Support for comment parsing

* origin/pr/110:
  Handle comments glued to block and rules
  Primary support for comment parsing

Author: sabberworm

lib/Sabberworm/CSS/CSSList/CSSList.php

@@ -6,17 +6,20 @@
 use Sabberworm\CSS\RuleSet\DeclarationBlock;
 use Sabberworm\CSS\RuleSet\RuleSet;
 use Sabberworm\CSS\Property\Selector;
+use Sabberworm\CSS\Comment\Commentable;
 
 /**
  * A CSSList is the most generic container available. Its contents include RuleSet as well as other CSSList objects.
  * Also, it may contain Import and Charset objects stemming from @-rules.
  */
-abstract class CSSList implements Renderable {
+abstract class CSSList implements Renderable, Commentable {
 
+	protected $aComments;
 	protected $aContents;
 	protected $iLineNo;
 
 	public function __construct($iLineNo = 0) {
+		$this->aComments = array();
 		$this->aContents = array();
 		$this->iLineNo = $iLineNo;
 	}
@@ -118,4 +121,26 @@ public abstract function isRootList();
 	public function getContents() {
 		return $this->aContents;
 	}
+
+	/**
+	 * @param array $aComments Array of comments.
+	 */
+	public function addComments(array $aComments) {
+		$this->aComments = array_merge($this->aComments, $aComments);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function getComments() {
+		return $this->aComments;
+	}
+
+	/**
+	 * @param array $aComments Array containing Comment objects.
+	 */
+	public function setComments(array $aComments) {
+		$this->aComments = $aComments;
+	}
+
 }

lib/Sabberworm/CSS/Comment/Comment.php

@@ -0,0 +1,51 @@
+<?php
+
+namespace Sabberworm\CSS\Comment;
+
+use Sabberworm\CSS\Renderable;
+
+class Comment implements Renderable {
+	protected $iLineNo;
+	protected $sComment;
+
+	public function __construct($sComment = '', $iLineNo = 0) {
+		$this->sComment = $sComment;
+		$this->iLineNo = $iLineNo;
+	}
+
+	/**
+	 * @return string
+	 */
+	public function getComment() {
+		return $this->sComment;
+	}
+
+	/**
+	 * @return int
+	 */
+	public function getLineNo() {
+		return $this->iLineNo;
+	}
+
+	/**
+	 * @return string
+	 */
+	public function setComment($sComment) {
+		$this->sComment = $sComment;
+	}
+
+	/**
+	 * @return string
+	 */
+	public function __toString() {
+		return $this->render(new \Sabberworm\CSS\OutputFormat());
+	}
+
+	/**
+	 * @return string
+	 */
+	public function render(\Sabberworm\CSS\OutputFormat $oOutputFormat) {
+		return '/*' . $this->sComment . '*/';
+	}
+
+}

lib/Sabberworm/CSS/Comment/Commentable.php

@@ -0,0 +1,23 @@
+<?php
+
+namespace Sabberworm\CSS\Comment;
+
+interface Commentable {
+
+	/**
+	 * @param array $aComments Array of comments.
+	 */
+	public function addComments(array $aComments);
+
+	/**
+	 * @return array
+	 */
+	public function getComments();
+
+	/**
+	 * @param array $aComments Array containing Comment objects.
+	 */
+	public function setComments(array $aComments);
+
+
+}

lib/Sabberworm/CSS/Parser.php

@@ -21,6 +21,7 @@
 use Sabberworm\CSS\Value\CSSString;
 use Sabberworm\CSS\Rule\Rule;
 use Sabberworm\CSS\Parsing\UnexpectedTokenException;
+use Sabberworm\CSS\Comment\Comment;
 
