parser_test.php.svn-base

来自「PHP 知识管理系统（基于树结构的知识管理系统）, 英文原版的PHP源码。」· SVN-BASE 代码 · 共 326 行

<?php
// $Id: parser_test.php 1608 2007-12-27 09:03:07Z pp11 $
require_once(dirname(__FILE__) . '/../autorun.php');
require_once(dirname(__FILE__) . '/../parser.php');
Mock::generate('SimpleHtmlSaxParser');
Mock::generate('SimpleSaxListener');

class TestOfParallelRegex extends UnitTestCase {
    
    function testNoPatterns() {
        $regex = &new ParallelRegex(false);
        $this->assertFalse($regex->match("Hello", $match));
        $this->assertEqual($match, "");
    }
    
    function testNoSubject() {
        $regex = &new ParallelRegex(false);
        $regex->addPattern(".*");
        $this->assertTrue($regex->match("", $match));
        $this->assertEqual($match, "");
    }
    
    function testMatchAll() {
        $regex = &new ParallelRegex(false);
        $regex->addPattern(".*");
        $this->assertTrue($regex->match("Hello", $match));
        $this->assertEqual($match, "Hello");
    }
    
    function testCaseSensitive() {
        $regex = &new ParallelRegex(true);
        $regex->addPattern("abc");
        $this->assertTrue($regex->match("abcdef", $match));
        $this->assertEqual($match, "abc");
        $this->assertTrue($regex->match("AAABCabcdef", $match));
        $this->assertEqual($match, "abc");
    }
    
    function testCaseInsensitive() {
        $regex = &new ParallelRegex(false);
        $regex->addPattern("abc");
        $this->assertTrue($regex->match("abcdef", $match));
        $this->assertEqual($match, "abc");
        $this->assertTrue($regex->match("AAABCabcdef", $match));
        $this->assertEqual($match, "ABC");
    }
    
    function testMatchMultiple() {
        $regex = &new ParallelRegex(true);
        $regex->addPattern("abc");
        $regex->addPattern("ABC");
        $this->assertTrue($regex->match("abcdef", $match));
        $this->assertEqual($match, "abc");
        $this->assertTrue($regex->match("AAABCabcdef", $match));
        $this->assertEqual($match, "ABC");
        $this->assertFalse($regex->match("Hello", $match));
    }
    
    function testPatternLabels() {
        $regex = &new ParallelRegex(false);
        $regex->addPattern("abc", "letter");
        $regex->addPattern("123", "number");
        $this->assertIdentical($regex->match("abcdef", $match), "letter");
        $this->assertEqual($match, "abc");
        $this->assertIdentical($regex->match("0123456789", $match), "number");
        $this->assertEqual($match, "123");
    }
}

class TestOfStateStack extends UnitTestCase {
    
    function testStartState() {
        $stack = &new SimpleStateStack("one");
        $this->assertEqual($stack->getCurrent(), "one");
    }
    
    function testExhaustion() {
        $stack = &new SimpleStateStack("one");
        $this->assertFalse($stack->leave());
    }
    
    function testStateMoves() {
        $stack = &new SimpleStateStack("one");
        $stack->enter("two");
        $this->assertEqual($stack->getCurrent(), "two");
        $stack->enter("three");
        $this->assertEqual($stack->getCurrent(), "three");
        $this->assertTrue($stack->leave());
        $this->assertEqual($stack->getCurrent(), "two");
        $stack->enter("third");
        $this->assertEqual($stack->getCurrent(), "third");
        $this->assertTrue($stack->leave());
        $this->assertTrue($stack->leave());
        $this->assertEqual($stack->getCurrent(), "one");
    }
}

class TestParser {
    
    function accept() {
    }
    
    function a() {
    }
    
    function b() {
    }
}
Mock::generate('TestParser');

class TestOfLexer extends UnitTestCase {
    
    function testEmptyPage() {
        $handler = &new MockTestParser();
        $handler->expectNever("accept");
        $handler->setReturnValue("accept", true);
        $handler->expectNever("accept");
        $handler->setReturnValue("accept", true);
        $lexer = &new SimpleLexer($handler);
        $lexer->addPattern("a+");
        $this->assertTrue($lexer->parse(""));
    }
    
