| 521 | } |
| 522 | v.push(Token::Char(c)); |
| 523 | } |
| 524 | ReservedChar::SemiColon if is_in_block == 0 => { |
| 525 | is_in_media = false; |
| 526 | v.push(Token::Char(c)); |
| 527 | } |
| 528 | ReservedChar::EqualSign => { |
| 529 | match match v |
| 530 | .last() |
| 531 | .unwrap_or(&Token::Char(ReservedChar::Space)) |
| 532 | .get_char() |
| 533 | .unwrap_or(ReservedChar::Space) |
| 534 | { |
| 535 | ReservedChar::Tilde => Some(SelectorOperator::OneAttributeEquals), |
| 536 | ReservedChar::Pipe => { |
| 537 | Some(SelectorOperator::EqualsOrStartsWithFollowedByDash) |
| 538 | } |
| 539 | ReservedChar::Dollar => Some(SelectorOperator::EndsWith), |
| 540 | ReservedChar::Circumflex => Some(SelectorOperator::FirstStartsWith), |
| 541 | ReservedChar::Star => Some(SelectorOperator::Contains), |
| 542 | _ => None, |
| 543 | } { |
| 544 | Some(r) => { |
| 545 | v.pop(); |
| 546 | v.push(Token::SelectorOperator(r)); |
| 547 | } |
| 548 | None => v.push(Token::Char(c)), |
| 549 | } |
| 550 | } |
| 551 | c if !c.is_useless() => { |
| 552 | v.push(Token::Char(c)); |
| 553 | } |
| 554 | c => { |
| 555 | if match v.last() { |
| 556 | Some(c) => { |
| 557 | !c.is_useless() |
| 558 | && (!c.is_char() |
| 559 | || c.is_operator() |
| 560 | || matches!( |
| 561 | c.get_char(), |
| 562 | Some( |
| 563 | ReservedChar::CloseParenthese |
| 564 | | ReservedChar::CloseBracket |
| 565 | ) |
| 566 | )) |
| 567 | } |
| 568 | _ => false, |
| 569 | } { |
| 570 | v.push(Token::Char(ReservedChar::Space)); |
| 571 | } else if let Ok(op) = Operator::try_from(c) { |
| 572 | v.push(Token::Operator(op)); |
| 573 | } |
| 574 | } |
| 575 | } |
| 576 | start = pos + 1; |
| 577 | } |
| 578 | } |
| 579 | Ok(Tokens(clean_tokens(v))) |
| 580 | } |
| 581 | |
| 582 | fn clean_tokens(mut v: Vec<Token<'_>>) -> Vec<Token<'_>> { |
| 583 | // This function may remove multiple elements from the vector. Ideally we'd |
| 584 | // use `Vec::retain`, but the traversal requires inspecting the previously |
| 585 | // retained token and the next token, which `Vec::retain` doesn't allow. So |
| 586 | // we have to use a lower-level mechanism. |
| 587 | let mut i = 0; |
| 588 | // Index of the previous retained token, if there is one. |
| 589 | let mut previous_element_index: Option<usize> = None; |
| 590 | let mut is_in_calc = false; |
| 591 | let mut paren = 0; |
| 592 | // A vector of bools indicating which elements are to be retained. |
| 593 | let mut b = Vec::with_capacity(v.len()); |
| 594 | |
| 595 | while i < v.len() { |
| 596 | if v[i] == Token::Other("calc" ) { |
| 597 | is_in_calc = true; |
| 598 | } else if is_in_calc { |
| 599 | if v[i] == Token::Char(ReservedChar::CloseParenthese) { |
| 600 | paren -= 1; |
| 601 | is_in_calc = paren != 0; |
| 602 | } else if v[i] == Token::Char(ReservedChar::OpenParenthese) { |
| 603 | paren += 1; |
| 604 | } |
| 605 | } |
| 606 | |
| 607 | let mut retain = true; |
| 608 | if v[i].is_useless() || v[i].is_comment() { |
| 609 | if let Some(previous_element_index) = previous_element_index { |
| 610 | #[allow (clippy::if_same_then_else)] |
| 611 | if v[previous_element_index] == Token::Char(ReservedChar::CloseBracket) { |
| 612 | if i + 1 < v.len() |
| 613 | && (v[i + 1].is_useless() |
