[Summary view]

Differences Between: [Versions 310 and 402] [Versions 311 and 402] [Versions 39 and 402] [Versions 400 and 402]

   1  <?php
   2  
   3  namespace Sabberworm\CSS\Parsing;
   4  
   5  use Sabberworm\CSS\Comment\Comment;
   6  use Sabberworm\CSS\Settings;
   7  
   8  class ParserState
   9  {
  10      /**
  11       * @var null
  12       */
  13      const EOF = null;
  14  
  15      /**
  16       * @var Settings
  17       */
  18      private $oParserSettings;
  19  
  20      /**
  21       * @var string
  22       */
  23      private $sText;
  24  
  25      /**
  26       * @var array<int, string>
  27       */
  28      private $aText;
  29  
  30      /**
  31       * @var int
  32       */
  33      private $iCurrentPosition;
  34  
  35      /**
  36       * @var string
  37       */
  38      private $sCharset;
  39  
  40      /**
  41       * @var int
  42       */
  43      private $iLength;
  44  
  45      /**
  46       * @var int
  47       */
  48      private $iLineNo;
  49  
  50      /**
  51       * @param string $sText
  52       * @param int $iLineNo
  53       */
  54      public function __construct($sText, Settings $oParserSettings, $iLineNo = 1)
  55      {
  56          $this->oParserSettings = $oParserSettings;
  57          $this->sText = $sText;
  58          $this->iCurrentPosition = 0;
  59          $this->iLineNo = $iLineNo;
  60          $this->setCharset($this->oParserSettings->sDefaultCharset);
  61      }
  62  
  63      /**
  64       * @param string $sCharset
  65       *
  66       * @return void
  67       */
  68      public function setCharset($sCharset)
  69      {
  70          $this->sCharset = $sCharset;
  71          $this->aText = $this->strsplit($this->sText);
  72          if (is_array($this->aText)) {
  73              $this->iLength = count($this->aText);
  74          }
  75      }
  76  
  77      /**
  78       * @return string
  79       */
  80      public function getCharset()
  81      {
  82          return $this->sCharset;
  83      }
  84  
  85      /**
  86       * @return int
  87       */
  88      public function currentLine()
  89      {
  90          return $this->iLineNo;
  91      }
  92  
  93      /**
  94       * @return int
  95       */
  96      public function currentColumn()
  97      {
  98          return $this->iCurrentPosition;
  99      }
 100  
 101      /**
 102       * @return Settings
 103       */
 104      public function getSettings()
 105      {
 106          return $this->oParserSettings;
 107      }
 108  
 109      /**
 110       * @param bool $bIgnoreCase
 111       *
 112       * @return string
 113       *
 114       * @throws UnexpectedTokenException
 115       */
 116      public function parseIdentifier($bIgnoreCase = true)
 117      {
 118          $sResult = $this->parseCharacter(true);
 119          if ($sResult === null) {
 120              throw new UnexpectedTokenException($sResult, $this->peek(5), 'identifier', $this->iLineNo);
 121          }
 122          $sCharacter = null;
 123          while (($sCharacter = $this->parseCharacter(true)) !== null) {
 124              if (preg_match('/[a-zA-Z0-9\x{00A0}-\x{FFFF}_-]/Sux', $sCharacter)) {
 125                  $sResult .= $sCharacter;
 126              } else {
 127                  $sResult .= '\\' . $sCharacter;
 128              }
 129          }
 130          if ($bIgnoreCase) {
 131              $sResult = $this->strtolower($sResult);
 132          }
 133          return $sResult;
 134      }
 135  
 136      /**
 137       * @param bool $bIsForIdentifier
 138       *
 139       * @return string|null
 140       *
 141       * @throws UnexpectedEOFException
 142       * @throws UnexpectedTokenException
 143       */
 144      public function parseCharacter($bIsForIdentifier)
 145      {
 146          if ($this->peek() === '\\') {
 147              if (
 148                  $bIsForIdentifier && $this->oParserSettings->bLenientParsing
 149                  && ($this->comes('\0') || $this->comes('\9'))
 150              ) {
 151                  // Non-strings can contain \0 or \9 which is an IE hack supported in lenient parsing.
