| Index: trunk/extensions/SemanticMediaWiki/includes/SMW_QueryProcessor.php |
| — | — | @@ -33,33 +33,18 @@ |
| 34 | 34 | * as a string. Otherwise an object of type SMWQuery is returned. |
| 35 | 35 | */ |
| 36 | 36 | static public function createQuery($querystring, $params, $inline = true) { |
| 37 | | - /// TODO implement |
| 38 | | - // DEBUG: |
| 39 | | - $o_desc = new SMWNominalDescription(Title::newFromText("Africa")); |
| 40 | | - $value = SMWDataValue::newAttributeValue('Population','5853000'); |
| 41 | | - $t_desc = new SMWThingDescription(); |
| 42 | | - $v_desc = new SMWValueDescription($value, SMW_CMP_GEQ); |
| 43 | | - $a_desc = new SMWSomeAttribute(Title::newFromText('Attribute:Population'), $v_desc); |
| 44 | | - $r_desc = new SMWSomeRelation(Title::newFromText("Relation:located in"), $o_desc); |
| 45 | | - $r_desc2 = new SMWSomeRelation(Title::newFromText("Relation:borders"), $r_desc); |
| 46 | | - $r_desc3 = new SMWSomeRelation(Title::newFromText("Relation:located in"), $t_desc); |
| 47 | | - $c_desc = new SMWClassDescription(Title::newFromText("Category:Country")); |
| 48 | | - $desc = new SMWConjunction(array($c_desc, $r_desc)); |
| 49 | | - $desc2 = new SMWConjunction(array($c_desc, $a_desc, $r_desc2, $r_desc)); |
| 50 | | - $pr1 = new SMWPrintrequest(SMW_PRINT_THIS, 'Country'); |
| 51 | | - $desc->addPrintRequest($pr1); |
| 52 | | - $desc2->addPrintRequest($pr1); |
| 53 | | - $pr2 = new SMWPrintrequest(SMW_PRINT_RELS, 'Borders', Title::newFromText('Relation:Borders')); |
| 54 | | - $desc->addPrintRequest($pr2); |
| 55 | | - $desc2->addPrintRequest($pr2); |
| 56 | | - $pr3 = new SMWPrintrequest(SMW_PRINT_ATTS, 'Population', Title::newFromText('Attribute:Population')); |
| 57 | | - $desc->addPrintRequest($pr3); |
| 58 | | - $desc2->addPrintRequest($pr3); |
| 59 | | - $pr4 = new SMWPrintrequest(SMW_PRINT_CATS, 'Categories'); |
| 60 | | - $desc->addPrintRequest($pr4); |
| 61 | | - //$query = new SMWQuery($desc); |
| 62 | | - $query = new SMWQuery($desc2); |
| | 37 | + // parse query: |
| | 38 | + $qp = new SMWQueryParser(); |
| | 39 | + $desc = $qp->getQueryDescription($querystring); |
| | 40 | + /// TODO check for errors |
| 63 | 41 | |
| | 42 | + ///TODO do this only when wanted, use given label: |
| | 43 | + $desc->addPrintRequest(new SMWPrintrequest(SMW_PRINT_THIS, 'Mainlabel')); |
| | 44 | + |
| | 45 | + $query = new SMWQuery($desc); |
| | 46 | + |
| | 47 | + print '### Query:' . $desc->getQueryString() . ' ###'; |
| | 48 | + |
| 64 | 49 | // set query parameters: |
| 65 | 50 | global $smwgIQMaxLimit, $smwgIQMaxInlineLimit; |
| 66 | 51 | if ($inline) |
| — | — | @@ -144,5 +129,231 @@ |
| 145 | 130 | } |
| 146 | 131 | |
| 147 | 132 | } |
| | 133 | + |
| | 134 | + |
| | 135 | +/** |
| | 136 | + * Objects of this class are in charge of parsing a query string in order |
| | 137 | + * to create an SMWDescription. The class and methods are not static in order |
| | 138 | + * to more cleanly store the intermediate state and progress of the parser. |
| | 139 | + */ |
| | 140 | +class SMWQueryParser { |
| | 141 | + |
| | 142 | + protected $m_sepstack; // list of open blocks ("parentheses") that need closing at current step |
| | 143 | + protected $m_curstring; // remaining string to be parsed (parsing eats query string from the front) |
| | 144 | + protected $m_error; // false if all went right, string otherwise |
| | 145 | + |
| | 146 | + protected $m_categoryprefix; // cache label of category namespace . ':' |
| | 147 | + |
| | 148 | + public function SMWQueryParser() { |
