Move parser settings to external settings class

Author: sabberworm

lib/Sabberworm/CSS/Parser.php

@@ -18,6 +18,7 @@
 use Sabberworm\CSS\Value\URL;
 use Sabberworm\CSS\Value\String;
 use Sabberworm\CSS\Rule\Rule;
+use Sabberworm\CSS\Parsing\UnexpectedTokenException;
 
 /**
  * Parser class parses CSS from text into a data structure.
@@ -26,19 +27,17 @@ class Parser {
 
 	private $sText;
 	private $iCurrentPosition;
+	private $oParserSettings;
+	private $sCharset;
 	private $iLength;
 
-	/**
-	 * Should we use mb_* string functions
-	 *
-	 * @var bool
-	 */
-	private $bUseMbFunctions = TRUE;
-
-	public function __construct($sText, $sDefaultCharset = 'utf-8') {
+	public function __construct($sText, Settings $oParserSettings = null) {
 		$this->sText = $sText;
 		$this->iCurrentPosition = 0;
-		$this->setCharset($sDefaultCharset);
+		if ($oParserSettings === null) {
+			$oParserSettings = Settings::create();
+		}
+		$this->oParserSettings = $oParserSettings;
 	}
 
 	public function setCharset($sCharset) {
@@ -51,15 +50,12 @@ public function getCharset() {
 	}
 
 	public function parse() {
+		$this->setCharset($this->oParserSettings->sDefaultCharset);
 		$oResult = new Document();
 		$this->parseDocument($oResult);
 		return $oResult;
 	}
 
-	public function setUseMbFlag($bFlag) {
-		$this->bUseMbFunctions = (bool) $bFlag;
-	}
-
 	private function parseDocument(Document $oDocument) {
 		$this->consumeWhiteSpace();
 		$this->parseList($oDocument, true);
@@ -112,7 +108,7 @@ private function parseAtRule() {
 			$this->consume(';');
 			$this->setCharset($sCharset->getString());
 			return new Charset($sCharset);
-		} else if(preg_match('/^(-\\w+-)?keyframes$/', $sIdentifier) === 1) {
+		} else if (preg_match('/^(-\\w+-)?keyframes$/', $sIdentifier) === 1) {
 			$oResult = new KeyFrame();
 			$oResult->setVendorKeyFrame($sIdentifier);
 			$oResult->setAnimationName(trim($this->consumeUntil('{')));
@@ -123,15 +119,15 @@ private function parseAtRule() {
 		} else if ($sIdentifier === 'namespace') {
 			$sPrefix = null;
 			$mUrl = $this->parsePrimitiveValue();
-			if(!$this->comes(';')) {
+			if (!$this->comes(';')) {
 				$sPrefix = $mUrl;
 				$mUrl = $this->parsePrimitiveValue();
 			}
 			$this->consume(';');
-			if($sPrefix !== null && !is_string($sPrefix)) {
+			if ($sPrefix !== null && !is_string($sPrefix)) {
 				throw new \Exception('Wrong namespace prefix '.$sPrefix);
 			}
-			if(!($mUrl instanceof String || $mUrl instanceof URL)) {
+			if (!($mUrl instanceof String || $mUrl instanceof URL)) {
 				throw new \Exception('Wrong namespace url of invalid type '.$mUrl);
 			}
 			return new CSSNamespace($mUrl, $sPrefix);
@@ -148,9 +144,9 @@ private function parseAtRule() {
 	private function parseIdentifier($bAllowFunctions = true) {
 		$sResult = $this->parseCharacter(true);
 		if ($sResult === null) {
-			throw new \Exception("Identifier expected, got {$this->peek(5)}");
+			throw new UnexpectedTokenException($sResult, $this->peek(5), 'identifier');
 		}
-		$sCharacter;
+		$sCharacter = null;
 		while (($sCharacter = $this->parseCharacter(true)) !== null) {
 			$sResult .= $sCharacter;
 		}
@@ -265,7 +261,7 @@ private function parseRule() {
 			$this->consume('!');
 			$this->consumeWhiteSpace();
 			$sImportantMarker = $this->consume(strlen('important'));
-			if (mb_convert_case($sImportantMarker, MB_CASE_LOWER) !== 'important') {
+			if (mb_convert_case($sImportantMarker, MB_CASE_LOWER, $this->sCharset) !== 'important') {
 				throw new \Exception("! was followed by “" . $sImportantMarker . "”. Expected “important”");
 			}
 			$oRule->setIsImportant(true);
@@ -451,20 +447,25 @@ private function peek($iLength = 1, $iOffset = 0) {
 		if (is_string($iOffset)) {
 			$iOffset = $this->strlen($iOffset);
 		}
-		return $this->substr($this->sText, $this->iCurrentPosition + $iOffset, $iLength);
+		$iOffset = $this->iCurrentPosition + $iOffset;
+		if ($iOffset >= $this->iLength) {
+			return '';
+		}
+		$iLength = min($iLength, $this->iLength-$iOffset);
+		return $this->substr($this->sText, $iOffset, $iLength);
 	}
 
