X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=tests%2FWallabag%2FCoreBundle%2FHelper%2FContentProxyTest.php;h=40a6cc802d0f6701fd12e83c4a1f9ba95c5e6577;hb=5f08426201c336f96d593954fb45b284d7e60f4a;hp=398592e1fa519577d1ff90fd4d4686c6c7d21b7d;hpb=5661e8d42af5d0557bf8017b515c51335757d6f2;p=github%2Fwallabag%2Fwallabag.git
diff --git a/tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php b/tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php
index 398592e1..40a6cc80 100644
--- a/tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php
+++ b/tests/Wallabag/CoreBundle/Helper/ContentProxyTest.php
@@ -11,7 +11,6 @@ use Symfony\Component\Validator\ConstraintViolation;
use Symfony\Component\Validator\ConstraintViolationList;
use Symfony\Component\Validator\Validator\RecursiveValidator;
use Wallabag\CoreBundle\Entity\Entry;
-use Wallabag\CoreBundle\Entity\Tag;
use Wallabag\CoreBundle\Helper\ContentProxy;
use Wallabag\CoreBundle\Helper\RuleBasedTagger;
use Wallabag\UserBundle\Entity\User;
@@ -105,15 +104,12 @@ class ContentProxyTest extends TestCase
->method('fetchContent')
->willReturn([
'html' => false,
- 'title' => '',
+ 'title' => 'my title',
'url' => '',
'content_type' => '',
'language' => '',
'status' => '',
- 'open_graph' => [
- 'og_title' => 'my title',
- 'og_description' => 'desc',
- ],
+ 'description' => 'desc',
]);
$proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
@@ -148,13 +144,12 @@ class ContentProxyTest extends TestCase
'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' => 'http://3.3.3.3/cover.jpg',
+ 'description' => 'OG desc',
+ 'image' => 'http://3.3.3.3/cover.jpg',
+ 'headers' => [
+ 'content-type' => 'text/html',
],
]);
@@ -164,7 +159,7 @@ class ContentProxyTest extends TestCase
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertSame('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
@@ -190,6 +185,47 @@ class ContentProxyTest extends TestCase
'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, $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->assertContains('content', $entry->getContent());
+ $this->assertNull($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());
+ }
+
+ public function testWithContentAndContentImage()
+ {
+ $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' => "
Test
",
+ 'title' => 'this is my title',
+ 'url' => 'http://1.1.1.1',
'content_type' => 'text/html',
'language' => 'fr',
'status' => '200',
@@ -206,12 +242,54 @@ class ContentProxyTest extends TestCase
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
- $this->assertNull($entry->getPreviewPicture());
+ $this->assertSame("Test
", $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(4.0, $entry->getReadingTime());
+ $this->assertSame(0.0, $entry->getReadingTime());
+ $this->assertSame('1.1.1.1', $entry->getDomainName());
+ }
+
+ public function testWithContentImageAndOgImage()
+ {
+ $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' => "Test
",
+ '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',
+ ],
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $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("Test
", $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());
}
@@ -237,9 +315,11 @@ class ContentProxyTest extends TestCase
'html' => str_repeat('this is my content', 325),
'title' => 'this is my title',
'url' => 'http://1.1.1.1',
- 'content_type' => 'text/html',
'language' => 'dontexist',
'status' => '200',
+ 'headers' => [
+ 'content-type' => 'text/html',
+ ],
]);
$proxy = new ContentProxy($graby, $tagger, $validator, $this->getLogger(), $this->fetchingErrorMessage);
@@ -248,7 +328,7 @@ class ContentProxyTest extends TestCase
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertNull($entry->getLanguage());
$this->assertSame('200', $entry->getHttpStatus());
@@ -284,10 +364,10 @@ class ContentProxyTest extends TestCase
'content_type' => 'text/html',
'language' => 'fr',
'status' => '200',
- 'open_graph' => [
- 'og_title' => 'my OG title',
- 'og_description' => 'OG desc',
- 'og_image' => 'https://',
+ 'description' => 'OG desc',
+ 'image' => 'https://',
+ 'headers' => [
+ 'content-type' => 'text/html',
],
]);
@@ -297,7 +377,7 @@ class ContentProxyTest extends TestCase
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertNull($entry->getPreviewPicture());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
