lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/NearPoint.php');
   6 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8
   9 class Geocode
  10 {
  11     protected $oDB;
  12
  13     protected $aLangPrefOrder = array();
  14
  15     protected $bIncludeAddressDetails = false;
  16     protected $bIncludeExtraTags = false;
  17     protected $bIncludeNameDetails = false;
  18
  19     protected $bIncludePolygonAsPoints = false;
  20     protected $bIncludePolygonAsText = false;
  21     protected $bIncludePolygonAsGeoJSON = false;
  22     protected $bIncludePolygonAsKML = false;
  23     protected $bIncludePolygonAsSVG = false;
  24     protected $fPolygonSimplificationThreshold = 0.0;
  25
  26     protected $aExcludePlaceIDs = array();
  27     protected $bDeDupe = true;
  28     protected $bReverseInPlan = false;
  29
  30     protected $iLimit = 20;
  31     protected $iFinalLimit = 10;
  32     protected $iOffset = 0;
  33     protected $bFallback = false;
  34
  35     protected $aCountryCodes = false;
  36
  37     protected $bBoundedSearch = false;
  38     protected $aViewBox = false;
  39     protected $sViewboxCentreSQL = false;
  40     protected $sViewboxSmallSQL = false;
  41     protected $sViewboxLargeSQL = false;
  42
  43     protected $iMaxRank = 20;
  44     protected $iMinAddressRank = 0;
  45     protected $iMaxAddressRank = 30;
  46     protected $aAddressRankList = array();
  47     protected $exactMatchCache = 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 = $iMax;
 184     }
 185
 186     public function setRoute($aRoutePoints, $fRouteWidth)
 187     {
 188         $this->aViewBox = false;
 189
 190         $this->sViewboxCentreSQL = "ST_SetSRID('LINESTRING(";
 191         $sSep = '';
 192         foreach ($aRoutePoints as $aPoint) {
 193             $fPoint = (float)$aPoint;
 194             $this->sViewboxCentreSQL .= $sSep.$fPoint;
 195             $sSep = ($sSep == ' ') ? ',' : ' ';
 196         }
 197         $this->sViewboxCentreSQL .= ")'::geometry,4326)";
 198
 199         $this->sViewboxSmallSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL;
 200         $this->sViewboxSmallSQL .= ','.($fRouteWidth/69).')';
 201
 202         $this->sViewboxLargeSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL;
 203         $this->sViewboxLargeSQL .= ','.($fRouteWidth/30).')';
 204     }
 205
 206     public function setViewbox($aViewbox)
 207     {
 208         $this->aViewBox = array_map('floatval', $aViewbox);
 209
 210         $this->aViewBox[0] = max(-180.0, min(180, $this->aViewBox[0]));
 211         $this->aViewBox[1] = max(-90.0, min(90, $this->aViewBox[1]));
 212         $this->aViewBox[2] = max(-180.0, min(180, $this->aViewBox[2]));
 213         $this->aViewBox[3] = max(-90.0, min(90, $this->aViewBox[3]));
 214
 215         if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001
 216             || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001
 217         ) {
 218             userError("Bad parameter 'viewbox'. Not a box.");
 219         }
 220
 221         $fHeight = $this->aViewBox[0] - $this->aViewBox[2];
 222         $fWidth = $this->aViewBox[1] - $this->aViewBox[3];
 223         $aBigViewBox[0] = $this->aViewBox[0] + $fHeight;
 224         $aBigViewBox[2] = $this->aViewBox[2] - $fHeight;
 225         $aBigViewBox[1] = $this->aViewBox[1] + $fWidth;
 226         $aBigViewBox[3] = $this->aViewBox[3] - $fWidth;
 227
 228         $this->sViewboxCentreSQL = false;
 229         $this->sViewboxSmallSQL = sprintf(
 230             'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)',
 231             $this->aViewBox[0],
 232             $this->aViewBox[1],
 233             $this->aViewBox[2],
 234             $this->aViewBox[3]
 235         );
 236         $this->sViewboxLargeSQL = sprintf(
 237             'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)',
 238             $aBigViewBox[0],
 239             $aBigViewBox[1],
 240             $aBigViewBox[2],
 241             $aBigViewBox[3]
 242         );
 243     }
 244
 245     public function setQuery($sQueryString)
 246     {
 247         $this->sQuery = $sQueryString;
 248         $this->aStructuredQuery = false;
 249     }
 250
 251     public function getQueryString()
 252     {
 253         return $this->sQuery;
 254     }
 255
 256
 257     public function loadParamArray($oParams)
 258     {
 259         $this->bIncludeAddressDetails
 260          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 261         $this->bIncludeExtraTags
 262          = $oParams->getBool('extratags', $this->bIncludeExtraTags);
 263         $this->bIncludeNameDetails
 264          = $oParams->getBool('namedetails', $this->bIncludeNameDetails);
 265
 266         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 267         $this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe);
 268
 269         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 270         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 271
 272         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 273
 274         // List of excluded Place IDs - used for more acurate pageing
 275         $sExcluded = $oParams->getStringList('exclude_place_ids');
 276         if ($sExcluded) {
 277             foreach ($sExcluded as $iExcludedPlaceID) {
 278                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 279                 if ($iExcludedPlaceID)
 280                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 281             }
 282
 283             if (isset($aExcludePlaceIDs))
 284                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 285         }
 286
 287         // Only certain ranks of feature
 288         $sFeatureType = $oParams->getString('featureType');
 289         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 290         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 291
 292         // Country code list
 293         $sCountries = $oParams->getStringList('countrycodes');
 294         if ($sCountries) {
 295             foreach ($sCountries as $sCountryCode) {
 296                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 297                     $aCountries[] = strtolower($sCountryCode);
 298                 }
 299             }
 300             if (isset($aCountries))
 301                 $this->aCountryCodes = $aCountries;
 302         }
 303
 304         $aViewbox = $oParams->getStringList('viewboxlbrt');
 305         if ($aViewbox) {
 306             if (count($aViewbox) != 4) {
 307                 userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
 308             }
 309             $this->setViewbox($aViewbox);
 310         } else {
 311             $aViewbox = $oParams->getStringList('viewbox');
 312             if ($aViewbox) {
 313                 if (count($aViewbox) != 4) {
 314                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
 315                 }
 316                 $this->setViewBox($aViewbox);
 317             } else {
 318                 $aRoute = $oParams->getStringList('route');
 319                 $fRouteWidth = $oParams->getFloat('routewidth');
 320                 if ($aRoute && $fRouteWidth) {
 321                     $this->setRoute($aRoute, $fRouteWidth);
 322                 }
 323             }
 324         }
 325     }
