NatLibFi /
Skosmos
| Total Complexity | 320 |
| Total Lines | 2301 |
| Duplicated Lines | 0 % |
| Changes | 13 | ||
| Bugs | 4 | Features | 1 |
Complex classes like GenericSparql often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use GenericSparql, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 6 | class GenericSparql { |
||
| 7 | /** |
||
| 8 | * A SPARQL Client eg. an EasyRDF instance. |
||
| 9 | * @property EasyRdf\Sparql\Client $client |
||
| 10 | */ |
||
| 11 | protected $client; |
||
| 12 | /** |
||
| 13 | * Graph uri. |
||
| 14 | * @property object $graph |
||
| 15 | */ |
||
| 16 | protected $graph; |
||
| 17 | /** |
||
| 18 | * A SPARQL query graph part template. |
||
| 19 | * @property string $graphClause |
||
| 20 | */ |
||
| 21 | protected $graphClause; |
||
| 22 | /** |
||
| 23 | * Model instance. |
||
| 24 | * @property Model $model |
||
| 25 | */ |
||
| 26 | protected $model; |
||
| 27 | |||
| 28 | /** |
||
| 29 | * Cache used to avoid expensive shorten() calls |
||
| 30 | * @property array $qnamecache |
||
| 31 | */ |
||
| 32 | private $qnamecache = array(); |
||
| 33 | |||
| 34 | /** |
||
| 35 | * Requires the following three parameters. |
||
| 36 | * @param string $endpoint SPARQL endpoint address. |
||
| 37 | * @param string|null $graph Which graph to query: Either an URI, the special value "?graph" |
||
| 38 | * to use the default graph, or NULL to not use a GRAPH clause. |
||
| 39 | * @param object $model a Model instance. |
||
| 40 | */ |
||
| 41 | public function __construct($endpoint, $graph, $model) { |
||
| 42 | $this->graph = $graph; |
||
| 43 | $this->model = $model; |
||
| 44 | |||
| 45 | // create the EasyRDF SPARQL client instance to use |
||
| 46 | $this->initializeHttpClient(); |
||
| 47 | $this->client = new EasyRdf\Sparql\Client($endpoint); |
||
| 48 | |||
| 49 | // set graphClause so that it can be used by all queries |
||
| 50 | if ($this->isDefaultEndpoint()) // default endpoint; query any graph (and catch it in a variable) |
||
| 51 | { |
||
| 52 | $this->graphClause = "GRAPH $graph"; |
||
| 53 | } elseif ($graph !== null) // query a specific graph |
||
| 54 | { |
||
| 55 | $this->graphClause = "GRAPH <$graph>"; |
||
| 56 | } else // query the default graph |
||
| 57 | { |
||
| 58 | $this->graphClause = ""; |
||
| 59 | } |
||
| 60 | |||
| 61 | } |
||
| 62 | |||
| 63 | /** |
||
| 64 | * Returns prefix-definitions for a query |
||
| 65 | * |
||
| 66 | * @param string $query |
||
| 67 | * @return string |
||
| 68 | */ |
||
| 69 | protected function generateQueryPrefixes($query) |
||
| 70 | { |
||
| 71 | // Check for undefined prefixes |
||
| 72 | $prefixes = ''; |
||
| 73 | foreach (EasyRdf\RdfNamespace::namespaces() as $prefix => $uri) { |
||
| 74 | if (strpos($query, "{$prefix}:") !== false and |
||
| 75 | strpos($query, "PREFIX {$prefix}:") === false |
||
| 76 | ) { |
||
| 77 | $prefixes .= "PREFIX {$prefix}: <{$uri}>\n"; |
||
| 78 | } |
||
| 79 | } |
||
| 80 | return $prefixes; |
||
| 81 | } |
||
| 82 | |||
| 83 | /** |
||
| 84 | * Execute the SPARQL query using the SPARQL client, logging it as well. |
||
| 85 | * @param string $query SPARQL query to perform |
||
| 86 | * @return \EasyRdf\Sparql\Result|\EasyRdf\Graph query result |
||
| 87 | */ |
||
| 88 | protected function query($query) { |
||
| 89 | $queryId = sprintf("%05d", rand(0, 99999)); |
||
| 90 | $logger = $this->model->getLogger(); |
||
| 91 | $logger->info("[qid $queryId] SPARQL query:\n" . $this->generateQueryPrefixes($query) . "\n$query\n"); |
||
| 92 | $starttime = microtime(true); |
||
| 93 | $result = $this->client->query($query); |
||
| 94 | $elapsed = intval(round((microtime(true) - $starttime) * 1000)); |
||
| 95 | if(method_exists($result, 'numRows')) { |
||
| 96 | $numRows = $result->numRows(); |
||
| 97 | $logger->info("[qid $queryId] result: $numRows rows returned in $elapsed ms"); |
||
| 98 | } else { // graph result |
||
| 99 | $numTriples = $result->countTriples(); |
||
| 100 | $logger->info("[qid $queryId] result: $numTriples triples returned in $elapsed ms"); |
||
| 101 | } |
||
| 102 | return $result; |
||
| 103 | } |
||
| 104 | |||
| 105 | |||
| 106 | /** |
||
| 107 | * Generates FROM clauses for the queries |
||
| 108 | * @param Vocabulary[]|null $vocabs |
||
| 109 | * @return string |
||
| 110 | */ |
||
| 111 | protected function generateFromClause($vocabs=null) { |
||
| 112 | $clause = ''; |
||
| 113 | if (!$vocabs) { |
||
| 114 | return $this->graph !== '?graph' && $this->graph !== NULL ? "FROM <$this->graph>" : ''; |
||
| 115 | } |
||
| 116 | $graphs = $this->getVocabGraphs($vocabs); |
||
| 117 | foreach ($graphs as $graph) { |
||
| 118 | $clause .= "FROM NAMED <$graph> "; |
||
| 119 | } |
||
| 120 | return $clause; |
||
| 121 | } |
||
| 122 | |||
| 123 | protected function initializeHttpClient() { |
||
| 124 | // configure the HTTP client used by EasyRdf\Sparql\Client |
||
| 125 | $httpclient = EasyRdf\Http::getDefaultHttpClient(); |
||
| 126 | $httpclient->setConfig(array('timeout' => $this->model->getConfig()->getSparqlTimeout())); |
||
| 127 | |||
| 128 | // if special cache control (typically no-cache) was requested by the |
||
| 129 | // client, set the same type of cache control headers also in subsequent |
||
| 130 | // in the SPARQL requests (this is useful for performance testing) |
||
| 131 | // @codeCoverageIgnoreStart |
||
| 132 | $cacheControl = filter_input(INPUT_SERVER, 'HTTP_CACHE_CONTROL', FILTER_SANITIZE_STRING); |
||
| 133 | $pragma = filter_input(INPUT_SERVER, 'HTTP_PRAGMA', FILTER_SANITIZE_STRING); |
||
| 134 | if ($cacheControl !== null || $pragma !== null) { |
||
| 135 | $val = $pragma !== null ? $pragma : $cacheControl; |
||
| 136 | $httpclient->setHeaders('Cache-Control', $val); |
||
| 137 | } |
||
| 138 | // @codeCoverageIgnoreEnd |
||
| 139 | |||
| 140 | EasyRdf\Http::setDefaultHttpClient($httpclient); // actually redundant.. |
||
| 141 | } |
||
| 142 | |||
| 143 | /** |
||
| 144 | * Return true if this is the default SPARQL endpoint, used as the facade to query |
||
| 145 | * all vocabularies. |
||
| 146 | */ |
||
| 147 | |||
| 148 | protected function isDefaultEndpoint() { |
||
| 149 | return !is_null($this->graph) && $this->graph[0] == '?'; |
||
| 150 | } |
||
| 151 | |||
| 152 | /** |
||
| 153 | * Returns the graph instance |
||
| 154 | * @return object EasyRDF graph instance. |
||
| 155 | */ |
||
| 156 | public function getGraph() { |
||
| 157 | return $this->graph; |
||
| 158 | } |
||
| 159 | |||
| 160 | /** |
||
| 161 | * Shorten a URI |
||
| 162 | * @param string $uri URI to shorten |
||
| 163 | * @return string shortened URI, or original URI if it cannot be shortened |
||
| 164 | */ |
||
| 165 | private function shortenUri($uri) { |
||
| 166 | if (!array_key_exists($uri, $this->qnamecache)) { |
||
| 167 | $res = new EasyRdf\Resource($uri); |
||
| 168 | $qname = $res->shorten(); // returns null on failure |
||
| 169 | $this->qnamecache[$uri] = ($qname !== null) ? $qname : $uri; |
||
| 170 | } |
||
| 171 | return $this->qnamecache[$uri]; |
||
| 172 | } |
||
| 173 | |||
| 174 | |||
| 175 | /** |
||
| 176 | * Generates the sparql query for retrieving concept and collection counts in a vocabulary. |
||
| 177 | * @return string sparql query |
||
| 178 | */ |
||
| 179 | private function generateCountConceptsQuery($array, $group) { |
||
| 180 | $fcl = $this->generateFromClause(); |
||
| 181 | $optional = $array ? "(<$array>) " : ''; |
||
| 182 | $optional .= $group ? "(<$group>)" : ''; |
||
| 183 | $query = <<<EOQ |
||
| 184 | SELECT (COUNT(DISTINCT(?conc)) as ?c) ?type ?typelabel (COUNT(?depr) as ?deprcount) $fcl WHERE { |
||
| 185 | VALUES (?value) { (skos:Concept) (skos:Collection) $optional } |
||
| 186 | ?type rdfs:subClassOf* ?value |
||
| 187 | { ?type ^a ?conc . |
||
| 188 | OPTIONAL { ?conc owl:deprecated ?depr . |
||
| 189 | FILTER (?depr = True) |
||
| 190 | } |
||
| 191 | } UNION {SELECT * WHERE { |
||
| 192 | ?type rdfs:label ?typelabel |
||
| 193 | } |
||
| 194 | } |
||
| 195 | } GROUP BY ?type ?typelabel |
||
| 196 | EOQ; |
||
| 197 | return $query; |
||
| 198 | } |
||
| 199 | |||
| 200 | /** |
||
| 201 | * Used for transforming the concept count query results. |
||
| 202 | * @param EasyRdf\Sparql\Result $result query results to be transformed |
||
| 203 | * @param string $lang language of labels |
||
| 204 | * @return Array containing the label counts |
||
| 205 | */ |
||
| 206 | private function transformCountConceptsResults($result, $lang) { |
||
| 207 | $ret = array(); |
||
| 208 | foreach ($result as $row) { |
||
| 209 | if (!isset($row->type)) { |
||
| 210 | continue; |
||
| 211 | } |
||
| 212 | $typeURI = $row->type->getUri(); |
||
| 213 | $ret[$typeURI]['type'] = $typeURI; |
||
| 214 | |||
| 215 | if (!isset($row->typelabel)) { |
||
| 216 | $ret[$typeURI]['count'] = $row->c->getValue(); |
||
| 217 | $ret[$typeURI]['deprecatedCount'] = $row->deprcount->getValue(); |
||
| 218 | } |
||
| 219 | |||
| 220 | if (isset($row->typelabel) && $row->typelabel->getLang() === $lang) { |
||
| 221 | $ret[$typeURI]['label'] = $row->typelabel->getValue(); |
||
| 222 | } |
||
| 223 | |||
| 224 | } |
||
| 225 | return $ret; |
||
| 226 | } |
||
| 227 | |||
| 228 | /** |
||
| 229 | * Used for counting number of concepts and collections in a vocabulary. |
||
| 230 | * @param string $lang language of labels |
||
| 231 | * @param string $array the uri of the concept array class, eg. isothes:ThesaurusArray |
||
| 232 | * @param string $group the uri of the concept group class, eg. isothes:ConceptGroup |
||
| 233 | * @return array with number of concepts in this vocabulary per label |
||
| 234 | */ |
||
| 235 | public function countConcepts($lang = null, $array = null, $group = null) { |
||
| 236 | $query = $this->generateCountConceptsQuery($array, $group); |
||
| 237 | $result = $this->query($query); |
||
| 238 | return $this->transformCountConceptsResults($result, $lang); |
||
| 239 | } |
||
| 240 | |||
| 241 | /** |
||
| 242 | * @param array $langs Languages to query for |
||
| 243 | * @param string[] $props property names |
||
| 244 | * @return string sparql query |
||
| 245 | */ |
||
| 246 | private function generateCountLangConceptsQuery($langs, $classes, $props) { |
||
| 247 | $gcl = $this->graphClause; |
||
| 248 | $classes = ($classes) ? $classes : array('http://www.w3.org/2004/02/skos/core#Concept'); |
||
| 249 | |||
| 250 | $quote_string = function($val) { return "'$val'"; }; |
||
| 251 | $quoted_values = array_map($quote_string, $langs); |
||
| 252 | $langFilter = "FILTER(?lang IN (" . implode(',', $quoted_values) . "))"; |
||
| 253 | |||
| 254 | $values = $this->formatValues('?type', $classes, 'uri'); |
||
| 255 | $valuesProp = $this->formatValues('?prop', $props, null); |
||
| 256 | |||
| 257 | $query = <<<EOQ |
||
| 258 | SELECT ?lang ?prop |
||
| 259 | (COUNT(?label) as ?count) |
||
| 260 | WHERE { |
||
| 261 | $gcl { |
||
| 262 | $values |
||
| 263 | $valuesProp |
||
| 264 | ?conc a ?type . |
||
| 265 | ?conc ?prop ?label . |
||
| 266 | BIND(LANG(?label) AS ?lang) |
||
| 267 | $langFilter |
||
| 268 | } |
||
| 269 | } |
||
| 270 | GROUP BY ?lang ?prop ?type |
||
| 271 | EOQ; |
||
| 272 | return $query; |
||
| 273 | } |
||
| 274 | |||
| 275 | /** |
||
| 276 | * Transforms the CountLangConcepts results into an array of label counts. |
||
| 277 | * @param EasyRdf\Sparql\Result $result query results to be transformed |
||
| 278 | * @param array $langs Languages to query for |
||
| 279 | * @param string[] $props property names |
||
| 280 | */ |
||
| 281 | private function transformCountLangConceptsResults($result, $langs, $props) { |
||
| 282 | $ret = array(); |
||
| 283 | // set default count to zero; overridden below if query found labels |
||
| 284 | foreach ($langs as $lang) { |
||
| 285 | foreach ($props as $prop) { |
||
| 286 | $ret[$lang][$prop] = 0; |
||
| 287 | } |
||
| 288 | } |
||
| 289 | foreach ($result as $row) { |
||
| 290 | if (isset($row->lang) && isset($row->prop) && isset($row->count)) { |
||
| 291 | $ret[$row->lang->getValue()][$row->prop->shorten()] += |
||
| 292 | $row->count->getValue(); |
||
| 293 | } |
||
| 294 | |||
| 295 | } |
||
| 296 | ksort($ret); |
||
| 297 | return $ret; |
||
| 298 | } |
||
| 299 | |||
| 300 | /** |
||
| 301 | * Counts the number of concepts in a easyRDF graph with a specific language. |
