lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10
  11 class Geocode
  12 {
  13     protected $oDB;
  14
  15     protected $aLangPrefOrder = array();
  16
  17     protected $bIncludeAddressDetails = false;
  18     protected $bIncludeExtraTags = false;
  19     protected $bIncludeNameDetails = false;
  20
  21     protected $bIncludePolygonAsPoints = false;
  22     protected $bIncludePolygonAsText = false;
  23     protected $bIncludePolygonAsGeoJSON = false;
  24     protected $bIncludePolygonAsKML = false;
  25     protected $bIncludePolygonAsSVG = false;
  26     protected $fPolygonSimplificationThreshold = 0.0;
  27
  28     protected $aExcludePlaceIDs = array();
  29     protected $bDeDupe = true;
  30     protected $bReverseInPlan = false;
  31
  32     protected $iLimit = 20;
  33     protected $iFinalLimit = 10;
  34     protected $iOffset = 0;
  35     protected $bFallback = false;
  36
  37     protected $aCountryCodes = false;
  38
  39     protected $bBoundedSearch = false;
  40     protected $aViewBox = false;
  41     protected $aRoutePoints = false;
  42     protected $aRouteWidth = false;
  43
  44     protected $iMaxRank = 20;
  45     protected $iMinAddressRank = 0;
  46     protected $iMaxAddressRank = 30;
  47     protected $aAddressRankList = array();
  48
  49     protected $sAllowedTypesSQLList = false;
  50
  51     protected $sQuery = false;
  52     protected $aStructuredQuery = false;
  53
  54     protected $oNormalizer = null;
  55
  56
  57     public function __construct(&$oDB)
  58     {
  59         $this->oDB =& $oDB;
  60         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  61     }
  62
  63     private function normTerm($sTerm)
  64     {
  65         if ($this->oNormalizer === null) {
  66             return $sTerm;
  67         }
  68
  69         return $this->oNormalizer->transliterate($sTerm);
  70     }
  71
  72     public function setReverseInPlan($bReverse)
  73     {
  74         $this->bReverseInPlan = $bReverse;
  75     }
  76
  77     public function setLanguagePreference($aLangPref)
  78     {
  79         $this->aLangPrefOrder = $aLangPref;
  80     }
  81
  82     public function getMoreUrlParams()
  83     {
  84         if ($this->aStructuredQuery) {
  85             $aParams = $this->aStructuredQuery;
  86         } else {
  87             $aParams = array('q' => $this->sQuery);
  88         }
  89
  90         if ($this->aExcludePlaceIDs) {
  91             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  92         }
  93
  94         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  95         if ($this->bIncludeExtraTags) $aParams['extratags'] = '1';
  96         if ($this->bIncludeNameDetails) $aParams['namedetails'] = '1';
  97
  98         if ($this->bIncludePolygonAsPoints) $aParams['polygon'] = '1';
  99         if ($this->bIncludePolygonAsText) $aParams['polygon_text'] = '1';
 100         if ($this->bIncludePolygonAsGeoJSON) $aParams['polygon_geojson'] = '1';
 101         if ($this->bIncludePolygonAsKML) $aParams['polygon_kml'] = '1';
 102         if ($this->bIncludePolygonAsSVG) $aParams['polygon_svg'] = '1';
 103
 104         if ($this->fPolygonSimplificationThreshold > 0.0) {
 105             $aParams['polygon_threshold'] = $this->fPolygonSimplificationThreshold;
 106         }
 107
 108         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
 109         if (!$this->bDeDupe) $aParams['dedupe'] = '0';
 110
 111         if ($this->aCountryCodes) {
 112             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
 113         }
 114
 115         if ($this->aViewBox) {
 116             $aParams['viewbox'] = $this->aViewBox[0].','.$this->aViewBox[3]
 117                                   .','.$this->aViewBox[2].','.$this->aViewBox[1];
 118         }
 119
 120         return $aParams;
 121     }
 122
 123     public function setIncludePolygonAsPoints($b = true)
 124     {
 125         $this->bIncludePolygonAsPoints = $b;
 126     }
 127
 128     public function setIncludePolygonAsText($b = true)
 129     {
 130         $this->bIncludePolygonAsText = $b;
 131     }
 132
 133     public function setIncludePolygonAsGeoJSON($b = true)
 134     {
 135         $this->bIncludePolygonAsGeoJSON = $b;
 136     }
 137
 138     public function setIncludePolygonAsKML($b = true)
 139     {
 140         $this->bIncludePolygonAsKML = $b;
 141     }
 142
 143     public function setIncludePolygonAsSVG($b = true)
 144     {
 145         $this->bIncludePolygonAsSVG = $b;
 146     }
 147
 148     public function setPolygonSimplificationThreshold($f)
 149     {
 150         $this->fPolygonSimplificationThreshold = $f;
 151     }
 152
 153     public function setLimit($iLimit = 10)
 154     {
 155         if ($iLimit > 50) $iLimit = 50;
 156         if ($iLimit < 1) $iLimit = 1;
 157
 158         $this->iFinalLimit = $iLimit;
 159         $this->iLimit = $iLimit + min($iLimit, 10);
 160     }
 161
 162     public function setFeatureType($sFeatureType)
 163     {
 164         switch ($sFeatureType) {
 165             case 'country':
 166                 $this->setRankRange(4, 4);
 167                 break;
 168             case 'state':
 169                 $this->setRankRange(8, 8);
 170                 break;
 171             case 'city':
 172                 $this->setRankRange(14, 16);
 173                 break;
 174             case 'settlement':
 175                 $this->setRankRange(8, 20);
 176                 break;
 177         }
 178     }
 179
 180     public function setRankRange($iMin, $iMax)
 181     {
 182         $this->iMinAddressRank = $iMin;
 183         $this->iMaxAddressRank = $iMax;
 184     }
 185
 186     public function setViewbox($aViewbox)
 187     {
 188         $this->aViewBox = array_map('floatval', $aViewbox);
 189
 190         $this->aViewBox[0] = max(-180.0, min(180, $this->aViewBox[0]));
 191         $this->aViewBox[1] = max(-90.0, min(90, $this->aViewBox[1]));
 192         $this->aViewBox[2] = max(-180.0, min(180, $this->aViewBox[2]));
 193         $this->aViewBox[3] = max(-90.0, min(90, $this->aViewBox[3]));
 194
 195         if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001
 196             || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001
 197         ) {
 198             userError("Bad parameter 'viewbox'. Not a box.");
 199         }
 200     }
 201
 202     public function setQuery($sQueryString)
 203     {