@@ -321,19 +401,19 @@ class ContentProxyTest extends TestCase
'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',
+ 'headers' => [
+ 'cache-control' => 'no-cache',
+ 'content-type' => 'text/html',
],
]
);
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
@@ -364,15 +444,17 @@ class ContentProxyTest extends TestCase
'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',
+ 'headers' => [
+ 'content-type' => 'text/html',
+ ],
]
);
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
@@ -399,15 +481,17 @@ class ContentProxyTest extends TestCase
'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',
+ 'headers' => [
+ 'content-type' => 'text/html',
+ ],
]
);
$this->assertSame('http://1.1.1.1', $entry->getUrl());
$this->assertSame('this is my title', $entry->getTitle());
- $this->assertContains('this is my content', $entry->getContent());
+ $this->assertContains('content', $entry->getContent());
$this->assertSame('text/html', $entry->getMimetype());
$this->assertSame('fr', $entry->getLanguage());
$this->assertSame(4.0, $entry->getReadingTime());
@@ -416,7 +500,7 @@ class ContentProxyTest extends TestCase
$records = $handler->getRecords();
- $this->assertCount(1, $records);
+ $this->assertCount(3, $records);
$this->assertContains('Error while defining date', $records[0]['message']);
}
@@ -436,8 +520,10 @@ class ContentProxyTest extends TestCase
'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',
+ 'headers' => [
+ 'content-type' => 'text/html',
+ ],
]
);
@@ -476,13 +562,13 @@ class ContentProxyTest extends TestCase
'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',
+ //'og_title' => 'my OG title',
+ 'description' => 'OG desc',
+ 'image' => 'http://3.3.3.3/cover.jpg',
+ 'headers' => [
+ 'content-type' => 'text/html',
],
]
);
@@ -514,9 +600,10 @@ class ContentProxyTest extends TestCase
'html' => '',
'title' => 'this is my title',
'url' => 'http://1.1.1.1/image.jpg',
- 'content_type' => 'image/jpeg',
'status' => '200',
- 'open_graph' => [],
+ 'headers' => [
+ 'content-type' => 'image/jpeg',
+ ],
]);
$proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
@@ -532,6 +619,377 @@ class ContentProxyTest extends TestCase
$this->assertSame('1.1.1.1', $entry->getDomainName());
}
+ public function testWebsiteWithValidUTF8Title_doNothing()
+ {
+ // 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
+ // 'ð»â¤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');
+
+ $graby = $this->getMockBuilder('Graby\Graby')
+ ->setMethods(['fetchContent'])
+ ->disableOriginalConstructor()
+ ->getMock();
+
+ $graby->expects($this->any())
+ ->method('fetchContent')
+ ->willReturn([
+ 'html' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'text/html',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
+ $entry = new Entry(new User());
+ $proxy->updateEntry($entry, 'http://0.0.0.0');
+
+ // 'ð»â¤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()));
+ }
+
+ public function testWebsiteWithInvalidUTF8Title_removeInvalidCharacter()
+ {
+ // See http://graphemica.com for more info about the characters
+ // 'aâ¬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');
+
+ $graby = $this->getMockBuilder('Graby\Graby')
+ ->setMethods(['fetchContent'])
+ ->disableOriginalConstructor()
+ ->getMock();
+
+ $graby->expects($this->any())
+ ->method('fetchContent')
+ ->willReturn([
+ 'html' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'text/html',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $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()));
+ }
+
+ public function testPdfWithUTF16BETitle_convertToUTF8()
+ {
+ // See http://graphemica.com for more info about the characters
+ // 'ð»' (U+1F63B;D83DDE3B) in hexadecimal and as UTF16BE
+ $actualTitle = $this->hexToStr('D83DDE3B');
+
+ $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' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'application/pdf',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
+ $entry = new Entry(new User());
+ $proxy->updateEntry($entry, 'http://0.0.0.0');
+
+ // 'ð»' (U+1F63B or F09F98BB) in hexadecimal and UTF-8
+ $expectedTitle = 'F09F98BB';
+ $this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
+ }
+
+ public function testPdfWithUTF8Title_doNothing()