| | 149 | + global $wgContLang; |
| | 150 | + $this->m_categoryprefix = $wgContLang->getNsText(NS_CATEGORY) . ':'; |
| | 151 | + } |
| | 152 | + |
| | 153 | + /** |
| | 154 | + * Compute an SMWDescription from a query string. Return this description or |
| | 155 | + * false if there were errors. |
| | 156 | + */ |
| | 157 | + public function getQueryDescription($querystring) { |
| | 158 | + $this->m_curstring = $querystring; |
| | 159 | + $this->m_sepstack = array(); |
| | 160 | + return $this->getSubqueryDescription(); |
| | 161 | + } |
| | 162 | + |
| | 163 | + /** |
| | 164 | + * Compute an SMWDescription for current part of a query, which should |
| | 165 | + * be a standalone query (the main query or a subquery enclosed within |
| | 166 | + * "<q>...</q>". Recursively calls similar methods and returns false upon error. |
| | 167 | + */ |
| | 168 | + protected function getSubqueryDescription() { |
| | 169 | + $result = NULL; |
| | 170 | + while (($chunk = $this->readChunk()) != '') { |
| | 171 | + switch ($chunk) { |
| | 172 | + case '[[': // start new link block |
| | 173 | + $this->pushDelimiter(']]'); // expected termination symbol |
| | 174 | + $result = $this->addDescription($result,$this->getLinkDescription()); |
| | 175 | + break; |
| | 176 | + case '</q>': // exit current subquery |
| | 177 | + if ($this->popDelimiter('</q>')) { |
| | 178 | + //TODO: return computed description |
| | 179 | + } else { |
| | 180 | + $this->m_error = 'There appear to be too many occurences of \'' . $chunk . '\' in the query.'; |
| | 181 | + return false; |
| | 182 | + } |
| | 183 | + break; |
| | 184 | + default: // error: unexpected $chunk |
| | 185 | + $this->m_error = 'The part \'' . $chunk . '\' in the query was not understood. Results might not be as expected.'; // TODO: internationalise |
| | 186 | + return false; |
| | 187 | + } |
| | 188 | + } |
| | 189 | + return $result; |
| | 190 | + |
| | 191 | + /// TODO implement |
| | 192 | + // DEBUG: |
| | 193 | +// $o_desc = new SMWNominalDescription(Title::newFromText("Africa")); |
| | 194 | +// $value = SMWDataValueFactory::newAttributeValue('Population','5853000'); |
| | 195 | +// $t_desc = new SMWThingDescription(); |
| | 196 | +// $v_desc = new SMWValueDescription($value, SMW_CMP_GEQ); |
| | 197 | +// $a_desc = new SMWSomeAttribute(Title::newFromText('Attribute:Population'), $v_desc); |
| | 198 | +// $r_desc = new SMWSomeRelation(Title::newFromText("Relation:located in"), $o_desc); |
| | 199 | +// $r_desc2 = new SMWSomeRelation(Title::newFromText("Relation:borders"), $r_desc); |
| | 200 | +// $r_desc3 = new SMWSomeRelation(Title::newFromText("Relation:located in"), $t_desc); |
| | 201 | +// $c_desc = new SMWClassDescription(Title::newFromText("Category:Country")); |
| | 202 | +// $desc = new SMWConjunction(array($c_desc, $r_desc)); |
| | 203 | +// $desc2 = new SMWConjunction(array($c_desc, $a_desc, $r_desc2, $r_desc)); |
| | 204 | +// $pr1 = new SMWPrintrequest(SMW_PRINT_THIS, 'Country'); |
| | 205 | +// $desc->addPrintRequest($pr1); |
| | 206 | +// $desc2->addPrintRequest($pr1); |
| | 207 | +// $pr2 = new SMWPrintrequest(SMW_PRINT_RELS, 'Borders', Title::newFromText('Relation:Borders')); |
| | 208 | +// $desc->addPrintRequest($pr2); |
| | 209 | +// $desc2->addPrintRequest($pr2); |
| | 210 | +// $pr3 = new SMWPrintrequest(SMW_PRINT_ATTS, 'Population', Title::newFromText('Attribute:Population')); |
| | 211 | +// $desc->addPrintRequest($pr3); |
| | 212 | +// $desc2->addPrintRequest($pr3); |
