lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   7 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   8 require_once(CONST_BasePath.'/lib/SearchContext.php');
   9
  10 class Geocode
  11 {
  12     protected $oDB;
  13
  14     protected $aLangPrefOrder = array();
  15
  16     protected $bIncludeAddressDetails = false;
  17     protected $bIncludeExtraTags = false;
  18     protected $bIncludeNameDetails = false;
  19
  20     protected $bIncludePolygonAsPoints = false;
  21     protected $bIncludePolygonAsText = false;
  22     protected $bIncludePolygonAsGeoJSON = false;
  23     protected $bIncludePolygonAsKML = false;
  24     protected $bIncludePolygonAsSVG = false;
  25     protected $fPolygonSimplificationThreshold = 0.0;
  26
  27     protected $aExcludePlaceIDs = array();
  28     protected $bDeDupe = true;
  29     protected $bReverseInPlan = false;
  30
  31     protected $iLimit = 20;
  32     protected $iFinalLimit = 10;
  33     protected $iOffset = 0;
  34     protected $bFallback = false;
  35
  36     protected $aCountryCodes = false;
  37
  38     protected $bBoundedSearch = false;
  39     protected $aViewBox = false;
  40     protected $aRoutePoints = false;
  41     protected $aRouteWidth = 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 = $iMax;
 184     }
 185
 186     public function setViewbox($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($aPlaceIDs, $oCtx)
 368     {
 369         //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
 370         if (sizeof($aPlaceIDs) == 0) return array();
 371
 372         $sLanguagePrefArraySQL = getArraySQL(
 373             array_map("getDBQuoted", $this->aLangPrefOrder)
 374         );
 375
 376         // Get the details for display (is this a redundant extra step?)
 377         $sPlaceIDs = join(',', array_keys($aPlaceIDs));
 378
 379         $sImportanceSQL = $oCtx->viewboxImportanceSQL('ST_Collect(centroid)');
 380         $sImportanceSQLGeom = $oCtx->viewboxImportanceSQL('geometry');
 381
 382         $sSQL  = "SELECT ";
 383         $sSQL .= "    osm_type,";
 384         $sSQL .= "    osm_id,";
 385         $sSQL .= "    class,";
 386         $sSQL .= "    type,";
 387         $sSQL .= "    admin_level,";
 388         $sSQL .= "    rank_search,";
 389         $sSQL .= "    rank_address,";
 390         $sSQL .= "    min(place_id) AS place_id, ";
 391         $sSQL .= "    min(parent_place_id) AS parent_place_id, ";
 392         $sSQL .= "    country_code, ";
 393         $sSQL .= "    get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 394         $sSQL .= "    get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,";
 395         $sSQL .= "    get_name_by_language(name, ARRAY['ref']) AS ref,";
 396         if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,";
 397         if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,";
 398         $sSQL .= "    avg(ST_X(centroid)) AS lon, ";
 399         $sSQL .= "    avg(ST_Y(centroid)) AS lat, ";
 400         $sSQL .= "    COALESCE(importance,0.75-(rank_search::float/40)) $sImportanceSQL AS importance, ";
 401         if ($oCtx->hasNearPoint()) {
 402             $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 403         } else {
 404             $sSQL .= "    ( ";
 405             $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 406             $sSQL .= "       FROM ";
 407             $sSQL .= "         place_addressline s, ";
 408             $sSQL .= "         placex p";
 409             $sSQL .= "       WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
 410             $sSQL .= "         AND p.place_id = s.address_place_id ";
 411             $sSQL .= "         AND s.isaddress ";
 412             $sSQL .= "         AND p.importance is not null ";
 413             $sSQL .= "    ) AS addressimportance, ";
 414         }
 415         $sSQL .= "    (extratags->'place') AS extra_place ";
 416         $sSQL .= " FROM placex";
 417         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 418         $sSQL .= "   AND (";
 419         $sSQL .= "            placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 420         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 421             $sSQL .= "        OR (extratags->'place') = 'city'";
 422         }
 423         if ($this->aAddressRankList) {
 424             $sSQL .= "        OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 425         }
 426         $sSQL .= "       ) ";
 427         if ($this->sAllowedTypesSQLList) {
 428             $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
 429         }
 430         $sSQL .= "    AND linked_place_id is null ";
 431         $sSQL .= " GROUP BY ";
 432         $sSQL .= "     osm_type, ";
 433         $sSQL .= "     osm_id, ";
 434         $sSQL .= "     class, ";
 435         $sSQL .= "     type, ";
 436         $sSQL .= "     admin_level, ";
 437         $sSQL .= "     rank_search, ";
 438         $sSQL .= "     rank_address, ";
 439         $sSQL .= "     country_code, ";
 440         $sSQL .= "     importance, ";
 441         if (!$this->bDeDupe) $sSQL .= "place_id,";
 442         $sSQL .= "     langaddress, ";
 443         $sSQL .= "     placename, ";
 444         $sSQL .= "     ref, ";
 445         if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
 446         if ($this->bIncludeNameDetails) $sSQL .= "name, ";
 447         $sSQL .= "     extratags->'place' ";
 448
 449         // postcode table
 450         $sSQL .= "UNION ";
 451         $sSQL .= "SELECT";
 452         $sSQL .= "  'P' as osm_type,";
 453         $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
 454         $sSQL .= "  'place' as class, 'postcode' as type,";
 455         $sSQL .= "  null as admin_level, rank_search, rank_address,";
 456         $sSQL .= "  place_id, parent_place_id, country_code,";
 457         $sSQL .= "  get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 458         $sSQL .= "  postcode as placename,";
