wallabag/tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php

413 lines
15 KiB
PHP
Raw Normal View History

<?php
namespace Tests\Wallabag\CoreBundle\Helper;
use Psr\Log\NullLogger;
use Wallabag\CoreBundle\Helper\ContentProxy;
use Wallabag\CoreBundle\Entity\Entry;
use Wallabag\CoreBundle\Entity\Tag;
use Wallabag\UserBundle\Entity\User;
use Wallabag\CoreBundle\Helper\RuleBasedTagger;
use Graby\Graby;
use Monolog\Handler\TestHandler;
use Monolog\Logger;
class ContentProxyTest extends \PHPUnit_Framework_TestCase
{
private $fetchingErrorMessage = 'wallabag can\'t retrieve contents for this article. Please <a href="http://doc.wallabag.org/en/user/errors_during_fetching.html#how-can-i-help-to-fix-that">troubleshoot this issue</a>.';
2016-03-27 18:35:56 +00:00
public function testWithBadUrl()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$graby = $this->getMockBuilder('Graby\Graby')
->setMethods(['fetchContent'])
2016-03-27 18:35:56 +00:00
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
2016-03-27 18:35:56 +00:00
'html' => false,
'title' => '',
'url' => '',
'content_type' => '',
'language' => '',
]);
2016-03-27 18:35:56 +00:00
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://user@:80');
2016-03-27 18:35:56 +00:00
$this->assertEquals('http://user@:80', $entry->getUrl());
$this->assertEmpty($entry->getTitle());
2016-12-03 04:59:18 +00:00
$this->assertEquals($this->fetchingErrorMessage, $entry->getContent());
2016-03-27 18:35:56 +00:00
$this->assertEmpty($entry->getPreviewPicture());
$this->assertEmpty($entry->getMimetype());
$this->assertEmpty($entry->getLanguage());
$this->assertEquals(0.0, $entry->getReadingTime());
$this->assertEquals(false, $entry->getDomainName());
}
public function testWithEmptyContent()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$graby = $this->getMockBuilder('Graby\Graby')
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
2015-09-20 20:37:27 +00:00
'html' => false,
'title' => '',
'url' => '',
'content_type' => '',
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
$this->assertEquals('http://0.0.0.0', $entry->getUrl());
$this->assertEmpty($entry->getTitle());
2016-12-03 04:59:18 +00:00
$this->assertEquals($this->fetchingErrorMessage, $entry->getContent());
$this->assertEmpty($entry->getPreviewPicture());
$this->assertEmpty($entry->getMimetype());
2015-09-20 20:37:27 +00:00
$this->assertEmpty($entry->getLanguage());
$this->assertEquals(0.0, $entry->getReadingTime());
$this->assertEquals('0.0.0.0', $entry->getDomainName());
}
public function testWithEmptyContentButOG()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$graby = $this->getMockBuilder('Graby\Graby')
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
2015-09-20 20:37:27 +00:00
'html' => false,
'title' => '',
'url' => '',
'content_type' => '',
'language' => '',
2016-11-18 14:09:21 +00:00
'status' => '',
'open_graph' => [
2015-09-20 20:37:27 +00:00
'og_title' => 'my title',
'og_description' => 'desc',
],
]);
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://domain.io');
$this->assertEquals('http://domain.io', $entry->getUrl());
$this->assertEquals('my title', $entry->getTitle());
2017-05-05 12:33:36 +00:00
$this->assertEquals($this->fetchingErrorMessage.'<p><i>But we found a short description: </i></p>desc', $entry->getContent());
$this->assertEmpty($entry->getPreviewPicture());
2015-09-20 20:37:27 +00:00
$this->assertEmpty($entry->getLanguage());
2016-11-18 14:09:21 +00:00
$this->assertEmpty($entry->getHttpStatus());
$this->assertEmpty($entry->getMimetype());
$this->assertEquals(0.0, $entry->getReadingTime());
$this->assertEquals('domain.io', $entry->getDomainName());
}
public function testWithContent()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$graby = $this->getMockBuilder('Graby\Graby')
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
2015-09-20 20:37:27 +00:00
'language' => 'fr',
2016-11-18 14:09:21 +00:00
'status' => '200',
'open_graph' => [
'og_title' => 'my OG title',
'og_description' => 'OG desc',
2015-09-10 20:00:53 +00:00
'og_image' => 'http://3.3.3.3/cover.jpg',
],
]);
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertContains('this is my content', $entry->getContent());
$this->assertEquals('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertEquals('text/html', $entry->getMimetype());
2015-09-20 20:37:27 +00:00
$this->assertEquals('fr', $entry->getLanguage());
2016-11-18 14:09:21 +00:00
$this->assertEquals('200', $entry->getHttpStatus());
$this->assertEquals(4.0, $entry->getReadingTime());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
}
2015-10-11 20:27:47 +00:00
public function testWithContentAndNoOgImage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$graby = $this->getMockBuilder('Graby\Graby')
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'status' => '200',
'open_graph' => [
'og_title' => 'my OG title',
'og_description' => 'OG desc',
'og_image' => false,
],
]);
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = $proxy->updateEntry(new Entry(new User()), 'http://0.0.0.0');
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertContains('this is my content', $entry->getContent());
