/**
* Create a new attachment document
*
* @param MidgardAttachment $attachment The Attachment to index.
* @param MidgardObject $source The source objece to which the attachment is bound.
*/
public function __construct($attachment, $source)
{
//before doing anything else, verify that the attachment is readable, otherwise we might get stuck in endless loops later on
$test = $attachment->open('r');
if (!$test) {
debug_add('Attachment ' . $attachment->guid . ' cannot be read, aborting. Last midgard error: ' . midcom_connection::get_error_string(), MIDCOM_LOG_ERROR);
return false;
} else {
fclose($test);
}
parent::__construct();
$this->_set_type('midcom_attachment');
$this->_attachment = $attachment;
$this->_source = $source;
debug_print_r("Processing this attachment:", $attachment);
$this->source = $this->_source->guid;
$this->RI = $this->_attachment->guid;
$this->document_url = "{$GLOBALS['midcom_config']['midcom_site_url']}midcom-serveattachmentguid-{$this->RI}/{$this->_attachment->name}";
$this->_process_attachment();
$this->_process_topic();
}
/**
* Query the index and, if set, restrict the query by a given filter.
*
* @param string $query The query, which must suite the backends query syntax.
* @param midcom_services_indexer_filter $filter An optional filter used to restrict the query. This may be null indicating no filter.
* @return Array An array of documents matching the query, or false on a failure.
*/
public function query($query, midcom_services_indexer_filter $filter = null)
{
if ($filter !== null) {
$query .= ' AND ' . $filter->get_query_string();
}
$url = "http://{$GLOBALS['midcom_config']['indexer_xmltcp_host']}:{$GLOBALS['midcom_config']['indexer_xmltcp_port']}/solr/select";
$request = new HTTP_Request2($url, HTTP_Request2::METHOD_GET);
$url = $request->getUrl();
// FIXME: Make this configurable, even better: adapt the whole indexer system to fetching enable querying for counts and slices
$maxrows = 1000;
$url->setQueryVariables(array('q' => $query, 'fl' => '*,score', 'rows' => $maxrows));
if (!empty($this->_index_name)) {
$url->setQueryVariable('fq', '__INDEX_NAME:"' . rawurlencode($this->_index_name) . '"');
}
$request->setHeader('Accept-Charset', 'UTF-8');
$request->setHeader('Content-type', 'text/xml; charset=utf-8');
try {
$response = $request->send();
} catch (Exception $e) {
debug_add("Failed to execute request " . $url . ": " . $e->getMessage(), MIDCOM_LOG_WARN);
return false;
}
$this->code = $response->getStatus();
if ($this->code != 200) {
debug_print_r($url . " returned response code {$this->code}, body:", $response->getBody());
return false;
}
$body = $response->getBody();
$response = DomDocument::loadXML($body);
$xquery = new DomXPath($response);
$result = array();
$num = $xquery->query('/response/result')->item(0);
if ($num->getAttribute('numFound') == 0) {
return array();
}
foreach ($xquery->query('/response/result/doc') as $res) {
$doc = new midcom_services_indexer_document();
foreach ($res->childNodes as $str) {
$name = $str->getAttribute('name');
$doc->add_result($name, $str->tagName == 'float' ? (double) $str->nodeValue : (string) $str->nodeValue);
if ($name == 'RI') {
$doc->add_result('__RI', $str->nodeValue);
}
if ($name == 'score' && $filter == null) {
$doc->score = (double) $str->nodeValue;
}
}
/* FIXME: before result slicing is properly supported this can be too heavy
if ( isset($doc->source)
&& mgd_is_guid($doc->source))
{
midcom::get('cache')->content->register($doc->source);
}
*/
$result[] = $doc;
}
debug_add(sprintf('Returning %d results', count($result)), MIDCOM_LOG_INFO);
return $result;
}
/**
* This will translate all member variables into appropriate
* field records, missing topic data is auto-detected
*/
public function members_to_fields()
{
if (empty($this->topic_guid) || empty($this->topic_url) || empty($this->component)) {
//if one of those is missing, we override all three to ensure consistency
$this->_process_topic();
}
parent::members_to_fields();
}
/**
* @dataProvider provider_html2text
*/
public function test_html2text($in, $out)
{
$document = new midcom_services_indexer_document();
$this->assertEquals($out, $document->html2text($in));
}