 204         $this->sQuery = $sQueryString;
 205         $this->aStructuredQuery = false;
 206     }
 207
 208     public function getQueryString()
 209     {
 210         return $this->sQuery;
 211     }
 212
 213
 214     public function loadParamArray($oParams)
 215     {
 216         $this->bIncludeAddressDetails
 217          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 218         $this->bIncludeExtraTags
 219          = $oParams->getBool('extratags', $this->bIncludeExtraTags);
 220         $this->bIncludeNameDetails
 221          = $oParams->getBool('namedetails', $this->bIncludeNameDetails);
 222
 223         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 224         $this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe);
 225
 226         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 227         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 228
 229         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 230
 231         // List of excluded Place IDs - used for more acurate pageing
 232         $sExcluded = $oParams->getStringList('exclude_place_ids');
 233         if ($sExcluded) {
 234             foreach ($sExcluded as $iExcludedPlaceID) {
 235                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 236                 if ($iExcludedPlaceID)
 237                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 238             }
 239
 240             if (isset($aExcludePlaceIDs))
 241                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 242         }
 243
 244         // Only certain ranks of feature
 245         $sFeatureType = $oParams->getString('featureType');
 246         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 247         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 248
 249         // Country code list
 250         $sCountries = $oParams->getStringList('countrycodes');
 251         if ($sCountries) {
 252             foreach ($sCountries as $sCountryCode) {
 253                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 254                     $aCountries[] = strtolower($sCountryCode);
 255                 }
 256             }
 257             if (isset($aCountries))
 258                 $this->aCountryCodes = $aCountries;
 259         }
 260
 261         $aViewbox = $oParams->getStringList('viewboxlbrt');
 262         if ($aViewbox) {
 263             if (count($aViewbox) != 4) {
 264                 userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
 265             }
 266             $this->setViewbox($aViewbox);
 267         } else {
 268             $aViewbox = $oParams->getStringList('viewbox');
 269             if ($aViewbox) {
 270                 if (count($aViewbox) != 4) {
 271                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
 272                 }
 273                 $this->setViewBox($aViewbox);
 274             } else {
 275                 $aRoute = $oParams->getStringList('route');
 276                 $fRouteWidth = $oParams->getFloat('routewidth');
 277                 if ($aRoute && $fRouteWidth) {
 278                     $this->aRoutePoints = $aRoute;
 279                     $this->aRouteWidth = $fRouteWidth;
 280                 }
 281             }
 282         }
 283     }
 284
 285     public function setQueryFromParams($oParams)
 286     {
 287         // Search query
 288         $sQuery = $oParams->getString('q');
 289         if (!$sQuery) {
 290             $this->setStructuredQuery(
 291                 $oParams->getString('amenity'),
 292                 $oParams->getString('street'),
 293                 $oParams->getString('city'),
 294                 $oParams->getString('county'),
 295                 $oParams->getString('state'),
 296                 $oParams->getString('country'),
 297                 $oParams->getString('postalcode')
 298             );
 299             $this->setReverseInPlan(false);
 300         } else {
 301             $this->setQuery($sQuery);
 302         }
 303     }
 304
 305     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 306     {
 307         $sValue = trim($sValue);
 308         if (!$sValue) return false;
 309         $this->aStructuredQuery[$sKey] = $sValue;
 310         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 311             $this->iMinAddressRank = $iNewMinAddressRank;
 312             $this->iMaxAddressRank = $iNewMaxAddressRank;
 313         }
 314         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 315         return true;
 316     }
 317
 318     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 319     {
 320         $this->sQuery = false;
 321
 322         // Reset
 323         $this->iMinAddressRank = 0;
 324         $this->iMaxAddressRank = 30;
 325         $this->aAddressRankList = array();
 326
 327         $this->aStructuredQuery = array();
 328         $this->sAllowedTypesSQLList = false;
 329
 330         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 331         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 332         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 333         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 334         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 335         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 336         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 337
 338         if (sizeof($this->aStructuredQuery) > 0) {
 339             $this->sQuery = join(', ', $this->aStructuredQuery);
 340             if ($this->iMaxAddressRank < 30) {
 341                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 342             }
 343         }
 344     }
 345
 346     public function fallbackStructuredQuery()
 347     {
 348         if (!$this->aStructuredQuery) return false;
 349
 350         $aParams = $this->aStructuredQuery;
 351
 352         if (sizeof($aParams) == 1) return false;
 353
 354         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 355
 356         foreach ($aOrderToFallback as $sType) {
 357             if (isset($aParams[$sType])) {
 358                 unset($aParams[$sType]);
 359                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 360                 return true;
 361             }
 362         }
 363
 364         return false;
 365     }
 366
 367     public function getDetails($aResults, $oCtx)
 368     {
 369         // Get the details for display (is this a redundant extra step?)