$this->assertNull($entry->getPreviewPicture());
$this->assertEquals('text/html', $entry->getMimetype());
$this->assertEquals('fr', $entry->getLanguage());
$this->assertEquals('200', $entry->getHttpStatus());
$this->assertEquals(4.0, $entry->getReadingTime());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
}
2016-03-27 18:35:56 +00:00
public function testWithForcedContent()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry(
$entry,
'http://0.0.0.0',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'date' => '1395635872',
'authors' => ['Jeremy', 'Nico', 'Thomas'],
'all_headers' => [
'Cache-Control' => 'no-cache',
2017-05-29 08:14:01 +00:00
],
]
);
2016-03-27 18:35:56 +00:00
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertContains('this is my content', $entry->getContent());
$this->assertEquals('text/html', $entry->getMimetype());
$this->assertEquals('fr', $entry->getLanguage());
$this->assertEquals(4.0, $entry->getReadingTime());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
$this->assertEquals('24/03/2014', $entry->getPublishedAt()->format('d/m/Y'));
$this->assertContains('Jeremy', $entry->getPublishedBy());
$this->assertContains('Nico', $entry->getPublishedBy());
$this->assertContains('Thomas', $entry->getPublishedBy());
$this->assertContains('no-cache', $entry->getHeaders());
}
public function testWithForcedContentAndDatetime()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
2017-05-30 08:08:50 +00:00
$proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry(
$entry,
'http://0.0.0.0',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'date' => '2016-09-08T11:55:58+0200',
]
);
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertContains('this is my content', $entry->getContent());
$this->assertEquals('text/html', $entry->getMimetype());
$this->assertEquals('fr', $entry->getLanguage());
$this->assertEquals(4.0, $entry->getReadingTime());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
$this->assertEquals('08/09/2016', $entry->getPublishedAt()->format('d/m/Y'));
}
public function testWithForcedContentAndBadDate()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$logger = new Logger('foo');
$handler = new TestHandler();
$logger->pushHandler($handler);
2017-05-30 08:08:50 +00:00
$proxy = new ContentProxy((new Graby()), $tagger, $logger, $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry(
$entry,
'http://0.0.0.0',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'date' => '01 02 2012',
]
);
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertContains('this is my content', $entry->getContent());
$this->assertEquals('text/html', $entry->getMimetype());
$this->assertEquals('fr', $entry->getLanguage());
$this->assertEquals(4.0, $entry->getReadingTime());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
$this->assertNull($entry->getPublishedAt());
$records = $handler->getRecords();
$this->assertCount(1, $records);
$this->assertContains('Error while defining date', $records[0]['message']);
2016-03-27 18:35:56 +00:00
}
public function testTaggerThrowException()
{
$graby = $this->getMockBuilder('Graby\Graby')
->disableOriginalConstructor()
->getMock();
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag')
->will($this->throwException(new \Exception()));
$proxy = new ContentProxy($graby, $tagger, $this->getLogger(), $this->fetchingErrorMessage);
2016-03-27 18:35:56 +00:00
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0', [
2016-03-27 18:35:56 +00:00
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
]);
$this->assertCount(0, $entry->getTags());
}
public function dataForCrazyHtml()
{
return [
'script and comment' => [
'<strong>Script inside:</strong> <!--[if gte IE 4]><script>alert(\'lol\');</script><![endif]--><br />',
2017-05-29 08:14:01 +00:00
'lol',
],
'script' => [
'<strong>Script inside:</strong><script>alert(\'lol\');</script>',
2017-05-29 08:14:01 +00:00
'script',
],
];
}
/**
* @dataProvider dataForCrazyHtml
*/
public function testWithCrazyHtmlContent($html, $escapedString)
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$proxy = new ContentProxy((new Graby()), $tagger, $this->getLogger(), $this->fetchingErrorMessage);
$entry = $proxy->updateEntry(
new Entry(new User()),
'http://1.1.1.1',
[
'html' => $html,
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'status' => '200',
'open_graph' => [
'og_title' => 'my OG title',
'og_description' => 'OG desc',
'og_image' => 'http://3.3.3.3/cover.jpg',
],
]
);
$this->assertEquals('http://1.1.1.1', $entry->getUrl());
$this->assertEquals('this is my title', $entry->getTitle());
$this->assertNotContains($escapedString, $entry->getContent());
$this->assertEquals('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertEquals('text/html', $entry->getMimetype());
$this->assertEquals('fr', $entry->getLanguage());
$this->assertEquals('200', $entry->getHttpStatus());
$this->assertEquals('1.1.1.1', $entry->getDomainName());
}
2015-10-11 20:27:47 +00:00
private function getTaggerMock()
{
return $this->getMockBuilder(RuleBasedTagger::class)
->setMethods(['tag'])
2015-10-11 20:27:47 +00:00
->disableOriginalConstructor()
->getMock();
}
private function getLogger()
{
return new NullLogger();
}
}