/**
* Returns a document that can be persisted based on the resource.
*
* @param \VDB\Spider\Resource $resource
*
* @return \Simgroep\ConcurrentSpiderBundle\PersistableDocument
*/
public function getDocumentByResource(Resource $resource)
{
switch ($resource->getResponse()->getContentType()) {
case 'application/pdf':
case 'application/octet-stream':
$data = $this->pdf->getData($resource);
break;
case 'application/msword':
case 'application/vnd.openxmlformats-officedocument.wordprocessingml.document':
case 'application/vnd.openxmlformats-officedocument.wordprocessingml.template':
if (false !== stripos($resource->getUri()->toString(), '.docx')) {
$data = $this->word2007->getData($resource);
break;
}
$data = $this->msdoc->getData($resource);
break;
case 'application/rtf':
$data = $this->rtf->getData($resource);
break;
case 'application/vnd.oasis.opendocument.text':
$data = $this->odt->getData($resource);
break;
case 'text/html':
default:
$data = $this->html->getData($resource);
break;
}
return new PersistableDocument($data);
}
/**
* @test
*/
public function retrieveValidDataFromResource()
{
$response = $this->getMockBuilder('Guzzle\\Http\\Message\\Response')->disableOriginalConstructor()->setMethods(['getBody', 'getLastModified'])->getMock();
$response->expects($this->once())->method('getLastModified')->will($this->returnValue('2015-06-18T23:49:41Z'));
$response->expects($this->once())->method('getBody')->will($this->returnValue(file_get_contents(__DIR__ . '/../../Mock/Documents/sample.docx')));
$uri = $this->getMockBuilder('VDB\\Uri\\Uri')->disableOriginalConstructor()->setMethods(['toString'])->getMock();
$uri->expects($this->exactly(2))->method('toString')->will($this->returnValue('http://blabdummy.de/dummydir/sample.docx'));
$crawler = new Crawler('', 'http://blabdummy.de/dummydir/sample.docx');
$resource = $this->getMockBuilder('VDB\\Spider\\Resource')->disableOriginalConstructor()->setMethods(['getResponse', 'getUri', 'getBody'])->getMock();
$resource->expects($this->exactly(2))->method('getResponse')->will($this->returnValue($response));
$resource->expects($this->exactly(2))->method('getUri')->will($this->returnValue($uri));
$type = new Word2007();
$data = $type->getData($resource);
$this->assertEquals(10, count($data));
$expectedKeys = ['id', 'url', 'content', 'title', 'tstamp', 'contentLength', 'lastModified', 'date', 'publishedDate', 'updatedDate'];
foreach ($expectedKeys as $expectedKey) {
$this->assertArrayHasKey($expectedKey, $data);
}
$this->assertEquals('sample.docx', $data['title']);
$this->assertNotEmpty($data, $data['content']);
}