/**
This function modifies the $endpoint_arr and adds each
of the statistics found here https://github.com/bio2rdf/bio2rdf-scripts/wiki/Bio2RDF-Dataset-Metrics
to the array
**/
function retrieveStatistics(&$endpoint_arr)
{
if (count($endpoint_arr)) {
foreach ($endpoint_arr as $name => $details) {
$endpoint_url = $details["endpoint_url"];
$graph_uri = $details["graph_uri"];
if (strlen($endpoint_url) != 0 && strlen($graph_uri) != 0) {
//now retrieve each of the stats
//numOfTriples
$numOfTriplesJson = trim(@file_get_contents(q1($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["triples"] = getNumOfTriples($numOfTriplesJson);
//numOfSubjects
$numOfSubjectsJson = trim(@file_get_contents(q2($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_subjects"] = getNumOfSubjects($numOfSubjectsJson);
//numOfPredicates
$numOfPredicatesJson = trim(@file_get_contents(q3($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_predicates"] = getNumOfPredicates($numOfPredicatesJson);
//numOfUniqueObjects
$numOfObjectsJson = trim(@file_get_contents(q4($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_objects"] = getNumOfObjects($numOfObjectsJson);
//numOfTypes
$numOfTypesJson = trim(@file_get_contents(q5($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_types"] = getNumOfTypes($numOfTypesJson);
//unique predicate-object links and their frequencies
$numOfPredObjectFreqsJson = trim(@file_get_contents(q6($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["predicate_object_links"] = getPredObjFreq($numOfPredObjectFreqsJson);
//unique predicate-literal links and their frequencies
$numOfUniquePredicateLiteralLinksandFreqsJson = trim(@file_get_contents(q7($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["predicate_literals"] = getPredLitLinks($numOfUniquePredicateLiteralLinksandFreqsJson);
//unique subject-predicate-unique object links and their frequencies
$numOfSubjectPredicateUniqueObjectJson = trim(@file_get_contents(q8($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_count_predicate_object_count"] = getSubPredObjLinks($numOfSubjectPredicateUniqueObjectJson);
//unique subject-predicate-unique literal links and their frequencies
$numOfSubjectPredUniqueLitJson = trim(@file_get_contents(q9($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_count_predicate_literal_count"] = getSubPredLitLinks($numOfSubjectPredUniqueLitJson);
//unique subject type-predicate-object type links and their frequencies
$numOfSubjectTypePredicateObjectJson = trim(@file_get_contents(q10($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_type_predicate_object_type"] = getSubTypePredObjType($numOfSubjectTypePredicateObjectJson);
}
}
}
return $endpoint_arr;
}
/**
This function modifies the $endpoint_arr and adds each
of the statistics found here https://github.com/bio2rdf/bio2rdf-scripts/wiki/Bio2RDF-Dataset-Metrics
to the array
**/
function retrieveStatistics(&$endpoint_arr)
{
$warn = "";
if (count($endpoint_arr)) {
foreach ($endpoint_arr as $name => $details) {
$endpoint_url = $details["endpoint_url"];
$graph_uri = $details["graph_uri"];
if (strlen($endpoint_url) != 0 && strlen($graph_uri) != 0) {
//now retrieve each of the stats
//nsns counts
$nsnsJSON = trim(@file_get_contents(nsQ($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["nsnscounts"] = getNSNSCounts($nsnsJSON);
//numOfTriples
$numOfTriplesJson = trim(@file_get_contents(q1($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["triples"] = getNumOfTriples($numOfTriplesJson);
//get the date
$dateJson = trim(@file_get_contents(getDatasetDateQuery($endpoint_url)));
$endpoint_arr[$name]['date'] = getDate2($dateJson);
//numOfSubjects
$numOfSubjectsJson = trim(@file_get_contents(q2($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_subjects"] = getNumOfSubjects($numOfSubjectsJson);
//numOfPredicates
$numOfPredicatesJson = trim(@file_get_contents(q3($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_predicates"] = getNumOfPredicates($numOfPredicatesJson);
//numOfUniqueObjects
$numOfObjectsJson = trim(@file_get_contents(q4($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_objects"] = getNumOfObjects($numOfObjectsJson);
//numOfTypes
$numOfTypesJson = trim(@file_get_contents(q5($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["unique_types"] = getNumOfTypes($numOfTypesJson);
//unique predicate-object links and their frequencies
$numOfPredObjectFreqsJson = trim(@file_get_contents(q6($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["predicate_object_links"] = getPredObjFreq($numOfPredObjectFreqsJson);
//unique predicate-literal links and their frequencies
$numOfUniquePredicateLiteralLinksandFreqsJson = trim(@file_get_contents(q7($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["predicate_literals"] = getPredLitLinks($numOfUniquePredicateLiteralLinksandFreqsJson);
//unique subject-predicate-unique object links and their frequencies
$numOfSubjectPredicateUniqueObjectJson = trim(@file_get_contents(q8($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_count_predicate_object_count"] = getSubPredObjLinks($numOfSubjectPredicateUniqueObjectJson);
//unique subject-predicate-unique literal links and their frequencies
$numOfSubjectPredUniqueLitJson = trim(@file_get_contents(q9($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_count_predicate_literal_count"] = getSubPredLitLinks($numOfSubjectPredUniqueLitJson);
//unique subject type-predicate-object type links and their frequencies
$numOfSubjectTypePredicateObjectJson = trim(@file_get_contents(q10($endpoint_url, $graph_uri)));
$endpoint_arr[$name]["subject_type_predicate_object_type"] = getSubTypePredObjType($numOfSubjectTypePredicateObjectJson);
} else {
$warn .= "WARNING :: Endpoint " . $name . " does not have all of required information! (missing either the endpoint or graph uri!)!\n";
}
}
if (strlen($warn)) {
echo $warn;
}
}
return $endpoint_arr;
}
