This is an extract of the Readability class from this [full-text-rss](https://github.com/Dither/full-text-rss) fork. It can be defined as a better version of the original [php-readability](https://bitbucket.org/fivefilters/php-readability/overview).
This is an extract of the Readability class from this [full-text-rss](https://github.com/Dither/full-text-rss) fork. It can be defined as a better version of the original [php-readability](https://bitbucket.org/fivefilters/php-readability).
@ -77,7 +85,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('Sorry, Readability was unable to parse this page for content.', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('Sorry, Readability was unable to parse this page for content.', $readability->getContent()->getInnerHtml());
}
}
public function testInitP()
public function testInitP(): void
{
{
$readability = $this->getReadability(str_repeat('<p>This is the awesome content :)</p>', 7), 'http://0.0.0.0');
$readability = $this->getReadability(str_repeat('<p>This is the awesome content :)</p>', 7), 'http://0.0.0.0');
$res = $readability->init();
$res = $readability->init();
@ -90,7 +98,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
}
}
public function testInitDivP()
public function testInitDivP(): void
{
{
$readability = $this->getReadability('<div>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</div>', 'http://0.0.0.0');
$readability = $this->getReadability('<div>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</div>', 'http://0.0.0.0');
$res = $readability->init();
$res = $readability->init();
@ -103,7 +111,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
}
}
public function testInitDiv()
public function testInitDiv(): void
{
{
$readability = $this->getReadability('<div>' . str_repeat('This is the awesome content :)', 7) . '</div>', 'http://0.0.0.0');
$readability = $this->getReadability('<div>' . str_repeat('This is the awesome content :)', 7) . '</div>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -117,7 +125,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('This is the awesome content :)', $readability->getContent()->getInnerHtml());
}
}
public function testWithFootnotes()
public function testWithFootnotes(): void
{
{
$readability = $this->getReadability('<div>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '</div>', 'http://0.0.0.0');
$readability = $this->getReadability('<div>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '</div>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -134,7 +142,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$readability = $this->getReadability('<div><h2>Title</h2>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<ahref="#nofollow"rel="nofollow">will NOT be removed</a></div>', 'http://0.0.0.0');
$readability = $this->getReadability('<div><h2>Title</h2>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<ahref="#nofollow"rel="nofollow">will NOT be removed</a></div>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -151,7 +159,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$readability = $this->getReadability('<div><h2>Title</h2>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<p>This is an awesome text with some links, here there are <iframesrc="http://youtube.com/test"href="#nofollow"rel="nofollow"></iframe><iframe>http://soundcloud.com/test</iframe></p></div>', 'http://0.0.0.0');
$readability = $this->getReadability('<div><h2>Title</h2>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<p>This is an awesome text with some links, here there are <iframesrc="http://youtube.com/test"href="#nofollow"rel="nofollow"></iframe><iframe>http://soundcloud.com/test</iframe></p></div>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -166,7 +174,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$readability = $this->getReadability('<article><p>' . str_repeat('This is an awesome text with some links, here there are: the awesome', 20) . '</p><p>This is an awesome text with some links, here there are <iframesrc="http://youtube.com/test"href="#nofollow"rel="nofollow"></iframe></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<article><p>' . str_repeat('This is an awesome text with some links, here there are: the awesome', 20) . '</p><p>This is an awesome text with some links, here there are <iframesrc="http://youtube.com/test"href="#nofollow"rel="nofollow"></iframe></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -181,7 +189,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<footer><aside>' . str_repeat('<p>This is an awesome text with some links, here there are</p>', 8) . '</aside></footer></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<footer><aside>' . str_repeat('<p>This is an awesome text with some links, here there are</p>', 8) . '</aside></footer></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -196,7 +204,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<divstyle="display:none">' . str_repeat('<pclass="clock">This text should be removed</p>', 10) . '</div></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<divstyle="display:none">' . str_repeat('<pclass="clock">This text should be removed</p>', 10) . '</div></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -211,7 +219,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text should be removed', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text should be removed', $readability->getContent()->getInnerHtml());
}
}
public function testWithClassesWithoutLightClean()
public function testWithClassesWithoutLightClean(): void
{
{
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<divstyle="display:none">' . str_repeat('<pclass="clock">This text should be removed</p>', 10) . '</div></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<article>' . str_repeat('<p>This is an awesome text with some links, here there are: <ahref="http://0.0.0.0/test.html">the awesome</a></p>', 7) . '<divstyle="display:none">' . str_repeat('<pclass="clock">This text should be removed</p>', 10) . '</div></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -227,7 +235,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text should be removed', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text should be removed', $readability->getContent()->getInnerHtml());
}
}
public function testWithTd()
public function testWithTd(): void
{
{
$readability = $this->getReadability('<table><tr>' . str_repeat('<td><p>This is an awesome text with some links, here there are the awesome</td>', 7) . '</tr></table>', 'http://0.0.0.0');
$readability = $this->getReadability('<table><tr>' . str_repeat('<td><p>This is an awesome text with some links, here there are the awesome</td>', 7) . '</tr></table>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -240,7 +248,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('This is an awesome text with some links, here there are', $readability->getContent()->getInnerHtml());
}
}
public function testWithSameClasses()
public function testWithSameClasses(): void
{
{
$readability = $this->getReadability('<articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<divclass="awesomecontent">This text is also an awesome text and you should know that !</div></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<divclass="awesomecontent">This text is also an awesome text and you should know that !</div></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -254,7 +262,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
public function testWithScript()
public function testWithScript(): void
{
{
$readability = $this->getReadability('<articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p><script>Thistextisalsoanawesometextandyoushouldknowthat!</script></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p><script>Thistextisalsoanawesometextandyoushouldknowthat!</script></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -268,7 +276,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
public function testTitle()
public function testTitle(): void
{
{
$readability = $this->getReadability('<title>this is my title</title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<title>this is my title</title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -282,7 +290,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
public function testTitleWithDash()
public function testTitleWithDash(): void
{
{
$readability = $this->getReadability('<title> title2 - title3 </title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<title> title2 - title3 </title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -296,7 +304,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
public function testTitleWithDoubleDot()
public function testTitleWithDoubleDot(): void
{
{
$readability = $this->getReadability('<title> title2 : title3 </title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<title> title2 : title3 </title><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -310,7 +318,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
public function testTitleTooShortUseH1()
public function testTitleTooShortUseH1(): void
{
{
$readability = $this->getReadability('<title>too short</title><h1>this is my h1 title !</h1><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability = $this->getReadability('<title>too short</title><h1>this is my h1 title !</h1><articleclass="awesomecontent">' . str_repeat('<p>This is an awesome text with some links, here there are the awesome</p>', 7) . '<p></p></article>', 'http://0.0.0.0');
$readability->debug = true;
$readability->debug = true;
@ -324,17 +332,14 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
$this->assertStringNotContainsString('This text is also an awesome text and you should know that', $readability->getContent()->getInnerHtml());
}
}
// dummy function to be used to the next test
public function testAutoClosingIframeNotThrowingException(): void
public function error2Exception($code, $string, $file, $line, $context)
{
throw new \Exception($string, $code);
}
public function testAutoClosingIframeNotThrowingException()
Jérémy Benoist
4 years ago
committed by
GitHub