wallabag/tests/Helper/ContentProxyTest.php

1140 lines
43 KiB
PHP
Raw Normal View History

<?php
2024-02-19 00:30:12 +00:00
namespace Tests\Wallabag\Helper;
2017-07-01 07:52:38 +00:00
use Graby\Graby;
2017-05-30 15:48:24 +00:00
use Monolog\Handler\TestHandler;
2017-07-01 07:52:38 +00:00
use Monolog\Logger;
2017-12-16 21:17:42 +00:00
use PHPUnit\Framework\TestCase;
2017-07-01 07:52:38 +00:00
use Psr\Log\NullLogger;
use Symfony\Component\Validator\ConstraintViolation;
use Symfony\Component\Validator\ConstraintViolationList;
use Symfony\Component\Validator\Validator\RecursiveValidator;
2024-02-19 00:30:12 +00:00
use Wallabag\Entity\Entry;
use Wallabag\Entity\User;
use Wallabag\Helper\ContentProxy;
use Wallabag\Helper\RuleBasedIgnoreOriginProcessor;
use Wallabag\Helper\RuleBasedTagger;
2017-12-16 21:17:42 +00:00
class ContentProxyTest extends 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');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->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' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => '',
],
2016-03-27 18:35:56 +00:00
'language' => '',
]);
2016-03-27 18:35:56 +00:00
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://user@:80');
2016-03-27 18:35:56 +00:00
2017-07-01 07:52:38 +00:00
$this->assertSame('http://user@:80', $entry->getUrl());
2016-03-27 18:35:56 +00:00
$this->assertEmpty($entry->getTitle());
2017-07-01 07:52:38 +00:00
$this->assertSame($this->fetchingErrorMessage, $entry->getContent());
2016-03-27 18:35:56 +00:00
$this->assertEmpty($entry->getPreviewPicture());
$this->assertEmpty($entry->getMimetype());
$this->assertEmpty($entry->getLanguage());
2017-07-01 07:52:38 +00:00
$this->assertSame(0.0, $entry->getReadingTime());
$this->assertNull($entry->getDomainName());
$this->assertTrue($entry->isNotParsed());
2016-03-27 18:35:56 +00:00
}
public function testWithEmptyContent()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
2015-09-20 20:37:27 +00:00
'html' => false,
'title' => '',
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => '',
],
2015-09-20 20:37:27 +00:00
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://0.0.0.0', $entry->getUrl());
$this->assertEmpty($entry->getTitle());
2017-07-01 07:52:38 +00:00
$this->assertSame($this->fetchingErrorMessage, $entry->getContent());
$this->assertEmpty($entry->getPreviewPicture());
$this->assertEmpty($entry->getMimetype());
2015-09-20 20:37:27 +00:00
$this->assertEmpty($entry->getLanguage());
2017-07-01 07:52:38 +00:00
$this->assertSame(0.0, $entry->getReadingTime());
$this->assertSame('0.0.0.0', $entry->getDomainName());
$this->assertTrue($entry->isNotParsed());
}
public function testWithEmptyContentButOG()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
2015-09-20 20:37:27 +00:00
'html' => false,
'title' => 'my title',
2015-09-20 20:37:27 +00:00
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => '',
],
2015-09-20 20:37:27 +00:00
'language' => '',
2016-11-18 14:09:21 +00:00
'status' => '',
'description' => 'desc',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://domain.io');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://domain.io', $entry->getUrl());
$this->assertSame('my title', $entry->getTitle());
$this->assertSame($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());
2017-07-01 07:52:38 +00:00
$this->assertSame(0.0, $entry->getReadingTime());
$this->assertSame('domain.io', $entry->getDomainName());
$this->assertTrue($entry->isNotParsed());
}
public function testWithContent()
{
2015-10-11 20:27:47 +00:00
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->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',
2015-09-20 20:37:27 +00:00
'language' => 'fr',
2016-11-18 14:09:21 +00:00
'status' => '200',
'description' => 'OG desc',
'image' => 'http://3.3.3.3/cover.jpg',
'headers' => [
'content-type' => 'text/html',
],
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
2015-10-11 20:27:47 +00:00
public function testWithContentAndNoOgImage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->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',
'language' => 'fr',
'status' => '200',
'description' => 'OG desc',
'image' => null,
'headers' => [
'content-type' => 'text/html',
],
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
2017-05-30 15:48:24 +00:00
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-06-12 14:46:33 +00:00
