lib-php/SearchDescription.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_LibDir.'/SpecialSearchOperator.php');
   6 require_once(CONST_LibDir.'/SearchContext.php');
   7 require_once(CONST_LibDir.'/Result.php');
   8
   9 /**
  10  * Description of a single interpretation of a search query.
  11  */
  12 class SearchDescription
  13 {
  14     /// Ranking how well the description fits the query.
  15     private $iSearchRank = 0;
  16     /// Country code of country the result must belong to.
  17     private $sCountryCode = '';
  18     /// List of word ids making up the name of the object.
  19     private $aName = array();
  20     /// True if the name is rare enough to force index use on name.
  21     private $bRareName = false;
  22     /// List of word ids making up the address of the object.
  23     private $aAddress = array();
  24     /// List of word ids that appear in the name but should be ignored.
  25     private $aNameNonSearch = array();
  26     /// List of word ids that appear in the address but should be ignored.
  27     private $aAddressNonSearch = array();
  28     /// Kind of search for special searches, see Nominatim::Operator.
  29     private $iOperator = Operator::NONE;
  30     /// Class of special feature to search for.
  31     private $sClass = '';
  32     /// Type of special feature to search for.
  33     private $sType = '';
  34     /// Housenumber of the object.
  35     private $sHouseNumber = '';
  36     /// Postcode for the object.
  37     private $sPostcode = '';
  38     /// Global search constraints.
  39     private $oContext;
  40
  41     // Temporary values used while creating the search description.
  42
  43     /// Index of phrase currently processed.
  44     private $iNamePhrase = -1;
  45
  46     /**
  47      * Create an empty search description.
  48      *
  49      * @param object $oContext Global context to use. Will be inherited by
  50      *                         all derived search objects.
  51      */
  52     public function __construct($oContext)
  53     {
  54         $this->oContext = $oContext;
  55     }
  56
  57     /**
  58      * Get current search rank.
  59      *
  60      * The higher the search rank the lower the likelihood that the
  61      * search is a correct interpretation of the search query.
  62      *
  63      * @return integer Search rank.
  64      */
  65     public function getRank()
  66     {
  67         return $this->iSearchRank;
  68     }
  69
  70     /**
  71      * Make this search a POI search.
  72      *
  73      * In a POI search, objects are not (only) searched by their name
  74      * but also by the primary OSM key/value pair (class and type in Nominatim).
  75      *
  76      * @param integer $iOperator Type of POI search
  77      * @param string  $sClass    Class (or OSM tag key) of POI.
  78      * @param string  $sType     Type (or OSM tag value) of POI.
  79      *
  80      * @return void
  81      */
  82     public function setPoiSearch($iOperator, $sClass, $sType)
  83     {
  84         $this->iOperator = $iOperator;
  85         $this->sClass = $sClass;
  86         $this->sType = $sType;
  87     }
  88
  89     /**
  90      * Check if any operator is set.
  91      *
  92      * @return bool True, if this is a special search operation.
  93      */
  94     public function hasOperator()
  95     {
  96         return $this->iOperator != Operator::NONE;
  97     }
  98
  99     /**
 100      * Extract key/value pairs from a query.
 101      *
 102      * Key/value pairs are recognised if they are of the form [<key>=<value>].
 103      * If multiple terms of this kind are found then all terms are removed
 104      * but only the first is used for search.
 105      *
 106      * @param string $sQuery Original query string.
 107      *
 108      * @return string The query string with the special search patterns removed.
 109      */
 110     public function extractKeyValuePairs($sQuery)
 111     {
 112         // Search for terms of kind [<key>=<value>].
 113         preg_match_all(
 114             '/\\[([\\w_]*)=([\\w_]*)\\]/',
 115             $sQuery,
 116             $aSpecialTermsRaw,
 117             PREG_SET_ORDER
 118         );
 119
 120         foreach ($aSpecialTermsRaw as $aTerm) {
 121             $sQuery = str_replace($aTerm[0], ' ', $sQuery);
 122             if (!$this->hasOperator()) {
 123                 $this->setPoiSearch(Operator::TYPE, $aTerm[1], $aTerm[2]);
 124             }
 125         }
 126
 127         return $sQuery;
 128     }
 129
 130     /**
 131      * Check if the combination of parameters is sensible.
 132      *
 133      * @return bool True, if the search looks valid.
 134      */
 135     public function isValidSearch()
 136     {
 137         if (empty($this->aName)) {
 138             if ($this->sHouseNumber) {
 139                 return false;
 140             }
 141             if (!$this->sClass && !$this->sCountryCode) {
 142                 return false;
 143             }
 144         }
 145
 146         return true;
 147     }
 148
 149     /////////// Search building functions
 150
 151
 152     /**
 153      * Derive new searches by adding a full term to the existing search.
 154      *
 155      * @param object  $oSearchTerm  Description of the token.
 156      * @param object  $oPosition    Description of the token position within
 157                                     the query.
 158      *
 159      * @return SearchDescription[] List of derived search descriptions.