 459         $sSQL .= "  postcode as ref,";
 460         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 461         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 462         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
 463         $sSQL .= "  (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, ";
 464         if ($oCtx->hasNearPoint()) {
 465             $sSQL .= $oCtx->distanceSQL('geometry')." AS addressimportance,";
 466         } else {
 467             $sSQL .= "  (";
 468             $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
 469             $sSQL .= "     FROM ";
 470             $sSQL .= "       place_addressline s, ";
 471             $sSQL .= "       placex p";
 472             $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
 473             $sSQL .= "       AND p.place_id = s.address_place_id ";
 474             $sSQL .= "       AND s.isaddress";
 475             $sSQL .= "       AND p.importance is not null";
 476             $sSQL .= "  ) AS addressimportance, ";
 477         }
 478         $sSQL .= "  null AS extra_place ";
 479         $sSQL .= "FROM location_postcode lp";
 480         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 481
 482         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
 483             // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
 484             // with start- and endnumber, the common osm housenumbers are usually saved as points
 485             $sHousenumbers = "";
 486             $i = 0;
 487             $length = count($aPlaceIDs);
 488             foreach ($aPlaceIDs as $placeID => $housenumber) {
 489                 $i++;
 490                 $sHousenumbers .= "(".$placeID.", ".$housenumber.")";
 491                 if ($i<$length) $sHousenumbers .= ", ";
 492             }
 493
 494             if (CONST_Use_US_Tiger_Data) {
 495                 // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 496                 $sSQL .= " union";
 497                 $sSQL .= " SELECT ";
 498                 $sSQL .= "     'T' AS osm_type, ";
 499                 $sSQL .= "     (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
 500                 $sSQL .= "     'place' AS class, ";
 501                 $sSQL .= "     'house' AS type, ";
 502                 $sSQL .= "     null AS admin_level, ";
 503                 $sSQL .= "     30 AS rank_search, ";
 504                 $sSQL .= "     30 AS rank_address, ";
 505                 $sSQL .= "     min(place_id) AS place_id, ";
 506                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 507                 $sSQL .= "     'us' AS country_code, ";
 508                 $sSQL .= "     get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
 509                 $sSQL .= "     null AS placename, ";
 510                 $sSQL .= "     null AS ref, ";
 511                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 512                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 513                 $sSQL .= "     avg(st_x(centroid)) AS lon, ";
 514                 $sSQL .= "     avg(st_y(centroid)) AS lat,";
 515                 $sSQL .= "     -1.15".$sImportanceSQL." AS importance, ";
 516                 if ($oCtx->hasNearPoint()) {
 517                     $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 518                 } else {
 519                     $sSQL .= "     (";
 520                     $sSQL .= "        SELECT max(p.importance*(p.rank_address+2))";
 521                     $sSQL .= "        FROM ";
 522                     $sSQL .= "          place_addressline s, ";
 523                     $sSQL .= "          placex p";
 524                     $sSQL .= "        WHERE s.place_id = min(blub.parent_place_id)";
 525                     $sSQL .= "          AND p.place_id = s.address_place_id ";
 526                     $sSQL .= "          AND s.isaddress";
 527                     $sSQL .= "          AND p.importance is not null";
 528                     $sSQL .= "     ) AS addressimportance, ";
 529                 }
 530                 $sSQL .= "     null AS extra_place ";
 531                 $sSQL .= " FROM (";
 532                 $sSQL .= "     SELECT place_id, ";    // interpolate the Tiger housenumbers here
 533                 $sSQL .= "         ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
 534                 $sSQL .= "         parent_place_id, ";
 535                 $sSQL .= "         housenumber_for_place";
 536                 $sSQL .= "     FROM (";
 537                 $sSQL .= "            location_property_tiger ";
 538                 $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
 539                 $sSQL .= "     WHERE ";
 540                 $sSQL .= "         housenumber_for_place>=0";
 541                 $sSQL .= "         AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 542                 $sSQL .= " ) AS blub"; //postgres wants an alias here
 543                 $sSQL .= " GROUP BY";
 544                 $sSQL .= "      place_id, ";
 545                 $sSQL .= "      housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
 546                 if (!$this->bDeDupe) $sSQL .= ", place_id ";
 547             }
 548             // osmline
 549             // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 550             $sSQL .= " UNION ";
 551             $sSQL .= "SELECT ";
 552             $sSQL .= "  'W' AS osm_type, ";
 553             $sSQL .= "  osm_id, ";
 554             $sSQL .= "  'place' AS class, ";
 555             $sSQL .= "  'house' AS type, ";
 556             $sSQL .= "  null AS admin_level, ";
 557             $sSQL .= "  30 AS rank_search, ";
 558             $sSQL .= "  30 AS rank_address, ";
 559             $sSQL .= "  min(place_id) as place_id, ";
 560             $sSQL .= "  min(parent_place_id) AS parent_place_id, ";
 561             $sSQL .= "  country_code, ";
