Anchor.php
2 years ago
OutputException.php
4 years ago
ParserState.php
2 years ago
SourceException.php
4 years ago
UnexpectedEOFException.php
4 years ago
UnexpectedTokenException.php
4 years ago
index.php
3 years ago
ParserState.php
318 lines
| 1 | <?php |
| 2 | namespace MailPoetVendor\Sabberworm\CSS\Parsing; |
| 3 | if (!defined('ABSPATH')) exit; |
| 4 | use MailPoetVendor\Sabberworm\CSS\Comment\Comment; |
| 5 | use MailPoetVendor\Sabberworm\CSS\Settings; |
| 6 | class ParserState |
| 7 | { |
| 8 | const EOF = null; |
| 9 | private $oParserSettings; |
| 10 | private $sText; |
| 11 | private $aText; |
| 12 | private $iCurrentPosition; |
| 13 | private $sCharset; |
| 14 | private $iLength; |
| 15 | private $iLineNo; |
| 16 | public function __construct($sText, Settings $oParserSettings, $iLineNo = 1) |
| 17 | { |
| 18 | $this->oParserSettings = $oParserSettings; |
| 19 | $this->sText = $sText; |
| 20 | $this->iCurrentPosition = 0; |
| 21 | $this->iLineNo = $iLineNo; |
| 22 | $this->setCharset($this->oParserSettings->sDefaultCharset); |
| 23 | } |
| 24 | public function setCharset($sCharset) |
| 25 | { |
| 26 | $this->sCharset = $sCharset; |
| 27 | $this->aText = $this->strsplit($this->sText); |
| 28 | if (\is_array($this->aText)) { |
| 29 | $this->iLength = \count($this->aText); |
| 30 | } |
| 31 | } |
| 32 | public function getCharset() |
| 33 | { |
| 34 | return $this->sCharset; |
| 35 | } |
| 36 | public function currentLine() |
| 37 | { |
| 38 | return $this->iLineNo; |
| 39 | } |
| 40 | public function currentColumn() |
| 41 | { |
| 42 | return $this->iCurrentPosition; |
| 43 | } |
| 44 | public function getSettings() |
| 45 | { |
| 46 | return $this->oParserSettings; |
| 47 | } |
| 48 | public function anchor() |
| 49 | { |
| 50 | return new Anchor($this->iCurrentPosition, $this); |
| 51 | } |
| 52 | public function setPosition($iPosition) |
| 53 | { |
| 54 | $this->iCurrentPosition = $iPosition; |
| 55 | } |
| 56 | public function parseIdentifier($bIgnoreCase = \true) |
| 57 | { |
| 58 | if ($this->isEnd()) { |
| 59 | throw new UnexpectedEOFException('', '', 'identifier', $this->iLineNo); |
| 60 | } |
| 61 | $sResult = $this->parseCharacter(\true); |
| 62 | if ($sResult === null) { |
| 63 | throw new UnexpectedTokenException($sResult, $this->peek(5), 'identifier', $this->iLineNo); |
| 64 | } |
| 65 | $sCharacter = null; |
| 66 | while (!$this->isEnd() && ($sCharacter = $this->parseCharacter(\true)) !== null) { |
| 67 | if (\preg_match('/[a-zA-Z0-9\\x{00A0}-\\x{FFFF}_-]/Sux', $sCharacter)) { |
| 68 | $sResult .= $sCharacter; |
| 69 | } else { |
| 70 | $sResult .= '\\' . $sCharacter; |
| 71 | } |
| 72 | } |
| 73 | if ($bIgnoreCase) { |
| 74 | $sResult = $this->strtolower($sResult); |
| 75 | } |
| 76 | return $sResult; |
| 77 | } |
| 78 | public function parseCharacter($bIsForIdentifier) |
| 79 | { |
| 80 | if ($this->peek() === '\\') { |
| 81 | if ($bIsForIdentifier && $this->oParserSettings->bLenientParsing && ($this->comes('\\0') || $this->comes('\\9'))) { |
| 82 | // Non-strings can contain \0 or \9 which is an IE hack supported in lenient parsing. |
| 83 | return null; |
| 84 | } |
| 85 | $this->consume('\\'); |
| 86 | if ($this->comes('\\n') || $this->comes('\\r')) { |
| 87 | return ''; |
| 88 | } |
| 89 | if (\preg_match('/[0-9a-fA-F]/Su', $this->peek()) === 0) { |
| 90 | return $this->consume(1); |
| 91 | } |
| 92 | $sUnicode = $this->consumeExpression('/^[0-9a-fA-F]{1,6}/u', 6); |
| 93 | if ($this->strlen($sUnicode) < 6) { |
| 94 | // Consume whitespace after incomplete unicode escape |
| 95 | if (\preg_match('/\\s/isSu', $this->peek())) { |
| 96 | if ($this->comes('MailPoetVendor\\r\\n')) { |
| 97 | $this->consume(2); |
| 98 | } else { |
| 99 | $this->consume(1); |
| 100 | } |
| 101 | } |
| 102 | } |