 160      */
 161     public function extendWithSearchTerm($oSearchTerm, $oPosition)
 162     {
 163         $aNewSearches = array();
 164
 165         if ($oPosition->maybePhrase('country')
 166             && is_a($oSearchTerm, '\Nominatim\Token\Country')
 167         ) {
 168             if (!$this->sCountryCode) {
 169                 $oSearch = clone $this;
 170                 $oSearch->iSearchRank++;
 171                 $oSearch->sCountryCode = $oSearchTerm->sCountryCode;
 172                 // Country is almost always at the end of the string
 173                 // - increase score for finding it anywhere else (optimisation)
 174                 if (!$oPosition->isLastToken()) {
 175                     $oSearch->iSearchRank += 5;
 176                     $oSearch->iNamePhrase = -1;
 177                 }
 178                 $aNewSearches[] = $oSearch;
 179             }
 180         } elseif ($oPosition->maybePhrase('postalcode')
 181                   && is_a($oSearchTerm, '\Nominatim\Token\Postcode')
 182         ) {
 183             if (!$this->sPostcode) {
 184                 // If we have structured search or this is the first term,
 185                 // make the postcode the primary search element.
 186                 if ($this->iOperator == Operator::NONE && $oPosition->isFirstToken()) {
 187                     $oSearch = clone $this;
 188                     $oSearch->iSearchRank++;
 189                     $oSearch->iOperator = Operator::POSTCODE;
 190                     $oSearch->aAddress = array_merge($this->aAddress, $this->aName);
 191                     $oSearch->aName =
 192                         array($oSearchTerm->iId => $oSearchTerm->sPostcode);
 193                     $aNewSearches[] = $oSearch;
 194                 }
 195
 196                 // If we have a structured search or this is not the first term,
 197                 // add the postcode as an addendum.
 198                 if ($this->iOperator != Operator::POSTCODE
 199                     && ($oPosition->isPhrase('postalcode') || !empty($this->aName))
 200                 ) {
 201                     $oSearch = clone $this;
 202                     $oSearch->iSearchRank++;
 203                     $oSearch->iNamePhrase = -1;
 204                     if (strlen($oSearchTerm->sPostcode) < 4) {
 205                         $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
 206                     }
 207                     $oSearch->sPostcode = $oSearchTerm->sPostcode;
 208                     $aNewSearches[] = $oSearch;
 209                 }
 210             }
 211         } elseif ($oPosition->maybePhrase('street')
 212                  && is_a($oSearchTerm, '\Nominatim\Token\HouseNumber')
 213         ) {
 214             if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
 215                 // sanity check: if the housenumber is not mainly made
 216                 // up of numbers, add a penalty
 217                 $iSearchCost = 1;
 218                 if (preg_match('/\\d/', $oSearchTerm->sToken) === 0
 219                     || preg_match_all('/[^0-9]/', $oSearchTerm->sToken, $aMatches) > 2) {
 220                     $iSearchCost++;
 221                 }
 222                 if ($this->iOperator != Operator::NONE) {
 223                     $iSearchCost++;
 224                 }
 225                 if (empty($oSearchTerm->iId)) {
 226                     $iSearchCost++;
 227                 }
 228                 // also must not appear in the middle of the address
 229                 if (!empty($this->aAddress)
 230                     || (!empty($this->aAddressNonSearch))
 231                     || $this->sPostcode
 232                 ) {
 233                     $iSearchCost++;
 234                 }
 235
 236                 $oSearch = clone $this;
 237                 $oSearch->iSearchRank += $iSearchCost;
 238                 $oSearch->iNamePhrase = -1;
 239                 $oSearch->sHouseNumber = $oSearchTerm->sToken;
 240                 $aNewSearches[] = $oSearch;
 241
 242                 // Housenumbers may appear in the name when the place has its own
 243                 // address terms.
 244                 if ($oSearchTerm->iId !== null
 245                     && ($this->iNamePhrase >= 0 || empty($this->aName))
 246                     && empty($this->aAddress)
 247                    ) {
 248                     $oSearch = clone $this;
 249                     $oSearch->iSearchRank += $iSearchCost;
 250                     $oSearch->aAddress = $this->aName;
 251                     $oSearch->bRareName = false;
 252                     $oSearch->aName = array($oSearchTerm->iId => $oSearchTerm->iId);
 253                     $aNewSearches[] = $oSearch;
 254                 }
 255             }
 256         } elseif ($oPosition->isPhrase('')
 257                   && is_a($oSearchTerm, '\Nominatim\Token\SpecialTerm')
 258         ) {
 259             if ($this->iOperator == Operator::NONE) {
 260                 $oSearch = clone $this;
 261                 $oSearch->iSearchRank += 2;
 262                 $oSearch->iNamePhrase = -1;
 263
 264                 $iOp = $oSearchTerm->iOperator;
 265                 if ($iOp == Operator::NONE) {
 266                     if (!empty($this->aName) || $this->oContext->isBoundedSearch()) {
 267                         $iOp = Operator::NAME;
 268                     } else {
 269                         $iOp = Operator::NEAR;
 270                     }
 271                     $oSearch->iSearchRank += 2;
 272                 } elseif (!$oPosition->isFirstToken() && !$oPosition->isLastToken()) {
 273                     $oSearch->iSearchRank += 2;
 274                 }
 275                 if ($this->sHouseNumber) {
 276                     $oSearch->iSearchRank++;
 277                 }
 278
 279                 $oSearch->setPoiSearch(
 280                     $iOp,
 281                     $oSearchTerm->sClass,
 282                     $oSearchTerm->sType
 283                 );
 284                 $aNewSearches[] = $oSearch;
 285             }
 286         } elseif (!$oPosition->isPhrase('country')
 287                   && is_a($oSearchTerm, '\Nominatim\Token\Word')
 288         ) {
 289             $iWordID = $oSearchTerm->iId;
 290             // Full words can only be a name if they appear at the beginning
 291             // of the phrase. In structured search the name must forcably in
 292             // the first phrase. In unstructured search it may be in a later
 293             // phrase when the first phrase is a house number.