 152                  return null;
 153              }
 154              $this->consume('\\');
 155              if ($this->comes('\n') || $this->comes('\r')) {
 156                  return '';
 157              }
 158              if (preg_match('/[0-9a-fA-F]/Su', $this->peek()) === 0) {
 159                  return $this->consume(1);
 160              }
 161              $sUnicode = $this->consumeExpression('/^[0-9a-fA-F]{1,6}/u', 6);
 162              if ($this->strlen($sUnicode) < 6) {
 163                  // Consume whitespace after incomplete unicode escape
 164                  if (preg_match('/\\s/isSu', $this->peek())) {
 165                      if ($this->comes('\r\n')) {
 166                          $this->consume(2);
 167                      } else {
 168                          $this->consume(1);
 169                      }
 170                  }
 171              }
 172              $iUnicode = intval($sUnicode, 16);
 173              $sUtf32 = "";
 174              for ($i = 0; $i < 4; ++$i) {
 175                  $sUtf32 .= chr($iUnicode & 0xff);
 176                  $iUnicode = $iUnicode >> 8;
 177              }
 178              return iconv('utf-32le', $this->sCharset, $sUtf32);
 179          }
 180          if ($bIsForIdentifier) {
 181              $peek = ord($this->peek());
 182              // Ranges: a-z A-Z 0-9 - _
 183              if (
 184                  ($peek >= 97 && $peek <= 122)
 185                  || ($peek >= 65 && $peek <= 90)
 186                  || ($peek >= 48 && $peek <= 57)
 187                  || ($peek === 45)
 188                  || ($peek === 95)
 189                  || ($peek > 0xa1)
 190              ) {
 191                  return $this->consume(1);
 192              }
 193          } else {
 194              return $this->consume(1);
 195          }
 196          return null;
 197      }
 198  
 199      /**
 200       * @return array<int, Comment>|void
 201       *
 202       * @throws UnexpectedEOFException
 203       * @throws UnexpectedTokenException
 204       */
 205      public function consumeWhiteSpace()
 206      {
 207          $comments = [];
 208          do {
 209              while (preg_match('/\\s/isSu', $this->peek()) === 1) {
 210                  $this->consume(1);
 211              }
 212              if ($this->oParserSettings->bLenientParsing) {
 213                  try {
 214                      $oComment = $this->consumeComment();
 215                  } catch (UnexpectedEOFException $e) {
 216                      $this->iCurrentPosition = $this->iLength;
 217                      return;
 218                  }
 219              } else {
 220                  $oComment = $this->consumeComment();
 221              }
 222              if ($oComment !== false) {
 223                  $comments[] = $oComment;
 224              }
 225          } while ($oComment !== false);
 226          return $comments;
 227      }
 228  
 229      /**
 230       * @param string $sString
 231       * @param bool $bCaseInsensitive
 232       *
 233       * @return bool
 234       */
 235      public function comes($sString, $bCaseInsensitive = false)
 236      {
 237          $sPeek = $this->peek(strlen($sString));
 238          return ($sPeek == '')
 239              ? false
 240              : $this->streql($sPeek, $sString, $bCaseInsensitive);
 241      }
 242  
 243      /**
 244       * @param int $iLength
 245       * @param int $iOffset
 246       *
 247       * @return string
 248       */
 249      public function peek($iLength = 1, $iOffset = 0)
 250      {
 251          $iOffset += $this->iCurrentPosition;
 252          if ($iOffset >= $this->iLength) {
 253              return '';
 254          }
 255          return $this->substr($iOffset, $iLength);
 256      }
 257  
 258      /**
 259       * @param int $mValue
 260       *
 261       * @return string
 262       *
 263       * @throws UnexpectedEOFException
 264       * @throws UnexpectedTokenException
 265       */
 266      public function consume($mValue = 1)
 267      {
 268          if (is_string($mValue)) {
 269              $iLineCount = substr_count($mValue, "\n");