| 614 | || v[i + 1] == Token::Char(ReservedChar::OpenCurlyBrace)) |
| 615 | { |
| 616 | retain = false; |
| 617 | } |
| 618 | } else if matches!(v[previous_element_index], Token::Other(_)) |
| 619 | && matches!( |
| 620 | v.get(i + 1), |
| 621 | Some(Token::Other(_) | Token::Char(ReservedChar::OpenParenthese)) |
| 622 | ) |
| 623 | { |
| 624 | // retain the space between keywords |
| 625 | // and the space that disambiguates functions from keyword-plus-parens |
| 626 | } else if v[previous_element_index].is_a_selector_element() |
| 627 | && matches!(v.get(i + 1), Some(Token::Char(ReservedChar::Star))) |
| 628 | { |
| 629 | // retain the space before `*` if it's preceded by a selector. |
| 630 | } else if matches!(v[previous_element_index], Token::Char(ReservedChar::Star)) |
| 631 | && v.get(i + 1) |
| 632 | .is_some_and(|elem| elem.is_a_selector_element()) |
| 633 | { |
| 634 | // retain the space after `*` if it's followed by a selector. |
| 635 | } else if matches!( |
| 636 | v[previous_element_index], |
| 637 | Token::Char( |
| 638 | ReservedChar::Star |
| 639 | | ReservedChar::Circumflex |
| 640 | | ReservedChar::Dollar |
| 641 | | ReservedChar::Tilde |
| 642 | ) |
| 643 | ) && matches!(v.get(i + 1), Some(Token::Char(ReservedChar::EqualSign))) |
| 644 | { |
| 645 | // retain the space between an operator and an equal sign |
| 646 | } else if matches!(v[previous_element_index], Token::Char(ReservedChar::Slash)) |
| 647 | && matches!(v.get(i + 1), Some(Token::Char(ReservedChar::Star))) |
| 648 | { |
| 649 | // this looks like a comment, but it is not |
| 650 | // retain the space between `/` and `*` |
| 651 | } else if is_in_calc && v[previous_element_index].is_useless() { |
| 652 | retain = false; |
| 653 | } else if !is_in_calc { |
| 654 | let prev = &v[previous_element_index]; |
| 655 | if ((prev.is_char() && prev != &Token::Char(ReservedChar::CloseParenthese)) |
| 656 | || prev.is_a_media() |
| 657 | || prev.is_a_license()) |
| 658 | || (i < v.len() - 1 |
| 659 | && v[i + 1].is_char() |
| 660 | && v[i + 1] != Token::Char(ReservedChar::OpenBracket)) |
| 661 | { |
| 662 | retain = false; |
| 663 | } |
| 664 | } |
| 665 | } |
| 666 | if retain && v[i].is_comment() { |
| 667 | // convert comments to spaces when minifying |
| 668 | v[i] = Token::Char(ReservedChar::Space); |
| 669 | } |
| 670 | } |
| 671 | if retain { |
| 672 | previous_element_index = Some(i); |
| 673 | } |
| 674 | b.push(retain); |
| 675 | i += 1; |
| 676 | } |
| 677 | assert_eq!(v.len(), b.len()); |
| 678 | let mut b = b.into_iter(); |
| 679 | v.retain(|_| b.next().unwrap()); |
| 680 | v |
| 681 | } |
| 682 | |
| 683 | #[derive (Debug, PartialEq, Eq, Clone)] |
| 684 | pub(super) struct Tokens<'a>(Vec<Token<'a>>); |
| 685 | |
| 686 | impl Tokens<'_> { |
| 687 | pub(super) fn write<W: std::io::Write>(self, mut w: W) -> std::io::Result<()> { |
| 688 | for token: &Token<'_> in self.0.iter() { |
| 689 | write!(w, " {}" , token)?; |
| 690 | } |
| 691 | Ok(()) |
| 692 | } |
| 693 | } |
| 694 | |
| 695 | impl fmt::Display for Tokens<'_> { |
| 696 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 697 | for token: &Token<'_> in self.0.iter() { |
| 698 | write!(f, " {}" , token)?; |
| 699 | } |
| 700 | Ok(()) |
| 701 | } |
| 702 | } |
| 703 | |
| 704 | #[test ] |
