public function testExtractsWithDocument()
{
$extractFile = __FILE__;
$expectedUrl = "http://localhost:8180/solr/update/extract?wt=json&json.nl=map&resource.name=ServiceTest.php&boost.field=2&literal.field=literal+value";
$expectedPostData = file_get_contents($extractFile);
$expectedContentType = 'application/octet-stream';
// default for extract
$expectedTimeout = false;
// set a mock transport
$mockTransport = $this->getMockHttpTransportInterface();
// setup expected call and response
$mockTransport->expects($this->once())->method('performPostRequest')->with($this->equalTo($expectedUrl), $this->equalTo($expectedPostData), $this->equalTo($expectedContentType), $this->equalTo($expectedTimeout))->will($this->returnValue(Apache_Solr_HttpTransport_ResponseTest::get200Response()));
$fixture = new Apache_Solr_Service();
$fixture->setHttpTransport($mockTransport);
$literals = new Apache_Solr_Document();
$literals->field = "literal value";
$literals->setFieldBoost('field', 2);
$fixture->extract($extractFile, null, $literals);
}
public function testExtractWithUrlDefers()
{
$extractUrl = "http://example.com";
$expectedUrl = "http://localhost:8180/solr/update/extract?resource.name=http%3A%2F%2Fexample.com&wt=json&json.nl=map";
$expectedPostData = Apache_Solr_HttpTransport_ResponseTest::BODY_200;
$expectedContentType = 'application/octet-stream';
// default for extract
$expectedTimeout = false;
// set a mock transport
$mockTransport = $this->getMockHttpTransportInterface();
// setup expected call and response
$mockTransport->expects($this->once())->method('performGetRequest')->with($this->equalTo($extractUrl))->will($this->returnValue(Apache_Solr_HttpTransport_ResponseTest::get200Response()));
$mockTransport->expects($this->once())->method('performPostRequest')->with($this->equalTo($expectedUrl), $this->equalTo($expectedPostData), $this->equalTo($expectedContentType), $this->equalTo($expectedTimeout))->will($this->returnValue(Apache_Solr_HttpTransport_ResponseTest::get200Response()));
$fixture = new Apache_Solr_Service();
$fixture->setHttpTransport($mockTransport);
$fixture->extract($extractUrl);
}
$d = null;
try {
$d = new Opus_Document($id);
} catch (Opus_Model_NotFoundException $e) {
// document with id $id does not exist
continue;
}
$files = $d->getFile();
if (count($files) == 0) {
continue;
}
foreach ($files as $file) {
$overallNumOfFulltexts++;
$response = null;
try {
$response = $solrServer->extract($file->getPath(), array('extractOnly' => 'true', 'extractFormat' => 'text'));
} catch (Exception $e) {
echo "error while extracting full text for document # " . $d->getId() . " (file name : " . $file->getPath() . " )\n";
$numOfNonExtractableFulltexts++;
continue;
}
if (is_null($response->getRawResponse()) || strlen(trim($response->getRawResponse())) == 0) {
echo "non-extractable full text for document # " . $d->getId() . " (file name: " . $file->getPath() . " )\n";
$numOfNonExtractableFulltexts++;
}
}
}
echo "overall num of full texts: {$overallNumOfFulltexts}\n";
$errorRate = 100.0 * $numOfNonExtractableFulltexts / $overallNumOfFulltexts;
echo "num of non extractable full texts: {$numOfNonExtractableFulltexts} ({$errorRate} %)\n";
exit;