 270              $iLength = $this->strlen($mValue);
 271              if (!$this->streql($this->substr($this->iCurrentPosition, $iLength), $mValue)) {
 272                  throw new UnexpectedTokenException($mValue, $this->peek(max($iLength, 5)), $this->iLineNo);
 273              }
 274              $this->iLineNo += $iLineCount;
 275              $this->iCurrentPosition += $this->strlen($mValue);
 276              return $mValue;
 277          } else {
 278              if ($this->iCurrentPosition + $mValue > $this->iLength) {
 279                  throw new UnexpectedEOFException($mValue, $this->peek(5), 'count', $this->iLineNo);
 280              }
 281              $sResult = $this->substr($this->iCurrentPosition, $mValue);
 282              $iLineCount = substr_count($sResult, "\n");
 283              $this->iLineNo += $iLineCount;
 284              $this->iCurrentPosition += $mValue;
 285              return $sResult;
 286          }
 287      }
 288  
 289      /**
 290       * @param string $mExpression
 291       * @param int|null $iMaxLength
 292       *
 293       * @return string
 294       *
 295       * @throws UnexpectedEOFException
 296       * @throws UnexpectedTokenException
 297       */
 298      public function consumeExpression($mExpression, $iMaxLength = null)
 299      {
 300          $aMatches = null;
 301          $sInput = $iMaxLength !== null ? $this->peek($iMaxLength) : $this->inputLeft();
 302          if (preg_match($mExpression, $sInput, $aMatches, PREG_OFFSET_CAPTURE) === 1) {
 303              return $this->consume($aMatches[0][0]);
 304          }
 305          throw new UnexpectedTokenException($mExpression, $this->peek(5), 'expression', $this->iLineNo);
 306      }
 307  
 308      /**
 309       * @return Comment|false
 310       */
 311      public function consumeComment()
 312      {
 313          $mComment = false;
 314          if ($this->comes('/*')) {
 315              $iLineNo = $this->iLineNo;
 316              $this->consume(1);
 317              $mComment = '';
 318              while (($char = $this->consume(1)) !== '') {
 319                  $mComment .= $char;
 320                  if ($this->comes('*/')) {
 321                      $this->consume(2);
 322                      break;
 323                  }
 324              }
 325          }
 326  
 327          if ($mComment !== false) {
 328              // We skip the * which was included in the comment.
 329              return new Comment(substr($mComment, 1), $iLineNo);
 330          }
 331  
 332          return $mComment;
 333      }
 334  
 335      /**
 336       * @return bool
 337       */
 338      public function isEnd()
 339      {
 340          return $this->iCurrentPosition >= $this->iLength;
 341      }
 342  
 343      /**
 344       * @param array<array-key, string>|string $aEnd
 345       * @param string $bIncludeEnd
 346       * @param string $consumeEnd
 347       * @param array<int, Comment> $comments
 348       *
 349       * @return string
 350       *
 351       * @throws UnexpectedEOFException
 352       * @throws UnexpectedTokenException
 353       */
 354      public function consumeUntil($aEnd, $bIncludeEnd = false, $consumeEnd = false, array &$comments = [])
 355      {
 356          $aEnd = is_array($aEnd) ? $aEnd : [$aEnd];
 357          $out = '';
 358          $start = $this->iCurrentPosition;
 359  
 360          while (!$this->isEnd()) {
 361              $char = $this->consume(1);
 362              if (in_array($char, $aEnd)) {
 363                  if ($bIncludeEnd) {
 364                      $out .= $char;
 365                  } elseif (!$consumeEnd) {
 366                      $this->iCurrentPosition -= $this->strlen($char);
 367                  }
 368                  return $out;
 369              }
 370              $out .= $char;
 371              if ($comment = $this->consumeComment()) {
 372                  $comments[] = $comment;
 373              }
 374          }
 375  
 376          if (in_array(self::EOF, $aEnd)) {
 377              return $out;
 378          }
 379  
 380          $this->iCurrentPosition = $start;
