Massive overhaul of Mustache.php's whitespace handling.

Move regex preparation into methods, use named subpatterns so the regex makes more sense. Fix the logic around eating newlines adjacent to tags. Update tests to reflect the changes.

Mustache.php

@@ -258,6 +258,17 @@ class Mustache {
 		return $template;
 	}
 
+	const SECTION_TYPES = '^#/';
+
+	protected function _prepareSectionRegEx($otag, $ctag) {
+		return sprintf(
+			'/(?:(?<=\\n)[ \\t]*)?%s(?<type>[%s])(?<tag_name>.+?)%s\\n?/s',
+			preg_quote($otag, '/'),
+			preg_quote(self::SECTION_TYPES, '/'),
+			preg_quote($ctag, '/')
+		);
+	}
+
 	/**
 	 * Extract a section from $template.
 	 *
@@ -268,9 +279,7 @@ class Mustache {
 	 * @return array $section, $offset, $type, $tag_name and $content
 	 */
 	protected function _findSection($template) {
-		$otag  = preg_quote($this->_otag, '/');
+		$regex = $this->_prepareSectionRegEx($this->_otag, $this->_ctag);
-		$ctag  = preg_quote($this->_ctag, '/');
-		$regex = '/' . $otag . '([\\^\\#\\/])\\s*(.+?)\\s*' . $ctag . '\\s*/ms';
 
 		$section_start = null;
 		$section_type  = null;
@@ -284,8 +293,8 @@ class Mustache {
 
 			$match    = $matches[0][0];
 			$offset   = $matches[0][1];
-			$type     = $matches[1][0];
+			$type     = $matches['type'][0];
-			$tag_name = trim($matches[2][0]);
+			$tag_name = trim($matches['tag_name'][0]);
 
 			$search_offset = $offset + strlen($match);
 
@@ -323,6 +332,14 @@ class Mustache {
 		}
 	}
 
+	protected function _preparePragmaRegEx($otag, $ctag) {
+		return sprintf(
+			'/%s%%\\s*(?<pragma_name>[\\w_-]+)(?<options_string>(?: [\\w]+=[\\w]+)*)\\s*%s\\n?/s',
+			preg_quote($otag, '/'),
+			preg_quote($ctag, '/')
+		);
+	}
+
 	/**
 	 * Initialize pragmas and remove all pragma tags.
 	 *
@@ -338,9 +355,7 @@ class Mustache {
 			return $template;
 		}
 
-		$otag = preg_quote($this->_otag, '/');
+		$regex = $this->_preparePragmaRegEx($this->_otag, $this->_ctag);
-		$ctag = preg_quote($this->_ctag, '/');
-		$regex = '/' . $otag . '%\\s*([\\w_-]+)((?: [\\w]+=[\\w]+)*)\\s*' . $ctag . '\\n?/s';
 		return preg_replace_callback($regex, array($this, '_renderPragma'), $template);
 	}
 
@@ -354,8 +369,8 @@ class Mustache {
 	 */
 	protected function _renderPragma($matches) {
 		$pragma         = $matches[0];
-		$pragma_name    = $matches[1];
+		$pragma_name    = $matches['pragma_name'];
-		$options_string = $matches[2];
+		$options_string = $matches['options_string'];
 
 		if (!in_array($pragma_name, $this->_pragmasImplemented)) {
 			throw new MustacheException('Unknown pragma: ' . $pragma_name, MustacheException::UNKNOWN_PRAGMA);
@@ -422,6 +437,17 @@ class Mustache {
 		return (isset($this->_throwsExceptions[$exception]) && $this->_throwsExceptions[$exception]);
 	}
 
+	const TAG_TYPES = '#^/=!<>\\{&';
+
+	protected function _prepareTagRegex($otag, $ctag) {
+		return sprintf(
+			'/(?:(?<=\\n)[ \\t]*)?%s(?<type>[%s]?)(?<tag_name>.+?)(?:\\1|})?%s(?:\\s*(?=\\n))?/s',
+			preg_quote($otag, '/'),
+			preg_quote(self::TAG_TYPES, '/'),
+			preg_quote($ctag, '/')
+		);
+	}
+
 	/**
 	 * Loop through and render individual Mustache tags.
 	 *
@@ -437,22 +463,25 @@ class Mustache {
 		$otag_orig = $this->_otag;
 		$ctag_orig = $this->_ctag;
 