 	private function consume($mValue = 1) {
 		if (is_string($mValue)) {
 			$iLength = $this->strlen($mValue);
 			if ($this->substr($this->sText, $this->iCurrentPosition, $iLength) !== $mValue) {
-				throw new \Exception("Expected $mValue, got " . $this->peek(5));
+				throw new UnexpectedTokenException($mValue, $this->peek(max($iLength, 5)));
 			}
 			$this->iCurrentPosition += $this->strlen($mValue);
 			return $mValue;
 		} else {
 			if ($this->iCurrentPosition + $mValue > $this->iLength) {
-				throw new \Exception("Tried to consume $mValue chars, exceeded file end");
+				throw new UnexpectedTokenException($mValue, $this->peek(5), 'count');
 			}
 			$sResult = $this->substr($this->sText, $this->iCurrentPosition, $mValue);
 			$this->iCurrentPosition += $mValue;
@@ -477,7 +478,7 @@ private function consumeExpression($mExpression) {
 		if (preg_match($mExpression, $this->inputLeft(), $aMatches, PREG_OFFSET_CAPTURE) === 1) {
 			return $this->consume($aMatches[0][0]);
 		}
-		throw new \Exception("Expected pattern $mExpression not found, got: {$this->peek(5)}");
+		throw new UnexpectedTokenException($mExpression, $this->peek(5), 'expression');
 	}
 
 	private function consumeWhiteSpace() {
@@ -504,7 +505,7 @@ private function isEnd() {
 	private function consumeUntil($sEnd) {
 		$iEndPos = mb_strpos($this->sText, $sEnd, $this->iCurrentPosition, $this->sCharset);
 		if ($iEndPos === false) {
-			throw new \Exception("Required $sEnd not found, got {$this->peek(5)}");
+			throw new UnexpectedTokenException($sEnd, $this->peek(5), 'search');
 		}
 		return $this->consume($iEndPos - $this->iCurrentPosition);
 	}
@@ -514,15 +515,15 @@ private function inputLeft() {
 	}
 
 	private function substr($string, $start, $length) {
-		if ($this->bUseMbFunctions) {
+		if ($this->oParserSettings->bMultibyteSupport) {
 			return mb_substr($string, $start, $length, $this->sCharset);
 		} else {
 			return substr($string, $start, $length);
 		}
 	}
 
 	private function strlen($text) {
-		if ($this->bUseMbFunctions) {
+		if ($this->oParserSettings->bMultibyteSupport) {
 			return mb_strlen($text, $this->sCharset);
 		} else {
 			return strlen($text);