 381          throw new UnexpectedEOFException(
 382              'One of ("' . implode('","', $aEnd) . '")',
 383              $this->peek(5),
 384              'search',
 385              $this->iLineNo
 386          );
 387      }
 388  
 389      /**
 390       * @return string
 391       */
 392      private function inputLeft()
 393      {
 394          return $this->substr($this->iCurrentPosition, -1);
 395      }
 396  
 397      /**
 398       * @param string $sString1
 399       * @param string $sString2
 400       * @param bool $bCaseInsensitive
 401       *
 402       * @return bool
 403       */
 404      public function streql($sString1, $sString2, $bCaseInsensitive = true)
 405      {
 406          if ($bCaseInsensitive) {
 407              return $this->strtolower($sString1) === $this->strtolower($sString2);
 408          } else {
 409              return $sString1 === $sString2;
 410          }
 411      }
 412  
 413      /**
 414       * @param int $iAmount
 415       *
 416       * @return void
 417       */
 418      public function backtrack($iAmount)
 419      {
 420          $this->iCurrentPosition -= $iAmount;
 421      }
 422  
 423      /**
 424       * @param string $sString
 425       *
 426       * @return int
 427       */
 428      public function strlen($sString)
 429      {
 430          if ($this->oParserSettings->bMultibyteSupport) {
 431              return mb_strlen($sString, $this->sCharset);
 432          } else {
 433              return strlen($sString);
 434          }
 435      }
 436  
 437      /**
 438       * @param int $iStart
 439       * @param int $iLength
 440       *
 441       * @return string
 442       */
 443      private function substr($iStart, $iLength)
 444      {
 445          if ($iLength < 0) {
 446              $iLength = $this->iLength - $iStart + $iLength;
 447          }
 448          if ($iStart + $iLength > $this->iLength) {
 449              $iLength = $this->iLength - $iStart;
 450          }
 451          $sResult = '';
 452          while ($iLength > 0) {
 453              $sResult .= $this->aText[$iStart];
 454              $iStart++;
 455              $iLength--;
 456          }
 457          return $sResult;
 458      }
 459  
 460      /**
 461       * @param string $sString
 462       *
 463       * @return string
 464       */
 465      private function strtolower($sString)
 466      {
 467          if ($this->oParserSettings->bMultibyteSupport) {
 468              return mb_strtolower($sString, $this->sCharset);
 469          } else {
 470              return strtolower($sString);
 471          }
 472      }
 473  
 474      /**
 475       * @param string $sString
 476       *
 477       * @return array<int, string>
 478       */
 479      private function strsplit($sString)
 480      {
 481          if ($this->oParserSettings->bMultibyteSupport) {
 482              if ($this->streql($this->sCharset, 'utf-8')) {
 483                  return preg_split('//u', $sString, -1, PREG_SPLIT_NO_EMPTY);
 484              } else {
 485                  $iLength = mb_strlen($sString, $this->sCharset);
 486                  $aResult = [];
 487                  for ($i = 0; $i < $iLength; ++$i) {
 488                      $aResult[] = mb_substr($sString, $i, 1, $this->sCharset);
 489                  }
 490                  return $aResult;
 491              }
 492          } else {
 493              if ($sString === '') {
 494                  return [];
 495              } else {
 496                  return str_split($sString);
 497              }
 498          }
 499      }
 500  
 501      /**
 502       * @param string $sString
 503       * @param string $sNeedle
 504       * @param int $iOffset
 505       *
 506       * @return int|false
 507       */
 508      private function strpos($sString, $sNeedle, $iOffset)
 509      {
 510          if ($this->oParserSettings->bMultibyteSupport) {
 511              return mb_strpos($sString, $sNeedle, $iOffset, $this->sCharset);
 512          } else {
 513              return strpos($sString, $sNeedle, $iOffset);
 514          }
 515      }
 516  }