$this->assertNull($entry->getPreviewPicture());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
public function testWithContentAndContentImage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => "<h1>Test</h1><p><img src='http://3.3.3.3/cover.jpg'/></p>",
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
2019-05-28 10:02:17 +00:00
'headers' => [
'content-type' => 'text/html',
],
'language' => 'fr',
'status' => '200',
2019-05-28 10:02:17 +00:00
'image' => null,
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
$this->assertSame("<h1>Test</h1><p><img src='http://3.3.3.3/cover.jpg'/></p>", $entry->getContent());
$this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(0.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
public function testWithContentImageAndOgImage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => "<h1>Test</h1><p><img src='http://3.3.3.3/nevermind.jpg'/></p>",
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
2019-05-28 10:02:17 +00:00
'headers' => [
'content-type' => 'text/html',
],
'language' => 'fr',
'status' => '200',
2019-05-28 10:02:17 +00:00
'image' => 'http://3.3.3.3/cover.jpg',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
$this->assertSame("<h1>Test</h1><p><img src='http://3.3.3.3/nevermind.jpg'/></p>", $entry->getContent());
$this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(0.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
public function testWithContentAndBadLanguage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2017-12-18 09:14:00 +00:00
$validator = $this->getValidator(false);
$validator->expects($this->once())
->method('validate')
->willReturn(new ConstraintViolationList([new ConstraintViolation('oops', 'oops', [], 'oops', 'language', 'dontexist')]));
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->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',
'language' => 'dontexist',
'status' => '200',
'headers' => [
'content-type' => 'text/html',
],
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $validator, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
2017-06-12 14:46:33 +00:00
$this->assertNull($entry->getLanguage());
2017-07-01 07:52:38 +00:00
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
public function testWithContentAndBadOgImage()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2017-12-18 09:14:00 +00:00
$validator = $this->getValidator(false);
$validator->expects($this->exactly(2))
->method('validate')
->will($this->onConsecutiveCalls(
new ConstraintViolationList(),
new ConstraintViolationList([new ConstraintViolation('oops', 'oops', [], 'oops', 'url', 'https://')])
));
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->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',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'text/html',
],
'language' => 'fr',
'status' => '200',
'description' => 'OG desc',
'image' => 'https://',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $validator, $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-06-12 14:46:33 +00:00
$this->assertNull($entry->getPreviewPicture());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
2016-03-27 18:35:56 +00:00
public function testWithForcedContent()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process');
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage, true);
$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',
'language' => 'fr',
'date' => '1395635872',
'authors' => ['Jeremy', 'Nico', 'Thomas'],
'headers' => [
'cache-control' => 'no-cache',
'content-type' => 'text/html',
2017-05-29 08:14:01 +00:00
],
]
);
2016-03-27 18:35:56 +00:00
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertSame('24/03/2014', $entry->getPublishedAt()->format('d/m/Y'));
$this->assertContains('Jeremy', $entry->getPublishedBy());
$this->assertContains('Nico', $entry->getPublishedBy());
$this->assertContains('Thomas', $entry->getPublishedBy());
$this->assertNotNull($entry->getHeaders(), 'Headers are stored, so value is not null');
$this->assertContains('no-cache', $entry->getHeaders());
$this->assertFalse($entry->isNotParsed());
}
2022-05-08 12:12:35 +00:00
public function testWithForcedContentAndDateTime()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2017-05-30 15:48:24 +00:00
$logHandler = new TestHandler();
2017-06-01 09:31:45 +00:00
$logger = new Logger('test', [$logHandler]);
2017-05-30 15:48:24 +00:00
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $logger, $this->fetchingErrorMessage);