lib/Sabberworm/CSS/Parsing/OutputFormat.php

@@ -0,0 +1,55 @@
+<?php
+
+namespace Sabberworm\CSS;
+
+class OutputFormat {
+	public $sStringQuotingType = '"';
+	public $bNewlinesAfterRule = false;
+	public $bNewlinesStartingRuleSets = false;
+	public $sIndentationLevel = 0;
+	public $sIndentation = "\t";
+	
+	private $oFormatter = null;
+	
+	public function __construct() {
+	}
+	
+	public function nextLevel() {
+		$result = clone $this;
+		$result->sIndentationLevel++;
+		$result->oFormatter = null;
+		return $result;
+	}
+	
+	public static function create() {
+		return new OutputFormatter();
+	}
+	
+	public function getFormatter() {
+		if($this->oFormatter === null) {
+			$this->oFormatter = new OutputFormatter($this);
+		}
+		return $this->oFormatter;
+	}
+}
+
+class OutputFormatter {
+	private $oFormat;
+	
+	public function __construct(OutputFormat $oFormat) {
+		$this->oFormat = $oFormat;
+	}
+	
+	public function newline($bOutput = false) {
+		if($bOutput) {
+			return $this->indent()."\n";
+		}
+		return '';
+	}
+	
+	public function indent() {
+		return str_repeat($this->sIndentation, $this->sIndentationLevel);
+	}
+	
+	
+}

lib/Sabberworm/CSS/Parsing/UnexpectedTokenException.php

@@ -0,0 +1,28 @@
+<?php
+
+namespace Sabberworm\CSS\Parsing;
+
+/**
+* Thrown if the CSS parsers encounters a token it did not expect
+*/
+class UnexpectedTokenException extends \Exception {
+	private $sExpected;
+	private $sFound;
+	// Possible values: literal, identifier, count, expression, search
+	private $sMatchType;
+
+	public function __construct($sExpected, $sFound, $sMatchType = 'literal') {
+		$this->sExpected = $sExpected;
+		$this->sFound = $sFound;
+		$this->sMatchType = $sMatchType;
+		$sMessage = "Token “{$sExpected}” ({$sMatchType}) not found. Got “{$sFound}”.";
+		if($this->sMatchType === 'search') {
+			$sMessage = "Search for “{$sExpected}” returned no results. Context: “{$sFound}”.";
+		} else if($this->sMatchType === 'count') {
+			$sMessage = "Next token was expected to have {$sExpected} chars. Context: “{$sFound}”.";
+		} else if($this->sMatchType === 'identifier') {
+			$sMessage = "Identifier expected. Got “{$sFound}”";
+		}
+		parent::__construct($sMessage);
+	}
+}

lib/Sabberworm/CSS/Settings.php

@@ -0,0 +1,52 @@
+<?php
+
+namespace Sabberworm\CSS;
+
+use Sabberworm\CSS\Rule\Rule;
+
+/**
+ * Parser settings class.
+ *
+ * Configure parser behaviour here.
+ */
+class Settings {
+	/**
+	* Multi-byte string support. If true (default), will use (slower) mb_strlen, mb_convert_case, mb_substr and mb_strpos functions. Otherwise, the normal (ASCII-Only) functions will be used.
+	*/
+	public $bMultibyteSupport = true;
+
+	/**
+	* The default charset for the CSS if no `@charset` rule is found. Defaults to utf-8.
+	*/
+	public $sDefaultCharset = 'utf-8';
+
+	/**
+	* Lenient parsing. When used, the parser will not choke on unexpected tokens but simply ignore them.
+	*/
+	public $bLenientParsing = true;
+
+	private function __construct() {}
+
+	public static function create() {
+		return new Settings();
+	}
+	
+	public function withMultibyteSupport($bMultibyteSupport = true) {
+		$this->bMultibyteSupport = $bMultibyteSupport;
+		return $this;
+	}
+	
+	public function withDefaultCharset($sDefaultCharset) {
+		$this->sDefaultCharset = $sDefaultCharset;
+		return $this;
+	}
+	
+	public function withLenientParsing($bLenientParsing = true) {
+		$this->bLenientParsing = $bLenientParsing;
+		return $this;
+	}
+	
+	public function beStrict() {
+		return $this->withLenientParsing(false);
+	}
+}