| | 213 | +// $pr4 = new SMWPrintrequest(SMW_PRINT_CATS, 'Categories'); |
| | 214 | +// $desc->addPrintRequest($pr4); |
| | 215 | +// |
| | 216 | +// return $desc2; |
| | 217 | + } |
| | 218 | + |
| | 219 | + /** |
| | 220 | + * Compute an SMWDescription for current part of a query, which should |
| | 221 | + * be the content of "[[ ... ]]". Recursively calls similar methods and |
| | 222 | + * returns false upon error. |
| | 223 | + */ |
| | 224 | + protected function getLinkDescription() { |
| | 225 | + $result = NULL; |
| | 226 | + // This method is called when we encountered an opening '[['. The following |
| | 227 | + // block could be a Category-statement, fixed object, relation or attribute |
| | 228 | + // statements, or according print statements. |
| | 229 | + $chunk = $this->readChunk(); |
| | 230 | + |
| | 231 | + if ($chunk == $this->m_categoryprefix) { // category statement |
| | 232 | + // note: no subqueries allowed here, inline disjunction allowed, wildcards allowed |
| | 233 | + $continue = true; |
| | 234 | + while ($continue) { |
| | 235 | + $chunk = $this->readChunk(); |
| | 236 | + switch ($chunk) { |
| | 237 | + case '+': //wildcard |
| | 238 | + break; |
| | 239 | + case '*': //print statement |
| | 240 | + break; |
| | 241 | + default: //assume category title |
| | 242 | + $cat = Title::newFromText($chunk, NS_CATEGORY); |
| | 243 | + if ($cat !== NULL) { |
| | 244 | + $result = $this->addDescription($result, new SMWClassDescription($cat), false); |
| | 245 | + } |
| | 246 | + } |
| | 247 | + $chunk = $this->readChunk(); |
| | 248 | + if ($chunk == '||') { |
| | 249 | + $continue = true; |
| | 250 | + } else { |
| | 251 | + $continue = false; |
| | 252 | + } |
| | 253 | + } |
| | 254 | + } else { // fixed subject, property query, or subquery |
| | 255 | + |
| | 256 | + } |
| | 257 | + |
| | 258 | + // terminate link (assuming that next chunk was read already) |
| | 259 | + if ($chunk == '|') { // label, TODO |
| | 260 | + $chunk = $this->readChunk(); |
| | 261 | + $label = ''; |
| | 262 | + ///TODO: rather have a mode for readChunk that stops only on ']]' |
| | 263 | + /// (otherwise we kill spaces in the label) |
| | 264 | + while ( ($chunk != ']]') && ($chunk !== '') ) { |
| | 265 | + $label .= $chunk; |
| | 266 | + $chunk = $this->readChunk(); |
| | 267 | + } |
| | 268 | + } |
| | 269 | + if ($chunk == ']]') { // expected termination |
| | 270 | + $this->popDelimiter(']]'); |
| | 271 | + return $result; |
| | 272 | + } else { |
| | 273 | + // What happended? We found some chunk that could not be processed as |
| | 274 | + // link content (as in [[Category:Test<q>]]) and there was no label to |
| | 275 | + // eat it. Or the closing ]] are just missing entirely. |
| | 276 | + if ($chunk != '') { //TODO: internationalise errors |
| | 277 | + $this->m_error = 'The symbol \'' . $chunk . '\' was used in a place where it is not useful.'; |
| | 278 | + } else { |
| | 279 | + $this->m_error = 'Some use of \'[[\' in your query was not closed by a matching \']]\'.'; |
| | 280 | + } |
| | 281 | + return false; |
| | 282 | + } |
| | 283 | + |
| | 284 | + return $result; |
| | 285 | + } |
| | 286 | + |
| | 287 | + /** |
| | 288 | + * Get the next unstructured string chunk from the query string. |
| | 289 | + * Chunks are delimited by any of the special strings used in inline queries |
| | 290 | + * (such as [[, ]], <q>, ...). If the string starts with such a delimiter, |
| | 291 | + * this delimiter is returned. Otherwise the first string in front of such a |
