Improve parsing of calc and url by handling escapes and preserving them as non-function keywords

Author: raxbg

lib/Sabberworm/CSS/Parsing/ParserState.php

@@ -18,12 +18,14 @@ class ParserState {
 	private $sCharset;
 	private $iLength;
 	private $iLineNo;
+	private $iAnchor;
 
 	public function __construct($sText, Settings $oParserSettings, $iLineNo = 1) {
 		$this->oParserSettings = $oParserSettings;
 		$this->sText = $sText;
 		$this->iCurrentPosition = 0;
 		$this->iLineNo = $iLineNo;
+		$this->iAnchor = null;
 		$this->setCharset($this->oParserSettings->sDefaultCharset);
 	}
 
@@ -48,13 +50,26 @@ public function getSettings() {
 		return $this->oParserSettings;
 	}
 
+	public function setAnchor() {
+		$this->iAnchor = $this->iCurrentPosition;
+	}
+
+	public function backtrackToAnchor() {
+		if ($this->iAnchor !== null) {
+			$this->iCurrentPosition = $this->iAnchor;
+		}
+	}
+
 	public function parseIdentifier($bIgnoreCase = true) {
+		if ($this->isEnd()) {
+			throw new UnexpectedEOFException('', '', 'identifier', $this->iLineNo);
+		}
 		$sResult = $this->parseCharacter(true);
 		if ($sResult === null) {
 			throw new UnexpectedTokenException($sResult, $this->peek(5), 'identifier', $this->iLineNo);
 		}
 		$sCharacter = null;
-		while (($sCharacter = $this->parseCharacter(true)) !== null) {
+		while (!$this->isEnd() && ($sCharacter = $this->parseCharacter(true)) !== null) {
 			$sResult .= $sCharacter;
 		}
 		if ($bIgnoreCase) {

lib/Sabberworm/CSS/Value/CalcFunction.php

@@ -11,15 +11,15 @@ class CalcFunction extends CSSFunction {
 
 	public static function parse(ParserState $oParserState) {
 		$aOperators = array('+', '-', '*', '/');
-		$aTerminators = array('(', ';', "\n", "\r", ParserState::EOF);
-		$sFunction = $oParserState->consumeUntil($aTerminators, false, true);
-		if ($oParserState->peek(1, -1) != '(') {
+		$sFunction = $oParserState->parseIdentifier();
+		if ($oParserState->peek() != '(') {
 			// Found ; or end of line before an opening bracket
-			throw new UnexpectedTokenException('(', $oParserState->peek(1, -1), 'literal', $oParserState->currentLine());
+			throw new UnexpectedTokenException('(', $oParserState->peek(), 'literal', $oParserState->currentLine());
 		} else if (!in_array($sFunction, array('calc', '-moz-calc', '-webkit-calc'))) {
 			// Found invalid calc definition. Example calc (...
 			throw new UnexpectedTokenException('calc', $sFunction, 'literal', $oParserState->currentLine());
 		}
+		$oParserState->consume('(');
 		$oCalcList = new CalcRuleValueList($oParserState->currentLine());
 		$oList = new RuleValueList(',', $oParserState->currentLine());
 		$iNestingLevel = 0;

lib/Sabberworm/CSS/Value/URL.php

@@ -14,11 +14,21 @@ public function __construct(CSSString $oURL, $iLineNo = 0) {
 	}
 
 	public static function parse(ParserState $oParserState) {
-		$bUseUrl = $oParserState->comes('url', true);
+		$oParserState->setAnchor();
+		$sIdentifier = '';
+		for ($i = 0; $i < 3; $i++) {
+			$sChar = $oParserState->parseCharacter(true);
+			if ($sChar === null) {
+				break;
+			}
+			$sIdentifier .= $sChar;
+		}
+		$bUseUrl = $oParserState->streql($sIdentifier, 'url');
 		if ($bUseUrl) {
-			$oParserState->consume('url');
 			$oParserState->consumeWhiteSpace();
 			$oParserState->consume('(');
+		} else {
+			$oParserState->backtrackToAnchor();
 		}
 		$oParserState->consumeWhiteSpace();
 		$oResult = new URL(CSSString::parse($oParserState), $oParserState->currentLine());
@@ -46,4 +56,4 @@ public function render(\Sabberworm\CSS\OutputFormat $oOutputFormat) {
 		return "url({$this->oURL->render($oOutputFormat)})";
 	}
 
-}
+}

lib/Sabberworm/CSS/Value/Value.php

@@ -63,16 +63,25 @@ public static function parseValue(ParserState $oParserState, $aListDelimiters =
 	}
 
 	public static function parseIdentifierOrFunction(ParserState $oParserState, $bIgnoreCase = false) {
-		$sResult = $oParserState->parseIdentifier($bIgnoreCase);
+		$oParserState->setAnchor();
+		$mResult = $oParserState->parseIdentifier($bIgnoreCase);
 
 		if ($oParserState->comes('(')) {
-			$oParserState->consume('(');
-			$aArguments = Value::parseValue($oParserState, array('=', ' ', ','));
-			$sResult = new CSSFunction($sResult, $aArguments, ',', $oParserState->currentLine());
-			$oParserState->consume(')');
+			if ($oParserState->streql('url', $mResult)) {
+				$oParserState->backtrackToAnchor();
+				$mResult = URL::parse($oParserState);
+			} else if ($oParserState->streql('calc', $mResult) || $oParserState->streql('-webkit-calc', $mResult) || $oParserState->streql('-moz-calc', $mResult)) {
+				$oParserState->backtrackToAnchor();
+				$mResult = CalcFunction::parse($oParserState);
+			} else {
+				$oParserState->consume('(');
+				$aArguments = Value::parseValue($oParserState, array('=', ' ', ','));
+				$mResult = new CSSFunction($mResult, $aArguments, ',', $oParserState->currentLine());
+				$oParserState->consume(')');
+			}
 		}
 
-		return $sResult;
+		return $mResult;
 	}
 
 	public static function parsePrimitiveValue(ParserState $oParserState) {
@@ -82,10 +91,6 @@ public static function parsePrimitiveValue(ParserState $oParserState) {
 			$oValue = Size::parse($oParserState);
 		} else if ($oParserState->comes('#') || $oParserState->comes('rgb', true) || $oParserState->comes('hsl', true)) {
 			$oValue = Color::parse($oParserState);
-		} else if ($oParserState->comes('url', true)) {
-			$oValue = URL::parse($oParserState);
-		} else if ($oParserState->comes('calc', true) || $oParserState->comes('-webkit-calc', true) || $oParserState->comes('-moz-calc', true)) {
-			$oValue = CalcFunction::parse($oParserState);
 		} else if ($oParserState->comes("'") || $oParserState->comes('"')) {
 			$oValue = CSSString::parse($oParserState);
 		} else if ($oParserState->comes("progid:") && $oParserState->getSettings()->bLenientParsing) {

tests/Sabberworm/CSS/ParserTest.php

@@ -420,8 +420,8 @@ function testInvalidCalcInFile() {
 		$sExpected = 'div {}
 div {}
 div {}
-div {}
-div {}';
+div {height: -moz-calc;}
+div {height: calc;}';
 		$this->assertSame($sExpected, $oDoc->render());
 	}