$entry = new Entry(new User());
2017-06-01 09:31:45 +00:00
$proxy->updateEntry(
$entry,
2017-06-01 09:31:45 +00:00
'http://1.1.1.1',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'language' => 'fr',
'date' => '2016-09-08T11:55:58+0200',
'headers' => [
'content-type' => 'text/html',
],
]
);
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertSame('08/09/2016', $entry->getPublishedAt()->format('d/m/Y'));
$this->assertFalse($entry->isNotParsed());
}
public function testWithForcedContentAndBadDate()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$logger = new Logger('foo');
$handler = new TestHandler();
$logger->pushHandler($handler);
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $logger, $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry(
$entry,
2017-06-01 09:31:45 +00:00
'http://1.1.1.1',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'language' => 'fr',
'date' => '01 02 2012',
'headers' => [
'content-type' => 'text/html',
],
]
);
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('content', $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertNull($entry->getPublishedAt());
$this->assertFalse($entry->isNotParsed());
$records = $handler->getRecords();
$this->assertCount(3, $records);
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('Error while defining date', $records[0]['message']);
2016-03-27 18:35:56 +00:00
}
public function testTaggerThrowException()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag')
->will($this->throwException(new \Exception()));
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
2017-06-01 09:31:45 +00:00
$proxy->updateEntry(
$entry,
'http://1.1.1.1',
[
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'language' => 'fr',
'headers' => [
'content-type' => 'text/html',
],
2017-06-01 09:31:45 +00:00
]
);
2016-03-27 18:35:56 +00:00
$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');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
2017-06-01 09:31:45 +00:00
$entry = new Entry(new User());
$proxy->updateEntry(
$entry,
'http://1.1.1.1',
[
'html' => $html,
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
'language' => 'fr',
'status' => '200',
2024-01-01 18:11:01 +00:00
// 'og_title' => 'my OG title',
'description' => 'OG desc',
'image' => 'http://3.3.3.3/cover.jpg',
'headers' => [
'content-type' => 'text/html',
],
]
);
2017-07-01 07:52:38 +00:00
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringNotContainsString($escapedString, $entry->getContent());
2017-07-01 07:52:38 +00:00
$this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
public function testWithImageAsContent()
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => '<p><img src="http://1.1.1.1/image.jpg" /></p>',
'title' => 'this is my title',
'url' => 'http://1.1.1.1/image.jpg',
'status' => '200',
'headers' => [
'content-type' => 'image/jpeg',
],
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
2017-07-03 11:56:39 +00:00
$this->assertSame('http://1.1.1.1/image.jpg', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
2020-06-15 11:37:50 +00:00
$this->assertStringContainsString('http://1.1.1.1/image.jpg', $entry->getContent());
$this->assertSame('http://1.1.1.1/image.jpg', $entry->getPreviewPicture());
2017-07-03 11:56:39 +00:00
$this->assertSame('image/jpeg', $entry->getMimetype());
$this->assertSame('200', $entry->getHttpStatus());
$this->assertSame('1.1.1.1', $entry->getDomainName());
$this->assertFalse($entry->isNotParsed());
}
2020-12-08 08:17:10 +00:00
public function testWebsiteWithValidUTF8TitleDoNothing()
{
// You can use https://www.online-toolz.com/tools/text-hex-convertor.php to convert UTF-8 text <=> hex
// See http://graphemica.com for more info about the characters
// '😻z' (U+1F63B or F09F98BB; U+2124 or E284A4; U+007A or 7A) in hexadecimal and UTF-8
$actualTitle = $this->hexToStr('F09F98BB' . 'E284A4' . '7A');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'text/html',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// '😻z' (U+1F63B or F09F98BB; U+2124 or E284A4; U+007A or 7A) in hexadecimal and UTF-8
$expectedTitle = 'F09F98BB' . 'E284A4' . '7A';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
2020-12-08 08:17:10 +00:00
public function testWebsiteWithInvalidUTF8TitleRemoveInvalidCharacter()
{
// See http://graphemica.com for more info about the characters
// 'a€b' (61;80;62) in hexadecimal and WINDOWS-1252 - but 80 is a invalid UTF-8 character.