 370         //$aResults is an array of Result objects
 371         if (sizeof($aResults) == 0) return array();
 372
 373         $sLanguagePrefArraySQL = getArraySQL(
 374             array_map("getDBQuoted", $this->aLangPrefOrder)
 375         );
 376
 377         $sImportanceSQL = $oCtx->viewboxImportanceSQL('ST_Collect(centroid)');
 378         $sImportanceSQLGeom = $oCtx->viewboxImportanceSQL('geometry');
 379
 380         $aSubSelects = array();
 381
 382         $sPlaceIDs = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 383         if ($sPlaceIDs) {
 384             $sSQL  = "SELECT ";
 385             $sSQL .= "    osm_type,";
 386             $sSQL .= "    osm_id,";
 387             $sSQL .= "    class,";
 388             $sSQL .= "    type,";
 389             $sSQL .= "    admin_level,";
 390             $sSQL .= "    rank_search,";
 391             $sSQL .= "    rank_address,";
 392             $sSQL .= "    min(place_id) AS place_id, ";
 393             $sSQL .= "    min(parent_place_id) AS parent_place_id, ";
 394             $sSQL .= "    country_code, ";
 395             $sSQL .= "    get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 396             $sSQL .= "    get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,";
 397             $sSQL .= "    get_name_by_language(name, ARRAY['ref']) AS ref,";
 398             if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,";
 399             if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,";
 400             $sSQL .= "    avg(ST_X(centroid)) AS lon, ";
 401             $sSQL .= "    avg(ST_Y(centroid)) AS lat, ";
 402             $sSQL .= "    COALESCE(importance,0.75-(rank_search::float/40)) $sImportanceSQL AS importance, ";
 403             if ($oCtx->hasNearPoint()) {
 404                 $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 405             } else {
 406                 $sSQL .= "    ( ";
 407                 $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 408                 $sSQL .= "       FROM ";
 409                 $sSQL .= "         place_addressline s, ";
 410                 $sSQL .= "         placex p";
 411                 $sSQL .= "       WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
 412                 $sSQL .= "         AND p.place_id = s.address_place_id ";
 413                 $sSQL .= "         AND s.isaddress ";
 414                 $sSQL .= "         AND p.importance is not null ";
 415                 $sSQL .= "    ) AS addressimportance, ";
 416             }
 417             $sSQL .= "    (extratags->'place') AS extra_place ";
 418             $sSQL .= " FROM placex";
 419             $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 420             $sSQL .= "   AND (";
 421             $sSQL .= "            placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 422             if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 423                 $sSQL .= "        OR (extratags->'place') = 'city'";
 424             }
 425             if ($this->aAddressRankList) {
 426                 $sSQL .= "        OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 427             }
 428             $sSQL .= "       ) ";
 429             if ($this->sAllowedTypesSQLList) {
 430                 $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
 431             }
 432             $sSQL .= "    AND linked_place_id is null ";
 433             $sSQL .= " GROUP BY ";
 434             $sSQL .= "     osm_type, ";
 435             $sSQL .= "     osm_id, ";
 436             $sSQL .= "     class, ";
 437             $sSQL .= "     type, ";
 438             $sSQL .= "     admin_level, ";
 439             $sSQL .= "     rank_search, ";
 440             $sSQL .= "     rank_address, ";
 441             $sSQL .= "     country_code, ";
 442             $sSQL .= "     importance, ";
 443             if (!$this->bDeDupe) $sSQL .= "place_id,";
 444             $sSQL .= "     langaddress, ";
 445             $sSQL .= "     placename, ";
 446             $sSQL .= "     ref, ";
 447             if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
 448             if ($this->bIncludeNameDetails) $sSQL .= "name, ";
 449             $sSQL .= "     extratags->'place' ";
 450
 451             $aSubSelects[] = $sSQL;
 452         }
 453
 454         // postcode table
 455         $sPlaceIDs = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 456         if ($sPlaceIDs) {
 457             $sSQL = 'SELECT';
 458             $sSQL .= "  'P' as osm_type,";
 459             $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
 460             $sSQL .= "  'place' as class, 'postcode' as type,";
 461             $sSQL .= "  null as admin_level, rank_search, rank_address,";
 462             $sSQL .= "  place_id, parent_place_id, country_code,";
 463             $sSQL .= "  get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 464             $sSQL .= "  postcode as placename,";
 465             $sSQL .= "  postcode as ref,";
 466             if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 467             if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 468             $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
 469             $sSQL .= "  (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, ";
 470             if ($oCtx->hasNearPoint()) {
 471                 $sSQL .= $oCtx->distanceSQL('geometry')." AS addressimportance,";
 472             } else {
 473                 $sSQL .= "  (";
 474                 $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
 475                 $sSQL .= "     FROM ";
 476                 $sSQL .= "       place_addressline s, ";
 477                 $sSQL .= "       placex p";
 478                 $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
 479                 $sSQL .= "       AND p.place_id = s.address_place_id ";
 480                 $sSQL .= "       AND s.isaddress";
 481                 $sSQL .= "       AND p.importance is not null";
 482                 $sSQL .= "  ) AS addressimportance, ";
 483             }
 484             $sSQL .= "  null AS extra_place ";
 485             $sSQL .= "FROM location_postcode lp";
 486             $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 487
 488             $aSubSelects[] = $sSQL;
 489         }
 490
 491         // All other tables are rank 30 only.