 326
 327     public function setQueryFromParams($oParams)
 328     {
 329         // Search query
 330         $sQuery = $oParams->getString('q');
 331         if (!$sQuery) {
 332             $this->setStructuredQuery(
 333                 $oParams->getString('amenity'),
 334                 $oParams->getString('street'),
 335                 $oParams->getString('city'),
 336                 $oParams->getString('county'),
 337                 $oParams->getString('state'),
 338                 $oParams->getString('country'),
 339                 $oParams->getString('postalcode')
 340             );
 341             $this->setReverseInPlan(false);
 342         } else {
 343             $this->setQuery($sQuery);
 344         }
 345     }
 346
 347     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 348     {
 349         $sValue = trim($sValue);
 350         if (!$sValue) return false;
 351         $this->aStructuredQuery[$sKey] = $sValue;
 352         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 353             $this->iMinAddressRank = $iNewMinAddressRank;
 354             $this->iMaxAddressRank = $iNewMaxAddressRank;
 355         }
 356         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 357         return true;
 358     }
 359
 360     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 361     {
 362         $this->sQuery = false;
 363
 364         // Reset
 365         $this->iMinAddressRank = 0;
 366         $this->iMaxAddressRank = 30;
 367         $this->aAddressRankList = array();
 368
 369         $this->aStructuredQuery = array();
 370         $this->sAllowedTypesSQLList = False;
 371
 372         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 373         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 374         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 375         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 376         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 377         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 378         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 379
 380         if (sizeof($this->aStructuredQuery) > 0) {
 381             $this->sQuery = join(', ', $this->aStructuredQuery);
 382             if ($this->iMaxAddressRank < 30) {
 383                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 384             }
 385         }
 386     }
 387
 388     public function fallbackStructuredQuery()
 389     {
 390         if (!$this->aStructuredQuery) return false;
 391
 392         $aParams = $this->aStructuredQuery;
 393
 394         if (sizeof($aParams) == 1) return false;
 395
 396         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 397
 398         foreach ($aOrderToFallback as $sType) {
 399             if (isset($aParams[$sType])) {
 400                 unset($aParams[$sType]);
 401                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 402                 return true;
 403             }
 404         }
 405
 406         return false;
 407     }
 408
 409     public function getDetails($aPlaceIDs)
 410     {
 411         //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
 412         if (sizeof($aPlaceIDs) == 0) return array();
 413
 414         $sLanguagePrefArraySQL = getArraySQL(
 415             array_map("getDBQuoted",
 416             $this->aLangPrefOrder)
 417         );
 418
 419         // Get the details for display (is this a redundant extra step?)
 420         $sPlaceIDs = join(',', array_keys($aPlaceIDs));
 421
 422         $sImportanceSQL = '';
 423         $sImportanceSQLGeom = '';
 424         if ($this->sViewboxSmallSQL) {
 425             $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
 426             $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) THEN 1 ELSE 0.75 END * ";
 427         }
 428         if ($this->sViewboxLargeSQL) {
 429             $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
 430             $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * ";
 431         }
 432
 433         $sSQL  = "SELECT ";
 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 .= "    min(place_id) AS place_id, ";
 442         $sSQL .= "    min(parent_place_id) AS parent_place_id, ";
 443         $sSQL .= "    country_code, ";
 444         $sSQL .= "    get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 445         $sSQL .= "    get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,";
 446         $sSQL .= "    get_name_by_language(name, ARRAY['ref']) AS ref,";
 447         if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,";
 448         if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,";
 449         $sSQL .= "    avg(ST_X(centroid)) AS lon, ";
 450         $sSQL .= "    avg(ST_Y(centroid)) AS lat, ";
 451         $sSQL .= "    ".$sImportanceSQL."COALESCE(importance,0.75-(rank_search::float/40)) AS importance, ";
 452         $sSQL .= "    ( ";
 453         $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 454         $sSQL .= "       FROM ";
 455         $sSQL .= "         place_addressline s, ";
 456         $sSQL .= "         placex p";
 457         $sSQL .= "       WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
 458         $sSQL .= "         AND p.place_id = s.address_place_id ";
 459         $sSQL .= "         AND s.isaddress ";
 460         $sSQL .= "         AND p.importance is not null ";
 461         $sSQL .= "    ) AS addressimportance, ";
 462         $sSQL .= "    (extratags->'place') AS extra_place ";
 463         $sSQL .= " FROM placex";
 464         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 465         $sSQL .= "   AND (";
 466         $sSQL .= "            placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 467         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 468             $sSQL .= "        OR (extratags->'place') = 'city'";
 469         }
 470         if ($this->aAddressRankList) {
 471             $sSQL .= "        OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 472         }
 473         $sSQL .= "       ) ";
 474         if ($this->sAllowedTypesSQLList) {
 475             $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
 476         }
 477         $sSQL .= "    AND linked_place_id is null ";
 478         $sSQL .= " GROUP BY ";
 479         $sSQL .= "     osm_type, ";
 480         $sSQL .= "     osm_id, ";
 481         $sSQL .= "     class, ";
 482         $sSQL .= "     type, ";
 483         $sSQL .= "     admin_level, ";
 484         $sSQL .= "     rank_search, ";
 485         $sSQL .= "     rank_address, ";
 486         $sSQL .= "     country_code, ";
 487         $sSQL .= "     importance, ";
 488         if (!$this->bDeDupe) $sSQL .= "place_id,";
 489         $sSQL .= "     langaddress, ";
 490         $sSQL .= "     placename, ";
 491         $sSQL .= "     ref, ";
 492         if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
 493         if ($this->bIncludeNameDetails) $sSQL .= "name, ";
 494         $sSQL .= "     extratags->'place' ";
 495
 496         // postcode table
 497         $sSQL .= "UNION ";
 498         $sSQL .= "SELECT";
 499         $sSQL .= "  'P' as osm_type,";
 500         $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
 501         $sSQL .= "  'place' as class, 'postcode' as type,";
