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         $sSQL .= "    ( ";
 402         $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 403         $sSQL .= "       FROM ";
 404         $sSQL .= "         place_addressline s, ";
 405         $sSQL .= "         placex p";
 406         $sSQL .= "       WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
 407         $sSQL .= "         AND p.place_id = s.address_place_id ";
 408         $sSQL .= "         AND s.isaddress ";
 409         $sSQL .= "         AND p.importance is not null ";
 410         $sSQL .= "    ) AS addressimportance, ";
 411         $sSQL .= "    (extratags->'place') AS extra_place ";
 412         $sSQL .= " FROM placex";
 413         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 414         $sSQL .= "   AND (";
 415         $sSQL .= "            placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 416         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 417             $sSQL .= "        OR (extratags->'place') = 'city'";
 418         }
 419         if ($this->aAddressRankList) {
 420             $sSQL .= "        OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 421         }
 422         $sSQL .= "       ) ";
 423         if ($this->sAllowedTypesSQLList) {
 424             $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
 425         }
 426         $sSQL .= "    AND linked_place_id is null ";
 427         $sSQL .= " GROUP BY ";
 428         $sSQL .= "     osm_type, ";
 429         $sSQL .= "     osm_id, ";
 430         $sSQL .= "     class, ";
 431         $sSQL .= "     type, ";
 432         $sSQL .= "     admin_level, ";
 433         $sSQL .= "     rank_search, ";
 434         $sSQL .= "     rank_address, ";
 435         $sSQL .= "     country_code, ";
 436         $sSQL .= "     importance, ";
 437         if (!$this->bDeDupe) $sSQL .= "place_id,";
 438         $sSQL .= "     langaddress, ";
 439         $sSQL .= "     placename, ";
 440         $sSQL .= "     ref, ";
 441         if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
 442         if ($this->bIncludeNameDetails) $sSQL .= "name, ";
 443         $sSQL .= "     extratags->'place' ";
 444
 445         // postcode table
 446         $sSQL .= "UNION ";
 447         $sSQL .= "SELECT";
 448         $sSQL .= "  'P' as osm_type,";
 449         $sSQL .= "  (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
 450         $sSQL .= "  'place' as class, 'postcode' as type,";
 451         $sSQL .= "  null as admin_level, rank_search, rank_address,";
 452         $sSQL .= "  place_id, parent_place_id, country_code,";
 453         $sSQL .= "  get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
 454         $sSQL .= "  postcode as placename,";
 455         $sSQL .= "  postcode as ref,";
 456         if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 457         if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 458         $sSQL .= "  ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
 459         $sSQL .= "  (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, ";
 460         $sSQL .= "  (";
 461         $sSQL .= "     SELECT max(p.importance*(p.rank_address+2))";
 462         $sSQL .= "     FROM ";
 463         $sSQL .= "       place_addressline s, ";
 464         $sSQL .= "       placex p";
 465         $sSQL .= "     WHERE s.place_id = lp.parent_place_id";
 466         $sSQL .= "       AND p.place_id = s.address_place_id ";
 467         $sSQL .= "       AND s.isaddress";
 468         $sSQL .= "       AND p.importance is not null";
 469         $sSQL .= "  ) AS addressimportance, ";
 470         $sSQL .= "  null AS extra_place ";
 471         $sSQL .= "FROM location_postcode lp";
 472         $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
 473
 474         if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
 475             // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
 476             // with start- and endnumber, the common osm housenumbers are usually saved as points
 477             $sHousenumbers = "";
 478             $i = 0;
 479             $length = count($aPlaceIDs);
 480             foreach ($aPlaceIDs as $placeID => $housenumber) {
 481                 $i++;
 482                 $sHousenumbers .= "(".$placeID.", ".$housenumber.")";
 483                 if ($i<$length) $sHousenumbers .= ", ";
 484             }
 485
 486             if (CONST_Use_US_Tiger_Data) {
 487                 // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 488                 $sSQL .= " union";
 489                 $sSQL .= " SELECT ";
 490                 $sSQL .= "     'T' AS osm_type, ";
 491                 $sSQL .= "     (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
 492                 $sSQL .= "     'place' AS class, ";
 493                 $sSQL .= "     'house' AS type, ";
 494                 $sSQL .= "     null AS admin_level, ";
 495                 $sSQL .= "     30 AS rank_search, ";
 496                 $sSQL .= "     30 AS rank_address, ";
 497                 $sSQL .= "     min(place_id) AS place_id, ";
 498                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 499                 $sSQL .= "     'us' AS country_code, ";
 500                 $sSQL .= "     get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
 501                 $sSQL .= "     null AS placename, ";