 492         if ($this->iMaxAddressRank == 30) {
 493             // TIGER table
 494             if (CONST_Use_US_Tiger_Data) {
 495                 $sPlaceIDs = Result::joinIdsByTable($aResults, Result::TABLE_TIGER);
 496                 if ($sPlaceIDs) {
 497                     $sHousenumbers = Result::sqlHouseNumberTable($aResults, Result::TABLE_TIGER);
 498                     // Tiger search only if a housenumber was searched and if it was found
 499                     // (realized through a join)
 500                     $sSQL = " SELECT ";
 501                     $sSQL .= "     'T' AS osm_type, ";
 502                     $sSQL .= "     (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
 503                     $sSQL .= "     'place' AS class, ";
 504                     $sSQL .= "     'house' AS type, ";
 505                     $sSQL .= "     null AS admin_level, ";
 506                     $sSQL .= "     30 AS rank_search, ";
 507                     $sSQL .= "     30 AS rank_address, ";
 508                     $sSQL .= "     min(place_id) AS place_id, ";
 509                     $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 510                     $sSQL .= "     'us' AS country_code, ";
 511                     $sSQL .= "     get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
 512                     $sSQL .= "     null AS placename, ";
 513                     $sSQL .= "     null AS ref, ";
 514                     if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 515                     if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 516                     $sSQL .= "     avg(st_x(centroid)) AS lon, ";
 517                     $sSQL .= "     avg(st_y(centroid)) AS lat,";
 518                     $sSQL .= "     -1.15".$sImportanceSQL." AS importance, ";
 519                     if ($oCtx->hasNearPoint()) {
 520                         $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 521                     } else {
 522                         $sSQL .= "     (";
 523                         $sSQL .= "        SELECT max(p.importance*(p.rank_address+2))";
 524                         $sSQL .= "        FROM ";
 525                         $sSQL .= "          place_addressline s, ";
 526                         $sSQL .= "          placex p";
 527                         $sSQL .= "        WHERE s.place_id = min(blub.parent_place_id)";
 528                         $sSQL .= "          AND p.place_id = s.address_place_id ";
 529                         $sSQL .= "          AND s.isaddress";
 530                         $sSQL .= "          AND p.importance is not null";
 531                         $sSQL .= "     ) AS addressimportance, ";
 532                     }
 533                     $sSQL .= "     null AS extra_place ";
 534                     $sSQL .= " FROM (";
 535                     $sSQL .= "     SELECT place_id, ";    // interpolate the Tiger housenumbers here
 536                     $sSQL .= "         ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
 537                     $sSQL .= "         parent_place_id, ";
 538                     $sSQL .= "         housenumber_for_place";
 539                     $sSQL .= "     FROM (";
 540                     $sSQL .= "            location_property_tiger ";
 541                     $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
 542                     $sSQL .= "     WHERE ";
 543                     $sSQL .= "         housenumber_for_place >= startnumber";
 544                     $sSQL .= "         AND housenumber_for_place <= endnumber";
 545                     $sSQL .= " ) AS blub"; //postgres wants an alias here
 546                     $sSQL .= " GROUP BY";
 547                     $sSQL .= "      place_id, ";
 548                     $sSQL .= "      housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
 549                     if (!$this->bDeDupe) $sSQL .= ", place_id ";
 550
 551                     $aSubSelects[] = $sSQL;
 552                 }
 553             }
 554
 555             // osmline - interpolated housenumbers
 556             $sPlaceIDs = Result::joinIdsByTable($aResults, Result::TABLE_OSMLINE);
 557             if ($sPlaceIDs) {
 558                 $sHousenumbers = Result::sqlHouseNumberTable($aResults, Result::TABLE_OSMLINE);
 559                 // interpolation line search only if a housenumber was searched
 560                 // (realized through a join)
 561                 $sSQL = "SELECT ";
 562                 $sSQL .= "  'W' AS osm_type, ";
 563                 $sSQL .= "  osm_id, ";
 564                 $sSQL .= "  'place' AS class, ";
 565                 $sSQL .= "  'house' AS type, ";
 566                 $sSQL .= "  null AS admin_level, ";
 567                 $sSQL .= "  30 AS rank_search, ";
 568                 $sSQL .= "  30 AS rank_address, ";
 569                 $sSQL .= "  min(place_id) as place_id, ";
 570                 $sSQL .= "  min(parent_place_id) AS parent_place_id, ";
 571                 $sSQL .= "  country_code, ";
 572                 $sSQL .= "  get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
 573                 $sSQL .= "  null AS placename, ";
 574                 $sSQL .= "  null AS ref, ";
 575                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 576                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 577                 $sSQL .= "  AVG(st_x(centroid)) AS lon, ";
 578                 $sSQL .= "  AVG(st_y(centroid)) AS lat, ";
 579                 $sSQL .= "  -0.1".$sImportanceSQL." AS importance, ";  // slightly smaller than the importance for normal houses with rank 30, which is 0
 580                 if ($oCtx->hasNearPoint()) {
 581                     $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 582                 } else {
 583                     $sSQL .= "  (";
 584                     $sSQL .= "     SELECT ";
 585                     $sSQL .= "       MAX(p.importance*(p.rank_address+2)) ";
 586                     $sSQL .= "     FROM";
 587                     $sSQL .= "       place_addressline s, ";
 588                     $sSQL .= "       placex p";
 589                     $sSQL .= "     WHERE s.place_id = min(blub.parent_place_id) ";
 590                     $sSQL .= "       AND p.place_id = s.address_place_id ";
 591                     $sSQL .= "       AND s.isaddress ";
 592                     $sSQL .= "       AND p.importance is not null";
 593                     $sSQL .= "  ) AS addressimportance,";
 594                 }
 595                 $sSQL .= "  null AS extra_place ";
 596                 $sSQL .= "  FROM (";
 597                 $sSQL .= "     SELECT ";
 598                 $sSQL .= "         osm_id, ";
 599                 $sSQL .= "         place_id, ";
 600                 $sSQL .= "         country_code, ";
 601                 $sSQL .= "         CASE ";             // interpolate the housenumbers here
 602                 $sSQL .= "           WHEN startnumber != endnumber ";
 603                 $sSQL .= "           THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
 604                 $sSQL .= "           ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
 605                 $sSQL .= "         END as centroid, ";
 606                 $sSQL .= "         parent_place_id, ";
 607                 $sSQL .= "         housenumber_for_place ";