 294             if (!empty($this->aName) || !($oPosition->isFirstPhrase() || $oPosition->isPhrase(''))) {
 295                 if (($oPosition->isPhrase('') || !$oPosition->isFirstPhrase())
 296                     && $oSearchTerm->iTermCount > 1
 297                 ) {
 298                     $oSearch = clone $this;
 299                     $oSearch->iNamePhrase = -1;
 300                     $oSearch->iSearchRank += 1;
 301                     $oSearch->aAddress[$iWordID] = $iWordID;
 302                     $aNewSearches[] = $oSearch;
 303                 }
 304             } elseif (empty($this->aNameNonSearch)) {
 305                 $oSearch = clone $this;
 306                 $oSearch->iSearchRank++;
 307                 $oSearch->aName = array($iWordID => $iWordID);
 308                 if (CONST_Search_NameOnlySearchFrequencyThreshold) {
 309                     $oSearch->bRareName =
 310                         $oSearchTerm->iSearchNameCount
 311                           < CONST_Search_NameOnlySearchFrequencyThreshold;
 312                 }
 313                 $aNewSearches[] = $oSearch;
 314             }
 315         } elseif (!$oPosition->isPhrase('country')
 316                   && is_a($oSearchTerm, '\Nominatim\Token\Partial')
 317         ) {
 318             $aNewSearches = $this->extendWithPartialTerm(
 319                 $oSearchTerm,
 320                 $oPosition
 321             );
 322         }
 323
 324         return $aNewSearches;
 325     }
 326
 327     /**
 328      * Derive new searches by adding a partial term to the existing search.
 329      *
 330      * @param object  $oSearchTerm  Description of the token.
 331      * @param object  $oPosition    Description of the token position within
 332                                     the query.
 333      *
 334      * @return SearchDescription[] List of derived search descriptions.
 335      */
 336     private function extendWithPartialTerm($oSearchTerm, $oPosition)
 337     {
 338         $aNewSearches = array();
 339         $iWordID = $oSearchTerm->iId;
 340
 341         if (($oPosition->isPhrase('') || !$oPosition->isFirstPhrase())
 342             && (!empty($this->aName))
 343         ) {
 344             $oSearch = clone $this;
 345             $oSearch->iSearchRank++;
 346             if (preg_match('#^[0-9 ]+$#', $oSearchTerm->sToken)) {
 347                 $oSearch->iSearchRank++;
 348             }
 349             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
 350                 $oSearch->aAddress[$iWordID] = $iWordID;
 351             } else {
 352                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
 353             }
 354             $aNewSearches[] = $oSearch;
 355         }
 356
 357         if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
 358             && ((empty($this->aName) && empty($this->aNameNonSearch))
 359                 || $this->iNamePhrase == $oPosition->getPhrase())
 360         ) {
 361             $oSearch = clone $this;
 362             $oSearch->iSearchRank++;
 363             if (empty($this->aName) && empty($this->aNameNonSearch)) {
 364                 $oSearch->iSearchRank++;
 365             }
 366             if (preg_match('#^[0-9 ]+$#', $oSearchTerm->sToken)) {
 367                 $oSearch->iSearchRank++;
 368             }
 369             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
 370                 if (empty($this->aName)
 371                     && CONST_Search_NameOnlySearchFrequencyThreshold
 372                 ) {
 373                     $oSearch->bRareName =
 374                         $oSearchTerm->iSearchNameCount
 375                           < CONST_Search_NameOnlySearchFrequencyThreshold;
 376                 } else {
 377                     $oSearch->bRareName = false;
 378                 }
 379                 $oSearch->aName[$iWordID] = $iWordID;
 380             } else {
 381                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
 382             }
 383             $oSearch->iNamePhrase = $oPosition->getPhrase();
 384             $aNewSearches[] = $oSearch;
 385         }
 386
 387         return $aNewSearches;
 388     }
 389
 390     /////////// Query functions
 391
 392
 393     /**
 394      * Query database for places that match this search.
 395      *
 396      * @param object  $oDB      Nominatim::DB instance to use.
 397      * @param integer $iMinRank Minimum address rank to restrict search to.
 398      * @param integer $iMaxRank Maximum address rank to restrict search to.
 399      * @param integer $iLimit   Maximum number of results.
 400      *
 401      * @return mixed[] An array with two fields: IDs contains the list of
 402      *                 matching place IDs and houseNumber the houseNumber
 403      *                 if appicable or -1 if not.
 404      */
 405     public function query(&$oDB, $iMinRank, $iMaxRank, $iLimit)
 406     {
 407         $aResults = array();
 408
 409         if ($this->sCountryCode
 410             && empty($this->aName)
 411             && !$this->iOperator
 412             && !$this->sClass
 413             && !$this->oContext->hasNearPoint()
 414         ) {
 415             // Just looking for a country - look it up
 416             if (4 >= $iMinRank && 4 <= $iMaxRank) {
 417                 $aResults = $this->queryCountry($oDB);
 418             }
 419         } elseif (empty($this->aName) && empty($this->aAddress)) {
 420             // Neither name nor address? Then we must be
 421             // looking for a POI in a geographic area.