// The correct UTF-8 € character (U+20AC) is E282AC
$actualTitle = $this->hexToStr('61' . '80' . '62');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'text/html',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// 'ab' (61;62) because all invalid UTF-8 character (like 80) are removed
$expectedTitle = '61' . '62';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
2020-12-08 08:17:10 +00:00
public function testPdfWithUTF16BETitleConvertToUTF8()
{
// See http://graphemica.com for more info about the characters
// '😻' (U+1F63B;D83DDE3B) in hexadecimal and as UTF16BE
$actualTitle = $this->hexToStr('D83DDE3B');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'application/pdf',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// '😻' (U+1F63B or F09F98BB) in hexadecimal and UTF-8
$expectedTitle = 'F09F98BB';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
2020-12-08 08:17:10 +00:00
public function testPdfWithUTF8TitleDoNothing()
{
// See http://graphemica.com for more info about the characters
// '😻' (U+1F63B;D83DDE3B) in hexadecimal and as UTF8
$actualTitle = $this->hexToStr('F09F98BB');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'application/pdf',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// '😻' (U+1F63B or F09F98BB) in hexadecimal and UTF-8
$expectedTitle = 'F09F98BB';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
2020-12-08 08:17:10 +00:00
public function testPdfWithWINDOWS1252TitleConvertToUTF8()
{
// See http://graphemica.com for more info about the characters
// '€' (80) in hexadecimal and WINDOWS-1252
$actualTitle = $this->hexToStr('80');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'application/pdf',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// '€' (U+20AC or E282AC) in hexadecimal and UTF-8
$expectedTitle = 'E282AC';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
2020-12-08 08:17:10 +00:00
public function testPdfWithInvalidCharacterInTitleRemoveInvalidCharacter()
{
/*
* I spend too much time on trying to solve the problem of that test.
* Starting with PHP 8.1 this test fails because the string with invalid character is detected as WINDOWS-1252 and then converted.
* In PHP < 8.1, the string encoding can't be detected and nothing is then converted.
* So the removal of the invalid char happens in `sanitizeUTF8Text`
*
* So, I don't understand why the string with invalid char is detected as WINDOWS-1252 in PHP 8.1 and not before.
*/
$this->markTestSkipped('Encoding issue in PHP >= 8.1');
// See http://graphemica.com for more info about the characters
// '😻<F09F98BB>z' (U+1F63B or F09F98BB; U+2124 or E284A4; invalid character 81; U+007A or 7A) in hexadecimal and UTF-8
// 0x81 is not a valid character for UTF16, UTF8 and WINDOWS-1252
$actualTitle = $this->hexToStr('F09F98BB' . 'E284A4' . '81' . '7A');
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
2022-09-01 18:54:56 +00:00
$graby = $this->getMockBuilder(Graby::class)
->setMethods(['fetchContent'])
->disableOriginalConstructor()
->getMock();
$graby->expects($this->any())
->method('fetchContent')
->willReturn([
'html' => false,
'title' => $actualTitle,
'url' => '',
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => 'application/pdf',
],
'language' => '',
]);
$proxy = new ContentProxy($graby, $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
$entry = new Entry(new User());
$proxy->updateEntry($entry, 'http://0.0.0.0');
// '😻z' (U+1F63B or F09F98BB; U+2124 or E284A4; U+007A or 7A) in hexadecimal and UTF-8
// the 0x81 (represented by <20>) is invalid for UTF16, UTF8 and WINDOWS-1252 and is removed
$expectedTitle = 'F09F98BB' . 'E284A4' . '7A';
$this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
}
/**
* Data provider for testWithChangedUrl.