    function testSinglePattern() {
        $handler = &new MockTestParser();
        $handler->expectArgumentsAt(0, "accept", array("aaa", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "accept", array("x", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(2, "accept", array("a", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "accept", array("yyy", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(4, "accept", array("a", LEXER_MATCHED));
        $handler->expectArgumentsAt(5, "accept", array("x", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(6, "accept", array("aaa", LEXER_MATCHED));
        $handler->expectArgumentsAt(7, "accept", array("z", LEXER_UNMATCHED));
        $handler->expectCallCount("accept", 8);
        $handler->setReturnValue("accept", true);
        $lexer = &new SimpleLexer($handler);
        $lexer->addPattern("a+");
        $this->assertTrue($lexer->parse("aaaxayyyaxaaaz"));
    }
    
    function testMultiplePattern() {
        $handler = &new MockTestParser();
        $target = array("a", "b", "a", "bb", "x", "b", "a", "xxxxxx", "a", "x");
        for ($i = 0; $i < count($target); $i++) {
            $handler->expectArgumentsAt($i, "accept", array($target[$i], '*'));
        }
        $handler->expectCallCount("accept", count($target));
        $handler->setReturnValue("accept", true);
        $lexer = &new SimpleLexer($handler);
        $lexer->addPattern("a+");
        $lexer->addPattern("b+");
        $this->assertTrue($lexer->parse("ababbxbaxxxxxxax"));
    }
}

class TestOfLexerModes extends UnitTestCase {
    
    function testIsolatedPattern() {
        $handler = &new MockTestParser();
        $handler->expectArgumentsAt(0, "a", array("a", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array("b", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(2, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "a", array("bxb", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(4, "a", array("aaa", LEXER_MATCHED));
        $handler->expectArgumentsAt(5, "a", array("x", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(6, "a", array("aaaa", LEXER_MATCHED));
        $handler->expectArgumentsAt(7, "a", array("x", LEXER_UNMATCHED));
        $handler->expectCallCount("a", 8);
        $handler->setReturnValue("a", true);
        $lexer = &new SimpleLexer($handler, "a");
        $lexer->addPattern("a+", "a");
        $lexer->addPattern("b+", "b");
        $this->assertTrue($lexer->parse("abaabxbaaaxaaaax"));
    }
    
    function testModeChange() {
        $handler = &new MockTestParser();
        $handler->expectArgumentsAt(0, "a", array("a", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array("b", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(2, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "a", array("b", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(4, "a", array("aaa", LEXER_MATCHED));
        $handler->expectArgumentsAt(0, "b", array(":", LEXER_ENTER));
        $handler->expectArgumentsAt(1, "b", array("a", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(2, "b", array("b", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "b", array("a", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(4, "b", array("bb", LEXER_MATCHED));
        $handler->expectArgumentsAt(5, "b", array("a", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(6, "b", array("bbb", LEXER_MATCHED));
        $handler->expectArgumentsAt(7, "b", array("a", LEXER_UNMATCHED));
        $handler->expectCallCount("a", 5);
        $handler->expectCallCount("b", 8);
        $handler->setReturnValue("a", true);
        $handler->setReturnValue("b", true);
        $lexer = &new SimpleLexer($handler, "a");
        $lexer->addPattern("a+", "a");
        $lexer->addEntryPattern(":", "a", "b");
        $lexer->addPattern("b+", "b");
        $this->assertTrue($lexer->parse("abaabaaa:ababbabbba"));
    }
    
    function testNesting() {
        $handler = &new MockTestParser();
        $handler->setReturnValue("a", true);
        $handler->setReturnValue("b", true);
        $handler->expectArgumentsAt(0, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array("b", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(2, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "a", array("b", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(0, "b", array("(", LEXER_ENTER));
        $handler->expectArgumentsAt(1, "b", array("bb", LEXER_MATCHED));
        $handler->expectArgumentsAt(2, "b", array("a", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(3, "b", array("bb", LEXER_MATCHED));
        $handler->expectArgumentsAt(4, "b", array(")", LEXER_EXIT));
        $handler->expectArgumentsAt(4, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(5, "a", array("b", LEXER_UNMATCHED));
        $handler->expectCallCount("a", 6);
        $handler->expectCallCount("b", 5);
        $lexer = &new SimpleLexer($handler, "a");
        $lexer->addPattern("a+", "a");
        $lexer->addEntryPattern("(", "a", "b");
        $lexer->addPattern("b+", "b");
        $lexer->addExitPattern(")", "b");
        $this->assertTrue($lexer->parse("aabaab(bbabb)aab"));
    }
    
