lib/SearchDescription.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/SpecialSearchOperator.php');
   6 require_once(CONST_BasePath.'/lib/SearchContext.php');
   7
   8 /**
   9  * Description of a single interpretation of a search query.
  10  */
  11 class SearchDescription
  12 {
  13     /// Ranking how well the description fits the query.
  14     private $iSearchRank = 0;
  15     /// Country code of country the result must belong to.
  16     private $sCountryCode = '';
  17     /// List of word ids making up the name of the object.
  18     private $aName = array();
  19     /// List of word ids making up the address of the object.
  20     private $aAddress = array();
  21     /// Subset of word ids of full words making up the address.
  22     private $aFullNameAddress = array();
  23     /// List of word ids that appear in the name but should be ignored.
  24     private $aNameNonSearch = array();
  25     /// List of word ids that appear in the address but should be ignored.
  26     private $aAddressNonSearch = array();
  27     /// Kind of search for special searches, see Nominatim::Operator.
  28     private $iOperator = Operator::NONE;
  29     /// Class of special feature to search for.
  30     private $sClass = '';
  31     /// Type of special feature to search for.
  32     private $sType = '';
  33     /// Housenumber of the object.
  34     private $sHouseNumber = '';
  35     /// Postcode for the object.
  36     private $sPostcode = '';
  37     /// Global search constraints.
  38     private $oContext;
  39
  40     // Temporary values used while creating the search description.
  41
  42     /// Index of phrase currently processed.
  43     private $iNamePhrase = -1;
  44
  45
  46     public function __construct($oContext)
  47     {
  48         $this->oContext = $oContext;
  49     }
  50
  51     public function getRank()
  52     {
  53         return $this->iSearchRank;
  54     }
  55
  56     public function addToRank($iAddRank)
  57     {
  58         $this->iSearchRank += $iAddRank;
  59         return $this->iSearchRank;
  60     }
  61
  62     public function getPostCode()
  63     {
  64         return $this->sPostcode;
  65     }
  66
  67     public function setPoiSearch($iOperator, $sClass, $sType)
  68     {
  69         $this->iOperator = $iOperator;
  70         $this->sClass = $sClass;
  71         $this->sType = $sType;
  72     }
  73
  74     public function isNamedSearch()
  75     {
  76         return sizeof($this->aName) > 0 || sizeof($this->aAddress) > 0;
  77     }
  78
  79     public function isCountrySearch()
  80     {
  81         return $this->sCountryCode && sizeof($this->aName) == 0
  82                && !$this->iOperator && !$this->oContext->hasNearPoint();
  83     }
  84
  85     public function isPoiSearch()
  86     {
  87         return (bool) $this->sClass;
  88     }
  89
  90     public function looksLikeFullAddress()
  91     {
  92         return sizeof($this->aName)
  93                && (sizeof($this->aAddress || $this->sCountryCode))
  94                && preg_match('/[0-9]+/', $this->sHouseNumber);
  95     }
  96
  97     public function isOperator($iType)
  98     {
  99         return $this->iOperator == $iType;
 100     }
 101
 102     public function hasHouseNumber()
 103     {
 104         return (bool) $this->sHouseNumber;
 105     }
 106
 107     private function poiTable()
 108     {
 109         return 'place_classtype_'.$this->sClass.'_'.$this->sType;
 110     }
 111
 112     public function countryCodeSQL($sVar, $sCountryList)
 113     {
 114         if ($this->sCountryCode) {
 115             return $sVar.' = \''.$this->sCountryCode."'";
 116         }
 117         if ($sCountryList) {
 118             return $sVar.' in ('.$sCountryList.')';
 119         }
 120
 121         return '';
 122     }
 123
 124     public function hasOperator()
 125     {
 126         return $this->iOperator != Operator::NONE;
 127     }
 128
 129     public function extractKeyValuePairs($sQuery)
 130     {
 131         // Search for terms of kind [<key>=<value>].