 562             $sSQL .= "  get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
 563             $sSQL .= "  null AS placename, ";
 564             $sSQL .= "  null AS ref, ";
 565             if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 566             if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 567             $sSQL .= "  AVG(st_x(centroid)) AS lon, ";
 568             $sSQL .= "  AVG(st_y(centroid)) AS lat, ";
 569             $sSQL .= "  -0.1".$sImportanceSQL." AS importance, ";  // slightly smaller than the importance for normal houses with rank 30, which is 0
 570             if ($oCtx->hasNearPoint()) {
 571                 $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 572             } else {
 573                 $sSQL .= "  (";
 574                 $sSQL .= "     SELECT ";
 575                 $sSQL .= "       MAX(p.importance*(p.rank_address+2)) ";
 576                 $sSQL .= "     FROM";
 577                 $sSQL .= "       place_addressline s, ";
 578                 $sSQL .= "       placex p";
 579                 $sSQL .= "     WHERE s.place_id = min(blub.parent_place_id) ";
 580                 $sSQL .= "       AND p.place_id = s.address_place_id ";
 581                 $sSQL .= "       AND s.isaddress ";
 582                 $sSQL .= "       AND p.importance is not null";
 583                 $sSQL .= "  ) AS addressimportance,";
 584             }
 585             $sSQL .= "  null AS extra_place ";
 586             $sSQL .= "  FROM (";
 587             $sSQL .= "     SELECT ";
 588             $sSQL .= "         osm_id, ";
 589             $sSQL .= "         place_id, ";
 590             $sSQL .= "         country_code, ";
 591             $sSQL .= "         CASE ";             // interpolate the housenumbers here
 592             $sSQL .= "           WHEN startnumber != endnumber ";
 593             $sSQL .= "           THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
 594             $sSQL .= "           ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
 595             $sSQL .= "         END as centroid, ";
 596             $sSQL .= "         parent_place_id, ";
 597             $sSQL .= "         housenumber_for_place ";
 598             $sSQL .= "     FROM (";
 599             $sSQL .= "            location_property_osmline ";
 600             $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
 601             $sSQL .= "          ) ";
 602             $sSQL .= "     WHERE housenumber_for_place>=0 ";
 603             $sSQL .= "       AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 604             $sSQL .= "  ) as blub"; //postgres wants an alias here
 605             $sSQL .= "  GROUP BY ";
 606             $sSQL .= "    osm_id, ";
 607             $sSQL .= "    place_id, ";
 608             $sSQL .= "    housenumber_for_place, ";
 609             $sSQL .= "    country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
 610             if (!$this->bDeDupe) $sSQL .= ", place_id ";
 611
 612             if (CONST_Use_Aux_Location_data) {
 613                 $sSQL .= " UNION ";
 614                 $sSQL .= "  SELECT ";
 615                 $sSQL .= "     'L' AS osm_type, ";
 616                 $sSQL .= "     place_id AS osm_id, ";
 617                 $sSQL .= "     'place' AS class,";
 618                 $sSQL .= "     'house' AS type, ";
 619                 $sSQL .= "     null AS admin_level, ";
 620                 $sSQL .= "     0 AS rank_search,";
 621                 $sSQL .= "     0 AS rank_address, ";
 622                 $sSQL .= "     min(place_id) AS place_id,";
 623                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 624                 $sSQL .= "     'us' AS country_code, ";
 625                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
 626                 $sSQL .= "     null AS placename, ";
 627                 $sSQL .= "     null AS ref, ";
 628                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 629                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 630                 $sSQL .= "     avg(ST_X(centroid)) AS lon, ";
 631                 $sSQL .= "     avg(ST_Y(centroid)) AS lat, ";
 632                 $sSQL .= "     -1.10".$sImportanceSQL." AS importance, ";
 633                 if ($oCtx->hasNearPoint()) {
 634                     $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,";
 635                 } else {
 636                     $sSQL .= "     ( ";
 637                     $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 638                     $sSQL .= "       FROM ";
 639                     $sSQL .= "          place_addressline s, ";
 640                     $sSQL .= "          placex p";
 641                     $sSQL .= "       WHERE s.place_id = min(location_property_aux.parent_place_id)";
 642                     $sSQL .= "         AND p.place_id = s.address_place_id ";
 643                     $sSQL .= "         AND s.isaddress";
 644                     $sSQL .= "         AND p.importance is not null";
 645                     $sSQL .= "     ) AS addressimportance, ";
 646                 }
 647                 $sSQL .= "     null AS extra_place ";
 648                 $sSQL .= "  FROM location_property_aux ";
 649                 $sSQL .= "  WHERE place_id in ($sPlaceIDs) ";
 650                 $sSQL .= "    AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
 651                 $sSQL .= "  GROUP BY ";
 652                 $sSQL .= "     place_id, ";
 653                 if (!$this->bDeDupe) $sSQL .= "place_id, ";
 654                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
 655             }
 656         }
 657
 658         $sSQL .= " order by importance desc";
 659         if (CONST_Debug) {
 660             echo "<hr>";
 661             var_dump($sSQL);
 662         }
 663         $aSearchResults = chksql(
 664             $this->oDB->getAll($sSQL),
 665             "Could not get details for place."
