'http://api.opencalais.com', 'contentType' => 'TEXT/HTML', 'outputFormat' => 'XML/RDF', 'externalID' => '', 'submitter' => 'Drupal', 'calculateRelevanceScore' => 'true', 'allowSearch' => 'false', 'allowDistribution' => 'false', 'caller' => 'Drupal', ); public $parameters; public $rdf; public $triples; public $flatTriples; public $keywords; /** * Constructs an instance of the Calais facade. * * Valid parameters are specified in the options array as key/value pairs with the * parameter name being the key and the parameter setting being the value * e.g. array('allowSearch' => 'false') * * @param options An array of parameter options for the Calais Web Service. * These will override the defaults. * * @see http://opencalais.com/APIcalls#inputparameters */ function __construct($options = array()) { $this->defaults['externalID'] = time(); $this->defaults['allowSearch'] = variable_get('calais_api_allow_searching', TRUE) ? 'true' : 'false'; $this->defaults['allowDistribution'] = variable_get('calais_api_allow_distribution', TRUE) ? 'true' : 'false'; $this->parameters = array_merge($this->defaults, $options); } /** * Analyze the provided content, passing it to Calais in XML format for more accurate data processing. * * @param $title The title of the content to process * @param $body The body ofd the content to process * @param $date The date of the content, if left blank/null analysis will use "today" * @return The processed Calais results. The raw RDF result is contained in the $this->rdf field. */ public function analyzeXML($title, $body, $date) { $content = $this->build_xml_content($title, $body, $date); $this->parameters['contentType'] = 'TEXT/XML'; return $this->analyze($content); } /** * Analyze the content via Calais. * * @param $content The content to ship off to Calais for analysis * @return The processed Calais results. The raw RDF result is contained in the $this->rdf field. */ public function analyze($content) { $headers = array('Content-Type' => 'application/x-www-form-urlencoded'); $data = array( 'licenseID' => variable_get('calais_api_key', NULL), 'content' => $content, 'paramsXML' => $this->build_xml_params(), ); $data_enc = http_build_query($data, '', '&'); $uri = $this->parameters['host'] . self::PATH; $ret = drupal_http_request($uri, $headers, 'POST', $data_enc); if (isset($ret->error)) { self::log_calais_error($ret); return array(); } $this->rdf = $ret->data; $keywords = $this->parse_rdf($this->rdf); return $keywords; } private static function log_calais_error($ret) { $msg = t('Calais processing error: @msg', array('@msg' => $ret->data)); drupal_set_message($msg, 'error'); watchdog('calais', 'Calais processing error: (@code - @error) @msg', array('@code' => $ret->code, '@error' => $ret->error, '@msg' => $ret->data), WATCHDOG_ERROR); } /** * Build the XML Parameters required by the Calais Web-Service * * @return XML document of Calais parameters. */ protected function build_xml_params() { $attributes = $this->parameters; $ret = << $attributes[caller] EOD; return $ret; } /** * Build the XML document request format expected by Calais * * @return an xml string to be submitted to Calais * @see http://opencalais.com/APIcalls#inputcontentformat */ protected function build_xml_content($title, $body, $date) { $req = ""; $req .= ""; $req .= "$date"; $req .= ""; $req .= ""; return $req; } /** * Parse the RDF. First render it into an indexed fashion, then have at it. We need to process * in 2 stages. The first stage identifies all of the entities, events, and facts. The second * stage then adds relevance and geo info to those previously identified terms. The 2nd pass i * is required b/c sometimes the relevance/geo data appears in the document before the term * has been identified. * * @param $rdf_xml * The RDF to parse * @return * An array of CalaisMetadata objects. * @see * CalaisMetadata.inc */ protected function parse_rdf($rdf_xml) { $parser = ARC2::getRDFXMLParser(); $parser->parse(NULL, $rdf_xml); $this->triples = $parser->getSimpleIndex(); // Use this method once http://drupal.org/node/348758 is resolved //$this->flatTriples = array_map('_rdf_deconstruct_arc2_triple', $parser->getTriples()); $this->flatTriples = array_map('_calais_api_deconstruct_arc2_triple', $parser->getTriples()); $this->keywords = new stdClass(); foreach ($this->triples as $guid => $indx) { $this->extract_entities($this->keywords, $guid, $indx); $this->extract_events($this->keywords, $guid, $indx); $this->extract_categories($this->keywords, $guid, $indx); } // Once we have the entities, then get additional metadata about them foreach ($this->triples as $guid => $indx) { $this->extract_relevance($this->keywords, $indx); $this->extract_geo($this->keywords, $indx); $this->disambiguate_company($this->keywords, $indx); } return $this->keywords; } /** * Extracts the entities from the returned data * * @param $keywords The result array for CalaisMetadata * @param $guid The guid for the current Calais Term * @param $indx The indexed triple for the current Calais Term/GUID */ protected function extract_entities(&$keywords, $guid, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && strpos($type[0], "http://s.opencalais.com/1/type/em/e") !