 132         preg_match_all(
 133             '/\\[([\\w_]*)=([\\w_]*)\\]/',
 134             $sQuery,
 135             $aSpecialTermsRaw,
 136             PREG_SET_ORDER
 137         );
 138
 139         foreach ($aSpecialTermsRaw as $aTerm) {
 140             $sQuery = str_replace($aTerm[0], ' ', $sQuery);
 141             if (!$this->hasOperator()) {
 142                 $this->setPoiSearch(Operator::TYPE, $aTerm[1], $aTerm[2]);
 143             }
 144         }
 145
 146         return $sQuery;
 147     }
 148
 149     public function isValidSearch(&$aCountryCodes)
 150     {
 151         if (!sizeof($this->aName)) {
 152             if ($this->sHouseNumber) {
 153                 return false;
 154             }
 155         }
 156         if ($aCountryCodes
 157             && $this->sCountryCode
 158             && !in_array($this->sCountryCode, $aCountryCodes)
 159         ) {
 160             return false;
 161         }
 162
 163         return true;
 164     }
 165
 166     /////////// Search building functions
 167
 168
 169     public function extendWithFullTerm($aSearchTerm, $bWordInQuery, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank)
 170     {
 171         $aNewSearches = array();
 172
 173         if (($sPhraseType == '' || $sPhraseType == 'country')
 174             && !empty($aSearchTerm['country_code'])
 175             && $aSearchTerm['country_code'] != '0'
 176         ) {
 177             if (!$this->sCountryCode) {
 178                 $oSearch = clone $this;
 179                 $oSearch->iSearchRank++;
 180                 $oSearch->sCountryCode = $aSearchTerm['country_code'];
 181                 // Country is almost always at the end of the string
 182                 // - increase score for finding it anywhere else (optimisation)
 183                 if (!$bLastToken) {
 184                     $oSearch->iSearchRank += 5;
 185                 }
 186                 $aNewSearches[] = $oSearch;
 187
 188                 // If it is at the beginning, we can be almost sure that
 189                 // the terms are in the wrong order. Increase score for all searches.
 190                 if ($bFirstToken) {
 191                     $iGlobalRank++;
 192                 }
 193             }
 194         } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode')
 195                   && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode'
 196         ) {
 197             // We need to try the case where the postal code is the primary element
 198             // (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode)
 199             // so try both.
 200             if (!$this->sPostcode && $bWordInQuery
 201                 && pg_escape_string($aSearchTerm['word']) == $aSearchTerm['word']
 202             ) {
 203                 // If we have structured search or this is the first term,
 204                 // make the postcode the primary search element.
 205                 if ($this->iOperator == Operator::NONE
 206                     && ($sPhraseType == 'postalcode' || $bFirstToken)
 207                 ) {
 208                     $oSearch = clone $this;
 209                     $oSearch->iSearchRank++;
 210                     $oSearch->iOperator = Operator::POSTCODE;
 211                     $oSearch->aAddress = array_merge($this->aAddress, $this->aName);
 212                     $oSearch->aName =
 213                         array($aSearchTerm['word_id'] => $aSearchTerm['word']);
 214                     $aNewSearches[] = $oSearch;
 215                 }
 216
 217                 // If we have a structured search or this is not the first term,
 218                 // add the postcode as an addendum.
 219                 if ($this->iOperator != Operator::POSTCODE
 220                     && ($sPhraseType == 'postalcode' || sizeof($this->aName))
 221                 ) {
 222                     $oSearch = clone $this;
 223                     $oSearch->iSearchRank++;
 224                     $oSearch->sPostcode = $aSearchTerm['word'];
 225                     $aNewSearches[] = $oSearch;
 226                 }
 227             }
 228         } elseif (($sPhraseType == '' || $sPhraseType == 'street')
 229                  && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house'
 230         ) {
 231             if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
 232                 $oSearch = clone $this;
 233                 $oSearch->iSearchRank++;
 234                 $oSearch->sHouseNumber = trim($aSearchTerm['word_token']);
 235                 // sanity check: if the housenumber is not mainly made