||
| 302 | * @param array $langs Languages to query for |
||
| 303 | * @return Array containing count of concepts for each language and property. |
||
| 304 | */ |
||
| 305 | public function countLangConcepts($langs, $classes = null) { |
||
| 306 | $props = array('skos:prefLabel', 'skos:altLabel', 'skos:hiddenLabel'); |
||
| 307 | $query = $this->generateCountLangConceptsQuery($langs, $classes, $props); |
||
| 308 | // Count the number of terms in each language |
||
| 309 | $result = $this->query($query); |
||
| 310 | return $this->transformCountLangConceptsResults($result, $langs, $props); |
||
| 311 | } |
||
| 312 | |||
| 313 | /** |
||
| 314 | * Formats a VALUES clause (SPARQL 1.1) which states that the variable should be bound to one |
||
| 315 | * of the constants given. |
||
| 316 | * @param string $varname variable name, e.g. "?uri" |
||
| 317 | * @param array $values the values |
||
| 318 | * @param string $type type of values: "uri", "literal" or null (determines quoting style) |
||
| 319 | */ |
||
| 320 | protected function formatValues($varname, $values, $type = null) { |
||
| 321 | $constants = array(); |
||
| 322 | foreach ($values as $val) { |
||
| 323 | if ($type == 'uri') { |
||
| 324 | $val = "<$val>"; |
||
| 325 | } |
||
| 326 | |||
| 327 | if ($type == 'literal') { |
||
| 328 | $val = "'$val'"; |
||
| 329 | } |
||
| 330 | |||
| 331 | $constants[] = "($val)"; |
||
| 332 | } |
||
| 333 | $values = implode(" ", $constants); |
||
| 334 | |||
| 335 | return "VALUES ($varname) { $values }"; |
||
| 336 | } |
||
| 337 | |||
| 338 | /** |
||
| 339 | * Filters multiple instances of the same vocabulary from the input array. |
||
| 340 | * @param \Vocabulary[]|null $vocabs array of Vocabulary objects |
||
| 341 | * @return \Vocabulary[] |
||
| 342 | */ |
||
| 343 | private function filterDuplicateVocabs($vocabs) { |
||
| 344 | // filtering duplicates |
||
| 345 | $uniqueVocabs = array(); |
||
| 346 | if ($vocabs !== null && sizeof($vocabs) > 0) { |
||
| 347 | foreach ($vocabs as $voc) { |
||
| 348 | $uniqueVocabs[$voc->getId()] = $voc; |
||
| 349 | } |
||
| 350 | } |
||
| 351 | |||
| 352 | return $uniqueVocabs; |
||
| 353 | } |
||
| 354 | |||
| 355 | /** |
||
| 356 | * Generates a sparql query for one or more concept URIs |
||
| 357 | * @param mixed $uris concept URI (string) or array of URIs |
||
| 358 | * @param string|null $arrayClass the URI for thesaurus array class, or null if not used |
||
| 359 | * @param \Vocabulary[]|null $vocabs array of Vocabulary objects |
||
| 360 | * @return string sparql query |
||
| 361 | */ |
||
| 362 | private function generateConceptInfoQuery($uris, $arrayClass, $vocabs) { |
||
| 363 | $gcl = $this->graphClause; |
||
| 364 | $fcl = empty($vocabs) ? '' : $this->generateFromClause($vocabs); |
||
| 365 | $values = $this->formatValues('?uri', $uris, 'uri'); |
||
| 366 | $uniqueVocabs = $this->filterDuplicateVocabs($vocabs); |
||
| 367 | $valuesGraph = empty($vocabs) ? $this->formatValuesGraph($uniqueVocabs) : ''; |
||
| 368 | |||
| 369 | if ($arrayClass === null) { |
||
| 370 | $construct = $optional = ""; |
||
| 371 | } else { |
||
| 372 | // add information that can be used to format narrower concepts by |
||
| 373 | // the array they belong to ("milk by source animal" use case) |
||
| 374 | $construct = "\n ?x skos:member ?o . ?x skos:prefLabel ?xl . ?x a <$arrayClass> ."; |
||
| 375 | $optional = "\n OPTIONAL { |
||
| 376 | ?x skos:member ?o . |
||
| 377 | ?x a <$arrayClass> . |
||
| 378 | ?x skos:prefLabel ?xl . |
||
| 379 | FILTER NOT EXISTS { |
||
| 380 | ?x skos:member ?other . |
||
| 381 | MINUS { ?other skos:broader ?uri } |
||
| 382 | } |
||
| 383 | }"; |
||
| 384 | } |
||
| 385 | $query = <<<EOQ |
||
| 386 | CONSTRUCT { |
||
| 387 | ?s ?p ?uri . |
||
| 388 | ?sp ?uri ?op . |
||
| 389 | ?uri ?p ?o . |
||
| 390 | ?p rdfs:label ?proplabel . |
||
| 391 | ?p rdfs:comment ?propcomm . |
||
| 392 | ?p skos:definition ?propdef . |
||
| 393 | ?p rdfs:subPropertyOf ?pp . |
||
| 394 | ?pp rdfs:label ?plabel . |
||
| 395 | ?o a ?ot . |
||
| 396 | ?o skos:prefLabel ?opl . |
||
| 397 | ?o rdfs:label ?ol . |
||
| 398 | ?o rdf:value ?ov . |
||
| 399 | ?o skos:notation ?on . |
||
| 400 | ?o ?oprop ?oval . |
||
| 401 | ?o ?xlprop ?xlval . |
||
| 402 | ?directgroup skos:member ?uri . |
||
| 403 | ?parent skos:member ?group . |
||
| 404 | ?group skos:prefLabel ?grouplabel . |
||
| 405 | ?b1 rdf:first ?item . |
||
| 406 | ?b1 rdf:rest ?b2 . |
||
| 407 | ?item a ?it . |
||
| 408 | ?item skos:prefLabel ?il . |
||
| 409 | ?group a ?grouptype . $construct |
||
| 410 | } $fcl WHERE { |
||
| 411 | $values |
||
| 412 | $gcl { |
||
| 413 | { |
||
| 414 | ?s ?p ?uri . |
||
| 415 | FILTER(!isBlank(?s)) |
||
| 416 | FILTER(?p != skos:inScheme) |
||
| 417 | FILTER NOT EXISTS { ?s owl:deprecated true . } |
||
| 418 | } |
||
| 419 | UNION |
||
| 420 | { ?sp ?uri ?op . } |
||
| 421 | UNION |
||
| 422 | { |
||
| 423 | ?directgroup skos:member ?uri . |
||
| 424 | ?group skos:member+ ?uri . |
||
| 425 | ?group skos:prefLabel ?grouplabel . |
||
| 426 | ?group a ?grouptype . |
||
| 427 | OPTIONAL { ?parent skos:member ?group } |
||
| 428 | } |
||
| 429 | UNION |
||
| 430 | { |
||
| 431 | ?uri ?p ?o . |
||
| 432 | OPTIONAL { |
||
| 433 | ?o rdf:rest* ?b1 . |
||
| 434 | ?b1 rdf:first ?item . |
||
| 435 | ?b1 rdf:rest ?b2 . |
||
| 436 | OPTIONAL { ?item a ?it . } |
||
| 437 | OPTIONAL { ?item skos:prefLabel ?il . } |
||
| 438 | } |
||
| 439 | OPTIONAL { |
||
| 440 | { ?p rdfs:label ?proplabel . } |
||
| 441 | UNION |
||
| 442 | { ?p rdfs:comment ?propcomm . } |
||
| 443 | UNION |
||
| 444 | { ?p skos:definition ?propdef . } |
||
| 445 | UNION |
||
| 446 | { ?p rdfs:subPropertyOf ?pp . } |
||
| 447 | } |
||
| 448 | OPTIONAL { |
||
| 449 | { ?o a ?ot . } |
||
| 450 | UNION |
||
| 451 | { ?o skos:prefLabel ?opl . } |
||
| 452 | UNION |
||
| 453 | { ?o rdfs:label ?ol . } |
||
| 454 | UNION |
||
| 455 | { ?o rdf:value ?ov . |
||
| 456 | OPTIONAL { ?o ?oprop ?oval . } |
||
| 457 | } |
||
| 458 | UNION |
||
| 459 | { ?o skos:notation ?on . } |
||
| 460 | UNION |
||
| 461 | { ?o a skosxl:Label . |
||
| 462 | ?o ?xlprop ?xlval } |
||
| 463 | } $optional |
||
| 464 | } |
||
| 465 | } |
||
| 466 | } |
||
| 467 | $valuesGraph |
||
| 468 | EOQ; |
||
| 469 | return $query; |
||
| 470 | } |
||
| 471 | |||
| 472 | /** |
||
| 473 | * Transforms ConceptInfo query results into an array of Concept objects |
||
| 474 | * @param EasyRdf\Graph $result query results to be transformed |
||
| 475 | * @param array $uris concept URIs |
||
| 476 | * @param \Vocabulary[] $vocabs array of Vocabulary object |
||
| 477 | * @param string|null $clang content language |
||
| 478 | * @return Concept[] array of Concept objects |
||
| 479 | */ |
||
| 480 | private function transformConceptInfoResults($result, $uris, $vocabs, $clang) { |
||
| 481 | $conceptArray = array(); |
||
| 482 | foreach ($uris as $index => $uri) { |
||
| 483 | $conc = $result->resource($uri); |
||
| 484 | if (is_array($vocabs)) { |
||
| 485 | $vocab = (sizeof($vocabs) == 1) ? $vocabs[0] : $vocabs[$index]; |
||
| 486 | } else { |
||
| 487 | $vocab = null; |
||
| 488 | } |
||
| 489 | $conceptArray[] = new Concept($this->model, $vocab, $conc, $result, $clang); |
||
| 490 | } |
||
| 491 | return $conceptArray; |
||
| 492 | } |
||
| 493 | |||
| 494 | /** |
||
| 495 | * Returns information (as a graph) for one or more concept URIs |
||
| 496 | * @param mixed $uris concept URI (string) or array of URIs |
||
| 497 | * @param string|null $arrayClass the URI for thesaurus array class, or null if not used |
||
| 498 | * @param \Vocabulary[]|null $vocabs vocabularies to target |
||
| 499 | * @return \EasyRdf\Graph |
||
| 500 | */ |
||
| 501 | public function queryConceptInfoGraph($uris, $arrayClass = null, $vocabs = array()) { |
||
| 502 | // if just a single URI is given, put it in an array regardless |
||
| 503 | if (!is_array($uris)) { |
||
| 504 | $uris = array($uris); |
||
| 505 | } |
||
| 506 | |||
| 507 | $query = $this->generateConceptInfoQuery($uris, $arrayClass, $vocabs); |
||
| 508 | $result = $this->query($query); |
||
| 509 | return $result; |
||
| 510 | } |
||
| 511 | |||
| 512 | /** |
||
| 513 | * Returns information (as an array of Concept objects) for one or more concept URIs |
||
| 514 | * @param mixed $uris concept URI (string) or array of URIs |
||
| 515 | * @param string|null $arrayClass the URI for thesaurus array class, or null if not used |
||
| 516 | * @param \Vocabulary[] $vocabs vocabularies to target |
||
| 517 | * @param string|null $clang content language |
||
| 518 | * @return Concept[] |
||
| 519 | */ |
||
| 520 | public function queryConceptInfo($uris, $arrayClass = null, $vocabs = array(), $clang = null) { |
||
| 521 | // if just a single URI is given, put it in an array regardless |
||
| 522 | if (!is_array($uris)) { |
||
| 523 | $uris = array($uris); |
||
| 524 | } |
||
| 525 | $result = $this->queryConceptInfoGraph($uris, $arrayClass, $vocabs); |
||
| 526 | if ($result->isEmpty()) { |
||
| 527 | return []; |
||
| 528 | } |
||
| 529 | return $this->transformConceptInfoResults($result, $uris, $vocabs, $clang); |
||
| 530 | } |
||
| 531 | |||
| 532 | /** |
||
| 533 | * Generates the sparql query for queryTypes |
||
| 534 | * @param string $lang |
||
| 535 | * @return string sparql query |
||
| 536 | */ |
||
| 537 | private function generateQueryTypesQuery($lang) { |
||
| 538 | $fcl = $this->generateFromClause(); |
||
| 539 | $query = <<<EOQ |
||
| 540 | SELECT DISTINCT ?type ?label ?superclass $fcl |
||
| 541 | WHERE { |
||
| 542 | { |
||
| 543 | { BIND( skos:Concept as ?type ) } |
||
| 544 | UNION |
||
| 545 | { BIND( skos:Collection as ?type ) } |
||
| 546 | UNION |
||
| 547 | { BIND( isothes:ConceptGroup as ?type ) } |
||
| 548 | UNION |
||
| 549 | { BIND( isothes:ThesaurusArray as ?type ) } |
||
| 550 | UNION |
||
| 551 | { ?type rdfs:subClassOf/rdfs:subClassOf* skos:Concept . } |
||
| 552 | UNION |
||
| 553 | { ?type rdfs:subClassOf/rdfs:subClassOf* skos:Collection . } |
||
| 554 | } |
||
| 555 | OPTIONAL { |
||
| 556 | ?type rdfs:label ?label . |
||
| 557 | FILTER(langMatches(lang(?label), '$lang')) |
||
| 558 | } |
||
| 559 | OPTIONAL { |
||
| 560 | ?type rdfs:subClassOf ?superclass . |
||
| 561 | } |
||
| 562 | FILTER EXISTS { |
||
| 563 | ?s a ?type . |
||
| 564 | ?s skos:prefLabel ?prefLabel . |
||
| 565 | } |
||
| 566 | } |
||
| 567 | EOQ; |
||
| 568 | return $query; |
||
| 569 | } |
||
| 570 | |||
| 571 | /** |
||
| 572 | * Transforms the results into an array format. |
||
| 573 | * @param EasyRdf\Sparql\Result $result |
||
| 574 | * @return array Array with URIs (string) as key and array of (label, superclassURI) as value |
||
| 575 | */ |
||
| 576 | private function transformQueryTypesResults($result) { |
||
| 577 | $ret = array(); |
||
| 578 | foreach ($result as $row) { |
||
| 579 | $type = array(); |
||
| 580 | if (isset($row->label)) { |
||
| 581 | $type['label'] = $row->label->getValue(); |
||
| 582 | } |
||
| 583 | |||
| 584 | if (isset($row->superclass)) { |
||
| 585 | $type['superclass'] = $row->superclass->getUri(); |
||
| 586 | } |
||
| 587 | |||
| 588 | $ret[$row->type->getURI()] = $type; |
||
| 589 | } |
||
| 590 | return $ret; |
||
| 591 | } |
||
| 592 | |||
| 593 | /** |
||
| 594 | * Retrieve information about types from the endpoint |
||
| 595 | * @param string $lang |
||
| 596 | * @return array Array with URIs (string) as key and array of (label, superclassURI) as value |
||
| 597 | */ |
||
| 598 | public function queryTypes($lang) { |
||
| 599 | $query = $this->generateQueryTypesQuery($lang); |
||
| 600 | $result = $this->query($query); |
||
| 601 | return $this->transformQueryTypesResults($result); |
||
| 602 | } |
||
| 603 | |||
| 604 | /** |
||
| 605 | * Generates the concept scheme query. |
||
| 606 | * @param string $conceptscheme concept scheme URI |
||
| 607 | * @return string sparql query |
||
| 608 | */ |
||
| 609 | private function generateQueryConceptSchemeQuery($conceptscheme) { |
||
| 610 | $fcl = $this->generateFromClause(); |
||
| 611 | $query = <<<EOQ |
||
| 612 | CONSTRUCT { |
||
| 613 | <$conceptscheme> ?property ?value . |
||
| 614 | } $fcl WHERE { |
||
| 615 | <$conceptscheme> ?property ?value . |
||
| 616 | FILTER (?property != skos:hasTopConcept) |
||
| 617 | } |
||
| 618 | EOQ; |
||