 422             if ($this->oContext->isBoundedSearch()) {
 423                 $aResults = $this->queryNearbyPoi($oDB, $iLimit);
 424             }
 425         } elseif ($this->iOperator == Operator::POSTCODE) {
 426             // looking for postcode
 427             $aResults = $this->queryPostcode($oDB, $iLimit);
 428         } else {
 429             // Ordinary search:
 430             // First search for places according to name and address.
 431             $aResults = $this->queryNamedPlace(
 432                 $oDB,
 433                 $iMinRank,
 434                 $iMaxRank,
 435                 $iLimit
 436             );
 437
 438             // Now search for housenumber, if housenumber provided. Can be zero.
 439             if (($this->sHouseNumber || $this->sHouseNumber === '0') && !empty($aResults)) {
 440                 $aHnResults = $this->queryHouseNumber($oDB, $aResults);
 441
 442                 // Downgrade the rank of the street results, they are missing
 443                 // the housenumber. Also drop POI places (rank 30) here, they
 444                 // cannot be a parent place and therefore must not be shown
 445                 // as a result for a search with a missing housenumber.
 446                 foreach ($aResults as $oRes) {
 447                     if ($oRes->iAddressRank < 28) {
 448                         if ($oRes->iAddressRank >= 26) {
 449                             $oRes->iResultRank++;
 450                         } else {
 451                             $oRes->iResultRank += 2;
 452                         }
 453                         $aHnResults[$oRes->iId] = $oRes;
 454                     }
 455                 }
 456
 457                 $aResults = $aHnResults;
 458             }
 459
 460             // finally get POIs if requested
 461             if ($this->sClass && !empty($aResults)) {
 462                 $aResults = $this->queryPoiByOperator($oDB, $aResults, $iLimit);
 463             }
 464         }
 465
 466         Debug::printDebugTable('Place IDs', $aResults);
 467
 468         if (!empty($aResults) && $this->sPostcode) {
 469             $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 470             if ($sPlaceIds) {
 471                 $sSQL = 'SELECT place_id FROM placex';
 472                 $sSQL .= ' WHERE place_id in ('.$sPlaceIds.')';
 473                 $sSQL .= " AND postcode != '".$this->sPostcode."'";
 474                 Debug::printSQL($sSQL);
 475                 $aFilteredPlaceIDs = $oDB->getCol($sSQL);
 476                 if ($aFilteredPlaceIDs) {
 477                     foreach ($aFilteredPlaceIDs as $iPlaceId) {
 478                         $aResults[$iPlaceId]->iResultRank++;
 479                     }
 480                 }
 481             }
 482         }
 483
 484         return $aResults;
 485     }
 486
 487
 488     private function queryCountry(&$oDB)
 489     {
 490         $sSQL = 'SELECT place_id FROM placex ';
 491         $sSQL .= "WHERE country_code='".$this->sCountryCode."'";
 492         $sSQL .= ' AND rank_search = 4';
 493         if ($this->oContext->bViewboxBounded) {
 494             $sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)';
 495         }
 496         $sSQL .= ' ORDER BY st_area(geometry) DESC LIMIT 1';
 497
 498         Debug::printSQL($sSQL);
 499
 500         $iPlaceId = $oDB->getOne($sSQL);
 501
 502         $aResults = array();
 503         if ($iPlaceId) {
 504             $aResults[$iPlaceId] = new Result($iPlaceId);
 505         }
 506
 507         return $aResults;
 508     }
 509
 510     private function queryNearbyPoi(&$oDB, $iLimit)
 511     {
 512         if (!$this->sClass) {
 513             return array();
 514         }
 515
 516         $aDBResults = array();
 517         $sPoiTable = $this->poiTable();
 518
 519         if ($oDB->tableExists($sPoiTable)) {
 520             $sSQL = 'SELECT place_id FROM '.$sPoiTable.' ct';
 521             if ($this->oContext->sqlCountryList) {
 522                 $sSQL .= ' JOIN placex USING (place_id)';
 523             }
 524             if ($this->oContext->hasNearPoint()) {
 525                 $sSQL .= ' WHERE '.$this->oContext->withinSQL('ct.centroid');
 526             } elseif ($this->oContext->bViewboxBounded) {
 527                 $sSQL .= ' WHERE ST_Contains('.$this->oContext->sqlViewboxSmall.', ct.centroid)';
 528             }
 529             if ($this->oContext->sqlCountryList) {
 530                 $sSQL .= ' AND country_code in '.$this->oContext->sqlCountryList;
 531             }
 532             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 533             if ($this->oContext->sqlViewboxCentre) {
 534                 $sSQL .= ' ORDER BY ST_Distance(';
 535                 $sSQL .= $this->oContext->sqlViewboxCentre.', ct.centroid) ASC';
 536             } elseif ($this->oContext->hasNearPoint()) {
 537                 $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('ct.centroid').' ASC';
 538             }
 539             $sSQL .= " LIMIT $iLimit";
 540             Debug::printSQL($sSQL);
 541             $aDBResults = $oDB->getCol($sSQL);
 542         }
 543
 544         if ($this->oContext->hasNearPoint()) {
 545             $sSQL = 'SELECT place_id FROM placex WHERE ';