 236                 // up of numbers, add a penalty
 237                 if (preg_match_all("/[^0-9]/", $oSearch->sHouseNumber, $aMatches) > 2) {
 238                     $oSearch->iSearchRank++;
 239                 }
 240                 if (!isset($aSearchTerm['word_id'])) {
 241                     $oSearch->iSearchRank++;
 242                 }
 243                 // also must not appear in the middle of the address
 244                 if (sizeof($this->aAddress) || sizeof($this->aAddressNonSearch)) {
 245                     $oSearch->iSearchRank++;
 246                 }
 247                 $aNewSearches[] = $oSearch;
 248             }
 249         } elseif ($sPhraseType == ''
 250                   && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null
 251         ) {
 252             // require a normalized exact match of the term
 253             // if we have the normalizer version of the query
 254             // available
 255             if ($this->iOperator == Operator::NONE
 256                 && (isset($aSearchTerm['word']) && $aSearchTerm['word'])
 257                 && $bWordInQuery
 258             ) {
 259                 $oSearch = clone $this;
 260                 $oSearch->iSearchRank++;
 261
 262                 $iOp = Operator::NEAR; // near == in for the moment
 263                 if ($aSearchTerm['operator'] == '') {
 264                     if (sizeof($this->aName)) {
 265                         $iOp = Operator::NAME;
 266                     }
 267                     $oSearch->iSearchRank += 2;
 268                 }
 269
 270                 $oSearch->setPoiSearch($iOp, $aSearchTerm['class'], $aSearchTerm['type']);
 271                 $aNewSearches[] = $oSearch;
 272             }
 273         } elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
 274             $iWordID = $aSearchTerm['word_id'];
 275             if (sizeof($this->aName)) {
 276                 if (($sPhraseType == '' || !$bFirstPhrase)
 277                     && $sPhraseType != 'country'
 278                     && !$bHasPartial
 279                 ) {
 280                     $oSearch = clone $this;
 281                     $oSearch->iSearchRank++;
 282                     $oSearch->aAddress[$iWordID] = $iWordID;
 283                     $aNewSearches[] = $oSearch;
 284                 } else {
 285                     $this->aFullNameAddress[$iWordID] = $iWordID;
 286                 }
 287             } else {
 288                 $oSearch = clone $this;
 289                 $oSearch->iSearchRank++;
 290                 $oSearch->aName = array($iWordID => $iWordID);
 291                 $aNewSearches[] = $oSearch;
 292             }
 293         }
 294
 295         return $aNewSearches;
 296     }
 297
 298     public function extendWithPartialTerm($aSearchTerm, $bStructuredPhrases, $iPhrase, &$aWordFrequencyScores, $aFullTokens)
 299     {
 300         // Only allow name terms.
 301         if (!(isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])) {
 302             return array();
 303         }
 304
 305         $aNewSearches = array();
 306         $iWordID = $aSearchTerm['word_id'];
 307
 308         if ((!$bStructuredPhrases || $iPhrase > 0)
 309             && sizeof($this->aName)
 310             && strpos($aSearchTerm['word_token'], ' ') === false
 311         ) {
 312             if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
 313                 $oSearch = clone $this;
 314                 $oSearch->iSearchRank++;
 315                 $oSearch->aAddress[$iWordID] = $iWordID;
 316                 $aNewSearches[] = $oSearch;
 317             } else {
 318                 $oSearch = clone $this;
 319                 $oSearch->iSearchRank++;
 320                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
 321                 if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
 322                     $oSearch->iSearchRank += 2;
 323                 }
 324                 if (sizeof($aFullTokens)) {
 325                     $oSearch->iSearchRank++;
 326                 }
 327                 $aNewSearches[] = $oSearch;
 328
 329                 // revert to the token version?
 330                 foreach ($aFullTokens as $aSearchTermToken) {
 331                     if (empty($aSearchTermToken['country_code'])
 332                         && empty($aSearchTermToken['lat'])
 333                         && empty($aSearchTermToken['class'])
 334                     ) {
 335                         $oSearch = clone $this;
 336                         $oSearch->iSearchRank++;
