add unit tests for correct pattern selection when patterns contain...

add unit tests for correct pattern selection when patterns contain non-captured elements (e.g. boundaries, lookaheads & lookbehinds)

darcs-hash:20090419134450-f07c6-4ff7d226fcba002c840828336e73fb89cf48e3db.gz

_test/cases/inc/parser/lexer.test.php

@@ -446,7 +446,7 @@ class TestOfLexerByteIndices extends UnitTestCase {
 		$handler->expectCallCount("caught", 5);
         
 		$lexer = &new Doku_Lexer($handler, "ignore");
-		$lexer->addEntryPattern('<file>(?=.*\x3C/file\x3E)', "ignore", "caught");
+		$lexer->addEntryPattern('<file>(?=.*</file>)', "ignore", "caught");
 		$lexer->addExitPattern("</file>", "caught");
         $lexer->addSpecialPattern('b','caught','special');
         $lexer->mapHandler('special','caught');
@@ -590,7 +590,36 @@ class TestOfLexerByteIndices extends UnitTestCase {
 		$this->assertTrue($lexer->parse($doc));
 		$handler->tally();
 	}
-    
+
+    /**
+     * This test is primarily to ensure the correct match is chosen
+     * when there are non-captured elements in the pattern.
+     */
+    function testIndexSelectCorrectMatch() {
+        $doc = "ALL FOOLS ARE FOO";
+        $pattern = '\bFOO\b';
+
+        $handler = &new MockTestParserByteIndex($this);
+        $handler->setReturnValue("ignore", true);
+        $handler->setReturnValue("caught", true);
+
+        $matches = array();
+        preg_match('/'.$pattern.'/',$doc,$matches,PREG_OFFSET_CAPTURE);
+
+        $handler->expectArgumentsAt(
+            0,
+            "caught",
+            array("FOO", DOKU_LEXER_SPECIAL, $matches[0][1])
+            );
+        $handler->expectCallCount("caught", 1);
+
+        $lexer = &new Doku_Lexer($handler, "ignore");
+        $lexer->addSpecialPattern($pattern,'ignore','caught');
+
+        $this->assertTrue($lexer->parse($doc));
+        $handler->tally();
+    }
+
 }
 
 ?>

_test/cases/inc/parser/parser.inc.php

@@ -14,6 +14,7 @@ require_once DOKU_INC . 'inc/parser/parser.php';
 require_once DOKU_INC . 'inc/parser/handler.php';
 require_once DOKU_INC . 'inc/events.php';
 require_once DOKU_INC . 'inc/mail.php';
+
 //require_once DOKU . 'parser/renderer.php';
 
 //Mock::generate('Doku_Renderer');

_test/cases/inc/parser/parser_replacements.test.php

@@ -40,6 +40,23 @@ class TestOfDoku_Parser_Replacements extends TestOfDoku_Parser {
         $this->assertEqual(array_map('stripbyteindex',$this->H->calls),$calls);
     }
 
+    function testPickAcronymCorrectly() {
+        $this->P->addMode('acronym',new Doku_Parser_Mode_Acronym(array('FOO')));
+        $this->P->parse('ALL FOOLS ARE FOO');
+
+        $calls = array (
+            array('document_start',array()),
+            array('p_open',array()),
+            array('cdata',array("\n".'ALL FOOLS ARE ')),
+            array('acronym',array('FOO')),
+            array('cdata',array("\n")),
+            array('p_close',array()),
+            array('document_end',array()),
+        );
+
+        $this->assertEqual(array_map('stripbyteindex',$this->H->calls),$calls);
+    }
+
     function testMultipleAcronyms() {
         $this->P->addMode('acronym',new Doku_Parser_Mode_Acronym(array('FOO','BAR')));
         $this->P->parse('abc FOO def BAR xyz');

inc/parser/lexer.php

@@ -136,10 +136,9 @@ class Doku_LexerParallelRegex {
         }
 
         $idx = count($matches)-2;
-
         list($pre, $post) = preg_split($this->_patterns[$idx].$this->_getPerlMatchingFlags(), $subject, 2);
-
         $split = array($pre, $matches[0], $post);
+
         return isset($this->_labels[$idx]) ? $this->_labels[$idx] : true;
     }