 502         $sSQL .= "  null as admin_level, rank_search, rank_address,";
 503         $sSQL .= "  place_id, parent_place_id, country_code,";
 504         $sSQL .= "  get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 505         $sSQL .= "  postcode as placename,";
 506         $sSQL .= "  postcode as ref,";
 507         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 508         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 509         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
 510         $sSQL .=    $sImportanceSQLGeom."(0.75-(rank_search::float/40)) AS importance, ";
 511         $sSQL .= "  (";
 512         $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
 513         $sSQL .= "     FROM ";
 514         $sSQL .= "       place_addressline s, ";
 515         $sSQL .= "       placex p";
 516         $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
 517         $sSQL .= "       AND p.place_id = s.address_place_id ";
 518         $sSQL .= "       AND s.isaddress";
 519         $sSQL .= "       AND p.importance is not null";
 520         $sSQL .= "  ) AS addressimportance, ";
 521         $sSQL .= "  null AS extra_place ";
 522         $sSQL .= "FROM location_postcode lp";
 523         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 524
 525         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
 526             // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
 527             // with start- and endnumber, the common osm housenumbers are usually saved as points
 528             $sHousenumbers = "";
 529             $i = 0;
 530             $length = count($aPlaceIDs);
 531             foreach ($aPlaceIDs as $placeID => $housenumber) {
 532                 $i++;
 533                 $sHousenumbers .= "(".$placeID.", ".$housenumber.")";
 534                 if ($i<$length) $sHousenumbers .= ", ";
 535             }
 536
 537             if (CONST_Use_US_Tiger_Data) {
 538                 // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 539                 $sSQL .= " union";
 540                 $sSQL .= " SELECT ";
 541                 $sSQL .= "     'T' AS osm_type, ";
 542                 $sSQL .= "     (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
 543                 $sSQL .= "     'place' AS class, ";
 544                 $sSQL .= "     'house' AS type, ";
 545                 $sSQL .= "     null AS admin_level, ";
 546                 $sSQL .= "     30 AS rank_search, ";
 547                 $sSQL .= "     30 AS rank_address, ";
 548                 $sSQL .= "     min(place_id) AS place_id, ";
 549                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 550                 $sSQL .= "     'us' AS country_code, ";
 551                 $sSQL .= "     get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
 552                 $sSQL .= "     null AS placename, ";
 553                 $sSQL .= "     null AS ref, ";
 554                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 555                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 556                 $sSQL .= "     avg(st_x(centroid)) AS lon, ";
 557                 $sSQL .= "     avg(st_y(centroid)) AS lat,";
 558                 $sSQL .= "     ".$sImportanceSQL."-1.15 AS importance, ";
 559                 $sSQL .= "     (";
 560                 $sSQL .= "        SELECT max(p.importance*(p.rank_address+2))";
 561                 $sSQL .= "        FROM ";
 562                 $sSQL .= "          place_addressline s, ";
 563                 $sSQL .= "          placex p";
 564                 $sSQL .= "        WHERE s.place_id = min(blub.parent_place_id)";
 565                 $sSQL .= "          AND p.place_id = s.address_place_id ";
 566                 $sSQL .= "          AND s.isaddress";
 567                 $sSQL .= "          AND p.importance is not null";
 568                 $sSQL .= "     ) AS addressimportance, ";
 569                 $sSQL .= "     null AS extra_place ";
 570                 $sSQL .= " FROM (";
 571                 $sSQL .= "     SELECT place_id, ";    // interpolate the Tiger housenumbers here
 572                 $sSQL .= "         ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
 573                 $sSQL .= "         parent_place_id, ";
 574                 $sSQL .= "         housenumber_for_place";
 575                 $sSQL .= "     FROM (";
 576                 $sSQL .= "            location_property_tiger ";
 577                 $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
 578                 $sSQL .= "     WHERE ";
 579                 $sSQL .= "         housenumber_for_place>=0";
 580                 $sSQL .= "         AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 581                 $sSQL .= " ) AS blub"; //postgres wants an alias here
 582                 $sSQL .= " GROUP BY";
 583                 $sSQL .= "      place_id, ";
 584                 $sSQL .= "      housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
 585                 if (!$this->bDeDupe) $sSQL .= ", place_id ";
 586             }
 587             // osmline
 588             // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 589             $sSQL .= " UNION ";
 590             $sSQL .= "SELECT ";
 591             $sSQL .= "  'W' AS osm_type, ";
 592             $sSQL .= "  osm_id, ";
 593             $sSQL .= "  'place' AS class, ";
 594             $sSQL .= "  'house' AS type, ";
 595             $sSQL .= "  null AS admin_level, ";
 596             $sSQL .= "  30 AS rank_search, ";
 597             $sSQL .= "  30 AS rank_address, ";
 598             $sSQL .= "  min(place_id) as place_id, ";
 599             $sSQL .= "  min(parent_place_id) AS parent_place_id, ";
 600             $sSQL .= "  country_code, ";
 601             $sSQL .= "  get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
 602             $sSQL .= "  null AS placename, ";
 603             $sSQL .= "  null AS ref, ";
 604             if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 605             if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 606             $sSQL .= "  AVG(st_x(centroid)) AS lon, ";
 607             $sSQL .= "  AVG(st_y(centroid)) AS lat, ";
 608             $sSQL .= "  ".$sImportanceSQL."-0.1 AS importance, ";  // slightly smaller than the importance for normal houses with rank 30, which is 0
 609             $sSQL .= "  (";
 610             $sSQL .= "     SELECT ";
 611             $sSQL .= "       MAX(p.importance*(p.rank_address+2)) ";
 612             $sSQL .= "     FROM";
 613             $sSQL .= "       place_addressline s, ";
 614             $sSQL .= "       placex p";
 615             $sSQL .= "     WHERE s.place_id = min(blub.parent_place_id) ";
 616             $sSQL .= "       AND p.place_id = s.address_place_id ";
 617             $sSQL .= "       AND s.isaddress ";
 618             $sSQL .= "       AND p.importance is not null";
 619             $sSQL .= "  ) AS addressimportance,";
 620             $sSQL .= "  null AS extra_place ";
 621             $sSQL .= "  FROM (";
 622             $sSQL .= "     SELECT ";
 623             $sSQL .= "         osm_id, ";
 624             $sSQL .= "         place_id, ";
 625             $sSQL .= "         country_code, ";
 626             $sSQL .= "         CASE ";             // interpolate the housenumbers here