    function testSingular() {
        $handler = &new MockTestParser();
        $handler->setReturnValue("a", true);
        $handler->setReturnValue("b", true);
        $handler->expectArgumentsAt(0, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(2, "a", array("xx", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(3, "a", array("xx", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(0, "b", array("b", LEXER_SPECIAL));
        $handler->expectArgumentsAt(1, "b", array("bbb", LEXER_SPECIAL));
        $handler->expectCallCount("a", 4);
        $handler->expectCallCount("b", 2);
        $lexer = &new SimpleLexer($handler, "a");
        $lexer->addPattern("a+", "a");
        $lexer->addSpecialPattern("b+", "a", "b");
        $this->assertTrue($lexer->parse("aabaaxxbbbxx"));
    }
    
    function testUnwindTooFar() {
        $handler = &new MockTestParser();
        $handler->setReturnValue("a", true);
        $handler->expectArgumentsAt(0, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array(")", LEXER_EXIT));
        $handler->expectCallCount("a", 2);
        $lexer = &new SimpleLexer($handler, "a");
        $lexer->addPattern("a+", "a");
        $lexer->addExitPattern(")", "a");
        $this->assertFalse($lexer->parse("aa)aa"));
    }
}

class TestOfLexerHandlers extends UnitTestCase {
    
    function testModeMapping() {
        $handler = &new MockTestParser();
        $handler->setReturnValue("a", true);
        $handler->expectArgumentsAt(0, "a", array("aa", LEXER_MATCHED));
        $handler->expectArgumentsAt(1, "a", array("(", LEXER_ENTER));
        $handler->expectArgumentsAt(2, "a", array("bb", LEXER_MATCHED));
        $handler->expectArgumentsAt(3, "a", array("a", LEXER_UNMATCHED));
        $handler->expectArgumentsAt(4, "a", array("bb", LEXER_MATCHED));
        $handler->expectArgumentsAt(5, "a", array(")", LEXER_EXIT));
        $handler->expectArgumentsAt(6, "a", array("b", LEXER_UNMATCHED));
        $handler->expectCallCount("a", 7);
        $lexer = &new SimpleLexer($handler, "mode_a");
        $lexer->addPattern("a+", "mode_a");
        $lexer->addEntryPattern("(", "mode_a", "mode_b");
        $lexer->addPattern("b+", "mode_b");
        $lexer->addExitPattern(")", "mode_b");
        $lexer->mapHandler("mode_a", "a");
        $lexer->mapHandler("mode_b", "a");
        $this->assertTrue($lexer->parse("aa(bbabb)b"));
    }
}

class TestOfSimpleHtmlLexer extends UnitTestCase {
    
    function &createParser() {
        $parser = &new MockSimpleHtmlSaxParser();
        $parser->setReturnValue('acceptStartToken', true);
        $parser->setReturnValue('acceptEndToken', true);
        $parser->setReturnValue('acceptAttributeToken', true);
        $parser->setReturnValue('acceptEntityToken', true);
        $parser->setReturnValue('acceptTextToken', true);
        $parser->setReturnValue('ignore', true);
        return $parser;
    }
    
    function testNoContent() {
        $parser = &$this->createParser();
        $parser->expectNever('acceptStartToken');
        $parser->expectNever('acceptEndToken');
        $parser->expectNever('acceptAttributeToken');
        $parser->expectNever('acceptEntityToken');
        $parser->expectNever('acceptTextToken');
        $lexer = &new SimpleHtmlLexer($parser);
        $this->assertTrue($lexer->parse(''));
    }
    
    function testUninteresting() {
        $parser = &$this->createParser();
        $parser->expectOnce('acceptTextToken', array('<html></html>', '*'));
        $lexer = &new SimpleHtmlLexer($parser);
        $this->assertTrue($lexer->parse('<html></html>'));
    }
    
    function testSkipCss() {
        $parser = &$this->createParser();
        $parser->expectNever('acceptTextToken');
        $parser->expectAtLeastOnce('ignore');
        $lexer = &new SimpleHtmlLexer($parser);
        $this->assertTrue($lexer->parse("<style>Lot's of styles</style>"));
    }
    
    function testSkipJavaScript() {
        $parser = &$this->createParser();
        $parser->expectNever('acceptTextToken');
        $parser->expectAtLeastOnce('ignore');
        $lexer = &new SimpleHtmlLexer($parser);
        $this->assertTrue($lexer->parse("<SCRIPT>Javascript code {';:^%^%