 666         );
 667
 668         return $aSearchResults;
 669     }
 670
 671     public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
 672     {
 673         /*
 674              Calculate all searches using aValidTokens i.e.
 675              'Wodsworth Road, Sheffield' =>
 676
 677              Phrase Wordset
 678              0      0       (wodsworth road)
 679              0      1       (wodsworth)(road)
 680              1      0       (sheffield)
 681
 682              Score how good the search is so they can be ordered
 683          */
 684         $iGlobalRank = 0;
 685
 686         foreach ($aPhrases as $iPhrase => $aPhrase) {
 687             $aNewPhraseSearches = array();
 688             if ($bStructuredPhrases) {
 689                 $sPhraseType = $aPhraseTypes[$iPhrase];
 690             } else {
 691                 $sPhraseType = '';
 692             }
 693
 694             foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) {
 695                 // Too many permutations - too expensive
 696                 if ($iWordSet > 120) break;
 697
 698                 $aWordsetSearches = $aSearches;
 699
 700                 // Add all words from this wordset
 701                 foreach ($aWordset as $iToken => $sToken) {
 702                     //echo "<br><b>$sToken</b>";
 703                     $aNewWordsetSearches = array();
 704
 705                     foreach ($aWordsetSearches as $oCurrentSearch) {
 706                         //echo "<i>";
 707                         //var_dump($oCurrentSearch);
 708                         //echo "</i>";
 709
 710                         // If the token is valid
 711                         if (isset($aValidTokens[' '.$sToken])) {
 712                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 713                                 // Recheck if the original word shows up in the query.
 714                                 $bWordInQuery = false;
 715                                 if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
 716                                     $bWordInQuery = strpos(
 717                                         $sNormQuery,
 718                                         $this->normTerm($aSearchTerm['word'])
 719                                     ) !== false;
 720                                 }
 721                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 722                                     $aSearchTerm,
 723                                     $bWordInQuery,
 724                                     isset($aValidTokens[$sToken])
 725                                       && strpos($sToken, ' ') === false,
 726                                     $sPhraseType,
 727                                     $iToken == 0 && $iPhrase == 0,
 728                                     $iPhrase == 0,
 729                                     $iToken + 1 == sizeof($aWordset)
 730                                       && $iPhrase + 1 == sizeof($aPhrases),
 731                                     $iGlobalRank
 732                                 );
 733
 734                                 foreach ($aNewSearches as $oSearch) {
 735                                     if ($oSearch->getRank() < $this->iMaxRank) {
 736                                         $aNewWordsetSearches[] = $oSearch;
 737                                     }
 738                                 }
 739                             }
 740                         }
 741                         // Look for partial matches.
 742                         // Note that there is no point in adding country terms here
 743                         // because country is omitted in the address.