 546             $sSQL .= 'class = :class and type = :type';
 547             $sSQL .= ' AND '.$this->oContext->withinSQL('geometry');
 548             $sSQL .= ' AND linked_place_id is null';
 549             if ($this->oContext->sqlCountryList) {
 550                 $sSQL .= ' AND country_code in '.$this->oContext->sqlCountryList;
 551             }
 552             $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid').' ASC';
 553             $sSQL .= " LIMIT $iLimit";
 554             Debug::printSQL($sSQL);
 555             $aDBResults = $oDB->getCol(
 556                 $sSQL,
 557                 array(':class' => $this->sClass, ':type' => $this->sType)
 558             );
 559         }
 560
 561         $aResults = array();
 562         foreach ($aDBResults as $iPlaceId) {
 563             $aResults[$iPlaceId] = new Result($iPlaceId);
 564         }
 565
 566         return $aResults;
 567     }
 568
 569     private function queryPostcode(&$oDB, $iLimit)
 570     {
 571         $sSQL = 'SELECT p.place_id FROM location_postcode p ';
 572
 573         if (!empty($this->aAddress)) {
 574             $sSQL .= ', search_name s ';
 575             $sSQL .= 'WHERE s.place_id = p.parent_place_id ';
 576             $sSQL .= 'AND array_cat(s.nameaddress_vector, s.name_vector)';
 577             $sSQL .= '      @> '.$oDB->getArraySQL($this->aAddress).' AND ';
 578         } else {
 579             $sSQL .= 'WHERE ';
 580         }
 581
 582         $sSQL .= "p.postcode = '".reset($this->aName)."'";
 583         $sSQL .= $this->countryCodeSQL(' AND p.country_code');
 584         if ($this->oContext->bViewboxBounded) {
 585             $sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)';
 586         }
 587         $sSQL .= $this->oContext->excludeSQL(' AND p.place_id');
 588         $sSQL .= " LIMIT $iLimit";
 589
 590         Debug::printSQL($sSQL);
 591
 592         $aResults = array();
 593         foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 594             $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_POSTCODE);
 595         }
 596
 597         return $aResults;
 598     }
 599
 600     private function queryNamedPlace(&$oDB, $iMinAddressRank, $iMaxAddressRank, $iLimit)
 601     {
 602         $aTerms = array();
 603         $aOrder = array();
 604
 605         // Sort by existence of the requested house number but only if not
 606         // too many results are expected for the street, i.e. if the result
 607         // will be narrowed down by an address. Remeber that with ordering
 608         // every single result has to be checked.
 609         if ($this->sHouseNumber && ($this->bRareName || !empty($this->aAddress) || $this->sPostcode)) {
 610             $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
 611             $aOrder[] = ' (';
 612             $aOrder[0] .= 'EXISTS(';
 613             $aOrder[0] .= '  SELECT place_id';
 614             $aOrder[0] .= '  FROM placex';
 615             $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
 616             $aOrder[0] .= "    AND housenumber ~* E'".$sHouseNumberRegex."'";
 617             $aOrder[0] .= '  LIMIT 1';
 618             $aOrder[0] .= ') ';
 619             // also housenumbers from interpolation lines table are needed
 620             if (preg_match('/[0-9]+/', $this->sHouseNumber)) {
 621                 $iHouseNumber = intval($this->sHouseNumber);
 622                 $aOrder[0] .= 'OR EXISTS(';
 623                 $aOrder[0] .= '  SELECT place_id ';
 624                 $aOrder[0] .= '  FROM location_property_osmline ';
 625                 $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
 626                 $aOrder[0] .= '    AND startnumber is not NULL';
 627                 $aOrder[0] .= '    AND '.$iHouseNumber.'>=startnumber ';
 628                 $aOrder[0] .= '    AND '.$iHouseNumber.'<=endnumber ';
 629                 $aOrder[0] .= '  LIMIT 1';
 630                 $aOrder[0] .= ')';
 631             }
 632             $aOrder[0] .= ') DESC';
 633         }
 634
 635         if (!empty($this->aName)) {
 636             $aTerms[] = 'name_vector @> '.$oDB->getArraySQL($this->aName);
 637         }
 638         if (!empty($this->aAddress)) {
 639             // For infrequent name terms disable index usage for address
 640             if ($this->bRareName) {
 641                 $aTerms[] = 'array_cat(nameaddress_vector,ARRAY[]::integer[]) @> '.$oDB->getArraySQL($this->aAddress);
 642             } else {
 643                 $aTerms[] = 'nameaddress_vector @> '.$oDB->getArraySQL($this->aAddress);
 644             }
 645         }
 646
 647         $sCountryTerm = $this->countryCodeSQL('country_code');
 648         if ($sCountryTerm) {
 649             $aTerms[] = $sCountryTerm;
 650         }
 651
 652         if ($this->sHouseNumber) {
 653             $aTerms[] = 'address_rank between 16 and 30';
 654         } elseif (!$this->sClass || $this->iOperator == Operator::NAME) {
 655             if ($iMinAddressRank > 0) {
 656                 $aTerms[] = "((address_rank between $iMinAddressRank and $iMaxAddressRank) or (search_rank between $iMinAddressRank and $iMaxAddressRank))";
 657             }
 658         }
 659
 660         if ($this->oContext->hasNearPoint()) {
 661             $aTerms[] = $this->oContext->withinSQL('centroid');