 608                 $sSQL .= "     FROM (";
 609                 $sSQL .= "            location_property_osmline ";
 610                 $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
 611                 $sSQL .= "          ) ";
 612                 $sSQL .= "     WHERE housenumber_for_place>=0 ";
 613                 $sSQL .= "       AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 614                 $sSQL .= "  ) as blub"; //postgres wants an alias here
 615                 $sSQL .= "  GROUP BY ";
 616                 $sSQL .= "    osm_id, ";
 617                 $sSQL .= "    place_id, ";
 618                 $sSQL .= "    housenumber_for_place, ";
 619                 $sSQL .= "    country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
 620                 if (!$this->bDeDupe) $sSQL .= ", place_id ";
 621
 622                 $aSubSelects[] = $sSQL;
 623             }
 624
 625             if (CONST_Use_Aux_Location_data) {
 626                 $sPlaceIDs = Result::joinIdsByTable($aResults, Result::TABLE_AUX);
 627                 if ($sPlaceIDs) {
 628                     $sHousenumbers = Result::sqlHouseNumberTable($aResults, Result::TABLE_AUX);
 629                     $sSQL = "  SELECT ";
 630                     $sSQL .= "     'L' AS osm_type, ";
 631                     $sSQL .= "     place_id AS osm_id, ";
 632                     $sSQL .= "     'place' AS class,";
 633                     $sSQL .= "     'house' AS type, ";
 634                     $sSQL .= "     null AS admin_level, ";
 635                     $sSQL .= "     0 AS rank_search,";
 636                     $sSQL .= "     0 AS rank_address, ";
 637                     $sSQL .= "     min(place_id) AS place_id,";
 638                     $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 639                     $sSQL .= "     'us' AS country_code, ";
 640                     $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
 641                     $sSQL .= "     null AS placename, ";
 642                     $sSQL .= "     null AS ref, ";
 643                     if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 644                     if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 645                     $sSQL .= "     avg(ST_X(centroid)) AS lon, ";
 646                     $sSQL .= "     avg(ST_Y(centroid)) AS lat, ";
 647                     $sSQL .= "     -1.10".$sImportanceSQL." AS importance, ";
 648                     if ($oCtx->hasNearPoint()) {
 649                         $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 650                     } else {
 651                         $sSQL .= "     ( ";
 652                         $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 653                         $sSQL .= "       FROM ";
 654                         $sSQL .= "          place_addressline s, ";
 655                         $sSQL .= "          placex p";
 656                         $sSQL .= "       WHERE s.place_id = min(location_property_aux.parent_place_id)";
 657                         $sSQL .= "         AND p.place_id = s.address_place_id ";
 658                         $sSQL .= "         AND s.isaddress";
 659                         $sSQL .= "         AND p.importance is not null";
 660                         $sSQL .= "     ) AS addressimportance, ";
 661                     }
 662                     $sSQL .= "     null AS extra_place ";
 663                     $sSQL .= "  FROM location_property_aux ";
 664                     $sSQL .= "  WHERE place_id in ($sPlaceIDs) ";
 665                     $sSQL .= "    AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
 666                     $sSQL .= "  GROUP BY ";
 667                     $sSQL .= "     place_id, ";
 668                     if (!$this->bDeDupe) $sSQL .= "place_id, ";
 669                     $sSQL .= "     langaddress ";
 670
 671                     $aSubSelects[] = $sSQL;
 672                 }
 673             }
 674         }
 675
 676         if (!sizeof($aSubSelects)) {
 677             return array();
 678         }
 679
 680         $sSQL = join(' UNION ', $aSubSelects)." order by importance desc";
 681         if (CONST_Debug) {
 682             echo "<hr>";
 683             var_dump($sSQL);
 684         }
 685         $aSearchResults = chksql(
 686             $this->oDB->getAll($sSQL),
 687             "Could not get details for place."
 688         );
 689
 690         return $aSearchResults;
 691     }
 692
 693     public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
 694     {
 695         /*
 696              Calculate all searches using aValidTokens i.e.
 697              'Wodsworth Road, Sheffield' =>
 698
 699              Phrase Wordset
 700              0      0       (wodsworth road)
 701              0      1       (wodsworth)(road)
 702              1      0       (sheffield)
 703
 704              Score how good the search is so they can be ordered
 705          */
 706         $iGlobalRank = 0;
 707
 708         foreach ($aPhrases as $iPhrase => $oPhrase) {
 709             $aNewPhraseSearches = array();
 710             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 711
 712             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 713                 // Too many permutations - too expensive
 714                 if ($iWordSet > 120) break;
 715
 716                 $aWordsetSearches = $aSearches;
 717
 718                 // Add all words from this wordset
 719                 foreach ($aWordset as $iToken => $sToken) {
 720                     //echo "<br><b>$sToken</b>";
 721                     $aNewWordsetSearches = array();
 722
 723                     foreach ($aWordsetSearches as $oCurrentSearch) {
 724                         //echo "<i>";
 725                         //var_dump($oCurrentSearch);
 726                         //echo "</i>";
 727
 728                         // If the token is valid
 729                         if (isset($aValidTokens[' '.$sToken])) {
 730                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 731                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 732                                     $aSearchTerm,
 733                                     isset($aValidTokens[$sToken])
 734                                       && strpos($sToken, ' ') === false,
 735                                     $sPhraseType,
 736                                     $iToken == 0 && $iPhrase == 0,
 737                                     $iPhrase == 0,
 738                                     $iToken + 1 == sizeof($aWordset)
 739                                       && $iPhrase + 1 == sizeof($aPhrases),
 740                                     $iGlobalRank
 741                                 );
 742
 743                                 foreach ($aNewSearches as $oSearch) {
 744                                     if ($oSearch->getRank() < $this->iMaxRank) {
 745                                         $aNewWordsetSearches[] = $oSearch;
 746                                     }
 747                                 }
 748                             }
 749                         }
 750                         // Look for partial matches.
 751                         // Note that there is no point in adding country terms here
 752                         // because country is omitted in the address.