| 705 | fn css_basic() { |
| 706 | let s = r#" |
| 707 | /*! just some license */ |
| 708 | .foo > #bar p:hover { |
| 709 | color: blue; |
| 710 | background: "blue"; |
| 711 | } |
| 712 | |
| 713 | /* a comment! */ |
| 714 | @media screen and (max-width: 640px) { |
| 715 | .block:hover { |
| 716 | display: block; |
| 717 | } |
| 718 | }"# ; |
| 719 | let expected = vec![ |
| 720 | Token::License(" just some license " ), |
| 721 | Token::SelectorElement(SelectorElement::Class("foo" )), |
| 722 | Token::Char(ReservedChar::SuperiorThan), |
| 723 | Token::SelectorElement(SelectorElement::Id("bar" )), |
| 724 | Token::Char(ReservedChar::Space), |
| 725 | Token::SelectorElement(SelectorElement::Tag("p" )), |
| 726 | Token::SelectorElement(SelectorElement::PseudoClass("hover" )), |
| 727 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 728 | Token::Other("color" ), |
| 729 | Token::Char(ReservedChar::Colon), |
| 730 | Token::Other("blue" ), |
| 731 | Token::Char(ReservedChar::SemiColon), |
| 732 | Token::Other("background" ), |
| 733 | Token::Char(ReservedChar::Colon), |
| 734 | Token::String(" \"blue \"" ), |
| 735 | Token::Char(ReservedChar::SemiColon), |
| 736 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 737 | Token::SelectorElement(SelectorElement::Media("media" )), |
| 738 | Token::Other("screen" ), |
| 739 | Token::Char(ReservedChar::Space), |
| 740 | Token::Other("and" ), |
| 741 | Token::Char(ReservedChar::Space), |
| 742 | Token::Char(ReservedChar::OpenParenthese), |
| 743 | Token::Other("max-width" ), |
| 744 | Token::Char(ReservedChar::Colon), |
| 745 | Token::Other("640px" ), |
| 746 | Token::Char(ReservedChar::CloseParenthese), |
| 747 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 748 | Token::SelectorElement(SelectorElement::Class("block" )), |
| 749 | Token::SelectorElement(SelectorElement::PseudoClass("hover" )), |
| 750 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 751 | Token::Other("display" ), |
| 752 | Token::Char(ReservedChar::Colon), |
| 753 | Token::Other("block" ), |
| 754 | Token::Char(ReservedChar::SemiColon), |
| 755 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 756 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 757 | ]; |
| 758 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 759 | } |
| 760 | |
| 761 | #[test ] |
| 762 | fn elem_selector() { |
| 763 | let s = r#" |
| 764 | /** just some license */ |
| 765 | a[href*="example"] { |
| 766 | background: yellow; |
| 767 | } |
| 768 | a[href$=".org"] { |
| 769 | font-style: italic; |
| 770 | } |
| 771 | span[lang|="zh"] { |
| 772 | color: red; |
| 773 | } |
| 774 | a[href^="/"] { |
| 775 | background-color: gold; |
| 776 | } |
| 777 | div[value~="test"] { |
| 778 | border-width: 1px; |
| 779 | } |
| 780 | span[lang="pt"] { |
| 781 | font-size: 12em; /* I love big fonts */ |
| 782 | } |
| 783 | "# ; |
| 784 | let expected = vec![ |
| 785 | Token::License(" just some license " ), |
| 786 | Token::SelectorElement(SelectorElement::Tag("a" )), |
| 787 | Token::Char(ReservedChar::OpenBracket), |
| 788 | Token::Other("href" ), |
| 789 | Token::SelectorOperator(SelectorOperator::Contains), |
| 790 | Token::String(" \"example \"" ), |
| 791 | Token::Char(ReservedChar::CloseBracket), |