 662             $aOrder[] = $this->oContext->distanceSQL('centroid');
 663         } elseif ($this->sPostcode) {
 664             if (empty($this->aAddress)) {
 665                 $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
 666             } else {
 667                 $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."')";
 668             }
 669         }
 670
 671         $sExcludeSQL = $this->oContext->excludeSQL('place_id');
 672         if ($sExcludeSQL) {
 673             $aTerms[] = $sExcludeSQL;
 674         }
 675
 676         if ($this->oContext->bViewboxBounded) {
 677             $aTerms[] = 'centroid && '.$this->oContext->sqlViewboxSmall;
 678         }
 679
 680         if ($this->oContext->hasNearPoint()) {
 681             $aOrder[] = $this->oContext->distanceSQL('centroid');
 682         }
 683
 684         if ($this->sHouseNumber) {
 685             $sImportanceSQL = '- abs(26 - address_rank) + 3';
 686         } else {
 687             $sImportanceSQL = '(CASE WHEN importance = 0 OR importance IS NULL THEN 0.75001-(search_rank::float/40) ELSE importance END)';
 688         }
 689         $sImportanceSQL .= $this->oContext->viewboxImportanceSQL('centroid');
 690         $aOrder[] = "$sImportanceSQL DESC";
 691
 692         $aFullNameAddress = $this->oContext->getFullNameTerms();
 693         if (!empty($aFullNameAddress)) {
 694             $sExactMatchSQL = ' ( ';
 695             $sExactMatchSQL .= ' SELECT count(*) FROM ( ';
 696             $sExactMatchSQL .= '  SELECT unnest('.$oDB->getArraySQL($aFullNameAddress).')';
 697             $sExactMatchSQL .= '    INTERSECT ';
 698             $sExactMatchSQL .= '  SELECT unnest(nameaddress_vector)';
 699             $sExactMatchSQL .= ' ) s';
 700             $sExactMatchSQL .= ') as exactmatch';
 701             $aOrder[] = 'exactmatch DESC';
 702         } else {
 703             $sExactMatchSQL = '0::int as exactmatch';
 704         }
 705
 706         if ($this->sHouseNumber || $this->sClass) {
 707             $iLimit = 40;
 708         }
 709
 710         $aResults = array();
 711
 712         if (!empty($aTerms)) {
 713             $sSQL = 'SELECT place_id, address_rank,'.$sExactMatchSQL;
 714             $sSQL .= ' FROM search_name';
 715             $sSQL .= ' WHERE '.join(' and ', $aTerms);
 716             $sSQL .= ' ORDER BY '.join(', ', $aOrder);
 717             $sSQL .= ' LIMIT '.$iLimit;
 718
 719             Debug::printSQL($sSQL);
 720
 721             $aDBResults = $oDB->getAll($sSQL, null, 'Could not get places for search terms.');
 722
 723             foreach ($aDBResults as $aResult) {
 724                 $oResult = new Result($aResult['place_id']);
 725                 $oResult->iExactMatches = $aResult['exactmatch'];
 726                 $oResult->iAddressRank = $aResult['address_rank'];
 727                 $aResults[$aResult['place_id']] = $oResult;
 728             }
 729         }
 730
 731         return $aResults;
 732     }
 733
 734     private function queryHouseNumber(&$oDB, $aRoadPlaceIDs)
 735     {
 736         $aResults = array();
 737         $sRoadPlaceIDs = Result::joinIdsByTableMaxRank(
 738             $aRoadPlaceIDs,
 739             Result::TABLE_PLACEX,
 740             27
 741         );
 742         $sPOIPlaceIDs = Result::joinIdsByTableMinRank(
 743             $aRoadPlaceIDs,
 744             Result::TABLE_PLACEX,
 745             30
 746         );
 747
 748         $aIDCondition = array();
 749         if ($sRoadPlaceIDs) {
 750             $aIDCondition[] = 'parent_place_id in ('.$sRoadPlaceIDs.')';
 751         }
 752         if ($sPOIPlaceIDs) {
 753             $aIDCondition[] = 'place_id in ('.$sPOIPlaceIDs.')';
 754         }
 755
 756         if (empty($aIDCondition)) {
 757             return $aResults;
 758         }
 759
 760         $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
 761         $sSQL = 'SELECT place_id FROM placex WHERE';
 762         $sSQL .= "  housenumber ~* E'".$sHouseNumberRegex."'";
 763         $sSQL .= ' AND ('.join(' OR ', $aIDCondition).')';
 764         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 765
 766         Debug::printSQL($sSQL);
 767
 768         // XXX should inherit the exactMatches from its parent
 769         foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 770             $aResults[$iPlaceId] = new Result($iPlaceId);
 771         }
 772
 773         $bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
 774         $iHousenumber = intval($this->sHouseNumber);
 775         if ($bIsIntHouseNumber && $sRoadPlaceIDs && empty($aResults)) {
 776             // if nothing found, search in the interpolation line table
 777             $sSQL = 'SELECT distinct place_id FROM location_property_osmline';
 778             $sSQL .= ' WHERE startnumber is not NULL';
 779             $sSQL .= '  AND parent_place_id in ('.$sRoadPlaceIDs.') AND (';
 780             if ($iHousenumber % 2 == 0) {
 781                 // If housenumber is even, look for housenumber in streets
 782                 // with interpolationtype even or all.
 783                 $sSQL .= "interpolationtype='even'";
 784             } else {
 785                 // Else look for housenumber with interpolationtype odd or all.