 753                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 754                             // Allow searching for a word - but at extra cost
 755                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 756                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 757                                     $aSearchTerm,
 758                                     $bIsStructured,
 759                                     $iPhrase,
 760                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 761                                 );
 762
 763                                 foreach ($aNewSearches as $oSearch) {
 764                                     if ($oSearch->getRank() < $this->iMaxRank) {
 765                                         $aNewWordsetSearches[] = $oSearch;
 766                                     }
 767                                 }
 768                             }
 769                         }
 770                     }
 771                     // Sort and cut
 772                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 773                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 774                 }
 775                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 776
 777                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 778                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 779
 780                 $aSearchHash = array();
 781                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 782                     $sHash = serialize($aSearch);
 783                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 784                     else $aSearchHash[$sHash] = 1;
 785                 }
 786
 787                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 788             }
 789
 790             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 791             $aGroupedSearches = array();
 792             foreach ($aNewPhraseSearches as $aSearch) {
 793                 $iRank = $aSearch->getRank();
 794                 if ($iRank < $this->iMaxRank) {
 795                     if (!isset($aGroupedSearches[$iRank])) {
 796                         $aGroupedSearches[$iRank] = array();
 797                     }
 798                     $aGroupedSearches[$iRank][] = $aSearch;
 799                 }
 800             }
 801             ksort($aGroupedSearches);
 802
 803             $iSearchCount = 0;
 804             $aSearches = array();
 805             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 806                 $iSearchCount += sizeof($aNewSearches);
 807                 $aSearches = array_merge($aSearches, $aNewSearches);
 808                 if ($iSearchCount > 50) break;
 809             }
 810
 811             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 812         }
 813
 814         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 815         $aGroupedSearches = array();
 816         foreach ($aSearches as $oSearch) {
 817             if (!$oSearch->isValidSearch()) {
 818                 continue;
 819             }
 820
 821             $iRank = $oSearch->addToRank($iGlobalRank);
 822             if (!isset($aGroupedSearches[$iRank])) {
 823                 $aGroupedSearches[$iRank] = array();
 824             }
 825             $aGroupedSearches[$iRank][] = $oSearch;
 826         }
 827         ksort($aGroupedSearches);
 828
 829         return $aGroupedSearches;
 830     }
 831
 832     /* Perform the actual query lookup.
 833
 834         Returns an ordered list of results, each with the following fields:
 835             osm_type: type of corresponding OSM object
 836                         N - node
 837                         W - way
 838                         R - relation
 839                         P - postcode (internally computed)
 840             osm_id: id of corresponding OSM object
 841             class: general object class (corresponds to tag key of primary OSM tag)
 842             type: subclass of object (corresponds to tag value of primary OSM tag)
 843             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 844             rank_search: rank in search hierarchy
 845                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 846             rank_address: rank in address hierarchy (determines orer in address)
 847             place_id: internal key (may differ between different instances)
 848             country_code: ISO country code
 849             langaddress: localized full address
 850             placename: localized name of object
 851             ref: content of ref tag (if available)
 852             lon: longitude
 853             lat: latitude
 854             importance: importance of place based on Wikipedia link count
 855             addressimportance: cumulated importance of address elements
 856             extra_place: type of place (for admin boundaries, if there is a place tag)
 857             aBoundingBox: bounding Box
 858             label: short description of the object class/type (English only)
 859             name: full name (currently the same as langaddress)
 860             foundorder: secondary ordering for places with same importance
 861     */
 862
 863
 864     public function lookup()
 865     {
 866         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 867
 868         $oCtx = new SearchContext();
 869
 870         if ($this->aRoutePoints) {
 871             $oCtx->setViewboxFromRoute(
 872                 $this->oDB,
 873                 $this->aRoutePoints,
 874                 $this->aRouteWidth,
 875                 $this->bBoundedSearch
 876             );
 877         } elseif ($this->aViewBox) {
 878             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 879         }
 880         if ($this->aExcludePlaceIDs) {
 881             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 882         }
 883         if ($this->aCountryCodes) {
 884             $oCtx->setCountryList($this->aCountryCodes);
 885         }
 886
 887         $sNormQuery = $this->normTerm($this->sQuery);
 888         $sLanguagePrefArraySQL = getArraySQL(
 889             array_map("getDBQuoted", $this->aLangPrefOrder)
 890         );
 891
 892         $sQuery = $this->sQuery;
 893         if (!preg_match('//u', $sQuery)) {
 894             userError("Query string is not UTF-8 encoded.");
 895         }
 896
 897         // Conflicts between US state abreviations and various words for 'the' in different languages
 898         if (isset($this->aLangPrefOrder['name:en'])) {
 899             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 900             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 901             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 902         }
 903
 904         // Do we have anything that looks like a lat/lon pair?
 905         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 906
 907         $aSearchResults = array();
 908         if ($sQuery || $this->aStructuredQuery) {
 909             // Start with a single blank search
 910             $aSearches = array(new SearchDescription($oCtx));
 911
 912             if ($sQuery) {
 913                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 914             }
 915
 916             $sSpecialTerm = '';
 917             if ($sQuery) {
 918                 preg_match_all(
 919                     '/\\[([\\w ]*)\\]/u',
 920                     $sQuery,
 921                     $aSpecialTermsRaw,
 922                     PREG_SET_ORDER
 923                 );
 924                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 925                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 926                     if (!$sSpecialTerm) {
 927                         $sSpecialTerm = $aSpecialTerm[1];
 928                     }
 929                 }
 930             }
 931             if (!$sSpecialTerm && $this->aStructuredQuery
 932                 && isset($this->aStructuredQuery['amenity'])) {
 933                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 934                 unset($this->aStructuredQuery['amenity']);
 935             }
 936
 937             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 938                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 939                 $sToken = chksql(
 940                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 941                     "Cannot decode query. Wrong encoding?"
 942                 );
 943                 $sSQL = 'SELECT class, type FROM word ';
 944                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 945                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 946                 if (CONST_Debug) var_Dump($sSQL);
 947                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 948                 $aNewSearches = array();
 949                 foreach ($aSearches as $oSearch) {
 950                     foreach ($aSearchWords as $aSearchTerm) {
 951                         $oNewSearch = clone $oSearch;
 952                         $oNewSearch->setPoiSearch(
 953                             Operator::TYPE,
 954                             $aSearchTerm['class'],
 955                             $aSearchTerm['type']
 956                         );
 957                         $aNewSearches[] = $oNewSearch;
 958                     }
 959                 }
 960                 $aSearches = $aNewSearches;
 961             }
 962
 963             // Split query into phrases
 964             // Commas are used to reduce the search space by indicating where phrases split
 965             if ($this->aStructuredQuery) {
 966                 $aInPhrases = $this->aStructuredQuery;
 967                 $bStructuredPhrases = true;
 968             } else {
 969                 $aInPhrases = explode(',', $sQuery);
 970                 $bStructuredPhrases = false;
 971             }
 972
 973             // Convert each phrase to standard form
 974             // Create a list of standard words
 975             // Get all 'sets' of words
 976             // Generate a complete list of all
 977             $aTokens = array();
 978             $aPhrases = array();
 979             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 980                 $sPhrase = chksql(
 981                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 982                     "Cannot normalize query string (is it a UTF-8 string?)"