| 619 | return $query; |
||
| 620 | } |
||
| 621 | |||
| 622 | /** |
||
| 623 | * Retrieves conceptScheme information from the endpoint. |
||
| 624 | * @param string $conceptscheme concept scheme URI |
||
| 625 | * @return \EasyRdf\Sparql\Result|\EasyRdf\Graph query result graph |
||
| 626 | */ |
||
| 627 | public function queryConceptScheme($conceptscheme) { |
||
| 628 | $query = $this->generateQueryConceptSchemeQuery($conceptscheme); |
||
| 629 | return $this->query($query); |
||
| 630 | } |
||
| 631 | |||
| 632 | /** |
||
| 633 | * Generates the queryConceptSchemes sparql query. |
||
| 634 | * @param string $lang language of labels |
||
| 635 | * @return string sparql query |
||
| 636 | */ |
||
| 637 | private function generateQueryConceptSchemesQuery($lang) { |
||
| 638 | $fcl = $this->generateFromClause(); |
||
| 639 | $query = <<<EOQ |
||
| 640 | SELECT ?cs ?label ?preflabel ?title ?domain ?domainLabel $fcl |
||
| 641 | WHERE { |
||
| 642 | ?cs a skos:ConceptScheme . |
||
| 643 | OPTIONAL{ |
||
| 644 | ?cs dcterms:subject ?domain. |
||
| 645 | ?domain skos:prefLabel ?domainLabel. |
||
| 646 | FILTER(langMatches(lang(?domainLabel), '$lang')) |
||
| 647 | } |
||
| 648 | OPTIONAL { |
||
| 649 | ?cs rdfs:label ?label . |
||
| 650 | FILTER(langMatches(lang(?label), '$lang')) |
||
| 651 | } |
||
| 652 | OPTIONAL { |
||
| 653 | ?cs skos:prefLabel ?preflabel . |
||
| 654 | FILTER(langMatches(lang(?preflabel), '$lang')) |
||
| 655 | } |
||
| 656 | OPTIONAL { |
||
| 657 | { ?cs dc11:title ?title } |
||
| 658 | UNION |
||
| 659 | { ?cs dc:title ?title } |
||
| 660 | FILTER(langMatches(lang(?title), '$lang')) |
||
| 661 | } |
||
| 662 | } |
||
| 663 | ORDER BY ?cs |
||
| 664 | EOQ; |
||
| 665 | return $query; |
||
| 666 | } |
||
| 667 | |||
| 668 | /** |
||
| 669 | * Transforms the queryConceptScheme results into an array format. |
||
| 670 | * @param EasyRdf\Sparql\Result $result |
||
| 671 | * @return array |
||
| 672 | */ |
||
| 673 | private function transformQueryConceptSchemesResults($result) { |
||
| 674 | $ret = array(); |
||
| 675 | foreach ($result as $row) { |
||
| 676 | $conceptscheme = array(); |
||
| 677 | if (isset($row->label)) { |
||
| 678 | $conceptscheme['label'] = $row->label->getValue(); |
||
| 679 | } |
||
| 680 | |||
| 681 | if (isset($row->preflabel)) { |
||
| 682 | $conceptscheme['prefLabel'] = $row->preflabel->getValue(); |
||
| 683 | } |
||
| 684 | |||
| 685 | if (isset($row->title)) { |
||
| 686 | $conceptscheme['title'] = $row->title->getValue(); |
||
| 687 | } |
||
| 688 | // add dct:subject and their labels in the result |
||
| 689 | if(isset($row->domain) && isset($row->domainLabel)){ |
||
| 690 | $conceptscheme['subject']['uri']=$row->domain->getURI(); |
||
| 691 | $conceptscheme['subject']['prefLabel']=$row->domainLabel->getValue(); |
||
| 692 | } |
||
| 693 | |||
| 694 | $ret[$row->cs->getURI()] = $conceptscheme; |
||
| 695 | } |
||
| 696 | return $ret; |
||
| 697 | } |
||
| 698 | |||
| 699 | /** |
||
| 700 | * return a list of skos:ConceptScheme instances in the given graph |
||
| 701 | * @param string $lang language of labels |
||
| 702 | * @return array Array with concept scheme URIs (string) as keys and labels (string) as values |
||
| 703 | */ |
||
| 704 | public function queryConceptSchemes($lang) { |
||
| 705 | $query = $this->generateQueryConceptSchemesQuery($lang); |
||
| 706 | $result = $this->query($query); |
||
| 707 | return $this->transformQueryConceptSchemesResults($result); |
||
| 708 | } |
||
| 709 | |||
| 710 | /** |
||
| 711 | * Generate a VALUES clause for limiting the targeted graphs. |
||
| 712 | * @param Vocabulary[]|null $vocabs the vocabularies to target |
||
| 713 | * @return string[] array of graph URIs |
||
| 714 | */ |
||
| 715 | protected function getVocabGraphs($vocabs) { |
||
| 716 | if ($vocabs === null || sizeof($vocabs) == 0) { |
||
| 717 | // searching from all vocabularies - limit to known graphs |
||
| 718 | $vocabs = $this->model->getVocabularies(); |
||
| 719 | } |
||
| 720 | $graphs = array(); |
||
| 721 | foreach ($vocabs as $voc) { |
||
| 722 | $graph = $voc->getGraph(); |
||
| 723 | if (!is_null($graph) && !in_array($graph, $graphs)) { |
||
| 724 | $graphs[] = $graph; |
||
| 725 | } |
||
| 726 | } |
||
| 727 | return $graphs; |
||
| 728 | } |
||
| 729 | |||
| 730 | /** |
||
| 731 | * Generate a VALUES clause for limiting the targeted graphs. |
||
| 732 | * @param Vocabulary[]|null $vocabs array of Vocabulary objects to target |
||
| 733 | * @return string VALUES clause, or "" if not necessary to limit |
||
| 734 | */ |
||
| 735 | protected function formatValuesGraph($vocabs) { |
||
| 736 | if (!$this->isDefaultEndpoint()) { |
||
| 737 | return ""; |
||
| 738 | } |
||
| 739 | $graphs = $this->getVocabGraphs($vocabs); |
||
| 740 | return $this->formatValues('?graph', $graphs, 'uri'); |
||
| 741 | } |
||
| 742 | |||
| 743 | /** |
||
| 744 | * Generate a FILTER clause for limiting the targeted graphs. |
||
| 745 | * @param array $vocabs array of Vocabulary objects to target |
||
| 746 | * @return string FILTER clause, or "" if not necessary to limit |
||
| 747 | */ |
||
| 748 | protected function formatFilterGraph($vocabs) { |
||
| 749 | if (!$this->isDefaultEndpoint()) { |
||
| 750 | return ""; |
||
| 751 | } |
||
| 752 | $graphs = $this->getVocabGraphs($vocabs); |
||
| 753 | $values = array(); |
||
| 754 | foreach ($graphs as $graph) { |
||
| 755 | $values[] = "<$graph>"; |
||
| 756 | } |
||
| 757 | if (count($values)) { |
||
| 758 | return "FILTER (?graph IN (" . implode(',', $values) . "))"; |
||
| 759 | } |
||
| 760 | } |
||
| 761 | |||
| 762 | /** |
||
| 763 | * Formats combined limit and offset clauses for the sparql query |
||
| 764 | * @param int $limit maximum number of hits to retrieve; 0 for unlimited |
||
| 765 | * @param int $offset offset of results to retrieve; 0 for beginning of list |
||
| 766 | * @return string sparql query clauses |
||
| 767 | */ |
||
| 768 | protected function formatLimitAndOffset($limit, $offset) { |
||
| 769 | $limit = ($limit) ? 'LIMIT ' . $limit : ''; |
||
| 770 | $offset = ($offset) ? 'OFFSET ' . $offset : ''; |
||
| 771 | // eliminating whitespace and line changes when the conditions aren't needed. |
||
| 772 | $limitandoffset = ''; |
||
| 773 | if ($limit && $offset) { |
||
| 774 | $limitandoffset = "\n" . $limit . "\n" . $offset; |
||
| 775 | } elseif ($limit) { |
||
| 776 | $limitandoffset = "\n" . $limit; |
||
| 777 | } elseif ($offset) { |
||
| 778 | $limitandoffset = "\n" . $offset; |
||
| 779 | } |
||
| 780 | |||
| 781 | return $limitandoffset; |
||
| 782 | } |
||
| 783 | |||
| 784 | /** |
||
| 785 | * Formats a sparql query clause for limiting the search to specific concept types. |
||
| 786 | * @param array $types limit search to concepts of the given type(s) |
||
| 787 | * @return string sparql query clause |
||
| 788 | */ |
||
| 789 | protected function formatTypes($types) { |
||
| 790 | $typePatterns = array(); |
||
| 791 | if (!empty($types)) { |
||
| 792 | foreach ($types as $type) { |
||
| 793 | $unprefixed = EasyRdf\RdfNamespace::expand($type); |
||
| 794 | $typePatterns[] = "{ ?s a <$unprefixed> }"; |
||
| 795 | } |
||
| 796 | } |
||
| 797 | |||
| 798 | return implode(' UNION ', $typePatterns); |
||
| 799 | } |
||
| 800 | |||
| 801 | /** |
||
| 802 | * @param string $prop property to include in the result eg. 'broader' or 'narrower' |
||
| 803 | * @return string sparql query clause |
||
| 804 | */ |
||
| 805 | private function formatPropertyCsvClause($prop) { |
||
| 806 | # This expression creates a CSV row containing pairs of (uri,prefLabel) values. |
||
| 807 | # The REPLACE is performed for quotes (" -> "") so they don't break the CSV format. |
||
| 808 | $clause = <<<EOV |
||
| 809 | (GROUP_CONCAT(DISTINCT CONCAT( |
||
| 810 | '"', IF(isIRI(?$prop),STR(?$prop),''), '"', ',', |
||
| 811 | '"', REPLACE(IF(BOUND(?{$prop}lab),?{$prop}lab,''), '"', '""'), '"', ',', |
||
| 812 | '"', REPLACE(IF(isLiteral(?{$prop}),?{$prop},''), '"', '""'), '"' |
||
| 813 | ); separator='\\n') as ?{$prop}s) |
||
| 814 | EOV; |
||
| 815 | return $clause; |
||
| 816 | } |
||
| 817 | |||
| 818 | /** |
||
| 819 | * @return string sparql query clause |
||
| 820 | */ |
||
| 821 | private function formatPrefLabelCsvClause() { |
||
| 822 | # This expression creates a CSV row containing pairs of (prefLabel, lang) values. |
||
| 823 | # The REPLACE is performed for quotes (" -> "") so they don't break the CSV format. |
||
| 824 | $clause = <<<EOV |
||
| 825 | (GROUP_CONCAT(DISTINCT CONCAT( |
||
| 826 | '"', STR(?pref), '"', ',', '"', lang(?pref), '"' |
||
| 827 | ); separator='\\n') as ?preflabels) |
||
| 828 | EOV; |
||
| 829 | return $clause; |
||
| 830 | } |
||
| 831 | |||
| 832 | /** |
||
| 833 | * @param string $lang language code of the returned labels |
||
| 834 | * @param array|null $fields extra fields to include in the result (array of strings). (default: null = none) |
||
| 835 | * @return array sparql query clause |
||
| 836 | */ |
||
| 837 | protected function formatExtraFields($lang, $fields) { |
||
| 838 | // extra variable expressions to request and extra fields to query for |
||
| 839 | $ret = array('extravars' => '', 'extrafields' => ''); |
||
| 840 | |||
| 841 | if ($fields === null) { |
||
| 842 | return $ret; |
||
| 843 | } |
||
| 844 | |||
| 845 | if (in_array('prefLabel', $fields)) { |
||
| 846 | $ret['extravars'] .= $this->formatPreflabelCsvClause(); |
||
| 847 | $ret['extrafields'] .= <<<EOF |
||
| 848 | OPTIONAL { |
||
| 849 | ?s skos:prefLabel ?pref . |
||
| 850 | } |
||
| 851 | EOF; |
||
| 852 | // removing the prefLabel from the fields since it has been handled separately |
||
| 853 | $fields = array_diff($fields, array('prefLabel')); |
||
| 854 | } |
||
| 855 | |||
| 856 | foreach ($fields as $field) { |
||
| 857 | $ret['extravars'] .= $this->formatPropertyCsvClause($field); |
||
| 858 | $ret['extrafields'] .= <<<EOF |
||
| 859 | OPTIONAL { |
||
| 860 | ?s skos:$field ?$field . |
||
| 861 | FILTER(!isLiteral(?$field)||langMatches(lang(?{$field}), '$lang')) |
||
| 862 | OPTIONAL { ?$field skos:prefLabel ?{$field}lab . FILTER(langMatches(lang(?{$field}lab), '$lang')) } |
||
| 863 | } |
||
| 864 | EOF; |
||
| 865 | } |
||
| 866 | |||
| 867 | return $ret; |
||
| 868 | } |
||
| 869 | |||
| 870 | /** |
||
| 871 | * Generate condition for matching labels in SPARQL |
||
| 872 | * @param string $term search term |
||
| 873 | * @param string $searchLang language code used for matching labels (null means any language) |
||
| 874 | * @return string sparql query snippet |
||
| 875 | */ |
||
| 876 | protected function generateConceptSearchQueryCondition($term, $searchLang) |
||
| 877 | { |
||
| 878 | # use appropriate matching function depending on query type: =, strstarts, strends or full regex |
||
| 879 | if (preg_match('/^[^\*]+$/', $term)) { // exact query |
||
| 880 | $term = str_replace('\\', '\\\\', $term); // quote slashes |
||
| 881 | $term = str_replace('\'', '\\\'', mb_strtolower($term, 'UTF-8')); // make lowercase and escape single quotes |
||
| 882 | $filtercond = "LCASE(STR(?match)) = '$term'"; |
||
| 883 | } elseif (preg_match('/^[^\*]+\*$/', $term)) { // prefix query |
||
| 884 | $term = substr($term, 0, -1); // remove the final asterisk |
||
| 885 | $term = str_replace('\\', '\\\\', $term); // quote slashes |
||
| 886 | $term = str_replace('\'', '\\\'', mb_strtolower($term, 'UTF-8')); // make lowercase and escape single quotes |
||
| 887 | $filtercond = "STRSTARTS(LCASE(STR(?match)), '$term')"; |
||
| 888 | } elseif (preg_match('/^\*[^\*]+$/', $term)) { // suffix query |
||
| 889 | $term = substr($term, 1); // remove the preceding asterisk |
||
| 890 | $term = str_replace('\\', '\\\\', $term); // quote slashes |
||
| 891 | $term = str_replace('\'', '\\\'', mb_strtolower($term, 'UTF-8')); // make lowercase and escape single quotes |
||
| 892 | $filtercond = "STRENDS(LCASE(STR(?match)), '$term')"; |
||
| 893 | } else { // too complicated - have to use a regex |
||
| 894 | # make sure regex metacharacters are not passed through |
||
| 895 | $term = str_replace('\\', '\\\\', preg_quote($term)); |
||
| 896 | $term = str_replace('\\\\*', '.*', $term); // convert asterisk to regex syntax |
||
| 897 | $term = str_replace('\'', '\\\'', $term); // ensure single quotes are quoted |
||