 786                 $sSQL .= "interpolationtype='odd'";
 787             }
 788             $sSQL .= " or interpolationtype='all') and ";
 789             $sSQL .= $iHousenumber.'>=startnumber and ';
 790             $sSQL .= $iHousenumber.'<=endnumber';
 791             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 792
 793             Debug::printSQL($sSQL);
 794
 795             foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 796                 $oResult = new Result($iPlaceId, Result::TABLE_OSMLINE);
 797                 $oResult->iHouseNumber = $iHousenumber;
 798                 $aResults[$iPlaceId] = $oResult;
 799             }
 800         }
 801
 802         // If nothing found then search in Tiger data (location_property_tiger)
 803         if (CONST_Use_US_Tiger_Data && $sRoadPlaceIDs && $bIsIntHouseNumber && empty($aResults)) {
 804             $sSQL = 'SELECT place_id FROM location_property_tiger';
 805             $sSQL .= ' WHERE parent_place_id in ('.$sRoadPlaceIDs.') and (';
 806             if ($iHousenumber % 2 == 0) {
 807                 $sSQL .= "interpolationtype='even'";
 808             } else {
 809                 $sSQL .= "interpolationtype='odd'";
 810             }
 811             $sSQL .= " or interpolationtype='all') and ";
 812             $sSQL .= $iHousenumber.'>=startnumber and ';
 813             $sSQL .= $iHousenumber.'<=endnumber';
 814             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 815
 816             Debug::printSQL($sSQL);
 817
 818             foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 819                 $oResult = new Result($iPlaceId, Result::TABLE_TIGER);
 820                 $oResult->iHouseNumber = $iHousenumber;
 821                 $aResults[$iPlaceId] = $oResult;
 822             }
 823         }
 824
 825         return $aResults;
 826     }
 827
 828
 829     private function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit)
 830     {
 831         $aResults = array();
 832         $sPlaceIDs = Result::joinIdsByTable($aParentIDs, Result::TABLE_PLACEX);
 833
 834         if (!$sPlaceIDs) {
 835             return $aResults;
 836         }
 837
 838         if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NAME) {
 839             // If they were searching for a named class (i.e. 'Kings Head pub')
 840             // then we might have an extra match
 841             $sSQL = 'SELECT place_id FROM placex ';
 842             $sSQL .= " WHERE place_id in ($sPlaceIDs)";
 843             $sSQL .= "   AND class='".$this->sClass."' ";
 844             $sSQL .= "   AND type='".$this->sType."'";
 845             $sSQL .= '   AND linked_place_id is null';
 846             $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 847             $sSQL .= ' ORDER BY rank_search ASC ';
 848             $sSQL .= " LIMIT $iLimit";
 849
 850             Debug::printSQL($sSQL);
 851
 852             foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 853                 $aResults[$iPlaceId] = new Result($iPlaceId);
 854             }
 855         }
 856
 857         // NEAR and IN are handled the same
 858         if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NEAR) {
 859             $sClassTable = $this->poiTable();
 860             $bCacheTable = $oDB->tableExists($sClassTable);
 861
 862             $sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
 863             Debug::printSQL($sSQL);
 864             $iMaxRank = (int) $oDB->getOne($sSQL);
 865
 866             // For state / country level searches the normal radius search doesn't work very well
 867             $sPlaceGeom = false;
 868             if ($iMaxRank < 9 && $bCacheTable) {
 869                 // Try and get a polygon to search in instead
 870                 $sSQL = 'SELECT geometry FROM placex';
 871                 $sSQL .= " WHERE place_id in ($sPlaceIDs)";
 872                 $sSQL .= "   AND rank_search < $iMaxRank + 5";
 873                 $sSQL .= "   AND ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon')";
 874                 $sSQL .= ' ORDER BY rank_search ASC ';
 875                 $sSQL .= ' LIMIT 1';
 876                 Debug::printSQL($sSQL);
 877                 $sPlaceGeom = $oDB->getOne($sSQL);
 878             }
 879
 880             if ($sPlaceGeom) {
 881                 $sPlaceIDs = false;
 882             } else {
 883                 $iMaxRank += 5;
 884                 $sSQL = 'SELECT place_id FROM placex';
 885                 $sSQL .= " WHERE place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
 886                 Debug::printSQL($sSQL);
 887                 $aPlaceIDs = $oDB->getCol($sSQL);
 888                 $sPlaceIDs = join(',', $aPlaceIDs);
 889             }
 890
 891             if ($sPlaceIDs || $sPlaceGeom) {
 892                 $fRange = 0.01;
 893                 if ($bCacheTable) {
 894                     // More efficient - can make the range bigger
 895                     $fRange = 0.05;
 896
 897                     $sOrderBySQL = '';
 898                     if ($this->oContext->hasNearPoint()) {
 899                         $sOrderBySQL = $this->oContext->distanceSQL('l.centroid');
 900                     } elseif ($sPlaceIDs) {
 901                         $sOrderBySQL = 'ST_Distance(l.centroid, f.geometry)';
 902                     } elseif ($sPlaceGeom) {
 903                         $sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
 904                     }
 905
 906                     $sSQL = 'SELECT distinct i.place_id';
 907                     if ($sOrderBySQL) {
 908                         $sSQL .= ', i.order_term';
 909                     }
