this is my title

debugText; } public function getDomainRegexp() { return $this->domainRegExp; } } class ReadabilityTest extends \PHPUnit_Framework_TestCase { public function testConstructDefault() { $readability = new ReadabilityTested(''); $this->assertNull($readability->url); $this->assertContains('Parsing URL', $readability->getDebugText()); $this->assertContains('Tidying document', $readability->getDebugText()); $this->assertNull($readability->getDomainRegexp()); $this->assertInstanceOf('DomDocument', $readability->dom); } public function testConstructSimple() { $readability = new ReadabilityTested('', 'http://0.0.0.0'); $this->assertEquals('http://0.0.0.0', $readability->url); $this->assertContains('Parsing URL', $readability->getDebugText()); $this->assertContains('Tidying document', $readability->getDebugText()); $this->assertEquals('/0\.0\.0\.0/', $readability->getDomainRegexp()); $this->assertInstanceOf('DomDocument', $readability->dom); } public function testInitNoContent() { $readability = new ReadabilityTested('', 'http://0.0.0.0'); $res = $readability->init(); $this->assertFalse($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('Sorry, Readability was unable to parse this page for content.', $readability->getContent()->innerHTML); } public function testInitP() { $readability = new ReadabilityTested(str_repeat('

This is the awesome content :)

', 7), 'http://0.0.0.0'); $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('

getContent()->innerHTML); } public function testInitDivP() { $readability = new ReadabilityTested('

'.str_repeat('

This is the awesome content :)

', 7).'

', 'http://0.0.0.0'); $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('

getContent()->innerHTML); } public function testInitDiv() { $readability = new ReadabilityTested('

'.str_repeat('This is the awesome content :)', 7).'

', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('

getContent()->innerHTML); } public function testWithFootnotes() { $readability = new ReadabilityTested('

'.str_repeat('

This is an awesome text with some links, here there are: the awesome

', 7).'

', 'http://0.0.0.0'); $readability->debug = true; $readability->convertLinksToFootnotes = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('

getContent()->innerHTML); $this->assertContains('readabilityFootnoteLink', $readability->getContent()->innerHTML); $this->assertContains('readabilityLink-3', $readability->getContent()->innerHTML); } public function testStandardClean() { $readability = new ReadabilityTested('

Title

'.str_repeat('

This is an awesome text with some links, here there are: the awesome

', 7).'will be removed

', 'http://0.0.0.0'); $readability->debug = true; $readability->lightClean = false; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('

getContent()->innerHTML); $this->assertNotContains('will be removed', $readability->getContent()->innerHTML); $this->assertNotContains('

', $readability->getContent()->innerHTML); } public function testWithIframe() { $readability = new ReadabilityTested('
Title
'.str_repeat('
This is an awesome text with some links, here there are: the awesome
', 7).'
This is an awesome text with some links, here there are http://soundcloud.com/test
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('
getContent()->innerHTML); $this->assertContains('nofollow', $readability->getContent()->innerHTML); } public function testWithArticle() { $readability = new ReadabilityTested('
'.str_repeat('This is an awesome text with some links, here there are: the awesome', 20).'
This is an awesome text with some links, here there are
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertContains('nofollow', $readability->getContent()->innerHTML); } public function testWithAside() { $readability = new ReadabilityTested('
'.str_repeat('
This is an awesome text with some links, here there are: the awesome
', 7).'
'.str_repeat('
This is an awesome text with some links, here there are
', 8).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('
', $readability->getContent()->innerHTML); $this->assertContains('
', $readability->getContent()->innerHTML); } public function testWithClasses() { $readability = new ReadabilityTested('
'.str_repeat('
This is an awesome text with some links, here there are: the awesome
', 7).'
'.str_repeat('
This text should be removed
', 10).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text should be removed', $readability->getContent()->innerHTML); } public function testWithTd() { $readability = new ReadabilityTested(''.str_repeat('', 7).'
This is an awesome text with some links, here there are the awesome
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="tr"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); } public function testWithSameClasses() { $readability = new ReadabilityTested('
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
This text is also an awesome text and you should know that !
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } public function testWithScript() { $readability = new ReadabilityTested('
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEmpty($readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } public function testTitle() { $readability = new ReadabilityTested('this is my title
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEquals('this is my title', $readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } public function testTitleWithDash() { $readability = new ReadabilityTested(' title2 - title3
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEquals('title2 - title3', $readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } public function testTitleWithDoubleDot() { $readability = new ReadabilityTested(' title2 : title3
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEquals('title2 : title3', $readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } public function testTitleTooShortUseH1() { $readability = new ReadabilityTested('too short
this is my h1 title !
'.str_repeat('
This is an awesome text with some links, here there are the awesome
', 7).'
', 'http://0.0.0.0'); $readability->debug = true; $res = $readability->init(); $this->assertTrue($res); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getContent()); $this->assertInstanceOf('Readability\JSLikeHTMLElement', $readability->getTitle()); $this->assertContains('alt="article"', $readability->getContent()->innerHTML); $this->assertEquals('this is my h1 title !', $readability->getTitle()->innerHTML); $this->assertContains('This is an awesome text with some links, here there are', $readability->getContent()->innerHTML); $this->assertNotContains('This text is also an awesome text and you should know that', $readability->getContent()->innerHTML); } // public function testConstructParser() // { // $readability = new ReadabilityTested('', 'http://0.0.0.0', 'html5lib'); // $this->assertEquals('http://0.0.0.0', $readability->url); // $this->assertContains('Parsing URL', $readability->getDebugText()); // $this->assertContains('Tidying document', $readability->getDebugText()); // $this->assertEquals('/0\.0\.0\.0/', $readability->getDomainRegexp()); // $this->assertInstanceOf('DomDocument', $readability->dom); // } }