| | 292 | + * delimiter is returned. |
| | 293 | + * Trailing and initial spaces are always ignored and chunks |
| | 294 | + * consisting only of spaces are not returned. |
| | 295 | + * If there is no more qurey string left to process, the empty string is |
| | 296 | + * returned (and in no other case). |
| | 297 | + */ |
| | 298 | + protected function readChunk() { |
| | 299 | + $chunks = preg_split('/[\s]*(\[\[|\]\]|::|:=|<q>|<\/q>|' . $this->m_categoryprefix . '|\|\||\|)[\s]*/', $this->m_curstring, 2, PREG_SPLIT_DELIM_CAPTURE); |
| | 300 | + if (count($chunks) == 1) { // no mathces anymore, strip spaces and finish |
| | 301 | + $this->m_curstring = ''; |
| | 302 | + return trim($chunks[0]); |
| | 303 | + } elseif (count($chunks) == 3) { // this chould generally happen if count is not 1 |
| | 304 | + if ($chunks[0] == '') { // string started with delimiter |
| | 305 | + $this->m_curstring = $chunks[2]; |
| | 306 | + return $chunks[1]; // spaces stripped already |
| | 307 | + } else { |
| | 308 | + $this->m_curstring = $chunks[1] . $chunks[2]; |
| | 309 | + return $chunks[0]; // spaces stripped already |
| | 310 | + } |
| | 311 | + } else { return false; } //should never happen |
| | 312 | + } |
| | 313 | + |
| | 314 | + /** |
| | 315 | + * Enter a new subblock in the query, which must at some time be terminated by the |
| | 316 | + * given $endstring delimiter calling popDelimiter(); |
| | 317 | + */ |
| | 318 | + protected function pushDelimiter($endstring) { |
| | 319 | + array_push($this->m_sepstack, $endstring); |
| | 320 | + } |
| | 321 | + |
| | 322 | + /** |
| | 323 | + * Exit a subblock in the query ending with the given delimiter. |
| | 324 | + * If the delimiter does not match the top-most open block, false |
| | 325 | + * will be returned. Otherwise return true. |
| | 326 | + */ |
| | 327 | + protected function popDelimiter($endstring) { |
| | 328 | + $topdelim = array_pop($this->m_sepstack); |
| | 329 | + return ($topdelim == $endstring); |
| | 330 | + } |
| | 331 | + |
| | 332 | + /** |
| | 333 | + * Extend a given description by a new one, either by adding the new description |
| | 334 | + * (if the old one is a container description) or by creating a new container. |
| | 335 | + * The parameter $conjunction determines whether the combination of both descriptions |
| | 336 | + * should be a disjunction or conjunction. |
| | 337 | + * |
| | 338 | + * In the special case that the current description is NULL, the new one will just |
| | 339 | + * replace the current one. |
| | 340 | + * |
| | 341 | + * The return value is the expected combined description. The object $curdesc will |
| | 342 | + * also be changed (if it was non-NULL). |
| | 343 | + */ |
| | 344 | + protected function addDescription($curdesc, $newdesc, $conjunction = true) { |
| | 345 | + if ($curdesc === NULL) { |
| | 346 | + return $newdesc; |
| | 347 | + } else { // we already found descriptions |
| | 348 | + if ( (($conjunction) && ($curdesc instanceof SMWConjunction)) || |
| | 349 | + ((!$conjunction) && ($curdesc instanceof SMWDisjunction)) ) { // use existing container |
| | 350 | + $curdesc->addDescription($newdesc); |
| | 351 | + } elseif ($conjunction) { // make new conjunction |
| | 352 | + return new SMWConjunction(array($curdesc,$newdesc)); |
| | 353 | + } else { // make new disjunction |
| | 354 | + return new SMWDisjunction(array($curdesc,$newdesc)); |
| | 355 | + } |
| | 356 | + } |
| | 357 | + } |
| | 358 | +} |
| 148 | 359 | |
| 149 | 360 | ?> |
| \ No newline at end of file |