 910                     $sSQL .= ' from (SELECT l.place_id';
 911                     if ($sOrderBySQL) {
 912                         $sSQL .= ','.$sOrderBySQL.' as order_term';
 913                     }
 914                     $sSQL .= ' from '.$sClassTable.' as l';
 915
 916                     if ($sPlaceIDs) {
 917                         $sSQL .= ',placex as f WHERE ';
 918                         $sSQL .= "f.place_id in ($sPlaceIDs) ";
 919                         $sSQL .= " AND ST_DWithin(l.centroid, f.centroid, $fRange)";
 920                     } elseif ($sPlaceGeom) {
 921                         $sSQL .= " WHERE ST_Contains('$sPlaceGeom', l.centroid)";
 922                     }
 923
 924                     $sSQL .= $this->oContext->excludeSQL(' AND l.place_id');
 925                     $sSQL .= 'limit 300) i ';
 926                     if ($sOrderBySQL) {
 927                         $sSQL .= 'order by order_term asc';
 928                     }
 929                     $sSQL .= " limit $iLimit";
 930
 931                     Debug::printSQL($sSQL);
 932
 933                     foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 934                         $aResults[$iPlaceId] = new Result($iPlaceId);
 935                     }
 936                 } else {
 937                     if ($this->oContext->hasNearPoint()) {
 938                         $fRange = $this->oContext->nearRadius();
 939                     }
 940
 941                     $sOrderBySQL = '';
 942                     if ($this->oContext->hasNearPoint()) {
 943                         $sOrderBySQL = $this->oContext->distanceSQL('l.geometry');
 944                     } else {
 945                         $sOrderBySQL = 'ST_Distance(l.geometry, f.geometry)';
 946                     }
 947
 948                     $sSQL = 'SELECT distinct l.place_id';
 949                     if ($sOrderBySQL) {
 950                         $sSQL .= ','.$sOrderBySQL.' as orderterm';
 951                     }
 952                     $sSQL .= ' FROM placex as l, placex as f';
 953                     $sSQL .= " WHERE f.place_id in ($sPlaceIDs)";
 954                     $sSQL .= "  AND ST_DWithin(l.geometry, f.centroid, $fRange)";
 955                     $sSQL .= "  AND l.class='".$this->sClass."'";
 956                     $sSQL .= "  AND l.type='".$this->sType."'";
 957                     $sSQL .= $this->oContext->excludeSQL(' AND l.place_id');
 958                     if ($sOrderBySQL) {
 959                         $sSQL .= 'ORDER BY orderterm ASC';
 960                     }
 961                     $sSQL .= " limit $iLimit";
 962
 963                     Debug::printSQL($sSQL);
 964
 965                     foreach ($oDB->getCol($sSQL) as $iPlaceId) {
 966                         $aResults[$iPlaceId] = new Result($iPlaceId);
 967                     }
 968                 }
 969             }
 970         }
 971
 972         return $aResults;
 973     }
 974
 975     private function poiTable()
 976     {
 977         return 'place_classtype_'.$this->sClass.'_'.$this->sType;
 978     }
 979
 980     private function countryCodeSQL($sVar)
 981     {
 982         if ($this->sCountryCode) {
 983             return $sVar.' = \''.$this->sCountryCode."'";
 984         }
 985         if ($this->oContext->sqlCountryList) {
 986             return $sVar.' in '.$this->oContext->sqlCountryList;
 987         }
 988
 989         return '';
 990     }
 991
 992     /////////// Sort functions
 993
 994
 995     public static function bySearchRank($a, $b)
 996     {
 997         if ($a->iSearchRank == $b->iSearchRank) {
 998             return $a->iOperator + strlen($a->sHouseNumber)
 999                      - $b->iOperator - strlen($b->sHouseNumber);
1000         }
1001
1002         return $a->iSearchRank < $b->iSearchRank ? -1 : 1;
1003     }
1004
1005     //////////// Debugging functions
1006
1007
1008     public function debugInfo()
1009     {
1010         return array(
1011                 'Search rank' => $this->iSearchRank,
1012                 'Country code' => $this->sCountryCode,
1013                 'Name terms' => $this->aName,
1014                 'Name terms (stop words)' => $this->aNameNonSearch,
1015                 'Address terms' => $this->aAddress,
1016                 'Address terms (stop words)' => $this->aAddressNonSearch,
1017                 'Address terms (full words)' => $this->aFullNameAddress ?? '',
1018                 'Special search' => $this->iOperator,
1019                 'Class' => $this->sClass,
1020                 'Type' => $this->sType,
1021                 'House number' => $this->sHouseNumber,
1022                 'Postcode' => $this->sPostcode
1023                );
1024     }
1025
1026     public function dumpAsHtmlTableRow(&$aWordIDs)
1027     {
1028         $kf = function ($k) use (&$aWordIDs) {
1029             return $aWordIDs[$k] ?? '['.$k.']';
1030         };
1031
1032         echo '<tr>';
1033         echo "<td>$this->iSearchRank</td>";
1034         echo '<td>'.join(', ', array_map($kf, $this->aName)).'</td>';
1035         echo '<td>'.join(', ', array_map($kf, $this->aNameNonSearch)).'</td>';
1036         echo '<td>'.join(', ', array_map($kf, $this->aAddress)).'</td>';
1037         echo '<td>'.join(', ', array_map($kf, $this->aAddressNonSearch)).'</td>';
1038         echo '<td>'.$this->sCountryCode.'</td>';
1039         echo '<td>'.Operator::toString($this->iOperator).'</td>';
1040         echo '<td>'.$this->sClass.'</td>';
1041         echo '<td>'.$this->sType.'</td>';
1042         echo '<td>'.$this->sPostcode.'</td>';
1043         echo '<td>'.$this->sHouseNumber.'</td>';
1044
1045         echo '</tr>';
1046     }
1047 }