-		$otag = preg_quote($this->_otag, '/');
+		$this->_tagRegEx = $this->_prepareTagRegEx($this->_otag, $this->_ctag);
-		$ctag = preg_quote($this->_ctag, '/');
-
-		$this->_tagRegEx = '/' . $otag . "([#\^\/=!<>\\{&])?(.+?)\\1?" . $ctag . "+/s";
 
 		$html = '';
 		$matches = array();
 		while (preg_match($this->_tagRegEx, $template, $matches, PREG_OFFSET_CAPTURE)) {
 			$tag      = $matches[0][0];
 			$offset   = $matches[0][1];
-			$modifier = $matches[1][0];
+			$modifier = $matches['type'][0];
-			$tag_name = trim($matches[2][0]);
+			$tag_name = trim($matches['tag_name'][0]);
 
 			$html .= substr($template, 0, $offset);
+
+			$next_offset = $offset + strlen($tag);
+			if ((substr($html, -1) == "\n") && (substr($template, $next_offset, 1) == "\n")) {
+				$next_offset++;
+			}
+			$template = substr($template, $next_offset);
+
 			$html .= $this->_renderTag($modifier, $tag_name);
-			$template = substr($template, $offset + strlen($tag));
 		}
 
 		$this->_otag = $otag_orig;
@@ -501,6 +530,10 @@ class Mustache {
 				return $this->_renderPartial($tag_name);
 				break;
 			case '{':
+				// strip the trailing } ...
+				if ($tag_name[(strlen($tag_name) - 1)] == '}') {
+					$tag_name = substr($tag_name, 0, -1);
+				}
 			case '&':
 				if ($this->_hasPragma(self::PRAGMA_UNESCAPED)) {
 					return $this->_renderEscaped($tag_name);
@@ -571,13 +604,12 @@ class Mustache {
 	 * @return string
 	 */
 	protected function _changeDelimiter($tag_name) {
-		$tags = explode(' ', $tag_name);
+		list($otag, $ctag) = explode(' ', $tag_name);
-		$this->_otag = $tags[0];
+		$this->_otag = $otag;
-		$this->_ctag = $tags[1];
+		$this->_ctag = $ctag;
+
+		$this->_tagRegEx = $this->_prepareTagRegEx($this->_otag, $this->_ctag);
 
-		$otag  = preg_quote($this->_otag, '/');
-		$ctag  = preg_quote($this->_ctag, '/');
-		$this->_tagRegEx = '/' . $otag . "([#\^\/=!<>\\{&])?(.+?)\\1?" . $ctag . "+/s";
 		return '';
 	}
 

README.markdown

@@ -83,7 +83,6 @@ Known Issues
 
  * Sections don't respect delimiter changes -- `delimiters` example currently fails with an
    "unclosed section" exception.
- * Mustache isn't always very good at whitespace.
 
 
 See Also

examples/complex/complex.mustache

@@ -1,16 +1,16 @@
 <h1>{{header}}</h1>
 {{#notEmpty}}
-  <ul>
+<ul>
-  {{#item}}
+{{#item}}
-    {{#current}}
+{{#current}}
-      <li><strong>{{name}}</strong></li>
+    <li><strong>{{name}}</strong></li>
-    {{/current}}
+{{/current}}
-    {{^current}}
+{{^current}}
-      <li><a href="{{url}}">{{name}}</a></li>
+    <li><a href="{{url}}">{{name}}</a></li>
-    {{/current}}
+{{/current}}
-  {{/item}}
+{{/item}}
-  </ul>
+</ul>
 {{/notEmpty}}
 {{#isEmpty}}
-  <p>The list is empty.</p>
+<p>The list is empty.</p>
 {{/isEmpty}}

examples/complex/complex.txt

@@ -1,6 +1,6 @@
 <h1>Colors</h1>
 <ul>
-  <li><strong>red</strong></li>
+    <li><strong>red</strong></li>
     <li><a href="#Green">green</a></li>
     <li><a href="#Blue">blue</a></li>
-    </ul>
+</ul>