| 792 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 793 | Token::Other("background" ), |
| 794 | Token::Char(ReservedChar::Colon), |
| 795 | Token::Other("yellow" ), |
| 796 | Token::Char(ReservedChar::SemiColon), |
| 797 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 798 | Token::SelectorElement(SelectorElement::Tag("a" )), |
| 799 | Token::Char(ReservedChar::OpenBracket), |
| 800 | Token::Other("href" ), |
| 801 | Token::SelectorOperator(SelectorOperator::EndsWith), |
| 802 | Token::String(" \".org \"" ), |
| 803 | Token::Char(ReservedChar::CloseBracket), |
| 804 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 805 | Token::Other("font-style" ), |
| 806 | Token::Char(ReservedChar::Colon), |
| 807 | Token::Other("italic" ), |
| 808 | Token::Char(ReservedChar::SemiColon), |
| 809 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 810 | Token::SelectorElement(SelectorElement::Tag("span" )), |
| 811 | Token::Char(ReservedChar::OpenBracket), |
| 812 | Token::Other("lang" ), |
| 813 | Token::SelectorOperator(SelectorOperator::EqualsOrStartsWithFollowedByDash), |
| 814 | Token::String(" \"zh \"" ), |
| 815 | Token::Char(ReservedChar::CloseBracket), |
| 816 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 817 | Token::Other("color" ), |
| 818 | Token::Char(ReservedChar::Colon), |
| 819 | Token::Other("red" ), |
| 820 | Token::Char(ReservedChar::SemiColon), |
| 821 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 822 | Token::SelectorElement(SelectorElement::Tag("a" )), |
| 823 | Token::Char(ReservedChar::OpenBracket), |
| 824 | Token::Other("href" ), |
| 825 | Token::SelectorOperator(SelectorOperator::FirstStartsWith), |
| 826 | Token::String(" \"/ \"" ), |
| 827 | Token::Char(ReservedChar::CloseBracket), |
| 828 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 829 | Token::Other("background-color" ), |
| 830 | Token::Char(ReservedChar::Colon), |
| 831 | Token::Other("gold" ), |
| 832 | Token::Char(ReservedChar::SemiColon), |
| 833 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 834 | Token::SelectorElement(SelectorElement::Tag("div" )), |
| 835 | Token::Char(ReservedChar::OpenBracket), |
| 836 | Token::Other("value" ), |
| 837 | Token::SelectorOperator(SelectorOperator::OneAttributeEquals), |
| 838 | Token::String(" \"test \"" ), |
| 839 | Token::Char(ReservedChar::CloseBracket), |
| 840 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 841 | Token::Other("border-width" ), |
| 842 | Token::Char(ReservedChar::Colon), |
| 843 | Token::Other("1px" ), |
| 844 | Token::Char(ReservedChar::SemiColon), |
| 845 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 846 | Token::SelectorElement(SelectorElement::Tag("span" )), |
| 847 | Token::Char(ReservedChar::OpenBracket), |
| 848 | Token::Other("lang" ), |
| 849 | Token::Char(ReservedChar::EqualSign), |
| 850 | Token::String(" \"pt \"" ), |
| 851 | Token::Char(ReservedChar::CloseBracket), |
| 852 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 853 | Token::Other("font-size" ), |
| 854 | Token::Char(ReservedChar::Colon), |
| 855 | Token::Other("12em" ), |
| 856 | Token::Char(ReservedChar::SemiColon), |
| 857 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 858 | ]; |
| 859 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 860 | } |