 744                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 745                             // Allow searching for a word - but at extra cost
 746                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 747                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 748                                     $aSearchTerm,
 749                                     $bStructuredPhrases,
 750                                     $iPhrase,
 751                                     $aWordFrequencyScores,
 752                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 753                                 );
 754
 755                                 foreach ($aNewSearches as $oSearch) {
 756                                     if ($oSearch->getRank() < $this->iMaxRank) {
 757                                         $aNewWordsetSearches[] = $oSearch;
 758                                     }
 759                                 }
 760                             }
 761                         }
 762                     }
 763                     // Sort and cut
 764                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 765                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 766                 }
 767                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 768
 769                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 770                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 771
 772                 $aSearchHash = array();
 773                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 774                     $sHash = serialize($aSearch);
 775                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 776                     else $aSearchHash[$sHash] = 1;
 777                 }
 778
 779                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 780             }
 781
 782             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 783             $aGroupedSearches = array();
 784             foreach ($aNewPhraseSearches as $aSearch) {
 785                 $iRank = $aSearch->getRank();
 786                 if ($iRank < $this->iMaxRank) {
 787                     if (!isset($aGroupedSearches[$iRank])) {
 788                         $aGroupedSearches[$iRank] = array();
 789                     }
 790                     $aGroupedSearches[$iRank][] = $aSearch;
 791                 }
 792             }
 793             ksort($aGroupedSearches);
 794
 795             $iSearchCount = 0;
 796             $aSearches = array();
 797             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 798                 $iSearchCount += sizeof($aNewSearches);
 799                 $aSearches = array_merge($aSearches, $aNewSearches);
 800                 if ($iSearchCount > 50) break;
 801             }
 802
 803             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 804         }
 805
 806         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 807         $aGroupedSearches = array();
 808         foreach ($aSearches as $oSearch) {
 809             if (!$oSearch->isValidSearch($this->aCountryCodes)) {
 810                 continue;
 811             }
 812
 813             $iRank = $oSearch->addToRank($iGlobalRank);
 814             if (!isset($aGroupedSearches[$iRank])) {
 815                 $aGroupedSearches[$iRank] = array();
 816             }
 817             $aGroupedSearches[$iRank][] = $oSearch;
 818         }
 819         ksort($aGroupedSearches);
 820
 821         return $aGroupedSearches;
 822     }
 823
 824     /* Perform the actual query lookup.
 825
 826         Returns an ordered list of results, each with the following fields:
 827             osm_type: type of corresponding OSM object
 828                         N - node
 829                         W - way
 830                         R - relation
 831                         P - postcode (internally computed)
 832             osm_id: id of corresponding OSM object
 833             class: general object class (corresponds to tag key of primary OSM tag)
 834             type: subclass of object (corresponds to tag value of primary OSM tag)
 835             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 836             rank_search: rank in search hierarchy
 837                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 838             rank_address: rank in address hierarchy (determines orer in address)
 839             place_id: internal key (may differ between different instances)
 840             country_code: ISO country code
 841             langaddress: localized full address
 842             placename: localized name of object
 843             ref: content of ref tag (if available)
 844             lon: longitude
 845             lat: latitude
 846             importance: importance of place based on Wikipedia link count
 847             addressimportance: cumulated importance of address elements
 848             extra_place: type of place (for admin boundaries, if there is a place tag)
 849             aBoundingBox: bounding Box
 850             label: short description of the object class/type (English only)
 851             name: full name (currently the same as langaddress)
 852             foundorder: secondary ordering for places with same importance
 853     */
 854
 855
 856     public function lookup()
 857     {
 858         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 859
 860         $oCtx = new SearchContext();
 861
 862         if ($this->aRoutePoints) {
 863             $oCtx->setViewboxFromRoute(
 864                 $this->oDB,
 865                 $this->aRoutePoints,
 866                 $this->aRouteWidth,
 867                 $this->bBoundedSearch
 868             );
 869         } elseif ($this->aViewBox) {
 870             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 871         }
 872         if ($this->aExcludePlaceIDs) {
 873             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 874         }
 875         if ($this->aCountryCodes) {
 876             $oCtx->setCountryList($this->aCountryCodes);
 877         }
 878
 879         $sNormQuery = $this->normTerm($this->sQuery);
 880         $sLanguagePrefArraySQL = getArraySQL(
 881             array_map("getDBQuoted", $this->aLangPrefOrder)
 882         );
 883
 884         $sQuery = $this->sQuery;
 885         if (!preg_match('//u', $sQuery)) {
 886             userError("Query string is not UTF-8 encoded.");
 887         }
 888
 889         // Conflicts between US state abreviations and various words for 'the' in different languages
 890         if (isset($this->aLangPrefOrder['name:en'])) {
 891             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 892             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 893             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 894         }
 895
 896         // Do we have anything that looks like a lat/lon pair?
 897         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 898
 899         $aSearchResults = array();
 900         if ($sQuery || $this->aStructuredQuery) {
 901             // Start with a single blank search
 902             $aSearches = array(new SearchDescription($oCtx));
 903
 904             if ($sQuery) {
 905                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 906             }
 907
 908             $sSpecialTerm = '';
 909             if ($sQuery) {
 910                 preg_match_all(
 911                     '/\\[([\\w ]*)\\]/u',
 912                     $sQuery,
 913                     $aSpecialTermsRaw,
 914                     PREG_SET_ORDER
 915                 );
 916                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 917                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 918                     if (!$sSpecialTerm) {
 919                         $sSpecialTerm = $aSpecialTerm[1];
 920                     }
 921                 }
 922             }
 923             if (!$sSpecialTerm && $this->aStructuredQuery
 924                 && isset($this->aStructuredQuery['amenity'])) {
 925                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 926                 unset($this->aStructuredQuery['amenity']);
 927             }
 928
 929             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 930                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 931                 $sToken = chksql(
 932                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 933                     "Cannot decode query. Wrong encoding?"
