Fix encoding issues and improve parsing performance

MyIntervals · Aug 24, 2016 · 2ddbdef · 2ddbdef
1 parent 50a802f
commit 2ddbdef
Show file tree

Hide file tree

Showing 24 changed files with 641 additions and 169 deletions.
diff --git a/lib/Sabberworm/CSS/Parser.php b/lib/Sabberworm/CSS/Parser.php
diff --git a/lib/Sabberworm/CSS/Property/Charset.php b/lib/Sabberworm/CSS/Property/Charset.php
@@ -1,6 +1,7 @@
 <?php
 
 namespace Sabberworm\CSS\Property;
+use Sabberworm\CSS\Value\CSSString;
 
 /**
  * Class representing an @charset rule.
@@ -11,12 +12,16 @@
  */
 class Charset implements AtRule {
 
-	private $sCharset;
+	private $oCharset;
 	protected $iLineNo;
-	protected $aComment;
+	protected $aComments;
 
-	public function __construct($sCharset, $iLineNo = 0) {
-		$this->sCharset = $sCharset;
+	/**
+	 * @param CSSString $oCharset
+	 * @param int $iLineNo
+	 */
+	public function __construct(CSSString $oCharset, $iLineNo = 0) {
+		$this->oCharset = $oCharset;
 		$this->iLineNo = $iLineNo;
 		$this->aComments = array();
 	}
@@ -28,28 +33,34 @@ public function getLineNo() {
 		return $this->iLineNo;
 	}
 
-	public function setCharset($sCharset) {
-		$this->sCharset = $sCharset;
+	/**
+	 * @param CSSString $oCharset
+	 */
+	public function setCharset(CSSString $oCharset) {
+		$this->oCharset = $oCharset;
 	}
 
+	/**
+	 * @return CSSString
+	 */
 	public function getCharset() {
-		return $this->sCharset;
+		return $this->oCharset;
 	}
 
 	public function __toString() {
 		return $this->render(new \Sabberworm\CSS\OutputFormat());
 	}
 
 	public function render(\Sabberworm\CSS\OutputFormat $oOutputFormat) {
-		return "@charset {$this->sCharset->render($oOutputFormat)};";
+		return "@charset {$this->oCharset->render($oOutputFormat)};";
 	}
 
 	public function atRuleName() {
 		return 'charset';
 	}
 
 	public function atRuleArgs() {
-		return $this->sCharset;
+		return $this->oCharset;
 	}
 
 	public function addComments(array $aComments) {

diff --git a/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php b/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php
@@ -226,7 +226,11 @@ public function expandFontShorthand() {
 		}
 		foreach ($aValues as $mValue) {
 			if (!$mValue instanceof Value) {
-				$mValue = mb_strtolower($mValue);
+				if (function_exists('mb_strtolower')) {
+					$mValue = mb_strtolower($mValue, 'utf-8');
+				} else {
+					$mValue = strtolower($mValue);
+				}
 			}
 			if (in_array($mValue, array('normal', 'inherit'))) {
 				foreach (array('font-style', 'font-weight', 'font-variant') as $sProperty) {
@@ -300,7 +304,11 @@ public function expandBackgroundShorthand() {
 		$iNumBgPos = 0;
 		foreach ($aValues as $mValue) {
 			if (!$mValue instanceof Value) {
-				$mValue = mb_strtolower($mValue);
+				if (function_exists('mb_strtolower')) {
+					$mValue = mb_strtolower($mValue, 'utf-8');
+				} else {
+					$mValue = strtolower($mValue);
+				}
 			}
 			if ($mValue instanceof URL) {
 				$aBgProperties['background-image'] = $mValue;
@@ -369,7 +377,11 @@ public function expandListStyleShorthand() {
 		}
 		foreach ($aValues as $mValue) {
 			if (!$mValue instanceof Value) {
-				$mValue = mb_strtolower($mValue);
+				if (function_exists('mb_strtolower')) {
+					$mValue = mb_strtolower($mValue, 'utf-8');
+				} else {
+					$mValue = strtolower($mValue);
+				}
 			}
 			if ($mValue instanceof Url) {
 				$aListProperties['list-style-image'] = $mValue;

diff --git a/lib/Sabberworm/CSS/Value/CSSString.php b/lib/Sabberworm/CSS/Value/CSSString.php
@@ -24,9 +24,40 @@ public function __toString() {
 	}
 
 	public function render(\Sabberworm\CSS\OutputFormat $oOutputFormat) {
-		$sString = addslashes($this->sString);
-		$sString = str_replace("\n", '\A', $sString);
-		return $oOutputFormat->getStringQuotingType() . $sString . $oOutputFormat->getStringQuotingType();
+		$sQuote = $oOutputFormat->getStringQuotingType();
+		$aString = preg_split('//u', $this->sString, null, PREG_SPLIT_NO_EMPTY);
+		$iLength = count($aString);
+		foreach ($aString as $i => $sChar) {
+			if (strlen($sChar) === 1) {
+				if ($sChar === $sQuote || $sChar === '\\') {
+					// Encode quoting related characters as hex values
+				} else {
+					$iOrd = ord($sChar);
+					if ($iOrd > 31 && $iOrd < 127) {
+						// Keep only human readable ascii characters
+						continue;
+					}
+				}
+			}
+
+			$sHex = '';
+			$sUtf32 = iconv('utf-8', 'utf-32le', $sChar);
+			$aBytes = str_split($sUtf32);
+			foreach (array_reverse($aBytes) as $sByte) {
+				$sHex .= str_pad(dechex(ord($sByte)), 2, '0', STR_PAD_LEFT);
+			}
+			$sHex = ltrim($sHex, '0');
+			if ($i + 1 < $iLength && strlen($sHex) < 6) {
+				// Add space after incomplete unicode escape if there can be any confusion
+				$sNextChar = $aString[$i + 1];
+				if (preg_match('/^[a-fA-F0-9\s]/u', $sNextChar)) {
+					$sHex .= ' ';
+				}
+			}
+			$aString[$i] = '\\' . $sHex;
+		}
+
+		return $sQuote . implode($aString) . $sQuote;
 	}
 
 }