| 898 | $filtercond = "REGEX(STR(?match), '^$term$', 'i')"; |
||
| 899 | } |
||
| 900 | |||
| 901 | $labelcondMatch = ($searchLang) ? "&& (?prop = skos:notation || LANGMATCHES(lang(?match), ?langParam))" : ""; |
||
| 902 | |||
| 903 | return "?s ?prop ?match . FILTER ($filtercond $labelcondMatch)"; |
||
| 904 | } |
||
| 905 | |||
| 906 | |||
| 907 | /** |
||
| 908 | * Inner query for concepts using a search term. |
||
| 909 | * @param string $term search term |
||
| 910 | * @param string $lang language code of the returned labels |
||
| 911 | * @param string $searchLang language code used for matching labels (null means any language) |
||
| 912 | * @param string[] $props properties to target e.g. array('skos:prefLabel','skos:altLabel') |
||
| 913 | * @param boolean $unique restrict results to unique concepts (default: false) |
||
| 914 | * @return string sparql query |
||
| 915 | */ |
||
| 916 | protected function generateConceptSearchQueryInner($term, $lang, $searchLang, $props, $unique, $filterGraph) |
||
| 917 | { |
||
| 918 | $valuesProp = $this->formatValues('?prop', $props); |
||
| 919 | $textcond = $this->generateConceptSearchQueryCondition($term, $searchLang); |
||
| 920 | |||
| 921 | $rawterm = str_replace(array('\\', '*', '"'), array('\\\\', '', '\"'), $term); |
||
| 922 | // graph clause, if necessary |
||
| 923 | $graphClause = $filterGraph != '' ? 'GRAPH ?graph' : ''; |
||
| 924 | |||
| 925 | // extra conditions for label language, if specified |
||
| 926 | $labelcondLabel = ($lang) ? "LANGMATCHES(lang(?label), '$lang')" : "lang(?match) = '' || LANGMATCHES(lang(?label), lang(?match))"; |
||
| 927 | // if search language and UI/display language differ, must also consider case where there is no prefLabel in |
||
| 928 | // the display language; in that case, should use the label with the same language as the matched label |
||
| 929 | $labelcondFallback = ($searchLang != $lang) ? |
||
| 930 | "OPTIONAL { # in case previous OPTIONAL block gives no labels\n" . |
||
| 931 | "?s skos:prefLabel ?label . FILTER (LANGMATCHES(LANG(?label), LANG(?match))) }" : ""; |
||
| 932 | |||
| 933 | // Including the labels if there is no query term given. |
||
| 934 | if ($rawterm === '') { |
||
| 935 | $labelClause = "?s skos:prefLabel ?label ."; |
||
| 936 | $labelClause = ($lang) ? $labelClause . " FILTER (LANGMATCHES(LANG(?label), '$lang'))" : $labelClause . ""; |
||
| 937 | return $labelClause . " BIND(?label AS ?match)"; |
||
| 938 | } |
||
| 939 | |||
| 940 | /* |
||
| 941 | * This query does some tricks to obtain a list of unique concepts. |
||
| 942 | * From each match generated by the text index, a string such as |
||
| 943 | * "1en@example" is generated, where the first character is a number |
||
| 944 | * encoding the property and priority, then comes the language tag and |
||
| 945 | * finally the original literal after an @ sign. Of these, the MIN |
||
| 946 | * function is used to pick the best match for each concept. Finally, |
||
| 947 | * the structure is unpacked to get back the original string. Phew! |
||
| 948 | */ |
||
| 949 | $hitvar = $unique ? '(MIN(?matchstr) AS ?hit)' : '(?matchstr AS ?hit)'; |
||
| 950 | $hitgroup = $unique ? 'GROUP BY ?s ?label ?notation' : ''; |
||
| 951 | |||
| 952 | $langClause = $this->generateLangClause($searchLang); |
||
| 953 | |||
| 954 | $query = <<<EOQ |
||
| 955 | SELECT DISTINCT ?s ?label ?notation $hitvar |
||
| 956 | WHERE { |
||
| 957 | $graphClause { |
||
| 958 | { |
||
| 959 | $valuesProp |
||
| 960 | VALUES (?prop ?pri ?langParam) { (skos:prefLabel 1 $langClause) (skos:altLabel 3 $langClause) (skos:notation 5 '') (skos:hiddenLabel 7 $langClause)} |
||
| 961 | $textcond |
||
| 962 | ?s ?prop ?match } |
||
| 963 | OPTIONAL { |
||
| 964 | ?s skos:prefLabel ?label . |
||
| 965 | FILTER ($labelcondLabel) |
||
| 966 | } $labelcondFallback |
||
| 967 | BIND(IF(langMatches(LANG(?match),'$lang'), ?pri, ?pri+1) AS ?npri) |
||
| 968 | BIND(CONCAT(STR(?npri), LANG(?match), '@', STR(?match)) AS ?matchstr) |
||
| 969 | OPTIONAL { ?s skos:notation ?notation } |
||
| 970 | } |
||
| 971 | $filterGraph |
||
| 972 | } |
||
| 973 | $hitgroup |
||
| 974 | EOQ; |
||
| 975 | |||
| 976 | return $query; |
||
| 977 | } |
||
| 978 | /** |
||
| 979 | * This function can be overwritten in other SPARQL dialects for the possibility of handling the different language clauses |
||
| 980 | * @param string $lang |
||
| 981 | * @return string formatted language clause |
||
| 982 | */ |
||
| 983 | protected function generateLangClause($lang) { |
||
| 984 | return "'$lang'"; |
||
| 985 | } |
||
| 986 | |||
| 987 | /** |
||
| 988 | * Query for concepts using a search term. |
||
| 989 | * @param array|null $fields extra fields to include in the result (array of strings). (default: null = none) |
||
| 990 | * @param boolean $unique restrict results to unique concepts (default: false) |
||
| 991 | * @param boolean $showDeprecated whether to include deprecated concepts in search results (default: false) |
||
| 992 | * @param ConceptSearchParameters $params |
||
| 993 | * @return string sparql query |
||
| 994 | */ |
||
| 995 | protected function generateConceptSearchQuery($fields, $unique, $params, $showDeprecated = false) { |
||
| 996 | $vocabs = $params->getVocabs(); |
||
| 997 | $gcl = $this->graphClause; |
||
| 998 | $fcl = empty($vocabs) ? '' : $this->generateFromClause($vocabs); |
||
| 999 | $formattedtype = $this->formatTypes($params->getTypeLimit()); |
||
| 1000 | $formattedfields = $this->formatExtraFields($params->getLang(), $fields); |
||
| 1001 | $extravars = $formattedfields['extravars']; |
||
| 1002 | $extrafields = $formattedfields['extrafields']; |
||
| 1003 | $schemes = $params->getSchemeLimit(); |
||
| 1004 | |||
| 1005 | // limit the search to only requested concept schemes |
||
| 1006 | $schemecond = ''; |
||
| 1007 | if (!empty($schemes)) { |
||
| 1008 | $conditions = array(); |
||
| 1009 | foreach($schemes as $scheme) { |
||
| 1010 | $conditions[] = "{?s skos:inScheme <$scheme>}"; |
||
| 1011 | } |
||
| 1012 | $schemecond = '{'.implode(" UNION ",$conditions).'}'; |
||
| 1013 | } |
||
| 1014 | $filterDeprecated=""; |
||
| 1015 | //show or hide deprecated concepts |
||
| 1016 | if(!$showDeprecated){ |
||
| 1017 | $filterDeprecated="FILTER NOT EXISTS { ?s owl:deprecated true }"; |
||
| 1018 | } |
||
| 1019 | // extra conditions for parent and group, if specified |
||
| 1020 | $parentcond = ($params->getParentLimit()) ? "?s skos:broader+ <" . $params->getParentLimit() . "> ." : ""; |
||
| 1021 | $groupcond = ($params->getGroupLimit()) ? "<" . $params->getGroupLimit() . "> skos:member ?s ." : ""; |
||
| 1022 | $pgcond = $parentcond . $groupcond; |
||
| 1023 | |||
| 1024 | $orderextra = $this->isDefaultEndpoint() ? $this->graph : ''; |
||
| 1025 | |||
| 1026 | # make VALUES clauses |
||
| 1027 | $props = array('skos:prefLabel', 'skos:altLabel'); |
||
| 1028 | |||
| 1029 | //add notation into searchable data for the vocabularies which have been configured for it |
||
| 1030 | if ($vocabs) { |
||
| 1031 | $searchByNotation = false; |
||
| 1032 | foreach ($vocabs as $vocab) { |
||
| 1033 | if ($vocab->getConfig()->searchByNotation()) { |
||
| 1034 | $searchByNotation = true; |
||
| 1035 | } |
||
| 1036 | } |
||
| 1037 | if ($searchByNotation) { |
||
| 1038 | $props[] = 'skos:notation'; |
||
| 1039 | } |
||
| 1040 | } |
||
| 1041 | |||
| 1042 | if ($params->getHidden()) { |
||
| 1043 | $props[] = 'skos:hiddenLabel'; |
||
| 1044 | } |
||
| 1045 | $filterGraph = empty($vocabs) ? $this->formatFilterGraph($vocabs) : ''; |
||
| 1046 | |||
| 1047 | // remove futile asterisks from the search term |
||
| 1048 | $term = $params->getSearchTerm(); |
||
| 1049 | while (strpos($term, '**') !== false) { |
||
| 1050 | $term = str_replace('**', '*', $term); |
||
| 1051 | } |
||
| 1052 | |||
| 1053 | $labelpriority = <<<EOQ |
||
| 1054 | FILTER(BOUND(?s)) |
||
| 1055 | BIND(STR(SUBSTR(?hit,1,1)) AS ?pri) |
||
| 1056 | BIND(IF((SUBSTR(STRBEFORE(?hit, '@'),1) != ?pri), STRLANG(STRAFTER(?hit, '@'), SUBSTR(STRBEFORE(?hit, '@'),2)), STRAFTER(?hit, '@')) AS ?match) |
||
| 1057 | BIND(IF((?pri = "1" || ?pri = "2") && ?match != ?label, ?match, ?unbound) as ?plabel) |
||
| 1058 | BIND(IF((?pri = "3" || ?pri = "4"), ?match, ?unbound) as ?alabel) |
||
| 1059 | BIND(IF((?pri = "7" || ?pri = "8"), ?match, ?unbound) as ?hlabel) |
||
| 1060 | EOQ; |
||
| 1061 | $innerquery = $this->generateConceptSearchQueryInner($params->getSearchTerm(), $params->getLang(), $params->getSearchLang(), $props, $unique, $filterGraph); |
||
| 1062 | if ($params->getSearchTerm() === '*' || $params->getSearchTerm() === '') { |
||
| 1063 | $labelpriority = ''; |
||
| 1064 | } |
||
| 1065 | $query = <<<EOQ |
||
| 1066 | SELECT DISTINCT ?s ?label ?plabel ?alabel ?hlabel ?graph ?notation (GROUP_CONCAT(DISTINCT STR(?type);separator=' ') as ?types) $extravars |
||
| 1067 | $fcl |
||
| 1068 | WHERE { |
||
| 1069 | $gcl { |
||
| 1070 | { |
||
| 1071 | $innerquery |
||
| 1072 | } |
||
| 1073 | $labelpriority |
||
| 1074 | $formattedtype |
||
| 1075 | { $pgcond |
||
| 1076 | ?s a ?type . |
||
| 1077 | $extrafields $schemecond |
||
| 1078 | } |
||
| 1079 | $filterDeprecated |
||
| 1080 | } |
||
| 1081 | $filterGraph |
||
| 1082 | } |
||
| 1083 | GROUP BY ?s ?match ?label ?plabel ?alabel ?hlabel ?notation ?graph |
||
| 1084 | ORDER BY LCASE(STR(?match)) LANG(?match) $orderextra |
||
| 1085 | EOQ; |
||
| 1086 | return $query; |
||
| 1087 | } |
||
| 1088 | |||
| 1089 | /** |
||
| 1090 | * Transform a single concept search query results into the skosmos desired return format. |
||
| 1091 | * @param $row SPARQL query result row |
||
| 1092 | * @param array $vocabs array of Vocabulary objects to search; empty for global search |
||
| 1093 | * @return array query result object |
||
| 1094 | */ |
||
| 1095 | private function transformConceptSearchResult($row, $vocabs, $fields) |
||
| 1096 | { |
||
| 1097 | $hit = array(); |
||
| 1098 | $hit['uri'] = $row->s->getUri(); |
||
| 1099 | |||
| 1100 | if (isset($row->graph)) { |
||
| 1101 | $hit['graph'] = $row->graph->getUri(); |
||
| 1102 | } |
||
| 1103 | |||
| 1104 | foreach (explode(" ", $row->types->getValue()) as $typeuri) { |
||
| 1105 | $hit['type'][] = $this->shortenUri($typeuri); |
||
| 1106 | } |
||
| 1107 | |||
| 1108 | if(!empty($fields)) { |
||
| 1109 | foreach ($fields as $prop) { |
||
| 1110 | $propname = $prop . 's'; |
||
| 1111 | if (isset($row->$propname)) { |
||
| 1112 | foreach (explode("\n", $row->$propname->getValue()) as $line) { |
||
| 1113 | $rdata = str_getcsv($line, ',', '"', '"'); |
||
| 1114 | $propvals = array(); |
||
| 1115 | if ($rdata[0] != '') { |
||
| 1116 | $propvals['uri'] = $rdata[0]; |
||
| 1117 | } |
||
| 1118 | if ($rdata[1] != '') { |
||
| 1119 | $propvals['prefLabel'] = $rdata[1]; |
||
| 1120 | } |
||
| 1121 | if ($rdata[2] != '') { |
||
| 1122 | $propvals = $rdata[2]; |
||
| 1123 | } |
||
| 1124 | |||
| 1125 | $hit['skos:' . $prop][] = $propvals; |
||
| 1126 | } |
||
| 1127 | } |
||
| 1128 | } |
||
| 1129 | } |
||
| 1130 | |||
| 1131 | |||
| 1132 | if (isset($row->preflabels)) { |
||
| 1133 | foreach (explode("\n", $row->preflabels->getValue()) as $line) { |
||
| 1134 | $pref = str_getcsv($line, ',', '"', '"'); |
||
| 1135 | $hit['prefLabels'][$pref[1]] = $pref[0]; |
||
| 1136 | } |
||
| 1137 | } |
||
| 1138 | |||
| 1139 | foreach ($vocabs as $vocab) { // looping the vocabulary objects and asking these for a localname for the concept. |
||
| 1140 | $localname = $vocab->getLocalName($hit['uri']); |
||
| 1141 | if ($localname !== $hit['uri']) { // only passing the result forward if the uri didn't boomerang right back. |
||
| 1142 | $hit['localname'] = $localname; |
||
| 1143 | break; // stopping the search when we find one that returns something valid. |
||
| 1144 | } |
||
| 1145 | } |
||
| 1146 | |||
| 1147 | if (isset($row->label)) { |
||
| 1148 | $hit['prefLabel'] = $row->label->getValue(); |
||
| 1149 | } |
||
| 1150 | |||
| 1151 | if (isset($row->label)) { |
||
| 1152 | $hit['lang'] = $row->label->getLang(); |
||
| 1153 | } |
||
| 1154 | |||
| 1155 | if (isset($row->notation)) { |
||
| 1156 | $hit['notation'] = $row->notation->getValue(); |
||
| 1157 | } |
||
| 1158 | |||
| 1159 | if (isset($row->plabel)) { |
||
| 1160 | $hit['matchedPrefLabel'] = $row->plabel->getValue(); |