 /**
  * Parser class parses CSS from text into a data structure.
@@ -83,12 +84,12 @@ public function parse() {
 	}
 
 	private function parseDocument(Document $oDocument) {
-		$this->consumeWhiteSpace();
 		$this->parseList($oDocument, true);
 	}
 
 	private function parseList(CSSList $oList, $bIsRoot = false) {
 		while (!$this->isEnd()) {
+			$comments = $this->consumeWhiteSpace();
 			$oListItem = null;
 			if($this->oParserSettings->bLenientParsing) {
 				try {
@@ -104,9 +105,9 @@ private function parseList(CSSList $oList, $bIsRoot = false) {
 				return;
 			}
 			if($oListItem) {
+				$oListItem->setComments($comments);
 				$oList->append($oListItem);
 			}
-			$this->consumeWhiteSpace();
 		}
 		if (!$bIsRoot) {
 			throw new SourceException("Unexpected end of document", $this->iLineNo);
@@ -161,7 +162,6 @@ private function parseAtRule() {
 			$oResult = new KeyFrame($iIdentifierLineNum);
 			$oResult->setVendorKeyFrame($sIdentifier);
 			$oResult->setAnimationName(trim($this->consumeUntil('{', false, true)));
-			$this->consumeWhiteSpace();
 			$this->parseList($oResult);
 			return $oResult;
 		} else if ($sIdentifier === 'namespace') {
@@ -182,7 +182,6 @@ private function parseAtRule() {
 		} else {
 			//Unknown other at rule (font-face or such)
 			$sArgs = trim($this->consumeUntil('{', false, true));
-			$this->consumeWhiteSpace();
 			$bUseRuleSet = true;
 			foreach($this->blockRules as $sBlockRuleName) {
 				if($this->identifierIs($sIdentifier, $sBlockRuleName)) {
@@ -303,17 +302,17 @@ private function parseCharacter($bIsForIdentifier) {
 	}
 
 	private function parseSelector() {
+		$aComments = array();
 		$oResult = new DeclarationBlock($this->iLineNo);
-		$oResult->setSelector($this->consumeUntil('{', false, true));
-		$this->consumeWhiteSpace();
+		$oResult->setSelector($this->consumeUntil('{', false, true, $aComments));
+		$oResult->setComments($aComments);
 		$this->parseRuleSet($oResult);
 		return $oResult;
 	}
 
 	private function parseRuleSet($oRuleSet) {
 		while ($this->comes(';')) {
 			$this->consume(';');
-			$this->consumeWhiteSpace();
 		}
 		while (!$this->comes('}')) {
 			$oRule = null;
@@ -327,7 +326,6 @@ private function parseRuleSet($oRuleSet) {
 						if($this->streql(substr($sConsume, -1), '}')) {
 							--$this->iCurrentPosition;
 						} else {
-							$this->consumeWhiteSpace();
 							while ($this->comes(';')) {
 								$this->consume(';');
 							}
@@ -343,14 +341,15 @@ private function parseRuleSet($oRuleSet) {
 			if($oRule) {
 				$oRuleSet->addRule($oRule);
 			}
-			$this->consumeWhiteSpace();
 		}
 		$this->consume('}');
 	}
 
 	private function parseRule() {
+		$aComments = $this->consumeWhiteSpace();
 		$oRule = new Rule($this->parseIdentifier(), $this->iLineNo);
-		$this->consumeWhiteSpace();
+		$oRule->setComments($aComments);
+		$oRule->addComments($this->consumeWhiteSpace());
 		$this->consume(':');
 		$oValue = $this->parseValue(self::listDelimiterForRule($oRule->getRule()));
 		$oRule->setValue($oValue);
@@ -369,7 +368,6 @@ private function parseRule() {
 		}
 		while ($this->comes(';')) {
 			$this->consume(';');
-			$this->consumeWhiteSpace();
 		}
 		return $oRule;
 	}
@@ -568,48 +566,65 @@ private function consumeExpression($mExpression) {
 	}
 
 	private function consumeWhiteSpace() {
+		$comments = array();
 		do {
 			while (preg_match('/\\s/isSu', $this->peek()) === 1) {
 				$this->consume(1);
 			}
 			if($this->oParserSettings->bLenientParsing) {
 				try {
-					$bHasComment = $this->consumeComment();
+					$oComment = $this->consumeComment();
 				} catch(UnexpectedTokenException $e) {
 					// When we can’t find the end of a comment, we assume the document is finished.
 					$this->iCurrentPosition = $this->iLength;
 					return;
 				}
 			} else {
-				$bHasComment = $this->consumeComment();
+				$oComment = $this->consumeComment();
 			}
-		} while($bHasComment);
+			if ($oComment !== false) {
+				$comments[] = $oComment;
+			}
+		} while($oComment !== false);
+		return $comments;
 	}
 
