Parser: macros are parsed in two phases [Closes nette/nette#711]

src/Latte/Parser.php

@@ -36,8 +36,8 @@ class Parser extends Object
 		'off' => array('[^\x00-\xFF]', ''),
 	);
 
-	/** @var string */
-	private $macroRe;
+	/** @var string[] */
+	private $delimiters;
 
 	/** @var string source template */
 	private $input;
@@ -66,7 +66,8 @@ class Parser extends Object
 		CONTEXT_HTML_TAG = 'htmlTag',
 		CONTEXT_HTML_ATTRIBUTE = 'htmlAttribute',
 		CONTEXT_RAW = 'raw',
-		CONTEXT_HTML_COMMENT = 'htmlComment';
+		CONTEXT_HTML_COMMENT = 'htmlComment',
+		CONTEXT_MACRO = 'macro';
 
 
 	/**
@@ -92,19 +93,9 @@ public function parse($input)
 		$this->lastHtmlTag = $this->syntaxEndTag = NULL;
 
 		while ($this->offset < strlen($input)) {
-			$matches = $this->{"context".$this->context[0]}();
-
-			if (!$matches) { // EOF
+			if ($this->{"context".$this->context[0]}() === FALSE) {
 				break;
-
-			} elseif (!empty($matches['comment'])) { // {* *}
-				$this->addToken(Token::COMMENT, $matches[0]);
-
-			} elseif (!empty($matches['macro'])) { // {macro}
-				$token = $this->addToken(Token::MACRO_TAG, $matches[0]);
-				list($token->name, $token->value, $token->modifiers, $token->empty) = $this->parseMacroTag($matches['macro']);
 			}
-
 			$this->filter();
 		}
 
@@ -123,7 +114,7 @@ private function contextHtmlText()
 		$matches = $this->match('~
 			(?:(?<=\n|^)[ \t]*)?<(?P<closing>/?)(?P<tag>[a-z0-9:]+)|  ##  begin of HTML tag <tag </tag - ignores <!DOCTYPE
 			<(?P<htmlcomment>!--(?!>))|     ##  begin of HTML comment <!--, but not <!-->
-			'.$this->macroRe.'              ##  macro tag
+			(?P<macro>' . $this->delimiters[0] . ')
 		~xsi');
 
 		if (!empty($matches['htmlcomment'])) { // <!--
@@ -136,8 +127,10 @@ private function contextHtmlText()
 			$token->closing = (bool) $matches['closing'];
 			$this->lastHtmlTag = $matches['closing'] . strtolower($matches['tag']);
 			$this->setContext(self::CONTEXT_HTML_TAG);
+
+		} else {
+			return $this->processMacro($matches);
 		}
-		return $matches;
 	}
 
 
@@ -147,8 +140,8 @@ private function contextHtmlText()
 	private function contextCData()
 	{
 		$matches = $this->match('~
-			</(?P<tag>'.$this->lastHtmlTag.')(?![a-z0-9:])| ##  end HTML tag </tag
-			'.$this->macroRe.'              ##  macro tag
+			</(?P<tag>' . $this->lastHtmlTag . ')(?![a-z0-9:])| ##  end HTML tag </tag
+			(?P<macro>' . $this->delimiters[0] . ')
 		~xsi');
 
 		if (!empty($matches['tag'])) { // </tag
@@ -157,8 +150,9 @@ private function contextCData()
 			$token->closing = TRUE;
 			$this->lastHtmlTag = '/' . $this->lastHtmlTag;
 			$this->setContext(self::CONTEXT_HTML_TAG);
+		} else {
+			return $this->processMacro($matches);
 		}
-		return $matches;
 	}
 
 
@@ -169,7 +163,7 @@ private function contextHtmlTag()
 	{
 		$matches = $this->match('~
 			(?P<end>\ ?/?>)([ \t]*\n)?|  ##  end of HTML tag
-			'.$this->macroRe.'|          ##  macro tag
+			(?P<macro>' . $this->delimiters[0] . ')|
 			\s*(?P<attr>[^\s/>={]+)(?:\s*=\s*(?P<value>["\']|[^\s/>{]+))? ## beginning of HTML attribute
 		~xsi');
 
@@ -193,8 +187,9 @@ private function contextHtmlTag()
 					$this->setContext(self::CONTEXT_HTML_ATTRIBUTE, $matches['value']);
 				}
 			}
+		} else {
+			return $this->processMacro($matches);
 		}
-		return $matches;
 	}
 
 
@@ -204,15 +199,16 @@ private function contextHtmlTag()
 	private function contextHtmlAttribute()
 	{
 		$matches = $this->match('~
-			(?P<quote>'.$this->context[1].')|  ##  end of HTML attribute
-			'.$this->macroRe.'                 ##  macro tag
+			(?P<quote>' . $this->context[1] . ')|  ##  end of HTML attribute
+			(?P<macro>' . $this->delimiters[0] . ')
 		~xsi');
 
 		if (!empty($matches['quote'])) { // (attribute end) '"
 			$this->addToken(Token::TEXT, $matches[0]);
 			$this->setContext(self::CONTEXT_HTML_TAG);
+		} else {
+			return $this->processMacro($matches);
 		}
-		return $matches;
 	}
 