 983                 );
 984                 if (trim($sPhrase)) {
 985                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 986                     $oPhrase->addTokens($aTokens);
 987                     $aPhrases[] = $oPhrase;
 988                 }
 989             }
 990
 991             if (sizeof($aTokens)) {
 992                 // Check which tokens we have, get the ID numbers
 993                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 994                 $sSQL .= ' FROM word ';
 995                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
 996
 997                 if (CONST_Debug) var_Dump($sSQL);
 998
 999                 $aValidTokens = array();
1000                 $aDatabaseWords = chksql(
1001                     $this->oDB->getAll($sSQL),
1002                     "Could not get word tokens."
1003                 );
1004                 $aWordFrequencyScores = array();
1005                 foreach ($aDatabaseWords as $aToken) {
1006                     // Filter country tokens that do not match restricted countries.
1007                     if ($this->aCountryCodes
1008                         && $aToken['country_code']
1009                         && !in_array($aToken['country_code'], $this->aCountryCodes)
1010                     ) {
1011                         continue;
1012                     }
1013
1014                     // Special terms need to appear in their normalized form.
1015                     if ($aToken['word'] && $aToken['class']) {
1016                         $sNormWord = $this->normTerm($aToken['word']);
1017                         if (strpos($sNormQuery, $sNormWord) === false) {
1018                             continue;
1019                         }
1020                     }
1021
1022                     if (isset($aValidTokens[$aToken['word_token']])) {
1023                         $aValidTokens[$aToken['word_token']][] = $aToken;
1024                     } else {
1025                         $aValidTokens[$aToken['word_token']] = array($aToken);
1026                     }
1027                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
1028                 }
1029                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
1030
1031                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
1032                 foreach ($aTokens as $sToken) {
1033                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
1034                         if (isset($aValidTokens[$aData[1]])) {
1035                             foreach ($aValidTokens[$aData[1]] as $aToken) {
1036                                 if (!$aToken['class']) {
1037                                     if (isset($aValidTokens[$sToken])) {
1038                                         $aValidTokens[$sToken][] = $aToken;
1039                                     } else {
1040                                         $aValidTokens[$sToken] = array($aToken);
1041                                     }
1042                                 }
1043                             }
1044                         }
1045                     }
1046                 }
1047
1048                 foreach ($aTokens as $sToken) {
1049                     // Unknown single word token with a number - assume it is a house number
1050                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
1051                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
1052                     }
1053                 }
1054
1055                 // Any words that have failed completely?
1056                 // TODO: suggestions
1057
1058                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
1059
1060                 if ($this->bReverseInPlan) {
1061                     // Reverse phrase array and also reverse the order of the wordsets in
1062                     // the first and final phrase. Don't bother about phrases in the middle
1063                     // because order in the address doesn't matter.
1064                     $aPhrases = array_reverse($aPhrases);
1065                     $aPhrases[0]->invertWordSets();
1066                     if (sizeof($aPhrases) > 1) {
1067                         $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
1068                     }
1069                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
1070
1071                     foreach ($aGroupedSearches as $aSearches) {
1072                         foreach ($aSearches as $aSearch) {
1073                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
1074                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
1075                             }
1076                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
1077                         }
1078                     }
1079
1080                     $aGroupedSearches = $aReverseGroupedSearches;
1081                     ksort($aGroupedSearches);
1082                 }
1083             } else {
1084                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
1085                 $aGroupedSearches = array();
1086                 foreach ($aSearches as $aSearch) {
1087                     if ($aSearch->getRank() < $this->iMaxRank) {
1088                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
1089                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
1090                     }
1091                 }
1092                 ksort($aGroupedSearches);
1093             }
1094
1095             // Filter out duplicate searches
1096             $aSearchHash = array();
1097             foreach ($aGroupedSearches as $iGroup => $aSearches) {
1098                 foreach ($aSearches as $iSearch => $aSearch) {
1099                     $sHash = serialize($aSearch);
1100                     if (isset($aSearchHash[$sHash])) {
1101                         unset($aGroupedSearches[$iGroup][$iSearch]);
1102                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
1103                     } else {
1104                         $aSearchHash[$sHash] = 1;
1105                     }
1106                 }
1107             }
1108
1109             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
1110
1111             // Start the search process
1112             $aResults = array();
1113             $iGroupLoop = 0;
1114             $iQueryLoop = 0;
1115             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
1116                 $iGroupLoop++;
1117                 foreach ($aSearches as $oSearch) {
1118                     $iQueryLoop++;
1119
1120                     if (CONST_Debug) {
1121                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
1122                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
1123                     }
1124
1125                     $aResults += $oSearch->query(
1126                         $this->oDB,
1127                         $aWordFrequencyScores,
1128                         $this->iMinAddressRank,
1129                         $this->iMaxAddressRank,
1130                         $this->iLimit
1131                     );
1132
1133                     if ($iQueryLoop > 20) break;
1134                 }
1135
1136                 if (sizeof($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
1137                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
1138                     // reduces the number of place ids, like a filter
1139                     // rank_address is 30 for interpolated housenumbers
1140                     $aFilterSql = array();
1141                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
1142                     if ($sPlaceIds) {
1143                         $sSQL = 'SELECT place_id FROM placex ';
1144                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
1145                         $sSQL .= "  AND (";
1146                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1147                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
1148                             $sSQL .= "     OR (extratags->'place') = 'city'";
1149                         }
1150                         if ($this->aAddressRankList) {
1151                             $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
1152                         }
1153                         $sSQL .= ")";
1154                         $aFilterSql[] = $sSQL;
1155                     }
1156                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
1157                     if ($sPlaceIds) {
1158                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
1159                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
1160                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1161                         if ($this->aAddressRankList) {
1162                             $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