||
| 1161 | $hit['lang'] = $row->plabel->getLang(); |
||
| 1162 | } elseif (isset($row->alabel)) { |
||
| 1163 | $hit['altLabel'] = $row->alabel->getValue(); |
||
| 1164 | $hit['lang'] = $row->alabel->getLang(); |
||
| 1165 | } elseif (isset($row->hlabel)) { |
||
| 1166 | $hit['hiddenLabel'] = $row->hlabel->getValue(); |
||
| 1167 | $hit['lang'] = $row->hlabel->getLang(); |
||
| 1168 | } |
||
| 1169 | return $hit; |
||
| 1170 | } |
||
| 1171 | |||
| 1172 | /** |
||
| 1173 | * Transform the concept search query results into the skosmos desired return format. |
||
| 1174 | * @param EasyRdf\Sparql\Result $results |
||
| 1175 | * @param array $vocabs array of Vocabulary objects to search; empty for global search |
||
| 1176 | * @return array query result object |
||
| 1177 | */ |
||
| 1178 | private function transformConceptSearchResults($results, $vocabs, $fields) { |
||
| 1179 | $ret = array(); |
||
| 1180 | |||
| 1181 | foreach ($results as $row) { |
||
| 1182 | if (!isset($row->s)) { |
||
| 1183 | // don't break if query returns a single dummy result |
||
| 1184 | continue; |
||
| 1185 | } |
||
| 1186 | $ret[] = $this->transformConceptSearchResult($row, $vocabs, $fields); |
||
| 1187 | } |
||
| 1188 | return $ret; |
||
| 1189 | } |
||
| 1190 | |||
| 1191 | /** |
||
| 1192 | * Query for concepts using a search term. |
||
| 1193 | * @param array $vocabs array of Vocabulary objects to search; empty for global search |
||
| 1194 | * @param array $fields extra fields to include in the result (array of strings). (default: null = none) |
||
| 1195 | * @param boolean $unique restrict results to unique concepts (default: false) |
||
| 1196 | * @param boolean $showDeprecated whether to include deprecated concepts in the result (default: false) |
||
| 1197 | * @param ConceptSearchParameters $params |
||
| 1198 | * @return array query result object |
||
| 1199 | */ |
||
| 1200 | public function queryConcepts($vocabs, $fields = null, $unique = false, $params, $showDeprecated = false) { |
||
| 1201 | $query = $this->generateConceptSearchQuery($fields, $unique, $params,$showDeprecated); |
||
| 1202 | $results = $this->query($query); |
||
| 1203 | return $this->transformConceptSearchResults($results, $vocabs, $fields); |
||
| 1204 | } |
||
| 1205 | |||
| 1206 | /** |
||
| 1207 | * Generates sparql query clauses used for creating the alphabetical index. |
||
| 1208 | * @param string $letter the letter (or special class) to search for |
||
| 1209 | * @return array of sparql query clause strings |
||
| 1210 | */ |
||
| 1211 | private function formatFilterConditions($letter, $lang) { |
||
| 1212 | $useRegex = false; |
||
| 1213 | |||
| 1214 | if ($letter == '*') { |
||
| 1215 | $letter = '.*'; |
||
| 1216 | $useRegex = true; |
||
| 1217 | } elseif ($letter == '0-9') { |
||
| 1218 | $letter = '[0-9].*'; |
||
| 1219 | $useRegex = true; |
||
| 1220 | } elseif ($letter == '!*') { |
||
| 1221 | $letter = '[^\\\\p{L}\\\\p{N}].*'; |
||
| 1222 | $useRegex = true; |
||
| 1223 | } |
||
| 1224 | |||
| 1225 | # make text query clause |
||
| 1226 | $lcletter = mb_strtolower($letter, 'UTF-8'); // convert to lower case, UTF-8 safe |
||
| 1227 | if ($useRegex) { |
||
| 1228 | $filtercondLabel = $lang ? "regex(str(?label), '^$letter$', 'i') && langMatches(lang(?label), '$lang')" : "regex(str(?label), '^$letter$', 'i')"; |
||
| 1229 | $filtercondALabel = $lang ? "regex(str(?alabel), '^$letter$', 'i') && langMatches(lang(?alabel), '$lang')" : "regex(str(?alabel), '^$letter$', 'i')"; |
||
| 1230 | } else { |
||
| 1231 | $filtercondLabel = $lang ? "strstarts(lcase(str(?label)), '$lcletter') && langMatches(lang(?label), '$lang')" : "strstarts(lcase(str(?label)), '$lcletter')"; |
||
| 1232 | $filtercondALabel = $lang ? "strstarts(lcase(str(?alabel)), '$lcletter') && langMatches(lang(?alabel), '$lang')" : "strstarts(lcase(str(?alabel)), '$lcletter')"; |
||
| 1233 | } |
||
| 1234 | return array('filterpref' => $filtercondLabel, 'filteralt' => $filtercondALabel); |
||
| 1235 | } |
||
| 1236 | |||
| 1237 | /** |
||
| 1238 | * Generates the sparql query used for rendering the alphabetical index. |
||
| 1239 | * @param string $letter the letter (or special class) to search for |
||
| 1240 | * @param string $lang language of labels |
||
| 1241 | * @param integer $limit limits the amount of results |
||
| 1242 | * @param integer $offset offsets the result set |
||
| 1243 | * @param array|null $classes |
||
| 1244 | * @param boolean $showDeprecated whether to include deprecated concepts in the result (default: false) |
||
| 1245 | * @param \EasyRdf\Resource|null $qualifier alphabetical list qualifier resource or null (default: null) |
||
| 1246 | * @return string sparql query |
||
| 1247 | */ |
||
| 1248 | protected function generateAlphabeticalListQuery($letter, $lang, $limit, $offset, $classes, $showDeprecated = false, $qualifier = null) { |
||
| 1249 | $gcl = $this->graphClause; |
||
| 1250 | $classes = ($classes) ? $classes : array('http://www.w3.org/2004/02/skos/core#Concept'); |
||
| 1251 | $values = $this->formatValues('?type', $classes, 'uri'); |
||
| 1252 | $limitandoffset = $this->formatLimitAndOffset($limit, $offset); |
||
| 1253 | $conditions = $this->formatFilterConditions($letter, $lang); |
||
| 1254 | $filtercondLabel = $conditions['filterpref']; |
||
| 1255 | $filtercondALabel = $conditions['filteralt']; |
||
| 1256 | $qualifierClause = $qualifier ? "OPTIONAL { ?s <" . $qualifier->getURI() . "> ?qualifier }" : ""; |
||
| 1257 | $filterDeprecated=""; |
||
| 1258 | if(!$showDeprecated){ |
||
| 1259 | $filterDeprecated="FILTER NOT EXISTS { ?s owl:deprecated true }"; |
||
| 1260 | } |
||
| 1261 | $query = <<<EOQ |
||
| 1262 | SELECT DISTINCT ?s ?label ?alabel ?qualifier |
||
| 1263 | WHERE { |
||
| 1264 | $gcl { |
||
| 1265 | { |
||
| 1266 | ?s skos:prefLabel ?label . |
||
| 1267 | FILTER ( |
||
| 1268 | $filtercondLabel |
||
| 1269 | ) |
||
| 1270 | } |
||
| 1271 | UNION |
||
| 1272 | { |
||
| 1273 | { |
||
| 1274 | ?s skos:altLabel ?alabel . |
||
| 1275 | FILTER ( |
||
| 1276 | $filtercondALabel |
||
| 1277 | ) |
||
| 1278 | } |
||
| 1279 | { |
||
| 1280 | ?s skos:prefLabel ?label . |
||
| 1281 | FILTER (langMatches(lang(?label), '$lang')) |
||
| 1282 | } |
||
| 1283 | } |
||
| 1284 | ?s a ?type . |
||
| 1285 | $qualifierClause |
||
| 1286 | $filterDeprecated |
||
| 1287 | $values |
||
| 1288 | } |
||
| 1289 | } |
||
| 1290 | ORDER BY LCASE(STR(COALESCE(?alabel, ?label))) STR(?s) LCASE(STR(?qualifier)) $limitandoffset |
||
| 1291 | EOQ; |
||
| 1292 | return $query; |
||
| 1293 | } |
||
| 1294 | |||
| 1295 | /** |
||
| 1296 | * Transforms the alphabetical list query results into an array format. |
||
| 1297 | * @param EasyRdf\Sparql\Result $results |
||
| 1298 | * @return array |
||
| 1299 | */ |
||
| 1300 | private function transformAlphabeticalListResults($results) { |
||
| 1301 | $ret = array(); |
||
| 1302 | |||
| 1303 | foreach ($results as $row) { |
||
| 1304 | if (!isset($row->s)) { |
||
| 1305 | continue; |
||
| 1306 | } |
||
| 1307 | // don't break if query returns a single dummy result |
||
| 1308 | |||
| 1309 | $hit = array(); |
||
| 1310 | $hit['uri'] = $row->s->getUri(); |
||
| 1311 | |||
| 1312 | $hit['localname'] = $row->s->localName(); |
||
| 1313 | |||
| 1314 | $hit['prefLabel'] = $row->label->getValue(); |
||
| 1315 | $hit['lang'] = $row->label->getLang(); |
||
| 1316 | |||
| 1317 | if (isset($row->alabel)) { |
||
| 1318 | $hit['altLabel'] = $row->alabel->getValue(); |
||
| 1319 | $hit['lang'] = $row->alabel->getLang(); |
||
| 1320 | } |
||
| 1321 | |||
| 1322 | if (isset($row->qualifier)) { |
||
| 1323 | if ($row->qualifier instanceof EasyRdf\Literal) { |
||
| 1324 | $hit['qualifier'] = $row->qualifier->getValue(); |
||
| 1325 | } |
||
| 1326 | else { |
||
| 1327 | $hit['qualifier'] = $row->qualifier->localName(); |
||
| 1328 | } |
||
| 1329 | } |
||
| 1330 | |||
| 1331 | $ret[] = $hit; |
||
| 1332 | } |
||
| 1333 | |||
| 1334 | return $ret; |
||
| 1335 | } |
||
| 1336 | |||
| 1337 | /** |
||
| 1338 | * Query for concepts with a term starting with the given letter. Also special classes '0-9' (digits), |
||
| 1339 | * '*!' (special characters) and '*' (everything) are accepted. |
||
| 1340 | * @param string $letter the letter (or special class) to search for |
||
| 1341 | * @param string $lang language of labels |
||
| 1342 | * @param integer $limit limits the amount of results |
||
| 1343 | * @param integer $offset offsets the result set |
||
| 1344 | * @param array $classes |
||
| 1345 | * @param boolean $showDeprecated whether to include deprecated concepts in the result (default: false) |
||
| 1346 | * @param \EasyRdf\Resource|null $qualifier alphabetical list qualifier resource or null (default: null) |
||
| 1347 | */ |
||
| 1348 | public function queryConceptsAlphabetical($letter, $lang, $limit = null, $offset = null, $classes = null, $showDeprecated = false, $qualifier = null) { |
||
| 1349 | if ($letter === '') { |
||
| 1350 | return array(); // special case: no letter given, return empty list |
||
| 1351 | } |
||
| 1352 | $query = $this->generateAlphabeticalListQuery($letter, $lang, $limit, $offset, $classes, $showDeprecated, $qualifier); |
||
| 1353 | $results = $this->query($query); |
||
| 1354 | return $this->transformAlphabeticalListResults($results); |
||
| 1355 | } |
||
| 1356 | |||
| 1357 | /** |
||
| 1358 | * Creates the query used for finding out which letters should be displayed in the alphabetical index. |
||
| 1359 | * Note that we force the datatype of the result variable otherwise Virtuoso does not properly interpret the DISTINCT and we have duplicated results |
||
| 1360 | * @param string $lang language |
||
| 1361 | * @return string sparql query |
||
| 1362 | */ |
||
| 1363 | private function generateFirstCharactersQuery($lang, $classes) { |
||
| 1364 | $gcl = $this->graphClause; |
||
| 1365 | $classes = (isset($classes) && sizeof($classes) > 0) ? $classes : array('http://www.w3.org/2004/02/skos/core#Concept'); |
||
| 1366 | $values = $this->formatValues('?type', $classes, 'uri'); |
||
| 1367 | $query = <<<EOQ |
||
| 1368 | SELECT DISTINCT (ucase(str(substr(?label, 1, 1))) as ?l) WHERE { |
||
| 1369 | $gcl { |
||
| 1370 | ?c skos:prefLabel ?label . |
||
| 1371 | ?c a ?type |
||
| 1372 | FILTER(langMatches(lang(?label), '$lang')) |
||
| 1373 | $values |
||
| 1374 | } |
||
| 1375 | } |
||
| 1376 | EOQ; |
||
| 1377 | return $query; |
||
| 1378 | } |
||
| 1379 | |||
| 1380 | /** |
||
| 1381 | * Transforms the first characters query results into an array format. |
||
| 1382 | * @param EasyRdf\Sparql\Result $result |
||
| 1383 | * @return array |
||
| 1384 | */ |
||
| 1385 | private function transformFirstCharactersResults($result) { |
||
| 1386 | $ret = array(); |
||
| 1387 | foreach ($result as $row) { |
||
| 1388 | $ret[] = $row->l->getValue(); |
||
| 1389 | } |
||
| 1390 | return $ret; |
||
| 1391 | } |
||
| 1392 | |||
| 1393 | /** |
||
| 1394 | * Query for the first characters (letter or otherwise) of the labels in the particular language. |
||
| 1395 | * @param string $lang language |
||
| 1396 | * @return array array of characters |
||
| 1397 | */ |
||
| 1398 | public function queryFirstCharacters($lang, $classes = null) { |
||
| 1399 | $query = $this->generateFirstCharactersQuery($lang, $classes); |
||
| 1400 | $result = $this->query($query); |
||
| 1401 | return $this->transformFirstCharactersResults($result); |
||
| 1402 | } |
||
| 1403 | |||
| 1404 | /** |
||
| 1405 | * @param string $uri |
||
| 1406 | * @param string $lang |
||
| 1407 | * @return string sparql query string |
||
| 1408 | */ |
||
| 1409 | private function generateLabelQuery($uri, $lang) { |
||
| 1410 | $fcl = $this->generateFromClause(); |
||
| 1411 | $labelcondLabel = ($lang) ? "FILTER( langMatches(lang(?label), '$lang') )" : ""; |
||
| 1412 | $query = <<<EOQ |
||
| 1413 | SELECT ?label $fcl |
||
| 1414 | WHERE { |
||
| 1415 | <$uri> a ?type . |
||
| 1416 | OPTIONAL { |
||
| 1417 | <$uri> skos:prefLabel ?label . |
||
| 1418 | $labelcondLabel |
||
| 1419 | } |
||
| 1420 | OPTIONAL { |
||
| 1421 | <$uri> rdfs:label ?label . |
||
| 1422 | $labelcondLabel |
||
| 1423 | } |
||
| 1424 | OPTIONAL { |
||
| 1425 | <$uri> dc:title ?label . |
||
| 1426 | $labelcondLabel |
||
| 1427 | } |
||
| 1428 | OPTIONAL { |
||