 627             $sSQL .= "           WHEN startnumber != endnumber ";
 628             $sSQL .= "           THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
 629             $sSQL .= "           ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
 630             $sSQL .= "         END as centroid, ";
 631             $sSQL .= "         parent_place_id, ";
 632             $sSQL .= "         housenumber_for_place ";
 633             $sSQL .= "     FROM (";
 634             $sSQL .= "            location_property_osmline ";
 635             $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
 636             $sSQL .= "          ) ";
 637             $sSQL .= "     WHERE housenumber_for_place>=0 ";
 638             $sSQL .= "       AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 639             $sSQL .= "  ) as blub"; //postgres wants an alias here
 640             $sSQL .= "  GROUP BY ";
 641             $sSQL .= "    osm_id, ";
 642             $sSQL .= "    place_id, ";
 643             $sSQL .= "    housenumber_for_place, ";
 644             $sSQL .= "    country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
 645             if (!$this->bDeDupe) $sSQL .= ", place_id ";
 646
 647             if (CONST_Use_Aux_Location_data) {
 648                 $sSQL .= " UNION ";
 649                 $sSQL .= "  SELECT ";
 650                 $sSQL .= "     'L' AS osm_type, ";
 651                 $sSQL .= "     place_id AS osm_id, ";
 652                 $sSQL .= "     'place' AS class,";
 653                 $sSQL .= "     'house' AS type, ";
 654                 $sSQL .= "     null AS admin_level, ";
 655                 $sSQL .= "     0 AS rank_search,";
 656                 $sSQL .= "     0 AS rank_address, ";
 657                 $sSQL .= "     min(place_id) AS place_id,";
 658                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 659                 $sSQL .= "     'us' AS country_code, ";
 660                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
 661                 $sSQL .= "     null AS placename, ";
 662                 $sSQL .= "     null AS ref, ";
 663                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 664                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 665                 $sSQL .= "     avg(ST_X(centroid)) AS lon, ";
 666                 $sSQL .= "     avg(ST_Y(centroid)) AS lat, ";
 667                 $sSQL .= "     ".$sImportanceSQL."-1.10 AS importance, ";
 668                 $sSQL .= "     ( ";
 669                 $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 670                 $sSQL .= "       FROM ";
 671                 $sSQL .= "          place_addressline s, ";
 672                 $sSQL .= "          placex p";
 673                 $sSQL .= "       WHERE s.place_id = min(location_property_aux.parent_place_id)";
 674                 $sSQL .= "         AND p.place_id = s.address_place_id ";
 675                 $sSQL .= "         AND s.isaddress";
 676                 $sSQL .= "         AND p.importance is not null";
 677                 $sSQL .= "     ) AS addressimportance, ";
 678                 $sSQL .= "     null AS extra_place ";
 679                 $sSQL .= "  FROM location_property_aux ";
 680                 $sSQL .= "  WHERE place_id in ($sPlaceIDs) ";
 681                 $sSQL .= "    AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
 682                 $sSQL .= "  GROUP BY ";
 683                 $sSQL .= "     place_id, ";
 684                 if (!$this->bDeDupe) $sSQL .= "place_id, ";
 685                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
 686             }
 687         }
 688
 689         $sSQL .= " order by importance desc";
 690         if (CONST_Debug) {
 691             echo "<hr>";
 692             var_dump($sSQL);
 693         }
 694         $aSearchResults = chksql(
 695             $this->oDB->getAll($sSQL),
 696             "Could not get details for place."
 697         );
 698
 699         return $aSearchResults;
 700     }
 701
 702     public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
 703     {
 704         /*
 705              Calculate all searches using aValidTokens i.e.
 706              'Wodsworth Road, Sheffield' =>
 707
 708              Phrase Wordset
 709              0      0       (wodsworth road)
 710              0      1       (wodsworth)(road)
 711              1      0       (sheffield)
 712
 713              Score how good the search is so they can be ordered
 714          */
 715         $iGlobalRank = 0;
 716
 717         foreach ($aPhrases as $iPhrase => $aPhrase) {
 718             $aNewPhraseSearches = array();
 719             if ($bStructuredPhrases) {
 720                 $sPhraseType = $aPhraseTypes[$iPhrase];
 721             } else {
 722                 $sPhraseType = '';
 723             }
 724
 725             foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) {
 726                 // Too many permutations - too expensive
 727                 if ($iWordSet > 120) break;
 728
 729                 $aWordsetSearches = $aSearches;
 730
 731                 // Add all words from this wordset
 732                 foreach ($aWordset as $iToken => $sToken) {
 733                     //echo "<br><b>$sToken</b>";
 734                     $aNewWordsetSearches = array();
 735
 736                     foreach ($aWordsetSearches as $oCurrentSearch) {
 737                         //echo "<i>";
 738                         //var_dump($oCurrentSearch);
 739                         //echo "</i>";
 740
 741                         // If the token is valid
 742                         if (isset($aValidTokens[' '.$sToken])) {
 743                             // Recheck if the original word shows up in the query.
 744                             $bWordInQuery = false;
 745                             if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
 746                                 $bWordInQuery = $this->normTerm($aSearchTerm['word'])) !== false;
 747                             }
 748                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 749                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 750                                     $aSearchTerm,
 751                                     $bWordInQuery,
 752                                     isset($aValidTokens[$sToken])
 753                                       && strpos($sToken, ' ') === false,
 754                                     $sPhraseType,
 755                                     $iToken == 0 && $iPhrase == 0,
 756                                     $iPhrase == 0,
 757                                     $iToken + 1 == sizeof($aWordset)
 758                                       && $iPhrase + 1 == sizeof($aPhrases),
 759                                     $iGlobalRank
 760                                 );
 761
 762                                 foreach ($aNewSearches as $oSearch) {
 763                                     if ($oSearch->getRank() < $this->iMaxRank) {
 764                                         $aNewWordsetSearches[] = $oSearch;
 765                                     }
 766                                 }
 767                             }
 768                         }
 769                         // Look for partial matches.