| 103 | $iUnicode = \intval($sUnicode, 16); |
| 104 | $sUtf32 = ""; |
| 105 | for ($i = 0; $i < 4; ++$i) { |
| 106 | $sUtf32 .= \chr($iUnicode & 0xff); |
| 107 | $iUnicode = $iUnicode >> 8; |
| 108 | } |
| 109 | return \iconv('utf-32le', $this->sCharset, $sUtf32); |
| 110 | } |
| 111 | if ($bIsForIdentifier) { |
| 112 | $peek = \ord($this->peek()); |
| 113 | // Ranges: a-z A-Z 0-9 - _ |
| 114 | if ($peek >= 97 && $peek <= 122 || $peek >= 65 && $peek <= 90 || $peek >= 48 && $peek <= 57 || $peek === 45 || $peek === 95 || $peek > 0xa1) { |
| 115 | return $this->consume(1); |
| 116 | } |
| 117 | } else { |
| 118 | return $this->consume(1); |
| 119 | } |
| 120 | return null; |
| 121 | } |
| 122 | public function consumeWhiteSpace() |
| 123 | { |
| 124 | $aComments = []; |
| 125 | do { |
| 126 | while (\preg_match('/\\s/isSu', $this->peek()) === 1) { |
| 127 | $this->consume(1); |
| 128 | } |
| 129 | if ($this->oParserSettings->bLenientParsing) { |
| 130 | try { |
| 131 | $oComment = $this->consumeComment(); |
| 132 | } catch (UnexpectedEOFException $e) { |
| 133 | $this->iCurrentPosition = $this->iLength; |
| 134 | return $aComments; |
| 135 | } |
| 136 | } else { |
| 137 | $oComment = $this->consumeComment(); |
| 138 | } |
| 139 | if ($oComment !== \false) { |
| 140 | $aComments[] = $oComment; |
| 141 | } |
| 142 | } while ($oComment !== \false); |
| 143 | return $aComments; |
| 144 | } |
| 145 | public function comes($sString, $bCaseInsensitive = \false) |
| 146 | { |
| 147 | $sPeek = $this->peek(\strlen($sString)); |
| 148 | return $sPeek == '' ? \false : $this->streql($sPeek, $sString, $bCaseInsensitive); |
| 149 | } |
| 150 | public function peek($iLength = 1, $iOffset = 0) |
| 151 | { |
| 152 | $iOffset += $this->iCurrentPosition; |
| 153 | if ($iOffset >= $this->iLength) { |
| 154 | return ''; |
| 155 | } |
| 156 | return $this->substr($iOffset, $iLength); |
| 157 | } |
| 158 | public function consume($mValue = 1) |
| 159 | { |
| 160 | if (\is_string($mValue)) { |
| 161 | $iLineCount = \substr_count($mValue, "\n"); |
| 162 | $iLength = $this->strlen($mValue); |
| 163 | if (!$this->streql($this->substr($this->iCurrentPosition, $iLength), $mValue)) { |
| 164 | throw new UnexpectedTokenException($mValue, $this->peek(\max($iLength, 5)), $this->iLineNo); |
| 165 | } |
| 166 | $this->iLineNo += $iLineCount; |
| 167 | $this->iCurrentPosition += $this->strlen($mValue); |
| 168 | return $mValue; |
| 169 | } else { |
| 170 | if ($this->iCurrentPosition + $mValue > $this->iLength) { |
| 171 | throw new UnexpectedEOFException($mValue, $this->peek(5), 'count', $this->iLineNo); |
| 172 | } |
| 173 | $sResult = $this->substr($this->iCurrentPosition, $mValue); |
| 174 | $iLineCount = \substr_count($sResult, "\n"); |
| 175 | $this->iLineNo += $iLineCount; |
| 176 | $this->iCurrentPosition += $mValue; |
| 177 | return $sResult; |
| 178 | } |
| 179 | } |
| 180 | public function consumeExpression($mExpression, $iMaxLength = null) |
| 181 | { |
| 182 | $aMatches = null; |
| 183 | $sInput = $iMaxLength !== null ? $this->peek($iMaxLength) : $this->inputLeft(); |
| 184 | if (\preg_match($mExpression, $sInput, $aMatches, \PREG_OFFSET_CAPTURE) === 1) { |
| 185 | return $this->consume($aMatches[0][0]); |
| 186 | } |
| 187 | throw new UnexpectedTokenException($mExpression, $this->peek(5), 'expression', $this->iLineNo); |
| 188 | } |
| 189 | public function consumeComment() |
| 190 | { |
| 191 | $mComment = \false; |
| 192 | if ($this->comes('/*')) { |
| 193 | $iLineNo = $this->iLineNo; |
| 194 | $this->consume(1); |
| 195 | $mComment = ''; |
| 196 | while (($char = $this->consume(1)) !== '') { |
| 197 | $mComment .= $char; |
| 198 | if ($this->comes('*/')) { |
| 199 | $this->consume(2); |
| 200 | break; |
| 201 | } |
| 202 | } |
| 203 | } |
| 204 | if ($mComment !== \false) { |