| 1429 | <$uri> dc11:title ?label . |
||
| 1430 | $labelcondLabel |
||
| 1431 | } |
||
| 1432 | } |
||
| 1433 | EOQ; |
||
| 1434 | return $query; |
||
| 1435 | } |
||
| 1436 | |||
| 1437 | |||
| 1438 | /** |
||
| 1439 | * @param string $uri |
||
| 1440 | * @param string $lang |
||
| 1441 | * @return string sparql query string |
||
| 1442 | */ |
||
| 1443 | private function generateAllLabelsQuery($uri, $lang) { |
||
| 1444 | $fcl = $this->generateFromClause(); |
||
| 1445 | $labelcondLabel = ($lang) ? "FILTER( langMatches(lang(?val), '$lang') )" : ""; |
||
| 1446 | $query = <<<EOQ |
||
| 1447 | SELECT DISTINCT ?prop ?val $fcl |
||
| 1448 | WHERE { |
||
| 1449 | <$uri> a ?type . |
||
| 1450 | OPTIONAL { |
||
| 1451 | <$uri> ?prop ?val . |
||
| 1452 | $labelcondLabel |
||
| 1453 | } |
||
| 1454 | VALUES ?prop { skos:prefLabel skos:altLabel skos:hiddenLabel } |
||
| 1455 | } |
||
| 1456 | EOQ; |
||
| 1457 | return $query; |
||
| 1458 | } |
||
| 1459 | |||
| 1460 | /** |
||
| 1461 | * Query for a label (skos:prefLabel, rdfs:label, dc:title, dc11:title) of a resource. |
||
| 1462 | * @param string $uri |
||
| 1463 | * @param string $lang |
||
| 1464 | * @return array array of labels (key: lang, val: label), or null if resource doesn't exist |
||
| 1465 | */ |
||
| 1466 | public function queryLabel($uri, $lang) { |
||
| 1467 | $query = $this->generateLabelQuery($uri, $lang); |
||
| 1468 | $result = $this->query($query); |
||
| 1469 | $ret = array(); |
||
| 1470 | foreach ($result as $row) { |
||
| 1471 | if (!isset($row->label)) { |
||
| 1472 | // existing concept but no labels |
||
| 1473 | return array(); |
||
| 1474 | } |
||
| 1475 | $ret[$row->label->getLang()] = $row->label; |
||
| 1476 | } |
||
| 1477 | |||
| 1478 | if (sizeof($ret) > 0) { |
||
| 1479 | // existing concept, with label(s) |
||
| 1480 | return $ret; |
||
| 1481 | } else { |
||
| 1482 | // nonexistent concept |
||
| 1483 | return null; |
||
| 1484 | } |
||
| 1485 | } |
||
| 1486 | |||
| 1487 | /** |
||
| 1488 | * Query for skos:prefLabels, skos:altLabels and skos:hiddenLabels of a resource. |
||
| 1489 | * @param string $uri |
||
| 1490 | * @param string $lang |
||
| 1491 | * @return array array of prefLabels, altLabels and hiddenLabels - or null if resource doesn't exist |
||
| 1492 | */ |
||
| 1493 | public function queryAllConceptLabels($uri, $lang) { |
||
| 1494 | $query = $this->generateAllLabelsQuery($uri, $lang); |
||
| 1495 | $result = $this->query($query); |
||
| 1496 | |||
| 1497 | if ($result->numRows() == 0) { |
||
| 1498 | // nonexistent concept |
||
| 1499 | return null; |
||
| 1500 | } |
||
| 1501 | |||
| 1502 | $ret = array(); |
||
| 1503 | foreach ($result as $row) { |
||
| 1504 | $labelName = $row->prop->localName(); |
||
| 1505 | if (isset($row->val)) { |
||
| 1506 | $ret[$labelName][] = $row->val->getValue(); |
||
| 1507 | } |
||
| 1508 | } |
||
| 1509 | return $ret; |
||
| 1510 | } |
||
| 1511 | |||
| 1512 | /** |
||
| 1513 | * Generates a SPARQL query to retrieve the super properties of a given property URI. |
||
| 1514 | * Note this must be executed in the graph where this information is available. |
||
| 1515 | * @param string $uri |
||
| 1516 | * @return string sparql query string |
||
| 1517 | */ |
||
| 1518 | private function generateSubPropertyOfQuery($uri) { |
||
| 1519 | $fcl = $this->generateFromClause(); |
||
| 1520 | $query = <<<EOQ |
||
| 1521 | SELECT ?superProperty $fcl |
||
| 1522 | WHERE { |
||
| 1523 | <$uri> rdfs:subPropertyOf ?superProperty |
||
| 1524 | } |
||
| 1525 | EOQ; |
||
| 1526 | return $query; |
||
| 1527 | } |
||
| 1528 | |||
| 1529 | /** |
||
| 1530 | * Query the super properties of a provided property URI. |
||
| 1531 | * @param string $uri URI of a propertyes |
||
| 1532 | * @return array array super properties, or null if none exist |
||
| 1533 | */ |
||
| 1534 | public function querySuperProperties($uri) { |
||
| 1535 | $query = $this->generateSubPropertyOfQuery($uri); |
||
| 1536 | $result = $this->query($query); |
||
| 1537 | $ret = array(); |
||
| 1538 | foreach ($result as $row) { |
||
| 1539 | if (isset($row->superProperty)) { |
||
| 1540 | $ret[] = $row->superProperty->getUri(); |
||
| 1541 | } |
||
| 1542 | |||
| 1543 | } |
||
| 1544 | |||
| 1545 | if (sizeof($ret) > 0) { |
||
| 1546 | // return result |
||
| 1547 | return $ret; |
||
| 1548 | } else { |
||
| 1549 | // no result, return null |
||
| 1550 | return null; |
||
| 1551 | } |
||
| 1552 | } |
||
| 1553 | |||
| 1554 | |||
| 1555 | /** |
||
| 1556 | * Generates a sparql query for queryNotation. |
||
| 1557 | * @param string $uri |
||
| 1558 | * @return string sparql query |
||
| 1559 | */ |
||
| 1560 | private function generateNotationQuery($uri) { |
||
| 1561 | $fcl = $this->generateFromClause(); |
||
| 1562 | |||
| 1563 | $query = <<<EOQ |
||
| 1564 | SELECT * $fcl |
||
| 1565 | WHERE { |
||
| 1566 | <$uri> skos:notation ?notation . |
||
| 1567 | } |
||
| 1568 | EOQ; |
||
| 1569 | return $query; |
||
| 1570 | } |
||
| 1571 | |||
| 1572 | /** |
||
| 1573 | * Query for the notation of the concept (skos:notation) of a resource. |
||
| 1574 | * @param string $uri |
||
| 1575 | * @return string notation or null if it doesn't exist |
||
| 1576 | */ |
||
| 1577 | public function queryNotation($uri) { |
||
| 1578 | $query = $this->generateNotationQuery($uri); |
||
| 1579 | $result = $this->query($query); |
||
| 1580 | foreach ($result as $row) { |
||
| 1581 | if (isset($row->notation)) { |
||
| 1582 | return $row->notation->getValue(); |
||
| 1583 | } |
||
| 1584 | } |
||
| 1585 | return null; |
||
| 1586 | } |
||
| 1587 | |||
| 1588 | /** |
||
| 1589 | * Generates a sparql query for queryProperty. |
||
| 1590 | * @param string $uri |
||
| 1591 | * @param string $prop the name of the property eg. 'skos:broader'. |
||
| 1592 | * @param string $lang |
||
| 1593 | * @param boolean $anylang if you want a label even when it isn't available in the language you requested. |
||
| 1594 | * @return string sparql query |
||
| 1595 | */ |
||
| 1596 | private function generatePropertyQuery($uri, $prop, $lang, $anylang) { |
||
| 1597 | $fcl = $this->generateFromClause(); |
||
| 1598 | $anylang = $anylang ? "OPTIONAL { ?object skos:prefLabel ?label }" : ""; |
||
| 1599 | |||
| 1600 | $query = <<<EOQ |
||
| 1601 | SELECT * $fcl |
||
| 1602 | WHERE { |
||
| 1603 | <$uri> a skos:Concept . |
||
| 1604 | OPTIONAL { |
||
| 1605 | <$uri> $prop ?object . |
||
| 1606 | OPTIONAL { |
||
| 1607 | ?object skos:prefLabel ?label . |
||
| 1608 | FILTER (langMatches(lang(?label), "$lang")) |
||
| 1609 | } |
||
| 1610 | OPTIONAL { |
||
| 1611 | ?object skos:prefLabel ?label . |
||
| 1612 | FILTER (lang(?label) = "") |
||
| 1613 | } |
||
| 1614 | $anylang |
||
| 1615 | } |
||
| 1616 | } |
||
| 1617 | EOQ; |
||
| 1618 | return $query; |
||
| 1619 | } |
||
| 1620 | |||
| 1621 | /** |
||
| 1622 | * Transforms the sparql query result into an array or null if the concept doesn't exist. |
||
| 1623 | * @param EasyRdf\Sparql\Result $result |
||
| 1624 | * @param string $lang |
||
| 1625 | * @return array array of property values (key: URI, val: label), or null if concept doesn't exist |
||
| 1626 | */ |
||
| 1627 | private function transformPropertyQueryResults($result, $lang) { |
||
| 1628 | $ret = array(); |
||
| 1629 | foreach ($result as $row) { |
||
| 1630 | if (!isset($row->object)) { |
||
| 1631 | return array(); |
||
| 1632 | } |
||
| 1633 | // existing concept but no properties |
||
| 1634 | if (isset($row->label)) { |
||
| 1635 | if ($row->label->getLang() === $lang || array_key_exists($row->object->getUri(), $ret) === false) { |
||
| 1636 | $ret[$row->object->getUri()]['label'] = $row->label->getValue(); |
||
| 1637 | } |
||
| 1638 | |||
| 1639 | } else { |
||
| 1640 | $ret[$row->object->getUri()]['label'] = null; |
||
| 1641 | } |
||
| 1642 | } |
||
| 1643 | if (sizeof($ret) > 0) { |
||
| 1644 | return $ret; |
||
| 1645 | } |
||
| 1646 | // existing concept, with properties |
||
| 1647 | else { |
||
| 1648 | return null; |
||
| 1649 | } |
||
| 1650 | // nonexistent concept |
||
| 1651 | } |
||
| 1652 | |||
| 1653 | /** |
||
| 1654 | * Query a single property of a concept. |
||
| 1655 | * @param string $uri |
||
| 1656 | * @param string $prop the name of the property eg. 'skos:broader'. |
||
| 1657 | * @param string $lang |
||
| 1658 | * @param boolean $anylang if you want a label even when it isn't available in the language you requested. |
||
| 1659 | * @return array array of property values (key: URI, val: label), or null if concept doesn't exist |
||
| 1660 | */ |
||
| 1661 | public function queryProperty($uri, $prop, $lang, $anylang = false) { |
||
| 1662 | $uri = is_array($uri) ? $uri[0] : $uri; |
||
| 1663 | $query = $this->generatePropertyQuery($uri, $prop, $lang, $anylang); |
||
| 1664 | $result = $this->query($query); |
||
| 1665 | return $this->transformPropertyQueryResults($result, $lang); |
||
| 1666 | } |
||
| 1667 | |||
| 1668 | /** |
||
| 1669 | * Query a single transitive property of a concept. |
||
| 1670 | * @param string $uri |
||
| 1671 | * @param array $props the name of the property eg. 'skos:broader'. |
||
| 1672 | * @param string $lang |
||
| 1673 | * @param integer $limit |
||
| 1674 | * @param boolean $anylang if you want a label even when it isn't available in the language you requested. |
||
| 1675 | * @return string sparql query |
||
| 1676 | */ |
||
| 1677 | private function generateTransitivePropertyQuery($uri, $props, $lang, $limit, $anylang) { |
||
| 1678 | $uri = is_array($uri) ? $uri[0] : $uri; |
||
| 1679 | $fcl = $this->generateFromClause(); |
||
| 1680 | $propertyClause = implode('|', $props); |
||
| 1681 | $otherlang = $anylang ? "OPTIONAL { ?object skos:prefLabel ?label }" : ""; |
||
| 1682 | // need to do a SPARQL subquery because LIMIT needs to be applied /after/ |
||
| 1683 | // the direct relationships have been collapsed into one string |
||
| 1684 | $query = <<<EOQ |
||
| 1685 | SELECT * $fcl |
||
| 1686 | WHERE { |
||
| 1687 | SELECT ?object ?label (GROUP_CONCAT(STR(?dir);separator=' ') as ?direct) |
||
| 1688 | WHERE { |
||
| 1689 | <$uri> a skos:Concept . |
||
| 1690 | OPTIONAL { |
||
| 1691 | <$uri> $propertyClause* ?object . |
||
| 1692 | OPTIONAL { |
||
| 1693 | ?object $propertyClause ?dir . |
||
| 1694 | } |
||
| 1695 | } |
||
| 1696 | OPTIONAL { |
||
| 1697 | ?object skos:prefLabel ?label . |
||
| 1698 | FILTER (langMatches(lang(?label), "$lang")) |
||
| 1699 | } |
||
| 1700 | $otherlang |
||
| 1701 | } |
||
| 1702 | GROUP BY ?object ?label |
||
| 1703 | } |
||
| 1704 | LIMIT $limit |
||
| 1705 | EOQ; |
||
| 1706 | return $query; |
||
| 1707 | } |
||
| 1708 | |||
| 1709 | /** |
||
| 1710 | * Transforms the sparql query result object into an array. |
||
| 1711 | * @param EasyRdf\Sparql\Result $result |
||
| 1712 | * @param string $lang |
||
| 1713 | * @param string $fallbacklang language to use if label is not available in the preferred language |
||
| 1714 | * @return array of property values (key: URI, val: label), or null if concept doesn't exist |
||
| 1715 | */ |
||
| 1716 | private function transformTransitivePropertyResults($result, $lang, $fallbacklang) { |
||
| 1717 | $ret = array(); |
||
| 1718 | foreach ($result as $row) { |
||
| 1719 | if (!isset($row->object)) { |
||
| 1720 | return array(); |
||
| 1721 | } |
||
| 1722 | // existing concept but no properties |
||
| 1723 | if (isset($row->label)) { |
||
| 1724 | $val = array('label' => $row->label->getValue()); |
||
| 1725 | } else { |
||
| 1726 | $val = array('label' => null); |
||
| 1727 | } |
||
| 1728 | if (isset($row->direct) && $row->direct->getValue() != '') { |
||
| 1729 | $val['direct'] = explode(' ', $row->direct->getValue()); |
||
| 1730 | } |
||
| 1731 | // Preventing labels in a non preferred language overriding the preferred language. |
||
| 1732 | if (isset($row->label) && $row->label->getLang() === $lang || array_key_exists($row->object->getUri(), $ret) === false) { |
||
| 1733 | if (!isset($row->label) || $row->label->getLang() === $lang) { |
||
| 1734 | $ret[$row->object->getUri()] = $val; |
||
| 1735 | } elseif ($row->label->getLang() === $fallbacklang) { |
||
| 1736 | $val['label'] .= ' (' . $row->label->getLang() . ')'; |
||
| 1737 | $ret[$row->object->getUri()] = $val; |
||
| 1738 | } |
||
| 1739 | } |
||
| 1740 | } |
||
| 1741 | |||