+ {
+ // See http://graphemica.com for more info about the characters
+ // 'ð»' (U+1F63B;D83DDE3B) in hexadecimal and as UTF8
+ $actualTitle = $this->hexToStr('F09F98BB');
+
+ $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' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'application/pdf',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
+ $entry = new Entry(new User());
+ $proxy->updateEntry($entry, 'http://0.0.0.0');
+
+ // 'ð»' (U+1F63B or F09F98BB) in hexadecimal and UTF-8
+ $expectedTitle = 'F09F98BB';
+ $this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
+ }
+
+ public function testPdfWithWINDOWS1252Title_convertToUTF8()
+ {
+ // See http://graphemica.com for more info about the characters
+ // 'â¬' (80) in hexadecimal and WINDOWS-1252
+ $actualTitle = $this->hexToStr('80');
+
+ $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' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'application/pdf',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
+ $entry = new Entry(new User());
+ $proxy->updateEntry($entry, 'http://0.0.0.0');
+
+ // 'â¬' (U+20AC or E282AC) in hexadecimal and UTF-8
+ $expectedTitle = 'E282AC';
+ $this->assertSame($expectedTitle, $this->strToHex($entry->getTitle()));
+ }
+
+ public function testPdfWithInvalidCharacterInTitle_removeInvalidCharacter()
+ {
+ // See http://graphemica.com for more info about the characters
+ // 'ð»â¤ï¿½z' (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');
+
+ $graby = $this->getMockBuilder('Graby\Graby')
+ ->setMethods(['fetchContent'])
+ ->disableOriginalConstructor()
+ ->getMock();
+
+ $graby->expects($this->any())
+ ->method('fetchContent')
+ ->willReturn([
+ 'html' => false,
+ 'title' => $actualTitle,
+ 'url' => '',
+ 'content_type' => 'application/pdf',
+ 'language' => '',
+ ]);
+
+ $proxy = new ContentProxy($graby, $tagger, $this->getValidator(), $this->getLogger(), $this->fetchingErrorMessage);
+ $entry = new Entry(new User());
+ $proxy->updateEntry($entry, 'http://0.0.0.0');
+
+ // 'ð»â¤z' (U+1F63B or F09F98BB; U+2124 or E284A4; U+007A or 7A) in hexadecimal and UTF-8
+ // the 0x81 (represented by �) 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
+ */
+ 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',
+ ],
+ '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',
+ ],
+ 'trailing slash' => [
+ 'https://example.com/hello-world',
+ null,
+ 'https://example.com/hello-world/',
+ 'https://example.com/hello-world/',
+ null,
+ 'example.com',
+ ],
+ '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',
+ ],
+ 'fragment in fetched content' => [
+ 'https://example.org/hello',
+ null,
+ 'https://example.org/hello#world',
+ 'https://example.org/hello',
+ null,
+ 'example.org',
+ ],
+ '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',
+ ],
+ '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',
+ ],
+ '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',
+ ],
+ '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',
+ ],
+ 'lemonde ignore pattern test' => [
+ 'http://www.lemonde.fr/tiny/url',
+ null,
+ 'http://example.com/hello-world',
+ 'http://example.com/hello-world',
+ null,
+ 'example.com',
+ ],
+ ];
+ }
+
+ /**
+ * @dataProvider dataForChangedUrl
+ */
+ public function testWithChangedUrl($entry_url, $origin_url, $content_url, $expected_entry_url, $expected_origin_url, $expected_domain)
+ {
+ $tagger = $this->getTaggerMock();
+ $tagger->expects($this->once())
+ ->method('tag');
+
+ $proxy = new ContentProxy((new Graby()), $tagger, $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,
+ '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.
+ *
+ * @param $string
+ *
+ * @return string
+ */
+ private function strToHex($string)
+ {
+ $hex = '';
+ for ($i = 0; $i < \strlen($string); ++$i) {
+ $ord = \ord($string[$i]);
+ $hexCode = dechex($ord);
+ $hex .= substr('0' . $hexCode, -2);
+ }
+
+ return strtoupper($hex);
+ }
+
+ /**
+ * https://stackoverflow.com/a/18506801.
+ *
+ * @param $hex
+ *
+ * @return string
+ */
+ private function hexToStr($hex)
+ {
+ $string = '';
+ for ($i = 0; $i < \strlen($hex) - 1; $i += 2) {
+ $string .= \chr(hexdec($hex[$i] . $hex[$i + 1]));
+ }
+
+ return $string;
+ }
+
private function getTaggerMock()
{
return $this->getMockBuilder(RuleBasedTagger::class)