*
* Arrays contain the following values:
* $entry_url
* $origin_url
* $content_url
* $expected_entry_url
* $expected_origin_url
* $expected_domain
* $processor_result
*/
public function dataForChangedUrl()
{
return [
'normal' => [
'http://0.0.0.0',
null,
'http://1.1.1.1',
'http://1.1.1.1',
'http://0.0.0.0',
'1.1.1.1',
false,
],
'origin already set' => [
'http://0.0.0.0',
'http://hello',
'http://1.1.1.1',
'http://1.1.1.1',
'http://hello',
'1.1.1.1',
false,
],
'trailing slash' => [
'https://example.com/hello-world',
null,
'https://example.com/hello-world/',
'https://example.com/hello-world/',
null,
'example.com',
false,
],
'query string in fetched content' => [
'https://example.org/hello',
null,
'https://example.org/hello?world=1',
'https://example.org/hello?world=1',
'https://example.org/hello',
'example.org',
false,
],
'fragment in fetched content' => [
'https://example.org/hello',
null,
'https://example.org/hello#world',
'https://example.org/hello',
null,
'example.org',
false,
],
'fragment and query string in fetched content' => [
'https://example.org/hello',
null,
'https://example.org/hello?foo#world',
'https://example.org/hello?foo#world',
'https://example.org/hello',
'example.org',
false,
],
'different path and query string in fetch content' => [
'https://example.org/hello',
null,
'https://example.org/world?foo',
'https://example.org/world?foo',
'https://example.org/hello',
'example.org',
false,
],
'feedproxy ignore list test' => [
'http://feedproxy.google.com/~r/Wallabag/~3/helloworld',
null,
'https://example.org/hello-wallabag',
'https://example.org/hello-wallabag',
null,
'example.org',
true,
],
'feedproxy ignore list test with origin url already set' => [
'http://feedproxy.google.com/~r/Wallabag/~3/helloworld',
'https://example.org/this-is-source',
'https://example.org/hello-wallabag',
'https://example.org/hello-wallabag',
'https://example.org/this-is-source',
'example.org',
true,
],
'lemonde ignore pattern test' => [
'http://www.lemonde.fr/tiny/url',
null,
'http://example.com/hello-world',
'http://example.com/hello-world',
null,
'example.com',
true,
],
];
}
/**
* @dataProvider dataForChangedUrl
*/
public function testWithChangedUrl($entry_url, $origin_url, $content_url, $expected_entry_url, $expected_origin_url, $expected_domain, $processor_result)
{
$tagger = $this->getTaggerMock();
$tagger->expects($this->once())
->method('tag');
$ruleBasedIgnoreOriginProcessor = $this->getRuleBasedIgnoreOriginProcessorMock();
$ruleBasedIgnoreOriginProcessor->expects($this->once())
->method('process')
->willReturn($processor_result);
2024-01-01 18:11:01 +00:00
$proxy = new ContentProxy(new Graby(), $tagger, $ruleBasedIgnoreOriginProcessor, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage, true);
$entry = new Entry(new User());
$entry->setOriginUrl($origin_url);
$proxy->updateEntry(
$entry,
$entry_url,
[
'html' => false,
'title' => '',
'url' => $content_url,
2019-02-11 09:59:56 +00:00
'headers' => [
'content-type' => '',
],
'language' => '',
],
true
);
$this->assertSame($expected_entry_url, $entry->getUrl());
$this->assertSame($expected_domain, $entry->getDomainName());
$this->assertSame($expected_origin_url, $entry->getOriginUrl());
}
/**
* https://stackoverflow.com/a/18506801.
*
* @return string
*/
private function strToHex($string)
{
$hex = '';
for ($i = 0; $i < \strlen($string); ++$i) {
$ord = \ord($string[$i]);
$hexCode = dechex($ord);
$hex .= substr('0' . $hexCode, -2);
}
return strtoupper($hex);
}
/**
2019-02-11 09:59:56 +00:00
* Convert hex to string.
*
* @see https://stackoverflow.com/a/18506801
*
* @return string
*/
private function hexToStr($hex)
{
$string = '';
for ($i = 0; $i < \strlen($hex) - 1; $i += 2) {
$string .= \chr(hexdec($hex[$i] . $hex[$i + 1]));
}
return $string;
}
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 getRuleBasedIgnoreOriginProcessorMock()
{
return $this->getMockBuilder(RuleBasedIgnoreOriginProcessor::class)
->setMethods(['process'])
->disableOriginalConstructor()
->getMock();
}
private function getLogger()
{
return new NullLogger();
}
2017-12-18 09:14:00 +00:00
private function getValidator($withDefaultMock = true)
{
2017-12-18 09:14:00 +00:00
$mock = $this->getMockBuilder(RecursiveValidator::class)
->setMethods(['validate'])
->disableOriginalConstructor()
->getMock();
2017-12-18 09:14:00 +00:00
if ($withDefaultMock) {
$mock->expects($this->any())
->method('validate')
->willReturn(new ConstraintViolationList());
}
return $mock;
}
}