corpus_parole: comparison server/src/app/Repositories/RdfDocumentRepository.php

equal deleted inserted replaced

-:f2c2c80a49f7
+:5d2621f71f39
 * Implement the DocumentRepository using EasyRdf
 * TODO: certainly split the transaction management (+add, +delete +transaction ) to an external class -> for this extend the sparql client.
 */
 class RdfDocumentRepository implements DocumentRepository {
-const ALL_QUERIES = [
+const BASE_DOC_QUERY
-"SELECT".
+= "SELECT".
 "    ?uri".
 "    ?doc".
 "    ?title".
 "    ?issued".
 "    ?modified".
 "    ?lang".
 "    (group_concat(distinct ?publisher;separator=\", \") as ?publishers) ".
-"WHERE {".
+"  WHERE {".
-"GRAPH ?uri { ?doc a <http://www.europeana.eu/schemas/edm/ProvidedCHO>.".
+"  GRAPH ?uri { ?doc a <http://www.europeana.eu/schemas/edm/ProvidedCHO>.".
 "    ?doc <http://purl.org/dc/elements/1.1/title> ?title.".
 "    OPTIONAL {?doc <http://purl.org/dc/elements/1.1/language> ?lang.} ".
 "    OPTIONAL {?doc <http://purl.org/dc/terms/issued> ?issued.} ".
 "    OPTIONAL {?doc <http://purl.org/dc/terms/modified> ?modified.} ".
 "    OPTIONAL {?doc <http://purl.org/dc/elements/1.1/publisher> ?publisher.} }".
-"} ".
+"  } ".
-"GROUP BY ?uri ?doc ?title ?issued ?modified ?lang ".
+"  GROUP BY ?uri ?doc ?title ?issued ?modified ?lang ".
-"ORDER BY ?uri",
+"  ORDER BY ?uri";
+const ADDITIONAL_DOC_QUERIES = [
 "SELECT".
 "    ?uri".
 "    ?doc".
 "    (sample(distinct ?ext) as ?extent) ".
 "WHERE {".
 "    GRAPH ?uri {".
 "        ?s a <http://www.europeana.eu/schemas/edm/WebResource>. ".
 "        ?uri <http://www.europeana.eu/schemas/edm/isShownBy> ?s. ".
 "        ?uri <http://www.europeana.eu/schemas/edm/aggregatedCHO> ?doc. ".
 "        OPTIONAL {?s <http://purl.org/dc/terms/extent> ?ext.}".
-"    }".
+"    }. ".
+"    %s".
 "} ".
-"GROUP BY ?uri ?doc ".
+"GROUP BY ?uri ?doc",
-"ORDER BY ?uri"
+"SELECT".
+"    ?uri".
+"    ?doc".
+"    (sample(distinct str(?s)) as ?transcript_url) ".
+"WHERE {".
+"    GRAPH ?uri {".
+"        ?s a <http://www.europeana.eu/schemas/edm/WebResource>. ".
+"        ?uri <http://www.europeana.eu/schemas/edm/aggregatedCHO> ?doc. ".
+"        OPTIONAL {?s <http://purl.org/dc/elements/1.1/format> ?f.} ".
+"    }. ".
+"    FILTER(str(?f) IN ( \"application/xml\", \"application/pdf\" )). ".
+"    %s".
+"} ".
+"GROUP BY ?uri ?doc"
 ];
 private $sparqlClient;
 private $lexvoResolver;
 public function getSparqlClient() {
 return $this->sparqlClient;
 }
 private function getResGraph($doc) {
+if(empty((array)$doc)) {
+return null;
+}
 $newGraph = new Graph($doc->uri->getUri());
 $newGraph->add($doc->uri, "rdf:type", $newGraph->resource("http://www.openarchives.org/ore/terms/Aggregation"));
 $newGraph->add($doc->uri, "http://www.europeana.eu/schemas/edm/aggregatedCHO", $doc->doc);
 $newGraph->add($doc->doc, "rdf:type", $newGraph->resource("http://www.europeana.eu/schemas/edm/ProvidedCHO"));
 if(isset($doc->title)) {
 $newGraph->add($doc->doc, "http://purl.org/dc/elements/1.1/publisher", $doc->publishers);
 }
 if(isset($doc->extent)) {
 $newGraph->add($doc->doc, "http://purl.org/dc/terms/extent", $doc->extent);
 }
+if(isset($doc->transcript_url)) {
+$newGraph->add($doc->doc, config('corpusparole.corpus_ontology_url').'transcript', $doc->transcript_url);
+}
 return $newGraph;
 }
-private function queryDocs($queries) {
+private function queryDocs($offset=null, $limit=null) {
 $resDocs = [];
+$limitsClauses = [];
-foreach($queries as $query) {
+$limitsClausesStr = "";
-$docs = $this->sparqlClient->query($query);
+if(!is_null($offset)) {
+array_push($limitsClauses, "OFFSET $offset");
+}
+if(!is_null($limit)) {
+array_push($limitsClauses, "LIMIT $limit");
+}
+if(!empty($limitsClauses)) {
+$limitsClausesStr = "\n" . join(" ", $limitsClauses);
+}
+$docs = $this->sparqlClient->query(self::BASE_DOC_QUERY.$limitsClausesStr);
+foreach($docs as $doc) {
+$graph = $this->getResGraph($doc);
+if(is_null($graph)) {
+continue;
+}
+$uri = $doc->uri->getUri();
+$resDocs[$uri] = $graph;
+}
+if(count($resDocs) == 0) {
+return [];
+}
+$filterUris = "FILTER(?uri in (<".join(">, <" , array_keys($resDocs)).">)) ";
+foreach(self::ADDITIONAL_DOC_QUERIES as $query) {
+$docs = $this->sparqlClient->query(sprintf($query, $filterUris));
 foreach($docs as $doc) {
 $graph = $this->getResGraph($doc);
+if(is_null($graph)) {
+continue;
+}
 $uri = $doc->uri->getUri();
 if(array_key_exists($uri, $resDocs)) {
 $resDocs[$uri] = Utils::mergeGraphs($resDocs[$uri], $graph);
 } else {
 $resDocs[$uri] = $graph;
 }
 }
 }
 return array_map(function($g) { return new DocumentResult($g->getUri(), $g); }, array_values($resDocs));
 }
 public function all() {
-return $this->queryDocs(self::ALL_QUERIES);
+return $this->queryDocs();
 }
 public function get($id, bool $short=false) {
 if(strpos($id, config('corpusparole.corpus_id_scheme')) === 0) {
 $transactionStarted = $this->sparqlClient->startTransaction();
 try {
 foreach($doc->getDeltaList() as $delta) {
+$this->sparqlClient->deleteWhere($delta->getDeleteWhere(), $delta->getUri());
 $this->sparqlClient->delete($delta->getDeletedGraph());
 $this->sparqlClient->add($delta->getAddedGraph());
 }
 if($transactionStarted) {
 $transactionStarted = false;
 $total = $this->getCount();
 $offset = max(0,($page - 1) * $perPage);
-$results = $this->queryDocs(array_map(function($q) use ($offset, $perPage) { return $q . "\nOFFSET $offset LIMIT $perPage"; }, self::ALL_QUERIES));
+$results = $this->queryDocs($offset, $perPage);
 return new LengthAwarePaginator($results, $total, $perPage, $page, [
 'path' => Paginator::resolveCurrentPath(),
 'pageName' => $pageName,
 ]);

changeset 279	5d2621f71f39
parent 277	bd4bc1db4f40
child 306	3fccf43160a7