 934                 );
 935                 $sSQL = 'SELECT class, type FROM word ';
 936                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 937                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 938                 if (CONST_Debug) var_Dump($sSQL);
 939                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 940                 $aNewSearches = array();
 941                 foreach ($aSearches as $oSearch) {
 942                     foreach ($aSearchWords as $aSearchTerm) {
 943                         $oNewSearch = clone $oSearch;
 944                         $oNewSearch->setPoiSearch(
 945                             Operator::TYPE,
 946                             $aSearchTerm['class'],
 947                             $aSearchTerm['type']
 948                         );
 949                         $aNewSearches[] = $oNewSearch;
 950                     }
 951                 }
 952                 $aSearches = $aNewSearches;
 953             }
 954
 955             // Split query into phrases
 956             // Commas are used to reduce the search space by indicating where phrases split
 957             if ($this->aStructuredQuery) {
 958                 $aPhrases = $this->aStructuredQuery;
 959                 $bStructuredPhrases = true;
 960             } else {
 961                 $aPhrases = explode(',', $sQuery);
 962                 $bStructuredPhrases = false;
 963             }
 964
 965             // Convert each phrase to standard form
 966             // Create a list of standard words
 967             // Get all 'sets' of words
 968             // Generate a complete list of all
 969             $aTokens = array();
 970             foreach ($aPhrases as $iPhrase => $sPhrase) {
 971                 $aPhrase = chksql(
 972                     $this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"),
 973                     "Cannot normalize query string (is it a UTF-8 string?)"
 974                 );
 975                 if (trim($aPhrase['string'])) {
 976                     $aPhrases[$iPhrase] = $aPhrase;
 977                     $aPhrases[$iPhrase]['words'] = explode(' ', $aPhrases[$iPhrase]['string']);
 978                     $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0);
 979                     $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets']));
 980                 } else {
 981                     unset($aPhrases[$iPhrase]);
 982                 }
 983             }
 984
 985             // Reindex phrases - we make assumptions later on that they are numerically keyed in order
 986             $aPhraseTypes = array_keys($aPhrases);
 987             $aPhrases = array_values($aPhrases);
 988
 989             if (sizeof($aTokens)) {
 990                 // Check which tokens we have, get the ID numbers
 991                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 992                 $sSQL .= ' FROM word ';
 993                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
 994
 995                 if (CONST_Debug) var_Dump($sSQL);
 996
 997                 $aValidTokens = array();
 998                 $aDatabaseWords = chksql(
 999                     $this->oDB->getAll($sSQL),
1000                     "Could not get word tokens."
1001                 );
1002                 $aPossibleMainWordIDs = array();
1003                 $aWordFrequencyScores = array();
1004                 foreach ($aDatabaseWords as $aToken) {
1005                     // Very special case - require 2 letter country param to match the country code found
1006                     if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
1007                         && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
1008                     ) {
1009                         continue;
1010                     }
1011
1012                     if (isset($aValidTokens[$aToken['word_token']])) {
1013                         $aValidTokens[$aToken['word_token']][] = $aToken;
1014                     } else {
1015                         $aValidTokens[$aToken['word_token']] = array($aToken);
1016                     }
1017                     if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
1018                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
1019                 }
1020                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
1021
1022                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
1023                 foreach ($aTokens as $sToken) {
1024                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
1025                         if (isset($aValidTokens[$aData[1]])) {
1026                             foreach ($aValidTokens[$aData[1]] as $aToken) {
1027                                 if (!$aToken['class']) {
1028                                     if (isset($aValidTokens[$sToken])) {
1029                                         $aValidTokens[$sToken][] = $aToken;
1030                                     } else {
1031                                         $aValidTokens[$sToken] = array($aToken);
1032                                     }
1033                                 }
1034                             }
1035                         }
1036                     }
1037                 }
1038
1039                 foreach ($aTokens as $sToken) {
1040                     // Unknown single word token with a number - assume it is a house number
1041                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
1042                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
1043                     }
1044                 }
1045
1046                 // Any words that have failed completely?
1047                 // TODO: suggestions
1048
1049                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
1050
1051                 if ($this->bReverseInPlan) {
1052                     // Reverse phrase array and also reverse the order of the wordsets in
1053                     // the first and final phrase. Don't bother about phrases in the middle
1054                     // because order in the address doesn't matter.