 770                         // Note that there is no point in adding country terms here
 771                         // because country is omitted in the address.
 772                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 773                             // Allow searching for a word - but at extra cost
 774                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 775                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 776                                     $aSearchTerm,
 777                                     $bStructuredPhrases,
 778                                     $iPhrase,
 779                                     $aWordFrequencyScores,
 780                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 781                                 );
 782
 783                                 foreach ($aNewSearches as $oSearch) {
 784                                     if ($oSearch->getRank() < $this->iMaxRank) {
 785                                         $aNewWordsetSearches[] = $oSearch;
 786                                     }
 787                                 }
 788
 789                             }
 790                         }
 791                     }
 792                     // Sort and cut
 793                     usort($aNewWordsetSearches, 'bySearchRank');
 794                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 795                 }
 796                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 797
 798                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 799                 usort($aNewPhraseSearches, 'bySearchRank');
 800
 801                 $aSearchHash = array();
 802                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 803                     $sHash = serialize($aSearch);
 804                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 805                     else $aSearchHash[$sHash] = 1;
 806                 }
 807
 808                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 809             }
 810
 811             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 812             $aGroupedSearches = array();
 813             foreach ($aNewPhraseSearches as $aSearch) {
 814                 $iRank = $aSearch->getRank();
 815                 if ($iRank < $this->iMaxRank) {
 816                     if (!isset($aGroupedSearches[$iRank])) {
 817                         $aGroupedSearches[$iRank] = array();
 818                     }
 819                     $aGroupedSearches[$iRank][] = $aSearch;
 820                 }
 821             }
 822             ksort($aGroupedSearches);
 823
 824             $iSearchCount = 0;
 825             $aSearches = array();
 826             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 827                 $iSearchCount += sizeof($aNewSearches);
 828                 $aSearches = array_merge($aSearches, $aNewSearches);
 829                 if ($iSearchCount > 50) break;
 830             }
 831
 832             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 833         }
 834
 835         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 836         $aGroupedSearches = array();
 837         foreach ($aSearches as $oSearch) {
 838             if (!$oSearch->isValidSearch()) {
 839                 continue;
 840             }
 841
 842             $iRank = $oSearch->addToRank($iGlobalRank);
 843             if (!isset($aGroupedSearches[$iRank]) {
 844                 $aGroupedSearches[$iRank] = array();
 845             }
 846             $aGroupedSearches[$iRank][] = $oSearch;
 847         }
 848         ksort($aGroupedSearches);
 849
 850         return $aGroupedSearches;
 851     }
 852
 853     /* Perform the actual query lookup.
 854
 855         Returns an ordered list of results, each with the following fields:
 856             osm_type: type of corresponding OSM object
 857                         N - node
 858                         W - way
 859                         R - relation
 860                         P - postcode (internally computed)
 861             osm_id: id of corresponding OSM object
 862             class: general object class (corresponds to tag key of primary OSM tag)
 863             type: subclass of object (corresponds to tag value of primary OSM tag)
 864             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 865             rank_search: rank in search hierarchy
 866                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 867             rank_address: rank in address hierarchy (determines orer in address)
 868             place_id: internal key (may differ between different instances)
 869             country_code: ISO country code
 870             langaddress: localized full address
 871             placename: localized name of object
 872             ref: content of ref tag (if available)
 873             lon: longitude
 874             lat: latitude
 875             importance: importance of place based on Wikipedia link count
 876             addressimportance: cumulated importance of address elements
 877             extra_place: type of place (for admin boundaries, if there is a place tag)
 878             aBoundingBox: bounding Box
 879             label: short description of the object class/type (English only)
 880             name: full name (currently the same as langaddress)
 881             foundorder: secondary ordering for places with same importance
 882     */
 883
 884
 885     public function lookup()
 886     {
 887         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 888
 889         $sNormQuery = $this->normTerm($this->sQuery);
 890         $sLanguagePrefArraySQL = getArraySQL(
 891             array_map("getDBQuoted",
 892             $this->aLangPrefOrder)
 893         );
 894         $sCountryCodesSQL = false;
 895         if ($this->aCountryCodes) {
 896             $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes));
 897         }
 898
 899         $sQuery = $this->sQuery;
 900         if (!preg_match('//u', $sQuery)) {
 901             userError("Query string is not UTF-8 encoded.");
 902         }
 903
 904         // Conflicts between US state abreviations and various words for 'the' in different languages
 905         if (isset($this->aLangPrefOrder['name:en'])) {
 906             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 907             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 908             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 909         }
 910
 911         $bBoundingBoxSearch = $this->bBoundedSearch && $this->sViewboxSmallSQL;
 912         if ($this->sViewboxCentreSQL) {
 913             // For complex viewboxes (routes) precompute the bounding geometry
 914             $sGeom = chksql(
 915                 $this->oDB->getOne("select ".$this->sViewboxSmallSQL),
 916                 "Could not get small viewbox"
 917             );
 918             $this->sViewboxSmallSQL = "'".$sGeom."'::geometry";
 919
 920             $sGeom = chksql(
 921                 $this->oDB->getOne("select ".$this->sViewboxLargeSQL),
 922                 "Could not get large viewbox"
 923             );
 924             $this->sViewboxLargeSQL = "'".$sGeom."'::geometry";
 925         }
 926
 927         // Do we have anything that looks like a lat/lon pair?