| 205 | // We skip the * which was included in the comment. |
| 206 | return new Comment(\substr($mComment, 1), $iLineNo); |
| 207 | } |
| 208 | return $mComment; |
| 209 | } |
| 210 | public function isEnd() |
| 211 | { |
| 212 | return $this->iCurrentPosition >= $this->iLength; |
| 213 | } |
| 214 | public function consumeUntil($aEnd, $bIncludeEnd = \false, $consumeEnd = \false, array &$comments = []) |
| 215 | { |
| 216 | $aEnd = \is_array($aEnd) ? $aEnd : [$aEnd]; |
| 217 | $out = ''; |
| 218 | $start = $this->iCurrentPosition; |
| 219 | while (!$this->isEnd()) { |
| 220 | $char = $this->consume(1); |
| 221 | if (\in_array($char, $aEnd)) { |
| 222 | if ($bIncludeEnd) { |
| 223 | $out .= $char; |
| 224 | } elseif (!$consumeEnd) { |
| 225 | $this->iCurrentPosition -= $this->strlen($char); |
| 226 | } |
| 227 | return $out; |
| 228 | } |
| 229 | $out .= $char; |
| 230 | if ($comment = $this->consumeComment()) { |
| 231 | $comments[] = $comment; |
| 232 | } |
| 233 | } |
| 234 | if (\in_array(self::EOF, $aEnd)) { |
| 235 | return $out; |
| 236 | } |
| 237 | $this->iCurrentPosition = $start; |
| 238 | throw new UnexpectedEOFException('One of ("' . \implode('","', $aEnd) . '")', $this->peek(5), 'search', $this->iLineNo); |
| 239 | } |
| 240 | private function inputLeft() |
| 241 | { |
| 242 | return $this->substr($this->iCurrentPosition, -1); |
| 243 | } |
| 244 | public function streql($sString1, $sString2, $bCaseInsensitive = \true) |
| 245 | { |
| 246 | if ($bCaseInsensitive) { |
| 247 | return $this->strtolower($sString1) === $this->strtolower($sString2); |
| 248 | } else { |
| 249 | return $sString1 === $sString2; |
| 250 | } |
| 251 | } |
| 252 | public function backtrack($iAmount) |
| 253 | { |
| 254 | $this->iCurrentPosition -= $iAmount; |
| 255 | } |
| 256 | public function strlen($sString) |
| 257 | { |
| 258 | if ($this->oParserSettings->bMultibyteSupport) { |
| 259 | return \mb_strlen($sString, $this->sCharset); |
| 260 | } else { |
| 261 | return \strlen($sString); |
| 262 | } |
| 263 | } |
| 264 | private function substr($iStart, $iLength) |
| 265 | { |
| 266 | if ($iLength < 0) { |
| 267 | $iLength = $this->iLength - $iStart + $iLength; |
| 268 | } |
| 269 | if ($iStart + $iLength > $this->iLength) { |
| 270 | $iLength = $this->iLength - $iStart; |
| 271 | } |
| 272 | $sResult = ''; |
| 273 | while ($iLength > 0) { |
| 274 | $sResult .= $this->aText[$iStart]; |
| 275 | $iStart++; |
| 276 | $iLength--; |
| 277 | } |
| 278 | return $sResult; |
| 279 | } |
| 280 | private function strtolower($sString) |
| 281 | { |
| 282 | if ($this->oParserSettings->bMultibyteSupport) { |
| 283 | return \mb_strtolower($sString, $this->sCharset); |
| 284 | } else { |
| 285 | return \strtolower($sString); |
| 286 | } |
| 287 | } |
| 288 | private function strsplit($sString) |
| 289 | { |
| 290 | if ($this->oParserSettings->bMultibyteSupport) { |
| 291 | if ($this->streql($this->sCharset, 'utf-8')) { |
| 292 | return \preg_split('//u', $sString, -1, \PREG_SPLIT_NO_EMPTY); |
| 293 | } else { |
| 294 | $iLength = \mb_strlen($sString, $this->sCharset); |
| 295 | $aResult = []; |
| 296 | for ($i = 0; $i < $iLength; ++$i) { |
| 297 | $aResult[] = \mb_substr($sString, $i, 1, $this->sCharset); |
| 298 | } |
| 299 | return $aResult; |
| 300 | } |
| 301 | } else { |
| 302 | if ($sString === '') { |
| 303 | return []; |
| 304 | } else { |
| 305 | return \str_split($sString); |
| 306 | } |
| 307 | } |
| 308 | } |
| 309 | private function strpos($sString, $sNeedle, $iOffset) |
| 310 | { |
| 311 | if ($this->oParserSettings->bMultibyteSupport) { |
| 312 | return \mb_strpos($sString, $sNeedle, $iOffset, $this->sCharset); |
| 313 | } else { |
| 314 | return \strpos($sString, $sNeedle, $iOffset); |
| 315 | } |
| 316 | } |
| 317 | } |
| 318 |