 337                         $oSearch->aAddress[$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
 338                         $aNewSearches[] = $oSearch;
 339                     }
 340                 }
 341             }
 342         }
 343
 344         if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
 345             && (!sizeof($this->aName) || $this->iNamePhrase == $iPhrase)
 346         ) {
 347             $oSearch = clone $this;
 348             $oSearch->iSearchRank++;
 349             if (!sizeof($this->aName)) {
 350                 $oSearch->iSearchRank += 1;
 351             }
 352             if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
 353                 $oSearch->iSearchRank += 2;
 354             }
 355             if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
 356                 $oSearch->aName[$iWordID] = $iWordID;
 357             } else {
 358                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
 359             }
 360             $oSearch->iNamePhrase = $iPhrase;
 361             $aNewSearches[] = $oSearch;
 362         }
 363
 364         return $aNewSearches;
 365     }
 366
 367     /////////// Query functions
 368
 369
 370     public function queryCountry(&$oDB)
 371     {
 372         $sSQL = 'SELECT place_id FROM placex ';
 373         $sSQL .= "WHERE country_code='".$this->sCountryCode."'";
 374         $sSQL .= ' AND rank_search = 4';
 375         if ($this->oContext->bViewboxBounded) {
 376             $sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)';
 377         }
 378         $sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
 379
 380         if (CONST_Debug) var_dump($sSQL);
 381
 382         return chksql($oDB->getCol($sSQL));
 383     }
 384
 385     public function queryNearbyPoi(&$oDB, $sCountryList, $iLimit)
 386     {
 387         if (!$this->sClass) {
 388             return array();
 389         }
 390
 391         $sPoiTable = $this->poiTable();
 392
 393         $sSQL = 'SELECT count(*) FROM pg_tables WHERE tablename = \''.$sPoiTable."'";
 394         if (chksql($oDB->getOne($sSQL))) {
 395             $sSQL = 'SELECT place_id FROM '.$sPoiTable.' ct';
 396             if ($sCountryList) {
 397                 $sSQL .= ' JOIN placex USING (place_id)';
 398             }
 399             if ($this->oContext->hasNearPoint()) {
 400                 $sSQL .= ' WHERE '.$this->oContext->withinSQL('ct.centroid');
 401             } else if ($this->oContext->bViewboxBounded) {
 402                 $sSQL .= ' WHERE ST_Contains('.$this->oContext->sqlViewboxSmall.', ct.centroid)';
 403             }
 404             if ($sCountryList) {
 405                 $sSQL .= " AND country_code in ($sCountryList)";
 406             }
 407             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 408             if ($this->oContext->sqlViewboxCentre) {
 409                 $sSQL .= ' ORDER BY ST_Distance(';
 410                 $sSQL .= $this->oContext->sqlViewboxCentre.', ct.centroid) ASC';
 411             } elseif ($this->oContext->hasNearPoint()) {
 412                 $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('ct.centroid').' ASC';
 413             }
 414             $sSQL .= " limit $iLimit";
 415             if (CONST_Debug) var_dump($sSQL);
 416             return chksql($oDB->getCol($sSQL));
 417         }
 418
 419         if ($this->oContext->hasNearPoint()) {
 420             $sSQL = 'SELECT place_id FROM placex WHERE ';
 421             $sSQL .= 'class=\''.$this->sClass."' and type='".$this->sType."'";
 422             $sSQL .= ' AND '.$this->oContext->withinSQL('geometry');
 423             $sSQL .= ' AND linked_place_id is null';
 424             if ($sCountryList) {
 425                 $sSQL .= " AND country_code in ($sCountryList)";
 426             }
 427             $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid')." ASC";
 428             $sSQL .= " LIMIT $iLimit";
 429             if (CONST_Debug) var_dump($sSQL);
 430             return chksql($oDB->getCol($sSQL));
 431         }
 432
 433         return array();
 434     }
 435
 436     public function queryPostcode(&$oDB, $sCountryList, $iLimit)
 437     {
 438         $sSQL = 'SELECT p.place_id FROM location_postcode p ';
 439
 440         if (sizeof($this->aAddress)) {
 441             $sSQL .= ', search_name s ';
 442             $sSQL .= 'WHERE s.place_id = p.parent_place_id ';
 443             $sSQL .= 'AND array_cat(s.nameaddress_vector, s.name_vector)';
 444             $sSQL .= '      @> '.getArraySQL($this->aAddress).' AND ';
 445         } else {
 446             $sSQL .= 'WHERE ';