 
@@ -223,14 +219,15 @@ private function contextHtmlComment()
 	{
 		$matches = $this->match('~
 			(?P<htmlcomment>-->)|   ##  end of HTML comment
-			'.$this->macroRe.'      ##  macro tag
+			(?P<macro>' . $this->delimiters[0] . ')
 		~xsi');
 
 		if (!empty($matches['htmlcomment'])) { // -->
 			$this->addToken(Token::HTML_TAG_END, $matches[0]);
 			$this->setContext(self::CONTEXT_HTML_TEXT);
+		} else {
+			return $this->processMacro($matches);
 		}
-		return $matches;
 	}
 
 
@@ -240,9 +237,50 @@ private function contextHtmlComment()
 	private function contextRaw()
 	{
 		$matches = $this->match('~
-			'.$this->macroRe.'     ##  macro tag
+			(?P<macro>' . $this->delimiters[0] . ')
 		~xsi');
-		return $matches;
+		return $this->processMacro($matches);
+	}
+
+
+	/**
+	 * Handles CONTEXT_MACRO.
+	 */
+	private function contextMacro()
+	{
+		$matches = $this->match('~
+			(?P<comment>\\*.*?\\*' . $this->delimiters[1] . '\n{0,2})|
+			(?P<macro>(?:
+				' . self::RE_STRING . '|
+				\{(?:' . self::RE_STRING . '|[^\'"{}])*+\}|
+				[^\'"{}]
+			)+?)
+			' . $this->delimiters[1] . '
+			(?P<rmargin>[ \t]*(?=\n))?
+		~xsiA');
+
+		if (!empty($matches['macro'])) {
+			$token = $this->addToken(Token::MACRO_TAG, $this->context[1][1] . $matches[0]);
+			list($token->name, $token->value, $token->modifiers, $token->empty) = $this->parseMacroTag($matches['macro']);
+			$this->context = $this->context[1][0];
+
+		} elseif (!empty($matches['comment'])) {
+			$this->addToken(Token::COMMENT, $this->context[1][1] . $matches[0]);
+			$this->context = $this->context[1][0];
+
+		} else {
+			throw new CompileException("Malformed macro on line {$this->getLine()}.");
+		}
+	}
+
+
+	private function processMacro($matches)
+	{
+		if (!empty($matches['macro'])) { // {macro} or {* *}
+			$this->setContext(self::CONTEXT_MACRO, array($this->context, $matches['macro']));
+		} else {
+			return FALSE;
+		}
 	}
 
 
@@ -325,17 +363,7 @@ public function setSyntax($type)
 	 */
 	public function setDelimiters($left, $right)
 	{
-		$this->macroRe = '
-			(?P<comment>' . $left . '\\*.*?\\*' . $right . '\n{0,2})|
-			' . $left . '
-				(?P<macro>(?:
-					' . self::RE_STRING . '|
-					\{(?:' . self::RE_STRING . '|[^\'"{}])*+\}|
-					[^\'"{}]
-				)+?)
-			' . $right . '
-			(?P<rmargin>[ \t]*(?=\n))?
-		';
+		$this->delimiters = array($left, $right);
 		return $this;
 	}
 
@@ -396,7 +424,9 @@ private function getLine()
 	protected function filter()
 	{
 		$token = end($this->output);
-		if ($token->type === Token::MACRO_TAG && $token->name === '/syntax') {
+		if (!$token) {
+
+		} elseif ($token->type === Token::MACRO_TAG && $token->name === '/syntax') {
 			$this->setSyntax($this->defaultSyntax);
 			$token->type = Token::COMMENT;
 

tests/Latte/Parser.errors.phpt

@@ -0,0 +1,29 @@
+<?php
+
+/**
+ * Test: Latte\Parser errors.
+ */
+
+use Tester\Assert,
+	Latte\Parser;
+
+
+require __DIR__ . '/../bootstrap.php';
+
+
+Assert::exception(function() {
+	$parser = new Parser;
+	$parser->parse("\xA0\xA0");
+}, 'InvalidArgumentException', 'Template is not valid UTF-8 stream.');
+
+
+Assert::exception(function() {
+	$parser = new Parser;
+	$parser->parse("{var \n'abc}");
+}, 'Latte\CompileException', 'Malformed macro on line 1.');
+
+
+Assert::exception(function() {
+	$parser = new Parser;
+	$parser->parse("\n{* \n'abc}");
+}, 'Latte\CompileException', 'Malformed macro on line 2.');

tests/Latte/Parser.parseMacroTag.phpt

@@ -10,7 +10,7 @@ use Tester\Assert;
 require __DIR__ . '/../bootstrap.php';
 
 
-$parser = new Latte\Parser();
+$parser = new Latte\Parser;
 
 
 Assert::same( array('?', 'echo', '', FALSE), $parser->parseMacroTag('? echo') );