1055                     $aPhrases = array_reverse($aPhrases);
1056                     $aPhrases[0]['wordsets'] = getInverseWordSets($aPhrases[0]['words'], 0);
1057                     if (sizeof($aPhrases) > 1) {
1058                         $aFinalPhrase = end($aPhrases);
1059                         $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
1060                     }
1061                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
1062
1063                     foreach ($aGroupedSearches as $aSearches) {
1064                         foreach ($aSearches as $aSearch) {
1065                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
1066                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
1067                             }
1068                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
1069                         }
1070                     }
1071
1072                     $aGroupedSearches = $aReverseGroupedSearches;
1073                     ksort($aGroupedSearches);
1074                 }
1075             } else {
1076                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
1077                 $aGroupedSearches = array();
1078                 foreach ($aSearches as $aSearch) {
1079                     if ($aSearch->getRank() < $this->iMaxRank) {
1080                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
1081                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
1082                     }
1083                 }
1084                 ksort($aGroupedSearches);
1085             }
1086
1087             // Filter out duplicate searches
1088             $aSearchHash = array();
1089             foreach ($aGroupedSearches as $iGroup => $aSearches) {
1090                 foreach ($aSearches as $iSearch => $aSearch) {
1091                     $sHash = serialize($aSearch);
1092                     if (isset($aSearchHash[$sHash])) {
1093                         unset($aGroupedSearches[$iGroup][$iSearch]);
1094                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
1095                     } else {
1096                         $aSearchHash[$sHash] = 1;
1097                     }
1098                 }
1099             }
1100
1101             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
1102
1103             // Start the search process
1104             // array with: placeid => -1 | tiger-housenumber
1105             $aResultPlaceIDs = array();
1106             $iGroupLoop = 0;
1107             $iQueryLoop = 0;
1108             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
1109                 $iGroupLoop++;
1110                 foreach ($aSearches as $oSearch) {
1111                     $iQueryLoop++;
1112
1113                     if (CONST_Debug) {
1114                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
1115                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
1116                     }
1117
1118                     $aRes = $oSearch->query(
1119                         $this->oDB,
1120                         $aWordFrequencyScores,
1121                         $this->exactMatchCache,
1122                         $this->iMinAddressRank,
1123                         $this->iMaxAddressRank,
1124                         $this->iLimit
1125                     );
1126
1127                     foreach ($aRes['IDs'] as $iPlaceID) {
1128                         // array for placeID => -1 | Tiger housenumber
1129                         $aResultPlaceIDs[$iPlaceID] = $aRes['houseNumber'];
1130                     }
1131                     if ($iQueryLoop > 20) break;
1132                 }
1133
1134                 if (sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
1135                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
1136                     // reduces the number of place ids, like a filter
1137                     // rank_address is 30 for interpolated housenumbers
1138                     $sWherePlaceId = 'WHERE place_id in (';
1139                     $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
1140
1141                     $sSQL = "SELECT place_id ";
1142                     $sSQL .= "FROM placex ".$sWherePlaceId;
1143                     $sSQL .= "  AND (";
1144                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1145                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
1146                         $sSQL .= "     OR (extratags->'place') = 'city'";
1147                     }
1148                     if ($this->aAddressRankList) {
1149                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
1150                     }
1151                     $sSQL .= "  ) UNION ";
1152                     $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
1153                     $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1154                     if ($this->aAddressRankList) {
1155                         $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
1156                     }
1157                     $sSQL .= ") ";
1158                     if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
1159                         $sSQL .= "UNION ";
1160                         $sSQL .= "  SELECT place_id ";
1161                         $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
1162                     }
1163                     if ($this->iMaxAddressRank == 30) {
1164                         $sSQL .= "UNION ";
1165                         $sSQL .= "  SELECT place_id ";
1166                         $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;
1167                     }
1168                     if (CONST_Debug) var_dump($sSQL);
1169                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1170                     $tempIDs = array();
1171                     foreach ($aFilteredPlaceIDs as $placeID) {
1172                         $tempIDs[$placeID] = $aResultPlaceIDs[$placeID];  //assign housenumber to placeID
1173                     }
1174                     $aResultPlaceIDs = $tempIDs;
1175                 }
1176
1177                 if (sizeof($aResultPlaceIDs)) break;
1178                 if ($iGroupLoop > 4) break;
1179                 if ($iQueryLoop > 30) break;
1180             }
1181
1182             // Did we find anything?