examples/double_section/double_section.mustache

@@ -1,7 +1,7 @@
 {{#t}}
-  * first
+* first
 {{/t}}
 * {{two}}
 {{#t}}
-  * third
+* third
 {{/t}}

examples/grand_parent_context/grand_parent_context.mustache

@@ -1,10 +1,7 @@
 {{grand_parent_id}}
 {{#parent_contexts}}
-{{grand_parent_id}}
+  {{parent_id}} ({{grand_parent_id}})
-{{parent_id}}
+  {{#child_contexts}}
-{{#child_contexts}}
+    {{child_id}} ({{parent_id}} << {{grand_parent_id}})
-{{grand_parent_id}}
+  {{/child_contexts}}
-{{parent_id}}
-{{child_id}}
-{{/child_contexts}}
 {{/parent_contexts}}

examples/grand_parent_context/grand_parent_context.txt

@@ -1,17 +1,7 @@
 grand_parent1
-grand_parent1
+  parent1 (grand_parent1)
-parent1
+    parent1-child1 (parent1 << grand_parent1)
-grand_parent1
+    parent1-child2 (parent1 << grand_parent1)
-parent1
+  parent2 (grand_parent1)
-parent1-child1
+    parent2-child1 (parent2 << grand_parent1)
-grand_parent1
+    parent2-child2 (parent2 << grand_parent1)
-parent1
-parent1-child2
-grand_parent1
-parent2
-grand_parent1
-parent2
-parent2-child1
-grand_parent1
-parent2
-parent2-child2

examples/inverted_double_section/inverted_double_section.mustache

@@ -1,7 +1,7 @@
 {{^t}}
-  * first
+* first
 {{/t}}
 * {{two}}
 {{^t}}
-  * third
+* third
 {{/t}}

examples/sections_spaces/SectionsSpaces.php

@@ -1,14 +0,0 @@
-<?php
-
-class SectionsSpaces extends Mustache {
-	public $start = "It worked the first time.";
-
-	public function middle() {
-		return array(
-			array('item' => "And it worked the second time."),
-			array('item' => "As well as the third."),
-		);
-	}
-
-	public $final = "Then, surprisingly, it worked the final time.";
-}

examples/sections_spaces/sections_spaces.mustache

@@ -1,9 +0,0 @@
- * {{ start }}
-{{# middle }}
- * {{ item }}
-{{/ middle }}
- * {{ final }}
-
- * {{ start }}
-{{# middle }} * {{ item }}{{/ middle }}
- * {{ final }}

examples/sections_spaces/sections_spaces.txt

@@ -1,9 +0,0 @@
- * It worked the first time.
- * And it worked the second time.
- * As well as the third.
- * Then, surprisingly, it worked the final time.
-
- * It worked the first time.
- * And it worked the second time.
- * As well as the third.
- * Then, surprisingly, it worked the final time.

test/MustachePragmaImplicitIteratorTest.php

@@ -11,7 +11,11 @@ class MustachePragmaImplicitIteratorTest extends PHPUnit_Framework_TestCase {
 		$m = $this->getMock('Mustache', array('_renderPragma'), array('{{%IMPLICIT-ITERATOR}}'));
 		$m->expects($this->exactly(1))
 			->method('_renderPragma')
-			->with(array('{{%IMPLICIT-ITERATOR}}', 'IMPLICIT-ITERATOR', null));
+			->with(array(
+				0 => '{{%IMPLICIT-ITERATOR}}',
+				1 => 'IMPLICIT-ITERATOR', 'pragma_name' => 'IMPLICIT-ITERATOR',
+				2 => null, 'options_string' => null
+			));
 		$m->render();
 	}
 

test/MustacheTest.php

@@ -37,7 +37,6 @@ class MustacheTest extends PHPUnit_Framework_TestCase {
 
 	protected $knownIssues = array(
 		'Delimiters'     => "Known issue: sections don't respect delimiter changes",
-		'SectionsSpaces' => "Known issue: Mustache fails miserably at whitespace",
 	);
 
 	/**