== FALSE) { $keyword_type_guid = $type[0]; $keyword_type = preg_replace('/.*\//ims', '', $keyword_type_guid); $keyword_value = $indx['http://s.opencalais.com/1/pred/name'][0]; if (!property_exists($keywords, $keyword_type)) { $keywords->$keyword_type = new CalaisMetadata($keyword_type_guid, $keyword_type); } $metadata = &$keywords->$keyword_type; $metadata->add_term($guid, $keyword_value); } } /** * Extracts the events & facts from the returned data. For now it is considered best that all * Events & Facts are put into one Vocabulary and identified by their type. * * @param $keywords The result array for CalaisMetadata * @param $guid The guid for the current Calais Term * @param $indx The indexed triple for the current Calais Term/GUID */ protected function extract_events(&$keywords, $guid, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && strpos($type[0], 'http://s.opencalais.com/1/type/em/r') === 0) { $type_guid = $type[0]; $type_value = preg_replace('/.*\//ims', '', $type_guid); $type_value = calais_api_make_readable($type_value); if (!property_exists($keywords, 'EventsFacts')) { $keywords->EventsFacts = new CalaisMetadata('http://drupal.org/project/opencalais/EventsFacts', 'EventsFacts'); } $metadata = &$keywords->EventsFacts; // Not sure if the best number for relevance of an Event/Fact, for now 1 will always include $metadata->add_term($type_guid, $type_value, 1.000); } } /** * Extracts the document level categorization from the returned data. * * @param $keywords The result array for CalaisMetadata * @param $guid The guid for the current Calais Term * @param $indx The indexed triple for the current Calais Term/GUID */ protected function extract_categories(&$keywords, $guid, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && $type[0] == 'http://s.opencalais.com/1/type/cat/DocCat') { $cat_guid = $indx['http://s.opencalais.com/1/pred/category'][0]; $cat_val = $indx['http://s.opencalais.com/1/pred/categoryName'][0]; $cat_score = $indx['http://s.opencalais.com/1/pred/score'][0]; $cat_score = $cat_score ? $cat_score : 1.000; if (!property_exists($keywords, 'CalaisDocumentCategory')) { $keywords->CalaisDocumentCategory = new CalaisMetadata($type[0], 'CalaisDocumentCategory'); } $cat_val = preg_replace('/_.*/ims', '', $cat_val); // remove everything after the first underscore $keywords->CalaisDocumentCategory->add_term($cat_guid, $cat_val, $cat_score); } } /** * Extracts the relevance score from the returned data * * @param $keywords The result array for CalaisMetadata * @param $indx The indexed triple for the current Calais Term */ protected function extract_relevance(&$keywords, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && $type[0] == 'http://s.opencalais.com/1/type/sys/RelevanceInfo') { $subject = $indx[self::CALAIS_SUBJECT][0]; $relevance = $indx['http://s.opencalais.com/1/pred/relevance'][0]; $this->assign_relevance($keywords, $subject, $relevance); } } /** * Assign a relevance score to the supplied subject. */ protected function assign_relevance(&$keywords, $subject, $relevance) { foreach ($keywords as &$entity) { if ($entity->has_guid($subject)) { $entity->add_relevance($subject, $relevance); } } } /** * Extracts the geo disambiguation data * * @param $keywords The result array for CalaisMetadata * @param $indx The indexed triple for the current Calais Term */ protected function extract_geo(&$keywords, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && strpos($type[0], 'http://s.opencalais.com/1/type/er/Geo') === 0) { $subject = $indx[self::CALAIS_SUBJECT][0]; $normalized = $indx['http://s.opencalais.com/1/pred/name'][0]; $lat = $indx['http://s.opencalais.com/1/pred/latitude'][0]; $lon = $indx['http://s.opencalais.com/1/pred/longitude'][0]; foreach ($keywords as &$entity) { if ($entity->has_guid($subject)) { $entity->add_geo($subject, $normalized, $lat, $lon); } } } } /** * Extracts the disambiguated company name from the response * * @param $keywords The result array for CalaisMetadata * @param $indx The indexed triple for the current Calais Term */ protected function disambiguate_company(&$keywords, $indx) { $type = $indx[self::RDF_TYPE]; if (isset($type) && $type[0] == 'http://s.opencalais.com/1/type/er/Company') { $subject = $indx[self::CALAIS_SUBJECT][0]; $resolvedName = $indx['http://s.opencalais.com/1/pred/name'][0]; $keywords->Company->set_resolved_name($subject, $resolvedName); } } }