 447         }
 448
 449         $sSQL .= "p.postcode = '".reset($this->aName)."'";
 450         $sCountryTerm = $this->countryCodeSQL('p.country_code', $sCountryList);
 451         if ($sCountryTerm) {
 452             $sSQL .= ' AND '.$sCountryTerm;
 453         }
 454         $sSQL .= $this->oContext->excludeSQL(' AND p.place_id');
 455         $sSQL .= " LIMIT $iLimit";
 456
 457         if (CONST_Debug) var_dump($sSQL);
 458
 459         return chksql($oDB->getCol($sSQL));
 460     }
 461
 462     public function queryNamedPlace(&$oDB, $aWordFrequencyScores, $sCountryList, $iMinAddressRank, $iMaxAddressRank, $iLimit)
 463     {
 464         $aTerms = array();
 465         $aOrder = array();
 466
 467         if ($this->sHouseNumber && sizeof($this->aAddress)) {
 468             $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
 469             $aOrder[] = ' (';
 470             $aOrder[0] .= 'EXISTS(';
 471             $aOrder[0] .= '  SELECT place_id';
 472             $aOrder[0] .= '  FROM placex';
 473             $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
 474             $aOrder[0] .= "    AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
 475             $aOrder[0] .= '  LIMIT 1';
 476             $aOrder[0] .= ') ';
 477             // also housenumbers from interpolation lines table are needed
 478             if (preg_match('/[0-9]+/', $this->sHouseNumber)) {
 479                 $iHouseNumber = intval($this->sHouseNumber);
 480                 $aOrder[0] .= 'OR EXISTS(';
 481                 $aOrder[0] .= '  SELECT place_id ';
 482                 $aOrder[0] .= '  FROM location_property_osmline ';
 483                 $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
 484                 $aOrder[0] .= '    AND startnumber is not NULL';
 485                 $aOrder[0] .= '    AND '.$iHouseNumber.'>=startnumber ';
 486                 $aOrder[0] .= '    AND '.$iHouseNumber.'<=endnumber ';
 487                 $aOrder[0] .= '  LIMIT 1';
 488                 $aOrder[0] .= ')';
 489             }
 490             $aOrder[0] .= ') DESC';
 491         }
 492
 493         if (sizeof($this->aName)) {
 494             $aTerms[] = 'name_vector @> '.getArraySQL($this->aName);
 495         }
 496         if (sizeof($this->aAddress)) {
 497             // For infrequent name terms disable index usage for address
 498             if (CONST_Search_NameOnlySearchFrequencyThreshold
 499                 && sizeof($this->aName) == 1
 500                 && $aWordFrequencyScores[$this->aName[reset($this->aName)]]
 501                      < CONST_Search_NameOnlySearchFrequencyThreshold
 502             ) {
 503                 $aTerms[] = 'array_cat(nameaddress_vector,ARRAY[]::integer[]) @> '.getArraySQL($this->aAddress);
 504             } else {
 505                 $aTerms[] = 'nameaddress_vector @> '.getArraySQL($this->aAddress);
 506             }
 507         }
 508
 509         $sCountryTerm = $this->countryCodeSQL('country_code', $sCountryList);
 510         if ($sCountryTerm) {
 511             $aTerms[] = $sCountryTerm;
 512         }
 513
 514         if ($this->sHouseNumber) {
 515             $aTerms[] = "address_rank between 16 and 27";
 516         } elseif (!$this->sClass || $this->iOperator == Operator::NAME) {
 517             if ($iMinAddressRank > 0) {
 518                 $aTerms[] = "address_rank >= ".$iMinAddressRank;
 519             }
 520             if ($iMaxAddressRank < 30) {
 521                 $aTerms[] = "address_rank <= ".$iMaxAddressRank;
 522             }
 523         }
 524
 525         if ($this->oContext->hasNearPoint()) {
 526             $aTerms[] = $this->oContext->withinSQL('centroid');
 527             $aOrder[] = $this->oContext->distanceSQL('centroid');
 528         } elseif ($this->sPostcode) {
 529             if (!sizeof($this->aAddress)) {
 530                 $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
 531             } else {
 532                 $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."')";
 533             }
 534         }
 535
 536         $sExcludeSQL = $this->oContext->excludeSQL('place_id');
 537         if ($sExcludeSQL) {
 538             $aTerms[] = $sExcludeSQL;
 539         }
 540
 541         if ($this->oContext->bViewboxBounded) {
 542             $aTerms[] = 'centroid && '.$this->oContext->sqlViewboxSmall;
 543         }
 544
 545         if ($this->oContext->hasNearPoint()) {
 546             $aOrder[] = $this->oContext->distanceSQL('centroid');
 547         }
 548
 549         if ($this->sHouseNumber) {
 550             $sImportanceSQL = '- abs(26 - address_rank) + 3';