 928         $oNearPoint = false;
 929         if ($aLooksLike = NearPoint::extractFromQuery($sQuery)) {
 930             $oNearPoint = $aLooksLike['pt'];
 931             $sQuery = $aLooksLike['query'];
 932         }
 933
 934         $aSearchResults = array();
 935         if ($sQuery || $this->aStructuredQuery) {
 936             // Start with a single blank search
 937             $aSearches = array(new SearchDescription());
 938
 939             if ($oNearPoint) {
 940                 $aSearches[0]->setNear($oNearPoint);
 941             }
 942
 943             if ($sQuery) {
 944                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 945             }
 946
 947             $sSpecialTerm = '';
 948             if ($sQuery) {
 949                 preg_match_all(
 950                     '/\\[([\\w ]*)\\]/u',
 951                     $sQuery,
 952                     $aSpecialTermsRaw,
 953                     PREG_SET_ORDER
 954                 );
 955                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 956                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 957                     if (!$sSpecialTerm) {
 958                         $sSpecialTerm = $aSpecialTerm[1];
 959                     }
 960                 }
 961             }
 962             if (!$sSpecialTerm && $this->aStructuredQuery
 963                 && isset($this->aStructuredQuery['amenity'])) {
 964                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 965                 unset($this->aStructuredQuery['amenity']);
 966             }
 967
 968             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 969                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 970                 $sToken = chksql(
 971                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 972                     "Cannot decode query. Wrong encoding?"
 973                 );
 974                 $sSQL = 'SELECT class, type FROM word ';
 975                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 976                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 977                 if (CONST_Debug) var_Dump($sSQL);
 978                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 979                 $aNewSearches = array();
 980                 foreach ($aSearches as $oSearch) {
 981                     foreach ($aSearchWords as $aSearchTerm) {
 982                         $oNewSearch = clone $oSearch;
 983                         $oNewSearch->setPoiSearch(
 984                             Operator::TYPE,
 985                             $aSearchTerm['class'],
 986                             $aSearchTerm['type'],
 987                         );
 988                         $aNewSearches[] = $oNewSearch;
 989                     }
 990                 }
 991                 $aSearches = $aNewSearches;
 992             }
 993
 994             // Split query into phrases
 995             // Commas are used to reduce the search space by indicating where phrases split
 996             if ($this->aStructuredQuery) {
 997                 $aPhrases = $this->aStructuredQuery;
 998                 $bStructuredPhrases = true;
 999             } else {
1000                 $aPhrases = explode(',', $sQuery);
1001                 $bStructuredPhrases = false;
1002             }
1003
1004             // Convert each phrase to standard form
1005             // Create a list of standard words
1006             // Get all 'sets' of words
1007             // Generate a complete list of all
1008             $aTokens = array();
1009             foreach ($aPhrases as $iPhrase => $sPhrase) {
1010                 $aPhrase = chksql(
1011                     $this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"),
1012                     "Cannot normalize query string (is it a UTF-8 string?)"
1013                 );
1014                 if (trim($aPhrase['string'])) {
1015                     $aPhrases[$iPhrase] = $aPhrase;
1016                     $aPhrases[$iPhrase]['words'] = explode(' ', $aPhrases[$iPhrase]['string']);
1017                     $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0);
1018                     $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets']));
1019                 } else {
1020                     unset($aPhrases[$iPhrase]);
1021                 }
1022             }
1023
1024             // Reindex phrases - we make assumptions later on that they are numerically keyed in order
1025             $aPhraseTypes = array_keys($aPhrases);
1026             $aPhrases = array_values($aPhrases);
1027
1028             if (sizeof($aTokens)) {
1029                 // Check which tokens we have, get the ID numbers
1030                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
1031                 $sSQL .= ' FROM word ';
1032                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
1033
1034                 if (CONST_Debug) var_Dump($sSQL);
1035
1036                 $aValidTokens = array();
1037                 $aDatabaseWords = chksql(
1038                     $this->oDB->getAll($sSQL),
1039                     "Could not get word tokens."
1040                 );
1041                 $aPossibleMainWordIDs = array();
1042                 $aWordFrequencyScores = array();
1043                 foreach ($aDatabaseWords as $aToken) {
1044                     // Very special case - require 2 letter country param to match the country code found
1045                     if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
1046                         && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
1047                     ) {
1048                         continue;
1049                     }
1050
1051                     if (isset($aValidTokens[$aToken['word_token']])) {
1052                         $aValidTokens[$aToken['word_token']][] = $aToken;
1053                     } else {
1054                         $aValidTokens[$aToken['word_token']] = array($aToken);
1055                     }
1056                     if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
1057                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
1058                 }
1059                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
1060
1061                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
1062                 foreach ($aTokens as $sToken) {
1063                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
1064                         if (isset($aValidTokens[$aData[1]])) {
1065                             foreach ($aValidTokens[$aData[1]] as $aToken) {
1066                                 if (!$aToken['class']) {
1067                                     if (isset($aValidTokens[$sToken])) {
1068                                         $aValidTokens[$sToken][] = $aToken;
1069                                     } else {
1070                                         $aValidTokens[$sToken] = array($aToken);
1071                                     }
1072                                 }
1073                             }
1074                         }
1075                     }
1076                 }
1077
1078                 foreach ($aTokens as $sToken) {
1079                     // Unknown single word token with a number - assume it is a house number
1080                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/[0-9]/', $sToken)) {
1081                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house'));
1082                     }
1083                 }
1084
1085                 // Any words that have failed completely?
1086                 // TODO: suggestions
1087
1088                 // Start the search process
1089                 // array with: placeid => -1 | tiger-housenumber
1090                 $aResultPlaceIDs = array();
1091
1092                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
1093
1094                 if ($this->bReverseInPlan) {
1095                     // Reverse phrase array and also reverse the order of the wordsets in
1096                     // the first and final phrase. Don't bother about phrases in the middle
1097                     // because order in the address doesn't matter.
