76 lines
2.6 KiB
PHP
76 lines
2.6 KiB
PHP
|
<?php
|
||
|
|
|
||
|
|
namespace Tests\Unit\Services\Parsers;
|
||
|
|
|
||
|
|
use App\Services\Parsers\VrtArticlePageParser;
|
||
|
|
use Tests\TestCase;
|
||
|
|
|
||
|
|
class VrtArticlePageParserTest extends TestCase
|
||
|
|
{
|
||
|
|
public function test_extract_thumbnail_from_og_image(): void
|
||
|
|
{
|
||
|
|
$html = '<html><head><meta property="og:image" content="https://images.vrt.be/image.jpg"/></head></html>';
|
||
|
|
|
||
|
|
$this->assertEquals(
|
||
|
|
'https://images.vrt.be/image.jpg',
|
||
|
|
VrtArticlePageParser::extractThumbnail($html)
|
||
|
|
);
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_thumbnail_falls_back_to_img_tag(): void
|
||
|
|
{
|
||
|
|
$html = '<html><body><img src="https://images.vrt.be/body.png" alt="test"/></body></html>';
|
||
|
|
|
||
|
|
$this->assertEquals(
|
||
|
|
'https://images.vrt.be/body.png',
|
||
|
|
VrtArticlePageParser::extractThumbnail($html)
|
||
|
|
);
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_thumbnail_prefers_og_image(): void
|
||
|
|
{
|
||
|
|
$html = '<html><head><meta property="og:image" content="https://images.vrt.be/og.jpg"/></head>'
|
||
|
|
.'<body><img src="https://images.vrt.be/body.png" alt="test"/></body></html>';
|
||
|
|
|
||
|
|
$this->assertEquals(
|
||
|
|
'https://images.vrt.be/og.jpg',
|
||
|
|
VrtArticlePageParser::extractThumbnail($html)
|
||
|
|
);
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_thumbnail_returns_null_when_not_found(): void
|
||
|
|
{
|
||
|
|
$this->assertNull(VrtArticlePageParser::extractThumbnail('<html><body>No images</body></html>'));
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_thumbnail_decodes_html_entities(): void
|
||
|
|
{
|
||
|
|
// Entity-encoded ampersands make the URL unusable downstream. See #117.
|
||
|
|
$html = '<html><head><meta property="og:image" content="https://images.vrt.be/orig.jpg?crop=1&w=1200"/></head></html>';
|
||
|
|
|
||
|
|
$thumbnail = VrtArticlePageParser::extractThumbnail($html);
|
||
|
|
|
||
|
|
$this->assertEquals('https://images.vrt.be/orig.jpg?crop=1&w=1200', $thumbnail);
|
||
|
|
$this->assertStringNotContainsString('&', (string) $thumbnail);
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_thumbnail_decodes_html_entities_in_img_tag(): void
|
||
|
|
{
|
||
|
|
$html = '<html><body><img src="https://images.vrt.be/pic.png?a=1&b=2" alt="test"/></body></html>';
|
||
|
|
|
||
|
|
$this->assertEquals(
|
||
|
|
'https://images.vrt.be/pic.png?a=1&b=2',
|
||
|
|
VrtArticlePageParser::extractThumbnail($html)
|
||
|
|
);
|
||
|
|
}
|
||
|
|
|
||
|
|
public function test_extract_data_returns_decoded_thumbnail(): void
|
||
|
|
{
|
||
|
|
$html = '<html><head><meta property="og:image" content="https://images.vrt.be/x.jpg?v=1&m=2"/></head></html>';
|
||
|
|
|
||
|
|
$data = VrtArticlePageParser::extractData($html);
|
||
|
|
|
||
|
|
$this->assertEquals('https://images.vrt.be/x.jpg?v=1&m=2', $data['thumbnail']);
|
||
|
|
}
|
||
|
|
}
|