 551         } else {
 552             $sImportanceSQL = '(CASE WHEN importance = 0 OR importance IS NULL THEN 0.75-(search_rank::float/40) ELSE importance END)';
 553         }
 554         $sImportanceSQL .= $this->oContext->viewboxImportanceSQL('centroid');
 555         $aOrder[] = "$sImportanceSQL DESC";
 556
 557         if (sizeof($this->aFullNameAddress)) {
 558             $sExactMatchSQL = ' ( ';
 559             $sExactMatchSQL .= ' SELECT count(*) FROM ( ';
 560             $sExactMatchSQL .= '  SELECT unnest('.getArraySQL($this->aFullNameAddress).')';
 561             $sExactMatchSQL .= '    INTERSECT ';
 562             $sExactMatchSQL .= '  SELECT unnest(nameaddress_vector)';
 563             $sExactMatchSQL .= ' ) s';
 564             $sExactMatchSQL .= ') as exactmatch';
 565             $aOrder[] = 'exactmatch DESC';
 566         } else {
 567             $sExactMatchSQL = '0::int as exactmatch';
 568         }
 569
 570         if ($this->sHouseNumber || $this->sClass) {
 571             $iLimit = 20;
 572         }
 573
 574         if (sizeof($aTerms)) {
 575             $sSQL = 'SELECT place_id,'.$sExactMatchSQL;
 576             $sSQL .= ' FROM search_name';
 577             $sSQL .= ' WHERE '.join(' and ', $aTerms);
 578             $sSQL .= ' ORDER BY '.join(', ', $aOrder);
 579             $sSQL .= ' LIMIT '.$iLimit;
 580
 581             if (CONST_Debug) var_dump($sSQL);
 582
 583             return chksql(
 584                 $oDB->getAll($sSQL),
 585                 "Could not get places for search terms."
 586             );
 587         }
 588
 589         return array();
 590     }
 591
 592
 593     public function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $iLimit)
 594     {
 595         $sPlaceIDs = join(',', $aRoadPlaceIDs);
 596
 597         $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
 598         $sSQL = 'SELECT place_id FROM placex ';
 599         $sSQL .= 'WHERE parent_place_id in ('.$sPlaceIDs.')';
 600         $sSQL .= "  AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
 601         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 602         $sSQL .= " LIMIT $iLimit";
 603
 604         if (CONST_Debug) var_dump($sSQL);
 605
 606         $aPlaceIDs = chksql($oDB->getCol($sSQL));
 607
 608         if (sizeof($aPlaceIDs)) {
 609             return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => -1);
 610         }
 611
 612         $bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
 613         $iHousenumber = intval($this->sHouseNumber);
 614         if ($bIsIntHouseNumber) {
 615             // if nothing found, search in the interpolation line table
 616             $sSQL = 'SELECT distinct place_id FROM location_property_osmline';
 617             $sSQL .= ' WHERE startnumber is not NULL';
 618             $sSQL .= '  AND parent_place_id in ('.$sPlaceIDs.') AND (';
 619             if ($iHousenumber % 2 == 0) {
 620                 // If housenumber is even, look for housenumber in streets
 621                 // with interpolationtype even or all.
 622                 $sSQL .= "interpolationtype='even'";
 623             } else {
 624                 // Else look for housenumber with interpolationtype odd or all.
 625                 $sSQL .= "interpolationtype='odd'";
 626             }
 627             $sSQL .= " or interpolationtype='all') and ";
 628             $sSQL .= $iHousenumber.">=startnumber and ";
 629             $sSQL .= $iHousenumber."<=endnumber";
 630             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 631             $sSQL .= " limit $iLimit";
 632
 633             if (CONST_Debug) var_dump($sSQL);
 634
 635             $aPlaceIDs = chksql($oDB->getCol($sSQL, 0));
 636
 637             if (sizeof($aPlaceIDs)) {
 638                 return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => $iHousenumber);
 639             }
 640         }
 641
 642         // If nothing found try the aux fallback table
 643         if (CONST_Use_Aux_Location_data) {
 644             $sSQL = 'SELECT place_id FROM location_property_aux';
 645             $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.')';
 646             $sSQL .= " AND housenumber = '".$this->sHouseNumber."'";
 647             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 648             $sSQL .= " limit $iLimit";
 649
 650             if (CONST_Debug) var_dump($sSQL);
 651
 652             $aPlaceIDs = chksql($oDB->getCol($sSQL));
 653
 654             if (sizeof($aPlaceIDs)) {
 655                 return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => -1);