1098                     $aPhrases = array_reverse($aPhrases);
1099                     $aPhrases[0]['wordsets'] = getInverseWordSets($aPhrases[0]['words'], 0);
1100                     if (sizeof($aPhrases) > 1) {
1101                         $aFinalPhrase = end($aPhrases);
1102                         $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
1103                     }
1104                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
1105
1106                     foreach ($aGroupedSearches as $aSearches) {
1107                         foreach ($aSearches as $aSearch) {
1108                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
1109                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
1110                             }
1111                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
1112                         }
1113                     }
1114
1115                     $aGroupedSearches = $aReverseGroupedSearches;
1116                     ksort($aGroupedSearches);
1117                 }
1118             } else {
1119                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
1120                 $aGroupedSearches = array();
1121                 foreach ($aSearches as $aSearch) {
1122                     if ($aSearch->getRank() < $this->iMaxRank) {
1123                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
1124                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
1125                     }
1126                 }
1127                 ksort($aGroupedSearches);
1128             }
1129
1130             // Filter out duplicate searches
1131             $aSearchHash = array();
1132             foreach ($aGroupedSearches as $iGroup => $aSearches) {
1133                 foreach ($aSearches as $iSearch => $aSearch) {
1134                     $sHash = serialize($aSearch);
1135                     if (isset($aSearchHash[$sHash])) {
1136                         unset($aGroupedSearches[$iGroup][$iSearch]);
1137                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
1138                     } else {
1139                         $aSearchHash[$sHash] = 1;
1140                     }
1141                 }
1142             }
1143
1144             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
1145
1146             $iGroupLoop = 0;
1147             $iQueryLoop = 0;
1148             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
1149                 $iGroupLoop++;
1150                 foreach ($aSearches as $oSearch) {
1151                     $iQueryLoop++;
1152                     $searchedHousenumber = -1;
1153
1154                     if (CONST_Debug) echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
1155                     if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
1156
1157                     $aPlaceIDs = array();
1158                     if ($oSearch->isCountrySearch()) {
1159                         // Just looking for a country - look it up
1160                         if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
1161                             $aPlaceIDs = $oSearch->queryCountry(
1162                                 $this->oDB,
1163                                 $bBoundingBoxSearch ? $this->sViewboxSmallSQL : ''
1164                             );
1165                         }
1166                     } elseif (!$oSearch->isNamedSearch()) {
1167                         // looking for a POI in a geographic area
1168                         if (!$bBoundingBoxSearch && !$oSearch->isNearSearch()) {
1169                             continue;
1170                         }
1171
1172                         $aPlaceIDs = $oSearch->queryNearbyPoi(
1173                             $this->oDB,
1174                             $sCountryCodesSQL,
1175                             $bBoundingBoxSearch ? $this->sViewboxSmallSQL : '',
1176                             $sViewboxCentreSQL,
1177                             $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1178                             $this->iLimit
1179                         );
1180                     } elseif ($oSearch->isOperator(Operator::POSTCODE)) {
1181                         $aPlaceIDs = $oSearch->queryPostcode(
1182                             $oDB,
1183                             $sCountryCodesSQL,
1184                             $this->iLimit
1185                         );
1186                     } else {
1187                         // Ordinary search:
1188                         // First search for places according to name and address.
1189                         $aNamedPlaceIDs = $oSearch->queryNamedPlace(
1190                             $this->oDB,
1191                             $aWordFrequencyScores,
1192                             $sCountryCodesSQL,
1193                             $this->iMinAddressRank,
1194                             $this->iMaxAddressRank,
1195                             $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1196                             $bBoundingBoxSearch ? $this->sViewboxSmallSQL : '',
1197                             $bBoundingBoxSearch ? $this->sViewboxLargeSQL : '',
1198                             $this->iLimit
1199                         );
1200
1201                         if (sizeof($aNamedPlaceIDs)) {
1202                             foreach ($aNamedPlaceIDs as $aRow) {
1203                                 $aPlaceIDs[] = $aRow['place_id'];
1204                                 $this->exactMatchCache[$aRow['place_id']] = $aRow['exactmatch'];
1205                             }
1206                         }
1207
1208                         //now search for housenumber, if housenumber provided
1209                         if ($oSearch->hasHouseNumber() && sizeof($aPlaceIDs)) {
1210                             $aResult = $oSearch->queryHouseNumber(
1211                                 $this->oDB,
1212                                 $aPlaceIDs,
1213                                 $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : ''
1214                                 $this->iLimit
1215                             );
1216
1217                             if (sizeof($aResult)) {
1218                                 $searchedHousenumber = $aResult['iHouseNumber'];
1219                                 $aPlaceIDs = $aResults['aPlaceIDs'];
1220                             } elseif (!$oSearch->looksLikeFullAddress()) {
1221                                 $aPlaceIDs = array();
1222                             }
1223                         }
1224
1225                         // finally get POIs if requested
1226                         if ($oSearch->isPoiSearch() && sizeof($aPlaceIDs)) {
1227                             $aPlaceIDs = $oSearch->queryPoiByOperator(
1228                                 $this->oDB,
1229                                 $aPlaceIDs,
1230                                 $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : ''
1231                                 $this->iLimit
1232                             );
1233                         }
1234                     }
1235
1236                     if (CONST_Debug) {
1237                         echo "<br><b>Place IDs:</b> ";
1238                         var_Dump($aPlaceIDs);
1239                     }
1240
1241                     if (sizeof($aPlaceIDs) && $oSearch->getPostcode()) {
1242                         $sSQL = 'SELECT place_id FROM placex';
1243                         $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
1244                         $sSQL .= " AND postcode = '".$oSearch->getPostcode()."'";
1245                         if (CONST_Debug) var_dump($sSQL);
1246                         $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1247                         if ($aFilteredPlaceIDs) {
1248                             $aPlaceIDs = $aFilteredPlaceIDs;
1249                             if (CONST_Debug) {
1250                                 echo "<br><b>Place IDs after postcode filtering:</b> ";
1251                                 var_Dump($aPlaceIDs);
1252                             }
1253                         }
1254                     }
1255
1256                     foreach ($aPlaceIDs as $iPlaceID) {
1257                         // array for placeID => -1 | Tiger housenumber
1258                         $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
1259                     }
1260                     if ($iQueryLoop > 20) break;
1261                 }
1262
1263                 if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
1264                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
1265                     // reduces the number of place ids, like a filter
1266                     // rank_address is 30 for interpolated housenumbers
1267                     $sWherePlaceId = 'WHERE place_id in (';
1268                     $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
1269
1270                     $sSQL = "SELECT place_id ";
1271                     $sSQL .= "FROM placex ".$sWherePlaceId;
1272                     $sSQL .= "  AND (";
1273                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1274                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
1275                         $sSQL .= "     OR (extratags->'place') = 'city'";
1276                     }
1277                     if ($this->aAddressRankList) {
1278                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
1279                     }
1280                     $sSQL .= "  ) UNION ";
1281                     $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
1282                     $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1283                     if ($this->aAddressRankList) {
1284                         $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
1285                     }
1286                     $sSQL .= ") ";
1287                     if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
1288                         $sSQL .= "UNION ";
1289                         $sSQL .= "  SELECT place_id ";
1290                         $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
1291                     }
1292                     if ($this->iMaxAddressRank == 30) {
1293                         $sSQL .= "UNION ";
1294                         $sSQL .= "  SELECT place_id ";
1295                         $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;
1296                     }
1297                     if (CONST_Debug) var_dump($sSQL);
1298                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1299                     $tempIDs = array();
1300                     foreach ($aFilteredPlaceIDs as $placeID) {
1301                         $tempIDs[$placeID] = $aResultPlaceIDs[$placeID];  //assign housenumber to placeID
1302                     }
1303                     $aResultPlaceIDs = $tempIDs;
1304                 }
1305
1306                 //exit;
1307                 if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break;
1308                 if ($iGroupLoop > 4) break;
1309                 if ($iQueryLoop > 30) break;
1310             }
1311
1312             // Did we find anything?
