Throw Exception when @charset appears where it shouldn’t

UnexpectedTokenException

Author: sabberworm

lib/Sabberworm/CSS/Parser.php

@@ -78,30 +78,54 @@ private function parseDocument(Document $oDocument) {
 
 	private function parseList(CSSList $oList, $bIsRoot = false) {
 		while (!$this->isEnd()) {
-			if ($this->comes('@')) {
-				$oList->append($this->parseAtRule());
-			} else if ($this->comes('}')) {
-				$this->consume('}');
-				if ($bIsRoot) {
-					throw new \Exception("Unopened {");
-				} else {
-					return;
+			$oListItem = null;
+			if($this->oParserSettings->bLenientParsing) {
+				try {
+					$oListItem = $this->parseListItem($oList, $bIsRoot);
+				} catch (UnexpectedTokenException $e) {
+					$oListItem = false;
 				}
 			} else {
-				if($this->oParserSettings->bLenientParsing) {
-					try {
-						$oList->append($this->parseSelector());
-					} catch (UnexpectedTokenException $e) {}
-				} else {
-					$oList->append($this->parseSelector());
-				}
+				$oListItem = $this->parseListItem($oList, $bIsRoot);
+			}
+			if($oListItem === null) {
+				// List parsing finished
+				return;
+			}
+			if($oListItem) {
+				$oList->append($oListItem);
 			}
 			$this->consumeWhiteSpace();
 		}
 		if (!$bIsRoot) {
 			throw new \Exception("Unexpected end of document");
 		}
 	}
+	
+	private function parseListItem(CSSList $oList, $bIsRoot = false) {
+		if ($this->comes('@')) {
+			$oAtRule = $this->parseAtRule();
+			if($oAtRule instanceof Charset) {
+				if(!$bIsRoot) {
+					throw new UnexpectedTokenException('@charset may only occur in root document', '', 'custom');
+				}
+				if(count($oList->getContents()) > 0) {
+					throw new UnexpectedTokenException('@charset must be the first parseable token in a document', '', 'custom');
+				}
+				$this->setCharset($oAtRule->getCharset()->getString());
+			}
+			return $oAtRule;
+		} else if ($this->comes('}')) {
+			$this->consume('}');
+			if ($bIsRoot) {
+				throw new \Exception("Unopened {");
+			} else {
+				return null;
+			}
+		} else {
+			return $this->parseSelector();
+		}
+	}
 
 	private function parseAtRule() {
 		$this->consume('@');
@@ -120,7 +144,6 @@ private function parseAtRule() {
 			$sCharset = $this->parseStringValue();
 			$this->consumeWhiteSpace();
 			$this->consume(';');
-			$this->setCharset($sCharset->getString());
 			return new Charset($sCharset);
 		} else if ($this->identifierIs($sIdentifier, 'keyframes')) {
 			$oResult = new KeyFrame();

lib/Sabberworm/CSS/Parsing/UnexpectedTokenException.php

@@ -22,6 +22,8 @@ public function __construct($sExpected, $sFound, $sMatchType = 'literal') {
 			$sMessage = "Next token was expected to have {$sExpected} chars. Context: “{$sFound}”.";
 		} else if($this->sMatchType === 'identifier') {
 			$sMessage = "Identifier expected. Got “{$sFound}”";
+		} else if($this->sMatchType === 'custom') {
+			$sMessage = trim("$sExpected $sFound");
 		}
 		parent::__construct($sMessage);
 	}

tests/Sabberworm/CSS/ParserTest.php

@@ -350,7 +350,7 @@ function testListValueRemoval() {
 		$this->assertSame('@media screen {html {some: -test(val2);}}
 #unrelated {other: yes;}', $oDoc->render());
 	}
-  
+	
 	/**
 	* @expectedException Sabberworm\CSS\Parsing\OutputException
 	*/
@@ -377,17 +377,43 @@ function testComments() {
 		$this->assertSame($sExpected, $oDoc->render());
 	}
 
-  function testEmptyFile() {
-    $oDoc = $this->parsedStructureForFile('-empty', Settings::create()->withMultibyteSupport(true));
+	function testEmptyFile() {
+		$oDoc = $this->parsedStructureForFile('-empty', Settings::create()->withMultibyteSupport(true));
 		$sExpected = '';
 		$this->assertSame($sExpected, $oDoc->render());
-  }
+	}
 
-  function testEmptyFileMbOff() {
-    $oDoc = $this->parsedStructureForFile('-empty', Settings::create()->withMultibyteSupport(false));
+	function testEmptyFileMbOff() {
+		$oDoc = $this->parsedStructureForFile('-empty', Settings::create()->withMultibyteSupport(false));
 		$sExpected = '';
 		$this->assertSame($sExpected, $oDoc->render());
-  }
+	}
+
+	function testCharsetLenient1() {
+		$oDoc = $this->parsedStructureForFile('-charset-after-rule', Settings::create()->withLenientParsing(true));
+		$sExpected = '#id {prop: var(--val);}';
+		$this->assertSame($sExpected, $oDoc->render());
+	}
+
+	function testCharsetLenient2() {
+		$oDoc = $this->parsedStructureForFile('-charset-in-block', Settings::create()->withLenientParsing(true));
+		$sExpected = '@media print {}';
+		$this->assertSame($sExpected, $oDoc->render());
+	}
+
+	/**
+	* @expectedException Sabberworm\CSS\Parsing\UnexpectedTokenException
+	*/
+	function testCharsetFailure1() {
+		$this->parsedStructureForFile('-charset-after-rule', Settings::create()->withLenientParsing(false));
+	}
+
+	/**
+	* @expectedException Sabberworm\CSS\Parsing\UnexpectedTokenException
+	*/
+	function testCharsetFailure2() {
+		$this->parsedStructureForFile('-charset-in-block', Settings::create()->withLenientParsing(false));
+	}
 
 	function parsedStructureForFile($sFileName, $oSettings = null) {
 		$sFile = dirname(__FILE__) . '/../../files' . DIRECTORY_SEPARATOR . "$sFileName.css";