 656             }
 657         }
 658
 659         // If nothing found then search in Tiger data (location_property_tiger)
 660         if (CONST_Use_US_Tiger_Data && $bIsIntHouseNumber) {
 661             $sSQL = 'SELECT distinct place_id FROM location_property_tiger';
 662             $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.') and (';
 663             if ($iHousenumber % 2 == 0) {
 664                 $sSQL .= "interpolationtype='even'";
 665             } else {
 666                 $sSQL .= "interpolationtype='odd'";
 667             }
 668             $sSQL .= " or interpolationtype='all') and ";
 669             $sSQL .= $iHousenumber.">=startnumber and ";
 670             $sSQL .= $iHousenumber."<=endnumber";
 671             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 672             $sSQL .= " limit $iLimit";
 673
 674             if (CONST_Debug) var_dump($sSQL);
 675
 676             $aPlaceIDs = chksql($oDB->getCol($sSQL, 0));
 677
 678             if (sizeof($aPlaceIDs)) {
 679                 return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => $iHousenumber);
 680             }
 681         }
 682
 683         return array();
 684     }
 685
 686
 687     public function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit)
 688     {
 689         $sPlaceIDs = join(',', $aParentIDs);
 690         $aClassPlaceIDs = array();
 691
 692         if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NAME) {
 693             // If they were searching for a named class (i.e. 'Kings Head pub')
 694             // then we might have an extra match
 695             $sSQL = 'SELECT place_id FROM placex ';
 696             $sSQL .= " WHERE place_id in ($sPlaceIDs)";
 697             $sSQL .= "   AND class='".$this->sClass."' ";
 698             $sSQL .= "   AND type='".$this->sType."'";
 699             $sSQL .= "   AND linked_place_id is null";
 700             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 701             $sSQL .= " ORDER BY rank_search ASC ";
 702             $sSQL .= " LIMIT $iLimit";
 703
 704             if (CONST_Debug) var_dump($sSQL);
 705
 706             $aClassPlaceIDs = chksql($oDB->getCol($sSQL));
 707         }
 708
 709         // NEAR and IN are handled the same
 710         if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NEAR) {
 711             $sClassTable = $this->poiTable();
 712             $sSQL = "SELECT count(*) FROM pg_tables WHERE tablename = '$sClassTable'";
 713             $bCacheTable = (bool) chksql($oDB->getOne($sSQL));
 714
 715             $sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
 716             if (CONST_Debug) var_dump($sSQL);
 717             $iMaxRank = (int)chksql($oDB->getOne($sSQL));
 718
 719             // For state / country level searches the normal radius search doesn't work very well
 720             $sPlaceGeom = false;
 721             if ($iMaxRank < 9 && $bCacheTable) {
 722                 // Try and get a polygon to search in instead
 723                 $sSQL = 'SELECT geometry FROM placex';
 724                 $sSQL .= " WHERE place_id in ($sPlaceIDs)";
 725                 $sSQL .= "   AND rank_search < $iMaxRank + 5";
 726                 $sSQL .= "   AND ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon')";
 727                 $sSQL .= " ORDER BY rank_search ASC ";
 728                 $sSQL .= " LIMIT 1";
 729                 if (CONST_Debug) var_dump($sSQL);
 730                 $sPlaceGeom = chksql($oDB->getOne($sSQL));
 731             }
 732
 733             if ($sPlaceGeom) {
 734                 $sPlaceIDs = false;
 735             } else {
 736                 $iMaxRank += 5;
 737                 $sSQL = 'SELECT place_id FROM placex';
 738                 $sSQL .= " WHERE place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
 739                 if (CONST_Debug) var_dump($sSQL);
 740                 $aPlaceIDs = chksql($oDB->getCol($sSQL));
 741                 $sPlaceIDs = join(',', $aPlaceIDs);
 742             }
 743
 744             if ($sPlaceIDs || $sPlaceGeom) {
 745                 $fRange = 0.01;
 746                 if ($bCacheTable) {
 747                     // More efficient - can make the range bigger
 748                     $fRange = 0.05;
 749
 750                     $sOrderBySQL = '';
 751                     if ($this->oContext->hasNearPoint()) {
 752                         $sOrderBySQL = $this->oContext->distanceSQL('l.centroid');
 753                     } elseif ($sPlaceIDs) {
 754                         $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