| 1742 | // second iteration of results to find labels for the ones that didn't have one in the preferred languages |
||
| 1743 | foreach ($result as $row) { |
||
| 1744 | if (isset($row->object) && array_key_exists($row->object->getUri(), $ret) === false) { |
||
| 1745 | $val = array('label' => $row->label->getValue()); |
||
| 1746 | if (isset($row->direct) && $row->direct->getValue() != '') { |
||
| 1747 | $val['direct'] = explode(' ', $row->direct->getValue()); |
||
| 1748 | } |
||
| 1749 | $ret[$row->object->getUri()] = $val; |
||
| 1750 | } |
||
| 1751 | } |
||
| 1752 | |||
| 1753 | if (sizeof($ret) > 0) { |
||
| 1754 | return $ret; |
||
| 1755 | } |
||
| 1756 | // existing concept, with properties |
||
| 1757 | else { |
||
| 1758 | return null; |
||
| 1759 | } |
||
| 1760 | // nonexistent concept |
||
| 1761 | } |
||
| 1762 | |||
| 1763 | /** |
||
| 1764 | * Query a single transitive property of a concept. |
||
| 1765 | * @param string $uri |
||
| 1766 | * @param array $props the property/properties. |
||
| 1767 | * @param string $lang |
||
| 1768 | * @param string $fallbacklang language to use if label is not available in the preferred language |
||
| 1769 | * @param integer $limit |
||
| 1770 | * @param boolean $anylang if you want a label even when it isn't available in the language you requested. |
||
| 1771 | * @return array array of property values (key: URI, val: label), or null if concept doesn't exist |
||
| 1772 | */ |
||
| 1773 | public function queryTransitiveProperty($uri, $props, $lang, $limit, $anylang = false, $fallbacklang = '') { |
||
| 1774 | $query = $this->generateTransitivePropertyQuery($uri, $props, $lang, $limit, $anylang); |
||
| 1775 | $result = $this->query($query); |
||
| 1776 | return $this->transformTransitivePropertyResults($result, $lang, $fallbacklang); |
||
| 1777 | } |
||
| 1778 | |||
| 1779 | /** |
||
| 1780 | * Generates the query for a concepts skos:narrowers. |
||
| 1781 | * @param string $uri |
||
| 1782 | * @param string $lang |
||
| 1783 | * @param string $fallback |
||
| 1784 | * @return string sparql query |
||
| 1785 | */ |
||
| 1786 | private function generateChildQuery($uri, $lang, $fallback, $props) { |
||
| 1787 | $uri = is_array($uri) ? $uri[0] : $uri; |
||
| 1788 | $fcl = $this->generateFromClause(); |
||
| 1789 | $propertyClause = implode('|', $props); |
||
| 1790 | $query = <<<EOQ |
||
| 1791 | SELECT ?child ?label ?child ?grandchildren ?notation $fcl WHERE { |
||
| 1792 | <$uri> a skos:Concept . |
||
| 1793 | OPTIONAL { |
||
| 1794 | ?child $propertyClause <$uri> . |
||
| 1795 | OPTIONAL { |
||
| 1796 | ?child skos:prefLabel ?label . |
||
| 1797 | FILTER (langMatches(lang(?label), "$lang")) |
||
| 1798 | } |
||
| 1799 | OPTIONAL { |
||
| 1800 | ?child skos:prefLabel ?label . |
||
| 1801 | FILTER (langMatches(lang(?label), "$fallback")) |
||
| 1802 | } |
||
| 1803 | OPTIONAL { # other language case |
||
| 1804 | ?child skos:prefLabel ?label . |
||
| 1805 | } |
||
| 1806 | OPTIONAL { |
||
| 1807 | ?child skos:notation ?notation . |
||
| 1808 | } |
||
| 1809 | BIND ( EXISTS { ?a $propertyClause ?child . } AS ?grandchildren ) |
||
| 1810 | } |
||
| 1811 | } |
||
| 1812 | EOQ; |
||
| 1813 | return $query; |
||
| 1814 | } |
||
| 1815 | |||
| 1816 | /** |
||
| 1817 | * Transforms the sparql result object into an array. |
||
| 1818 | * @param EasyRdf\Sparql\Result $result |
||
| 1819 | * @param string $lang |
||
| 1820 | * @return array array of arrays describing each child concept, or null if concept doesn't exist |
||
| 1821 | */ |
||
| 1822 | private function transformNarrowerResults($result, $lang) { |
||
| 1823 | $ret = array(); |
||
| 1824 | foreach ($result as $row) { |
||
| 1825 | if (!isset($row->child)) { |
||
| 1826 | return array(); |
||
| 1827 | } |
||
| 1828 | // existing concept but no children |
||
| 1829 | |||
| 1830 | $label = null; |
||
| 1831 | if (isset($row->label)) { |
||
| 1832 | if ($row->label->getLang() == $lang || strpos($row->label->getLang(), $lang . "-") == 0) { |
||
| 1833 | $label = $row->label->getValue(); |
||
| 1834 | } else { |
||
| 1835 | $label = $row->label->getValue() . " (" . $row->label->getLang() . ")"; |
||
| 1836 | } |
||
| 1837 | |||
| 1838 | } |
||
| 1839 | $childArray = array( |
||
| 1840 | 'uri' => $row->child->getUri(), |
||
| 1841 | 'prefLabel' => $label, |
||
| 1842 | 'hasChildren' => filter_var($row->grandchildren->getValue(), FILTER_VALIDATE_BOOLEAN), |
||
| 1843 | ); |
||
| 1844 | if (isset($row->notation)) { |
||
| 1845 | $childArray['notation'] = $row->notation->getValue(); |
||
| 1846 | } |
||
| 1847 | |||
| 1848 | $ret[] = $childArray; |
||
| 1849 | } |
||
| 1850 | if (sizeof($ret) > 0) { |
||
| 1851 | return $ret; |
||
| 1852 | } |
||
| 1853 | // existing concept, with children |
||
| 1854 | else { |
||
| 1855 | return null; |
||
| 1856 | } |
||
| 1857 | // nonexistent concept |
||
| 1858 | } |
||
| 1859 | |||
| 1860 | /** |
||
| 1861 | * Query the narrower concepts of a concept. |
||
| 1862 | * @param string $uri |
||
| 1863 | * @param string $lang |
||
| 1864 | * @param string $fallback |
||
| 1865 | * @return array array of arrays describing each child concept, or null if concept doesn't exist |
||
| 1866 | */ |
||
| 1867 | public function queryChildren($uri, $lang, $fallback, $props) { |
||
| 1868 | $query = $this->generateChildQuery($uri, $lang, $fallback, $props); |
||
| 1869 | $result = $this->query($query); |
||
| 1870 | return $this->transformNarrowerResults($result, $lang); |
||
| 1871 | } |
||
| 1872 | |||
| 1873 | /** |
||
| 1874 | * Query the top concepts of a vocabulary. |
||
| 1875 | * @param string $conceptSchemes concept schemes whose top concepts to query for |
||
| 1876 | * @param string $lang language of labels |
||
| 1877 | * @param string $fallback language to use if label is not available in the preferred language |
||
| 1878 | */ |
||
| 1879 | public function queryTopConcepts($conceptSchemes, $lang, $fallback) { |
||
| 1880 | if (!is_array($conceptSchemes)) { |
||
| 1881 | $conceptSchemes = array($conceptSchemes); |
||
| 1882 | } |
||
| 1883 | |||
| 1884 | $values = $this->formatValues('?topuri', $conceptSchemes, 'uri'); |
||
| 1885 | |||
| 1886 | $fcl = $this->generateFromClause(); |
||
| 1887 | $query = <<<EOQ |
||
| 1888 | SELECT DISTINCT ?top ?topuri ?label ?notation ?children $fcl WHERE { |
||
| 1889 | ?top skos:topConceptOf ?topuri . |
||
| 1890 | OPTIONAL { |
||
| 1891 | ?top skos:prefLabel ?label . |
||
| 1892 | FILTER (langMatches(lang(?label), "$lang")) |
||
| 1893 | } |
||
| 1894 | OPTIONAL { |
||
| 1895 | ?top skos:prefLabel ?label . |
||
| 1896 | FILTER (langMatches(lang(?label), "$fallback")) |
||
| 1897 | } |
||
| 1898 | OPTIONAL { # fallback - other language case |
||
| 1899 | ?top skos:prefLabel ?label . |
||
| 1900 | } |
||
| 1901 | OPTIONAL { ?top skos:notation ?notation . } |
||
| 1902 | BIND ( EXISTS { ?top skos:narrower ?a . } AS ?children ) |
||
| 1903 | $values |
||
| 1904 | } |
||
| 1905 | EOQ; |
||
| 1906 | $result = $this->query($query); |
||
| 1907 | $ret = array(); |
||
| 1908 | foreach ($result as $row) { |
||
| 1909 | if (isset($row->top) && isset($row->label)) { |
||
| 1910 | $label = $row->label->getValue(); |
||
| 1911 | if ($row->label->getLang() && $row->label->getLang() !== $lang && strpos($row->label->getLang(), $lang . "-") !== 0) { |
||
| 1912 | $label .= ' (' . $row->label->getLang() . ')'; |
||
| 1913 | } |
||
| 1914 | $top = array('uri' => $row->top->getUri(), 'topConceptOf' => $row->topuri->getUri(), 'label' => $label, 'hasChildren' => filter_var($row->children->getValue(), FILTER_VALIDATE_BOOLEAN)); |
||
| 1915 | if (isset($row->notation)) { |
||
| 1916 | $top['notation'] = $row->notation->getValue(); |
||
| 1917 | } |
||
| 1918 | |||
| 1919 | $ret[] = $top; |
||
| 1920 | } |
||
| 1921 | } |
||
| 1922 | |||
| 1923 | return $ret; |
||
| 1924 | } |
||
| 1925 | |||
| 1926 | /** |
||
| 1927 | * Generates a sparql query for finding the hierarchy for a concept. |
||
| 1928 | * A concept may be a top concept in multiple schemes, returned as a single whitespace-separated literal. |
||
| 1929 | * @param string $uri concept uri. |
||
| 1930 | * @param string $lang |
||
| 1931 | * @param string $fallback language to use if label is not available in the preferred language |
||
| 1932 | * @return string sparql query |
||
| 1933 | */ |
||
| 1934 | private function generateParentListQuery($uri, $lang, $fallback, $props) { |
||
| 1935 | $fcl = $this->generateFromClause(); |
||
| 1936 | $propertyClause = implode('|', $props); |
||
| 1937 | $query = <<<EOQ |
||
| 1938 | SELECT ?broad ?parent ?children ?grandchildren |
||
| 1939 | (SAMPLE(?lab) as ?label) (SAMPLE(?childlab) as ?childlabel) (GROUP_CONCAT(?topcs; separator=" ") as ?tops) |
||
| 1940 | (SAMPLE(?nota) as ?notation) (SAMPLE(?childnota) as ?childnotation) $fcl |
||
| 1941 | WHERE { |
||
| 1942 | <$uri> a skos:Concept . |
||
| 1943 | OPTIONAL { |
||
| 1944 | <$uri> $propertyClause* ?broad . |
||
| 1945 | OPTIONAL { |
||
| 1946 | ?broad skos:prefLabel ?lab . |
||
| 1947 | FILTER (langMatches(lang(?lab), "$lang")) |
||
| 1948 | } |
||
| 1949 | OPTIONAL { |
||
| 1950 | ?broad skos:prefLabel ?lab . |
||
| 1951 | FILTER (langMatches(lang(?lab), "$fallback")) |
||
| 1952 | } |
||
| 1953 | OPTIONAL { # fallback - other language case |
||
| 1954 | ?broad skos:prefLabel ?lab . |
||
| 1955 | } |
||
| 1956 | OPTIONAL { ?broad skos:notation ?nota . } |
||
| 1957 | OPTIONAL { ?broad $propertyClause ?parent . } |
||
| 1958 | OPTIONAL { ?broad skos:narrower ?children . |
||
| 1959 | OPTIONAL { |
||
| 1960 | ?children skos:prefLabel ?childlab . |
||
| 1961 | FILTER (langMatches(lang(?childlab), "$lang")) |
||
| 1962 | } |
||
| 1963 | OPTIONAL { |
||
| 1964 | ?children skos:prefLabel ?childlab . |
||
| 1965 | FILTER (langMatches(lang(?childlab), "$fallback")) |
||
| 1966 | } |
||
| 1967 | OPTIONAL { # fallback - other language case |
||
| 1968 | ?children skos:prefLabel ?childlab . |
||
| 1969 | } |
||
| 1970 | OPTIONAL { |
||
| 1971 | ?children skos:notation ?childnota . |
||
| 1972 | } |
||
| 1973 | } |
||
| 1974 | BIND ( EXISTS { ?children skos:narrower ?a . } AS ?grandchildren ) |
||
| 1975 | OPTIONAL { ?broad skos:topConceptOf ?topcs . } |
||
| 1976 | } |
||
| 1977 | } |
||
| 1978 | GROUP BY ?broad ?parent ?member ?children ?grandchildren |
||
| 1979 | EOQ; |
||
| 1980 | return $query; |
||
| 1981 | } |
||
| 1982 | |||
| 1983 | /** |
||
| 1984 | * Transforms the result into an array. |
||
| 1985 | * @param EasyRdf\Sparql\Result |
||
| 1986 | * @param string $lang |
||
| 1987 | * @return array|null an array for the REST controller to encode. |
||
| 1988 | */ |
||
| 1989 | private function transformParentListResults($result, $lang) |
||
| 1990 | { |
||
| 1991 | $ret = array(); |
||
| 1992 | foreach ($result as $row) { |
||
| 1993 | if (!isset($row->broad)) { |
||
| 1994 | // existing concept but no broaders |
||
| 1995 | return array(); |
||
| 1996 | } |
||
| 1997 | $uri = $row->broad->getUri(); |
||
| 1998 | if (!isset($ret[$uri])) { |
||
| 1999 | $ret[$uri] = array('uri' => $uri); |
||
| 2000 | } |
||
| 2001 | if (isset($row->exact)) { |
||
| 2002 | $ret[$uri]['exact'] = $row->exact->getUri(); |
||
| 2003 | } |
||
| 2004 | if (isset($row->tops)) { |
||
| 2005 | $topConceptsList=explode(" ", $row->tops->getValue()); |
||
| 2006 | // sort to guarantee an alphabetical ordering of the URI |
||
| 2007 | sort($topConceptsList); |
||
| 2008 | $ret[$uri]['tops'] = $topConceptsList; |
||
| 2009 | } |
||
| 2010 | if (isset($row->children)) { |
||
| 2011 | if (!isset($ret[$uri]['narrower'])) { |
||
| 2012 | $ret[$uri]['narrower'] = array(); |
||
| 2013 | } |
||
| 2014 | |||
| 2015 | $label = null; |
||
| 2016 | if (isset($row->childlabel)) { |
||
| 2017 | $label = $row->childlabel->getValue(); |
||
| 2018 | if ($row->childlabel->getLang() !== $lang && strpos($row->childlabel->getLang(), $lang . "-") !== 0) { |
||
| 2019 | $label .= " (" . $row->childlabel->getLang() . ")"; |
||
| 2020 | } |
||
| 2021 | |||
| 2022 | } |
||
| 2023 | |||
| 2024 | $childArr = array( |
||
| 2025 | 'uri' => $row->children->getUri(), |
||
| 2026 | 'label' => $label, |
||
| 2027 | 'hasChildren' => filter_var($row->grandchildren->getValue(), FILTER_VALIDATE_BOOLEAN), |
||
| 2028 | ); |
||
| 2029 | if (isset($row->childnotation)) { |
||