+	/**
+	 * @return false|Comment
+	 */
 	private function consumeComment() {
+		$mComment = false;
 		if ($this->comes('/*')) {
+			$iLineNo = $this->iLineNo;
 			$this->consume(1);
-			while ($this->consume(1) !== '') {
+			$mComment = '';
+			while (($char = $this->consume(1)) !== '') {
+				$mComment .= $char;
 				if ($this->comes('*/')) {
 					$this->consume(2);
-					return true;
+					break;
 				}
 			}
 		}
-		return false;
+
+		if ($mComment !== false) {
+			// We skip the * which was included in the comment.
+			return new Comment(substr($mComment, 1), $iLineNo);
+		}
+
+		return $mComment;
 	}
 
 	private function isEnd() {
 		return $this->iCurrentPosition >= $this->iLength;
 	}
 
-	private function consumeUntil($aEnd, $bIncludeEnd = false, $consumeEnd = false) {
+	private function consumeUntil($aEnd, $bIncludeEnd = false, $consumeEnd = false, array &$comments = array()) {
 		$aEnd = is_array($aEnd) ? $aEnd : array($aEnd);
 		$out = '';
 		$start = $this->iCurrentPosition;
 
 		while (($char = $this->consume(1)) !== '') {
-			$this->consumeComment();
 			if (in_array($char, $aEnd)) {
 				if ($bIncludeEnd) {
 					$out .= $char;
@@ -619,6 +634,9 @@ private function consumeUntil($aEnd, $bIncludeEnd = false, $consumeEnd = false)
 				return $out;
 			}
 			$out .= $char;
+			if ($comment = $this->consumeComment()) {
+				$comments[] = $comment;
+			}
 		}
 
 		$this->iCurrentPosition = $start;

lib/Sabberworm/CSS/Property/AtRule.php

@@ -3,8 +3,9 @@
 namespace Sabberworm\CSS\Property;
 
 use Sabberworm\CSS\Renderable;
+use Sabberworm\CSS\Comment\Commentable;
 
-interface AtRule extends Renderable {
+interface AtRule extends Renderable, Commentable {
 	const BLOCK_RULES = 'media/document/supports/region-style/font-feature-values';
 	// Since there are more set rules than block rules, we’re whitelisting the block rules and have anything else be treated as a set rule.
 	const SET_RULES = 'font-face/counter-style/page/swash/styleset/annotation'; //…and more font-specific ones (to be used inside font-feature-values)

lib/Sabberworm/CSS/Property/CSSNamespace.php

@@ -9,11 +9,13 @@ class CSSNamespace implements AtRule {
 	private $mUrl;
 	private $sPrefix;
 	private $iLineNo;
+	protected $aComments;
 
 	public function __construct($mUrl, $sPrefix = null, $iLineNo = 0) {
 		$this->mUrl = $mUrl;
 		$this->sPrefix = $sPrefix;
 		$this->iLineNo = $iLineNo;
+		$this->aComments = array();
 	}
 
 	/**
@@ -58,4 +60,16 @@ public function atRuleArgs() {
 		}
 		return $aResult;
 	}
+
+	public function addComments(array $aComments) {
+		$this->aComments = array_merge($this->aComments, $aComments);
+	}
+
+	public function getComments() {
+		return $this->aComments;
+	}
+
+	public function setComments(array $aComments) {
+		$this->aComments = $aComments;
+	}
 }

lib/Sabberworm/CSS/Property/Charset.php

@@ -13,10 +13,12 @@ class Charset implements AtRule {
 
 	private $sCharset;
 	protected $iLineNo;
+	protected $aComment;
 
 	public function __construct($sCharset, $iLineNo = 0) {
 		$this->sCharset = $sCharset;
 		$this->iLineNo = $iLineNo;
+		$this->aComments = array();
 	}
 
 	/**
@@ -49,4 +51,16 @@ public function atRuleName() {
 	public function atRuleArgs() {
 		return $this->sCharset;
 	}
+
+	public function addComments(array $aComments) {
+		$this->aComments = array_merge($this->aComments, $aComments);
+	}
+
+	public function getComments() {
+		return $this->aComments;
+	}
+
+	public function setComments(array $aComments) {
+		$this->aComments = $aComments;
+	}
 }