1163                         }
1164                         $sSQL .= ") ";
1165                         $aFilterSql[] = $sSQL;
1166                     }
1167
1168                     $aFilteredIDs = array();
1169                     if ($aFilterSql) {
1170                         $sSQL = join(' UNION ', $aFilterSql);
1171                         if (CONST_Debug) var_dump($sSQL);
1172                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
1173                     }
1174
1175                     $tempIDs = array();
1176                     foreach ($aResults as $oResult) {
1177                         if (($this->iMaxAddressRank == 30 &&
1178                              ($oResult->iTable == Result::TABLE_OSMLINE
1179                               || $oResult->iTable == Result::TABLE_AUX
1180                               || $oResult->iTable == Result::TABLE_TIGER))
1181                             || in_array($oResult->iId, $aFilteredIDs)
1182                         ) {
1183                             $tempIDs[$oResult->iId] = $oResult;
1184                         }
1185                     }
1186                     $aResults = $tempIDs;
1187                 }
1188
1189                 if (sizeof($aResults)) break;
1190                 if ($iGroupLoop > 4) break;
1191                 if ($iQueryLoop > 30) break;
1192             }
1193
1194             $aSearchResults = $this->getDetails($aResults, $oCtx);
1195         } else {
1196             // Just interpret as a reverse geocode
1197             $oReverse = new ReverseGeocode($this->oDB);
1198             $oReverse->setZoom(18);
1199
1200             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
1201
1202             if (CONST_Debug) var_dump("Reverse search", $aLookup);
1203
1204             if ($oLookup) {
1205                 $aResults = array($oLookup->iId => $oLookup);
1206                 $aSearchResults = $this->getDetails($aResults, $oCtx);
1207             } else {
1208                 $aSearchResults = array();
1209             }
1210         }
1211
1212         // No results? Done
1213         if (!sizeof($aSearchResults)) {
1214             if ($this->bFallback) {
1215                 if ($this->fallbackStructuredQuery()) {
1216                     return $this->lookup();
1217                 }
1218             }
1219
1220             return array();
1221         }
1222
1223         $aClassType = getClassTypesWithImportance();
1224         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
1225         foreach ($aRecheckWords as $i => $sWord) {
1226             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
1227         }
1228
1229         if (CONST_Debug) {
1230             echo '<i>Recheck words:<\i>';
1231             var_dump($aRecheckWords);
1232         }
1233
1234         $oPlaceLookup = new PlaceLookup($this->oDB);
1235         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
1236         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
1237         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
1238         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
1239         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
1240         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
1241
1242         foreach ($aSearchResults as $iResNum => $aResult) {
1243             // Default
1244             $fDiameter = getResultDiameter($aResult);
1245
1246             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
1247             if ($aOutlineResult) {
1248                 $aResult = array_merge($aResult, $aOutlineResult);
1249             }
1250
1251             if ($aResult['extra_place'] == 'city') {
1252                 $aResult['class'] = 'place';
1253                 $aResult['type'] = 'city';
1254                 $aResult['rank_search'] = 16;
1255             }
1256
1257             // Is there an icon set for this type of result?
1258             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
1259                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
1260             ) {
1261                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
1262             }
1263
1264             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
1265                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
1266             ) {
1267                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
1268             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
1269                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
1270             ) {
1271                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
1272             }
1273             // if tag '&addressdetails=1' is set in query
1274             if ($this->bIncludeAddressDetails) {
1275                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
1276                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber);
1277                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
1278                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
1279                 }
1280             }
1281
1282             if ($this->bIncludeExtraTags) {
1283                 if ($aResult['extra']) {
1284                     $aResult['sExtraTags'] = json_decode($aResult['extra']);
1285                 } else {
1286                     $aResult['sExtraTags'] = (object) array();
1287                 }
1288             }
1289
1290             if ($this->bIncludeNameDetails) {
1291                 if ($aResult['names']) {
1292                     $aResult['sNameDetails'] = json_decode($aResult['names']);
1293                 } else {
1294                     $aResult['sNameDetails'] = (object) array();
1295                 }
1296             }
1297
1298             $aResult['name'] = $aResult['langaddress'];
1299
1300             if ($oCtx->hasNearPoint()) {
1301                 $aResult['importance'] = 0.001;
1302                 $aResult['foundorder'] = $aResult['addressimportance'];
1303             } else {
1304                 // Adjust importance for the number of exact string matches in the result
1305                 $aResult['importance'] = max(0.001, $aResult['importance']);
1306                 $iCountWords = 0;
1307                 $sAddress = $aResult['langaddress'];
1308                 foreach ($aRecheckWords as $i => $sWord) {
1309                     if (stripos($sAddress, $sWord)!==false) {
1310                         $iCountWords++;
1311                         if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
1312                     }
1313                 }
1314
1315                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
1316
1317                 // secondary ordering (for results with same importance (the smaller the better):
1318                 // - approximate importance of address parts
1319                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
1320                 // - number of exact matches from the query
1321                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
1322                 // - importance of the class/type
1323                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1324                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1325                 ) {
1326                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1327                 } else {
1328                     $aResult['foundorder'] += 0.01;
1329                 }
1330             }
1331             if (CONST_Debug) var_dump($aResult);
1332             $aSearchResults[$iResNum] = $aResult;
1333         }
1334         uasort($aSearchResults, 'byImportance');
1335
1336         $aOSMIDDone = array();
1337         $aClassTypeNameDone = array();
1338         $aToFilter = $aSearchResults;
1339         $aSearchResults = array();
1340
1341         $bFirst = true;
1342         foreach ($aToFilter as $iResNum => $aResult) {
1343             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1344             if ($bFirst) {
1345                 $fLat = $aResult['lat'];
1346                 $fLon = $aResult['lon'];
1347                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1348                 $bFirst = false;
1349             }
1350             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1351                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1352             ) {
1353                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1354                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1355                 $aSearchResults[] = $aResult;
1356             }
1357
1358             // Absolute limit on number of results
1359             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1360         }
1361
1362         return $aSearchResults;
1363     } // end lookup()
1364 } // end class