1313             if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) {
1314                 $aSearchResults = $this->getDetails($aResultPlaceIDs);
1315             }
1316         } else {
1317             // Just interpret as a reverse geocode
1318             $oReverse = new ReverseGeocode($this->oDB);
1319             $oReverse->setZoom(18);
1320
1321             $aLookup = $oReverse->lookup(
1322                 $oNearPoint->lat(),
1323                 $oNearPoint->lon(),
1324                 false
1325             );
1326
1327             if (CONST_Debug) var_dump("Reverse search", $aLookup);
1328
1329             if ($aLookup['place_id']) {
1330                 $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1));
1331                 $aResultPlaceIDs[$aLookup['place_id']] = -1;
1332             } else {
1333                 $aSearchResults = array();
1334             }
1335         }
1336
1337         // No results? Done
1338         if (!sizeof($aSearchResults)) {
1339             if ($this->bFallback) {
1340                 if ($this->fallbackStructuredQuery()) {
1341                     return $this->lookup();
1342                 }
1343             }
1344
1345             return array();
1346         }
1347
1348         $aClassType = getClassTypesWithImportance();
1349         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
1350         foreach ($aRecheckWords as $i => $sWord) {
1351             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
1352         }
1353
1354         if (CONST_Debug) {
1355             echo '<i>Recheck words:<\i>';
1356             var_dump($aRecheckWords);
1357         }
1358
1359         $oPlaceLookup = new PlaceLookup($this->oDB);
1360         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
1361         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
1362         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
1363         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
1364         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
1365         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
1366
1367         foreach ($aSearchResults as $iResNum => $aResult) {
1368             // Default
1369             $fDiameter = getResultDiameter($aResult);
1370
1371             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
1372             if ($aOutlineResult) {
1373                 $aResult = array_merge($aResult, $aOutlineResult);
1374             }
1375
1376             if ($aResult['extra_place'] == 'city') {
1377                 $aResult['class'] = 'place';
1378                 $aResult['type'] = 'city';
1379                 $aResult['rank_search'] = 16;
1380             }
1381
1382             // Is there an icon set for this type of result?
1383             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
1384                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
1385             ) {
1386                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
1387             }
1388
1389             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
1390                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
1391             ) {
1392                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
1393             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
1394                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
1395             ) {
1396                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
1397             }
1398             // if tag '&addressdetails=1' is set in query
1399             if ($this->bIncludeAddressDetails) {
1400                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
1401                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);
1402                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
1403                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
1404                 }
1405             }
1406
1407             if ($this->bIncludeExtraTags) {
1408                 if ($aResult['extra']) {
1409                     $aResult['sExtraTags'] = json_decode($aResult['extra']);
1410                 } else {
1411                     $aResult['sExtraTags'] = (object) array();
1412                 }
1413             }
1414
1415             if ($this->bIncludeNameDetails) {
1416                 if ($aResult['names']) {
1417                     $aResult['sNameDetails'] = json_decode($aResult['names']);
1418                 } else {
1419                     $aResult['sNameDetails'] = (object) array();
1420                 }
1421             }
1422
1423             // Adjust importance for the number of exact string matches in the result
1424             $aResult['importance'] = max(0.001, $aResult['importance']);
1425             $iCountWords = 0;
1426             $sAddress = $aResult['langaddress'];
1427             foreach ($aRecheckWords as $i => $sWord) {
1428                 if (stripos($sAddress, $sWord)!==false) {
1429                     $iCountWords++;
1430                     if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
1431                 }
1432             }
1433
1434             $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
1435
1436             $aResult['name'] = $aResult['langaddress'];
1437             // secondary ordering (for results with same importance (the smaller the better):
1438             // - approximate importance of address parts
1439             $aResult['foundorder'] = -$aResult['addressimportance']/10;
1440             // - number of exact matches from the query
1441             if (isset($this->exactMatchCache[$aResult['place_id']])) {
1442                 $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']];
1443             } elseif (isset($this->exactMatchCache[$aResult['parent_place_id']])) {
1444                 $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']];
1445             }
1446             // - importance of the class/type
1447             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1448                 && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1449             ) {
1450                 $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1451             } else {
1452                 $aResult['foundorder'] += 0.01;
1453             }
1454             if (CONST_Debug) var_dump($aResult);
1455             $aSearchResults[$iResNum] = $aResult;
1456         }
1457         uasort($aSearchResults, 'byImportance');
1458
1459         $aOSMIDDone = array();
1460         $aClassTypeNameDone = array();
1461         $aToFilter = $aSearchResults;
1462         $aSearchResults = array();
1463
1464         $bFirst = true;
1465         foreach ($aToFilter as $iResNum => $aResult) {
1466             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1467             if ($bFirst) {
1468                 $fLat = $aResult['lat'];
1469                 $fLon = $aResult['lon'];
1470                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1471                 $bFirst = false;
1472             }
1473             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1474                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1475             ) {
1476                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1477                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1478                 $aSearchResults[] = $aResult;
1479             }
1480
1481             // Absolute limit on number of results
1482             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1483         }
1484
1485         return $aSearchResults;
1486     } // end lookup()
1487 } // end class