| 2030 | $childArr['notation'] = $row->childnotation->getValue(); |
||
| 2031 | } |
||
| 2032 | |||
| 2033 | if (!in_array($childArr, $ret[$uri]['narrower'])) { |
||
| 2034 | $ret[$uri]['narrower'][] = $childArr; |
||
| 2035 | } |
||
| 2036 | |||
| 2037 | } |
||
| 2038 | if (isset($row->label)) { |
||
| 2039 | $preflabel = $row->label->getValue(); |
||
| 2040 | if ($row->label->getLang() && $row->label->getLang() !== $lang && strpos($row->label->getLang(), $lang . "-") !== 0) { |
||
| 2041 | $preflabel .= ' (' . $row->label->getLang() . ')'; |
||
| 2042 | } |
||
| 2043 | |||
| 2044 | $ret[$uri]['prefLabel'] = $preflabel; |
||
| 2045 | } |
||
| 2046 | if (isset($row->notation)) { |
||
| 2047 | $ret[$uri]['notation'] = $row->notation->getValue(); |
||
| 2048 | } |
||
| 2049 | |||
| 2050 | if (isset($row->parent) && (isset($ret[$uri]['broader']) && !in_array($row->parent->getUri(), $ret[$uri]['broader']))) { |
||
| 2051 | $ret[$uri]['broader'][] = $row->parent->getUri(); |
||
| 2052 | } elseif (isset($row->parent) && !isset($ret[$uri]['broader'])) { |
||
| 2053 | $ret[$uri]['broader'][] = $row->parent->getUri(); |
||
| 2054 | } |
||
| 2055 | } |
||
| 2056 | if (sizeof($ret) > 0) { |
||
| 2057 | // existing concept, with children |
||
| 2058 | return $ret; |
||
| 2059 | } |
||
| 2060 | else { |
||
| 2061 | // nonexistent concept |
||
| 2062 | return null; |
||
| 2063 | } |
||
| 2064 | } |
||
| 2065 | |||
| 2066 | /** |
||
| 2067 | * Query for finding the hierarchy for a concept. |
||
| 2068 | * @param string $uri concept uri. |
||
| 2069 | * @param string $lang |
||
| 2070 | * @param string $fallback language to use if label is not available in the preferred language |
||
| 2071 | * @param array $props the hierarchy property/properties to use |
||
| 2072 | * @return an array for the REST controller to encode. |
||
| 2073 | */ |
||
| 2074 | public function queryParentList($uri, $lang, $fallback, $props) { |
||
| 2075 | $query = $this->generateParentListQuery($uri, $lang, $fallback, $props); |
||
| 2076 | $result = $this->query($query); |
||
| 2077 | return $this->transformParentListResults($result, $lang); |
||
| 2078 | } |
||
| 2079 | |||
| 2080 | /** |
||
| 2081 | * return a list of concept group instances, sorted by label |
||
| 2082 | * @param string $groupClass URI of concept group class |
||
| 2083 | * @param string $lang language of labels to return |
||
| 2084 | * @return string sparql query |
||
| 2085 | */ |
||
| 2086 | private function generateConceptGroupsQuery($groupClass, $lang) { |
||
| 2087 | $fcl = $this->generateFromClause(); |
||
| 2088 | $query = <<<EOQ |
||
| 2089 | SELECT ?group (GROUP_CONCAT(DISTINCT STR(?child);separator=' ') as ?children) ?label ?members ?notation $fcl |
||
| 2090 | WHERE { |
||
| 2091 | ?group a <$groupClass> . |
||
| 2092 | OPTIONAL { ?group skos:member|isothes:subGroup ?child . |
||
| 2093 | ?child a <$groupClass> } |
||
| 2094 | BIND(EXISTS{?group skos:member ?submembers} as ?members) |
||
| 2095 | OPTIONAL { ?group skos:prefLabel ?label } |
||
| 2096 | OPTIONAL { ?group rdfs:label ?label } |
||
| 2097 | FILTER (langMatches(lang(?label), '$lang')) |
||
| 2098 | OPTIONAL { ?group skos:notation ?notation } |
||
| 2099 | } |
||
| 2100 | GROUP BY ?group ?label ?members ?notation |
||
| 2101 | ORDER BY lcase(?label) |
||
| 2102 | EOQ; |
||
| 2103 | return $query; |
||
| 2104 | } |
||
| 2105 | |||
| 2106 | /** |
||
| 2107 | * Transforms the sparql query result into an array. |
||
| 2108 | * @param EasyRdf\Sparql\Result $result |
||
| 2109 | * @return array |
||
| 2110 | */ |
||
| 2111 | private function transformConceptGroupsResults($result) { |
||
| 2112 | $ret = array(); |
||
| 2113 | foreach ($result as $row) { |
||
| 2114 | if (!isset($row->group)) { |
||
| 2115 | # no groups found, see issue #357 |
||
| 2116 | continue; |
||
| 2117 | } |
||
| 2118 | $group = array('uri' => $row->group->getURI()); |
||
| 2119 | if (isset($row->label)) { |
||
| 2120 | $group['prefLabel'] = $row->label->getValue(); |
||
| 2121 | } |
||
| 2122 | |||
| 2123 | if (isset($row->children)) { |
||
| 2124 | $group['childGroups'] = explode(' ', $row->children->getValue()); |
||
| 2125 | } |
||
| 2126 | |||
| 2127 | if (isset($row->members)) { |
||
| 2128 | $group['hasMembers'] = $row->members->getValue(); |
||
| 2129 | } |
||
| 2130 | |||
| 2131 | if (isset($row->notation)) { |
||
| 2132 | $group['notation'] = $row->notation->getValue(); |
||
| 2133 | } |
||
| 2134 | |||
| 2135 | $ret[] = $group; |
||
| 2136 | } |
||
| 2137 | return $ret; |
||
| 2138 | } |
||
| 2139 | |||
| 2140 | /** |
||
| 2141 | * return a list of concept group instances, sorted by label |
||
| 2142 | * @param string $groupClass URI of concept group class |
||
| 2143 | * @param string $lang language of labels to return |
||
| 2144 | * @return array Result array with group URI as key and group label as value |
||
| 2145 | */ |
||
| 2146 | public function listConceptGroups($groupClass, $lang) { |
||
| 2147 | $query = $this->generateConceptGroupsQuery($groupClass, $lang); |
||
| 2148 | $result = $this->query($query); |
||
| 2149 | return $this->transformConceptGroupsResults($result); |
||
| 2150 | } |
||
| 2151 | |||
| 2152 | /** |
||
| 2153 | * Generates the sparql query for listConceptGroupContents |
||
| 2154 | * @param string $groupClass URI of concept group class |
||
| 2155 | * @param string $group URI of the concept group instance |
||
| 2156 | * @param string $lang language of labels to return |
||
| 2157 | * @param boolean $showDeprecated whether to include deprecated in the result |
||
| 2158 | * @return string sparql query |
||
| 2159 | */ |
||
| 2160 | private function generateConceptGroupContentsQuery($groupClass, $group, $lang, $showDeprecated = false) { |
||
| 2161 | $fcl = $this->generateFromClause(); |
||
| 2162 | $filterDeprecated=""; |
||
| 2163 | if(!$showDeprecated){ |
||
| 2164 | $filterDeprecated=" FILTER NOT EXISTS { ?conc owl:deprecated true }"; |
||
| 2165 | } |
||
| 2166 | $query = <<<EOQ |
||
| 2167 | SELECT ?conc ?super ?label ?members ?type ?notation $fcl |
||
| 2168 | WHERE { |
||
| 2169 | <$group> a <$groupClass> . |
||
| 2170 | { <$group> skos:member ?conc . } UNION { ?conc isothes:superGroup <$group> } |
||
| 2171 | $filterDeprecated |
||
| 2172 | ?conc a ?type . |
||
| 2173 | OPTIONAL { ?conc skos:prefLabel ?label . |
||
| 2174 | FILTER (langMatches(lang(?label), '$lang')) |
||
| 2175 | } |
||
| 2176 | OPTIONAL { ?conc skos:prefLabel ?label . } |
||
| 2177 | OPTIONAL { ?conc skos:notation ?notation } |
||
| 2178 | BIND(EXISTS{?submembers isothes:superGroup ?conc} as ?super) |
||
| 2179 | BIND(EXISTS{?conc skos:member ?submembers} as ?members) |
||
| 2180 | } ORDER BY lcase(?label) |
||
| 2181 | EOQ; |
||
| 2182 | return $query; |
||
| 2183 | } |
||
| 2184 | |||
| 2185 | /** |
||
| 2186 | * Transforms the sparql query result into an array. |
||
| 2187 | * @param EasyRdf\Sparql\Result $result |
||
| 2188 | * @param string $lang language of labels to return |
||
| 2189 | * @return array |
||
| 2190 | */ |
||
| 2191 | private function transformConceptGroupContentsResults($result, $lang) { |
||
| 2192 | $ret = array(); |
||
| 2193 | $values = array(); |
||
| 2194 | foreach ($result as $row) { |
||
| 2195 | if (!array_key_exists($row->conc->getURI(), $values)) { |
||
| 2196 | $values[$row->conc->getURI()] = array( |
||
| 2197 | 'uri' => $row->conc->getURI(), |
||
| 2198 | 'isSuper' => $row->super->getValue(), |
||
| 2199 | 'hasMembers' => $row->members->getValue(), |
||
| 2200 | 'type' => array($row->type->shorten()), |
||
| 2201 | ); |
||
| 2202 | if (isset($row->label)) { |
||
| 2203 | if ($row->label->getLang() == $lang || strpos($row->label->getLang(), $lang . "-") == 0) { |
||
| 2204 | $values[$row->conc->getURI()]['prefLabel'] = $row->label->getValue(); |
||
| 2205 | } else { |
||
| 2206 | $values[$row->conc->getURI()]['prefLabel'] = $row->label->getValue() . " (" . $row->label->getLang() . ")"; |
||
| 2207 | } |
||
| 2208 | |||
| 2209 | } |
||
| 2210 | if (isset($row->notation)) { |
||
| 2211 | $values[$row->conc->getURI()]['notation'] = $row->notation->getValue(); |
||
| 2212 | } |
||
| 2213 | |||
| 2214 | } else { |
||
| 2215 | $values[$row->conc->getURI()]['type'][] = $row->type->shorten(); |
||
| 2216 | } |
||
| 2217 | } |
||
| 2218 | |||
| 2219 | foreach ($values as $val) { |
||
| 2220 | $ret[] = $val; |
||
| 2221 | } |
||
| 2222 | |||
| 2223 | return $ret; |
||
| 2224 | } |
||
| 2225 | |||
| 2226 | /** |
||
| 2227 | * return a list of concepts in a concept group |
||
| 2228 | * @param string $groupClass URI of concept group class |
||
| 2229 | * @param string $group URI of the concept group instance |
||
| 2230 | * @param string $lang language of labels to return |
||
| 2231 | * @param boolean $showDeprecated whether to include deprecated concepts in search results |
||
| 2232 | * @return array Result array with concept URI as key and concept label as value |
||
| 2233 | */ |
||
| 2234 | public function listConceptGroupContents($groupClass, $group, $lang,$showDeprecated = false) { |
||
| 2235 | $query = $this->generateConceptGroupContentsQuery($groupClass, $group, $lang,$showDeprecated); |
||
| 2236 | $result = $this->query($query); |
||
| 2237 | return $this->transformConceptGroupContentsResults($result, $lang); |
||
| 2238 | } |
||
| 2239 | |||
| 2240 | /** |
||
| 2241 | * Generates the sparql query for queryChangeList. |
||
| 2242 | * @param string $prop the property uri pointing to timestamps, eg. 'dc:modified' |
||
| 2243 | * @param string $lang language of labels to return |
||
| 2244 | * @param int $offset offset of results to retrieve; 0 for beginning of list |
||
| 2245 | * @param int $limit maximum number of results to return |
||
| 2246 | * @param boolean $showDeprecated whether to include deprecated concepts in the change list |
||
| 2247 | * @return string sparql query |
||
| 2248 | */ |
||
| 2249 | private function generateChangeListQuery($prop, $lang, $offset, $limit=200, $showDeprecated=false) { |
||
| 2250 | $fcl = $this->generateFromClause(); |
||
| 2251 | $offset = ($offset) ? 'OFFSET ' . $offset : ''; |
||
| 2252 | |||
| 2253 | //Additional clauses when deprecated concepts need to be included in the results |
||
| 2254 | $deprecatedOptions = ''; |
||
| 2255 | $deprecatedVars = ''; |
||
| 2256 | if ($showDeprecated) { |
||
| 2257 | $deprecatedVars = '?replacedBy ?deprecated ?replacingLabel'; |
||
| 2258 | $deprecatedOptions = |
||
| 2259 | 'UNION {'. |
||
| 2260 | '?concept dc:isReplacedBy ?replacedBy ; dc:modified ?date2 .'. |
||
| 2261 | 'BIND(COALESCE(?date2, ?date) AS ?date)'. |
||
| 2262 | 'OPTIONAL { ?replacedBy skos:prefLabel ?replacingLabel .'. |
||
| 2263 | 'FILTER (langMatches(lang(?replacingLabel), \''.$lang.'\')) }}'. |
||
| 2264 | 'OPTIONAL { ?concept owl:deprecated ?deprecated . }'; |
||
| 2265 | } |
||
| 2266 | |||
| 2267 | $query = <<<EOQ |
||
| 2268 | SELECT ?concept ?date ?label $deprecatedVars $fcl |
||
| 2269 | WHERE { |
||
| 2270 | ?concept a skos:Concept ; |
||
| 2271 | skos:prefLabel ?label . |
||
| 2272 | FILTER (langMatches(lang(?label), '$lang')) |
||
| 2273 | { |
||
| 2274 | ?concept $prop ?date . |
||
| 2275 | MINUS { ?concept owl:deprecated True . } |
||
| 2276 | } |
||
| 2277 | $deprecatedOptions |
||
| 2278 | } |
||
| 2279 | ORDER BY DESC(YEAR(?date)) DESC(MONTH(?date)) LCASE(?label) DESC(?concept) |
||
| 2280 | LIMIT $limit $offset |
||
| 2281 | EOQ; |
||
| 2282 | |||
| 2283 | return $query; |
||
| 2284 | } |
||
| 2285 | |||
| 2286 | /** |
||
| 2287 | * Transforms the sparql query result into an array. |
||
| 2288 | * @param EasyRdf\Sparql\Result $result |
||
| 2289 | * @return array |
||
| 2290 | */ |
||
| 2291 | private function transformChangeListResults($result) { |
||
| 2292 | $ret = array(); |
||
| 2293 | foreach ($result as $row) { |
||
| 2294 | $concept = array('uri' => $row->concept->getURI()); |
||
| 2295 | if (isset($row->label)) { |
||
| 2296 | $concept['prefLabel'] = $row->label->getValue(); |
||
| 2297 | } |
||
| 2298 | |||
| 2299 | if (isset($row->date)) { |
||
| 2300 | try { |
||
| 2301 | $concept['date'] = $row->date->getValue(); |
||
| 2302 | } catch (Exception $e) { |
||
| 2303 | //don't record concepts with malformed dates e.g. 1986-21-00 |
||
| 2304 | continue; |
||
| 2305 | } |
||
| 2306 | } |
||
| 2307 | |||
| 2336 |