 755                     } elseif ($sPlaceGeom) {
 756                         $sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
 757                     }
 758
 759                     $sSQL = 'SELECT distinct i.place_id';
 760                     if ($sOrderBySQL) {
 761                         $sSQL .= ', i.order_term';
 762                     }
 763                     $sSQL .= ' from (SELECT l.place_id';
 764                     if ($sOrderBySQL) {
 765                         $sSQL .= ','.$sOrderBySQL.' as order_term';
 766                     }
 767                     $sSQL .= ' from '.$sClassTable.' as l';
 768
 769                     if ($sPlaceIDs) {
 770                         $sSQL .= ",placex as f WHERE ";
 771                         $sSQL .= "f.place_id in ($sPlaceIDs) ";
 772                         $sSQL .= " AND ST_DWithin(l.centroid, f.centroid, $fRange)";
 773                     } elseif ($sPlaceGeom) {
 774                         $sSQL .= " WHERE ST_Contains('$sPlaceGeom', l.centroid)";
 775                     }
 776
 777                     $sSQL .= $this->oContext->excludeSQL(' AND l.place_id');
 778                     $sSQL .= 'limit 300) i ';
 779                     if ($sOrderBySQL) {
 780                         $sSQL .= 'order by order_term asc';
 781                     }
 782                     $sSQL .= " limit $iLimit";
 783
 784                     if (CONST_Debug) var_dump($sSQL);
 785
 786                     $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($oDB->getCol($sSQL)));
 787                 } else {
 788                     if ($this->oContext->hasNearPoint()) {
 789                         $fRange = $this->oContext->nearRadius();
 790                     }
 791
 792                     $sOrderBySQL = '';
 793                     if ($this->oContext->hasNearPoint()) {
 794                         $sOrderBySQL = $this->oContext->distanceSQL('l.geometry');
 795                     } else {
 796                         $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
 797                     }
 798
 799                     $sSQL = 'SELECT distinct l.place_id';
 800                     if ($sOrderBySQL) {
 801                         $sSQL .= ','.$sOrderBySQL.' as orderterm';
 802                     }
 803                     $sSQL .= ' FROM placex as l, placex as f';
 804                     $sSQL .= " WHERE f.place_id in ($sPlaceIDs)";
 805                     $sSQL .= "  AND ST_DWithin(l.geometry, f.centroid, $fRange)";
 806                     $sSQL .= "  AND l.class='".$this->sClass."'";
 807                     $sSQL .= "  AND l.type='".$this->sType."'";
 808                     $sSQL .= $this->oContext->excludeSQL(' AND l.place_id');
 809                     if ($sOrderBySQL) {
 810                         $sSQL .= "ORDER BY orderterm ASC";
 811                     }
 812                     $sSQL .= " limit $iLimit";
 813
 814                     if (CONST_Debug) var_dump($sSQL);
 815
 816                     $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($oDB->getCol($sSQL)));
 817                 }
 818             }
 819         }
 820
 821         return $aClassPlaceIDs;
 822     }
 823
 824
 825     /////////// Sort functions
 826
 827
 828     public static function bySearchRank($a, $b)
 829     {
 830         if ($a->iSearchRank == $b->iSearchRank) {
 831             return $a->iOperator + strlen($a->sHouseNumber)
 832                      - $b->iOperator - strlen($b->sHouseNumber);
 833         }
 834
 835         return $a->iSearchRank < $b->iSearchRank ? -1 : 1;
 836     }
 837
 838     //////////// Debugging functions
 839
 840
 841     public function dumpAsHtmlTableRow(&$aWordIDs)
 842     {
 843         $kf = function ($k) use (&$aWordIDs) {
 844             return $aWordIDs[$k];
 845         };
 846
 847         echo "<tr>";
 848         echo "<td>$this->iSearchRank</td>";
 849         echo "<td>".join(', ', array_map($kf, $this->aName))."</td>";
 850         echo "<td>".join(', ', array_map($kf, $this->aNameNonSearch))."</td>";
 851         echo "<td>".join(', ', array_map($kf, $this->aAddress))."</td>";
 852         echo "<td>".join(', ', array_map($kf, $this->aAddressNonSearch))."</td>";
 853         echo "<td>".$this->sCountryCode."</td>";
 854         echo "<td>".Operator::toString($this->iOperator)."</td>";
 855         echo "<td>".$this->sClass."</td>";
 856         echo "<td>".$this->sType."</td>";
 857         echo "<td>".$this->sPostcode."</td>";
 858         echo "<td>".$this->sHouseNumber."</td>";
 859
 860         echo "</tr>";
 861     }
 862 }