| 861 | |
| 862 | #[test ] |
| 863 | fn check_media() { |
| 864 | let s = "@media (max-width: 700px) { color: red; }" ; |
| 865 | |
| 866 | let expected = vec![ |
| 867 | Token::SelectorElement(SelectorElement::Media("media" )), |
| 868 | Token::Char(ReservedChar::OpenParenthese), |
| 869 | Token::Other("max-width" ), |
| 870 | Token::Char(ReservedChar::Colon), |
| 871 | Token::Other("700px" ), |
| 872 | Token::Char(ReservedChar::CloseParenthese), |
| 873 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 874 | Token::SelectorElement(SelectorElement::Tag("color" )), |
| 875 | Token::Char(ReservedChar::Colon), |
| 876 | Token::Other("red" ), |
| 877 | Token::Char(ReservedChar::SemiColon), |
| 878 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 879 | ]; |
| 880 | |
| 881 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 882 | } |
| 883 | |
| 884 | #[test ] |
| 885 | fn check_supports() { |
| 886 | let s = "@supports not (display: grid) { div { float: right; } }" ; |
| 887 | |
| 888 | let expected = vec![ |
| 889 | Token::SelectorElement(SelectorElement::Media("supports" )), |
| 890 | Token::Other("not" ), |
| 891 | Token::Char(ReservedChar::Space), |
| 892 | Token::Char(ReservedChar::OpenParenthese), |
| 893 | Token::Other("display" ), |
| 894 | Token::Char(ReservedChar::Colon), |
| 895 | Token::Other("grid" ), |
| 896 | Token::Char(ReservedChar::CloseParenthese), |
| 897 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 898 | Token::SelectorElement(SelectorElement::Tag("div" )), |
| 899 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 900 | Token::Other("float" ), |
| 901 | Token::Char(ReservedChar::Colon), |
| 902 | Token::Other("right" ), |
| 903 | Token::Char(ReservedChar::SemiColon), |
| 904 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 905 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 906 | ]; |
| 907 | |
| 908 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 909 | } |
| 910 | |
| 911 | #[test ] |
| 912 | fn check_calc() { |
| 913 | let s = ".foo { width: calc(100% - 34px); }" ; |
| 914 | |
| 915 | let expected = vec![ |
| 916 | Token::SelectorElement(SelectorElement::Class("foo" )), |
| 917 | Token::Char(ReservedChar::OpenCurlyBrace), |
| 918 | Token::Other("width" ), |
| 919 | Token::Char(ReservedChar::Colon), |
| 920 | Token::Other("calc" ), |
| 921 | Token::Char(ReservedChar::OpenParenthese), |
| 922 | Token::Other("100%" ), |
| 923 | Token::Char(ReservedChar::Space), |
| 924 | Token::Other("-" ), |
| 925 | Token::Char(ReservedChar::Space), |
| 926 | Token::Other("34px" ), |
| 927 | Token::Char(ReservedChar::CloseParenthese), |
| 928 | Token::Char(ReservedChar::SemiColon), |
| 929 | Token::Char(ReservedChar::CloseCurlyBrace), |
| 930 | ]; |
| 931 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 932 | } |
| 933 | |
| 934 | #[test ] |
| 935 | fn check_attr_ast() { |
| 936 | let s = "x [y]" ; |
| 937 | |
| 938 | let expected = vec![ |
| 939 | Token::SelectorElement(SelectorElement::Tag("x" )), |
| 940 | Token::Char(ReservedChar::Space), |
| 941 | Token::Char(ReservedChar::OpenBracket), |
| 942 | Token::Other("y" ), |
| 943 | Token::Char(ReservedChar::CloseBracket), |
| 944 | ]; |
| 945 | assert_eq!(tokenize(s), Ok(Tokens(expected))); |
| 946 | } |
| 947 | |