1183             if (sizeof($aResultPlaceIDs)) {
1184                 $aSearchResults = $this->getDetails($aResultPlaceIDs, $oCtx);
1185             }
1186         } else {
1187             // Just interpret as a reverse geocode
1188             $oReverse = new ReverseGeocode($this->oDB);
1189             $oReverse->setZoom(18);
1190
1191             $aLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
1192
1193             if (CONST_Debug) var_dump("Reverse search", $aLookup);
1194
1195             if ($aLookup['place_id']) {
1196                 $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1), $oCtx);
1197                 $aResultPlaceIDs[$aLookup['place_id']] = -1;
1198             } else {
1199                 $aSearchResults = array();
1200             }
1201         }
1202
1203         // No results? Done
1204         if (!sizeof($aSearchResults)) {
1205             if ($this->bFallback) {
1206                 if ($this->fallbackStructuredQuery()) {
1207                     return $this->lookup();
1208                 }
1209             }
1210
1211             return array();
1212         }
1213
1214         $aClassType = getClassTypesWithImportance();
1215         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
1216         foreach ($aRecheckWords as $i => $sWord) {
1217             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
1218         }
1219
1220         if (CONST_Debug) {
1221             echo '<i>Recheck words:<\i>';
1222             var_dump($aRecheckWords);
1223         }
1224
1225         $oPlaceLookup = new PlaceLookup($this->oDB);
1226         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
1227         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
1228         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
1229         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
1230         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
1231         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
1232
1233         foreach ($aSearchResults as $iResNum => $aResult) {
1234             // Default
1235             $fDiameter = getResultDiameter($aResult);
1236
1237             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
1238             if ($aOutlineResult) {
1239                 $aResult = array_merge($aResult, $aOutlineResult);
1240             }
1241
1242             if ($aResult['extra_place'] == 'city') {
1243                 $aResult['class'] = 'place';
1244                 $aResult['type'] = 'city';
1245                 $aResult['rank_search'] = 16;
1246             }
1247
1248             // Is there an icon set for this type of result?
1249             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
1250                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
1251             ) {
1252                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
1253             }
1254
1255             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
1256                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
1257             ) {
1258                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
1259             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
1260                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
1261             ) {
1262                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
1263             }
1264             // if tag '&addressdetails=1' is set in query
1265             if ($this->bIncludeAddressDetails) {
1266                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
1267                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);
1268                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
1269                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
1270                 }
1271             }
1272
1273             if ($this->bIncludeExtraTags) {
1274                 if ($aResult['extra']) {
1275                     $aResult['sExtraTags'] = json_decode($aResult['extra']);
1276                 } else {
1277                     $aResult['sExtraTags'] = (object) array();
1278                 }
1279             }
1280
1281             if ($this->bIncludeNameDetails) {
1282                 if ($aResult['names']) {
1283                     $aResult['sNameDetails'] = json_decode($aResult['names']);
1284                 } else {
1285                     $aResult['sNameDetails'] = (object) array();
1286                 }
1287             }
1288
1289             $aResult['name'] = $aResult['langaddress'];
1290
1291             if ($oCtx->hasNearPoint())
1292             {
1293                 $aResult['importance'] = 0.001;
1294                 $aResult['foundorder'] = $aResult['addressimportance'];
1295             } else {
1296                 // Adjust importance for the number of exact string matches in the result
1297                 $aResult['importance'] = max(0.001, $aResult['importance']);
1298                 $iCountWords = 0;
1299                 $sAddress = $aResult['langaddress'];
1300                 foreach ($aRecheckWords as $i => $sWord) {
1301                     if (stripos($sAddress, $sWord)!==false) {
1302                         $iCountWords++;
1303                         if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
1304                     }
1305                 }
1306
1307                 $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
1308
1309                 // secondary ordering (for results with same importance (the smaller the better):
1310                 // - approximate importance of address parts
1311                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
1312                 // - number of exact matches from the query
1313                 if (isset($this->exactMatchCache[$aResult['place_id']])) {
1314                     $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']];
1315                 } elseif (isset($this->exactMatchCache[$aResult['parent_place_id']])) {
1316                     $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']];
1317                 }
1318                 // - importance of the class/type
1319                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1320                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1321                 ) {
1322                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1323                 } else {
1324                     $aResult['foundorder'] += 0.01;
1325                 }
1326             }
1327             if (CONST_Debug) var_dump($aResult);
1328             $aSearchResults[$iResNum] = $aResult;
1329         }
1330         uasort($aSearchResults, 'byImportance');
1331
1332         $aOSMIDDone = array();
1333         $aClassTypeNameDone = array();
1334         $aToFilter = $aSearchResults;
1335         $aSearchResults = array();
1336
1337         $bFirst = true;
1338         foreach ($aToFilter as $iResNum => $aResult) {
1339             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1340             if ($bFirst) {
1341                 $fLat = $aResult['lat'];
1342                 $fLon = $aResult['lon'];
1343                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1344                 $bFirst = false;
1345             }
1346             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1347                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1348             ) {
1349                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1350                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1351                 $aSearchResults[] = $aResult;
1352             }
1353
1354             // Absolute limit on number of results
1355             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1356         }
1357
1358         return $aSearchResults;
1359     } // end lookup()
1360 } // end class