 502                 $sSQL .= "     null AS ref, ";
 503                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
 504                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
 505                 $sSQL .= "     avg(st_x(centroid)) AS lon, ";
 506                 $sSQL .= "     avg(st_y(centroid)) AS lat,";
 507                 $sSQL .= "     -1.15".$sImportanceSQL." AS importance, ";
 508                 $sSQL .= "     (";
 509                 $sSQL .= "        SELECT max(p.importance*(p.rank_address+2))";
 510                 $sSQL .= "        FROM ";
 511                 $sSQL .= "          place_addressline s, ";
 512                 $sSQL .= "          placex p";
 513                 $sSQL .= "        WHERE s.place_id = min(blub.parent_place_id)";
 514                 $sSQL .= "          AND p.place_id = s.address_place_id ";
 515                 $sSQL .= "          AND s.isaddress";
 516                 $sSQL .= "          AND p.importance is not null";
 517                 $sSQL .= "     ) AS addressimportance, ";
 518                 $sSQL .= "     null AS extra_place ";
 519                 $sSQL .= " FROM (";
 520                 $sSQL .= "     SELECT place_id, ";    // interpolate the Tiger housenumbers here
 521                 $sSQL .= "         ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
 522                 $sSQL .= "         parent_place_id, ";
 523                 $sSQL .= "         housenumber_for_place";
 524                 $sSQL .= "     FROM (";
 525                 $sSQL .= "            location_property_tiger ";
 526                 $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
 527                 $sSQL .= "     WHERE ";
 528                 $sSQL .= "         housenumber_for_place>=0";
 529                 $sSQL .= "         AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 530                 $sSQL .= " ) AS blub"; //postgres wants an alias here
 531                 $sSQL .= " GROUP BY";
 532                 $sSQL .= "      place_id, ";
 533                 $sSQL .= "      housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
 534                 if (!$this->bDeDupe) $sSQL .= ", place_id ";
 535             }
 536             // osmline
 537             // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
 538             $sSQL .= " UNION ";
 539             $sSQL .= "SELECT ";
 540             $sSQL .= "  'W' AS osm_type, ";
 541             $sSQL .= "  osm_id, ";
 542             $sSQL .= "  'place' AS class, ";
 543             $sSQL .= "  'house' AS type, ";
 544             $sSQL .= "  null AS admin_level, ";
 545             $sSQL .= "  30 AS rank_search, ";
 546             $sSQL .= "  30 AS rank_address, ";
 547             $sSQL .= "  min(place_id) as place_id, ";
 548             $sSQL .= "  min(parent_place_id) AS parent_place_id, ";
 549             $sSQL .= "  country_code, ";
 550             $sSQL .= "  get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
 551             $sSQL .= "  null AS placename, ";
 552             $sSQL .= "  null AS ref, ";
 553             if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 554             if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 555             $sSQL .= "  AVG(st_x(centroid)) AS lon, ";
 556             $sSQL .= "  AVG(st_y(centroid)) AS lat, ";
 557             $sSQL .= "  -0.1".$sImportanceSQL." AS importance, ";  // slightly smaller than the importance for normal houses with rank 30, which is 0
 558             $sSQL .= "  (";
 559             $sSQL .= "     SELECT ";
 560             $sSQL .= "       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 ";
 572             $sSQL .= "         osm_id, ";
 573             $sSQL .= "         place_id, ";
 574             $sSQL .= "         country_code, ";
 575             $sSQL .= "         CASE ";             // interpolate the housenumbers here
 576             $sSQL .= "           WHEN startnumber != endnumber ";
 577             $sSQL .= "           THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
 578             $sSQL .= "           ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
 579             $sSQL .= "         END as centroid, ";
 580             $sSQL .= "         parent_place_id, ";
 581             $sSQL .= "         housenumber_for_place ";
 582             $sSQL .= "     FROM (";
 583             $sSQL .= "            location_property_osmline ";
 584             $sSQL .= "            JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
 585             $sSQL .= "          ) ";
 586             $sSQL .= "     WHERE housenumber_for_place>=0 ";
 587             $sSQL .= "       AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
 588             $sSQL .= "  ) as blub"; //postgres wants an alias here
 589             $sSQL .= "  GROUP BY ";
 590             $sSQL .= "    osm_id, ";
 591             $sSQL .= "    place_id, ";
 592             $sSQL .= "    housenumber_for_place, ";
 593             $sSQL .= "    country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
 594             if (!$this->bDeDupe) $sSQL .= ", place_id ";
 595
 596             if (CONST_Use_Aux_Location_data) {
 597                 $sSQL .= " UNION ";
 598                 $sSQL .= "  SELECT ";
 599                 $sSQL .= "     'L' AS osm_type, ";
 600                 $sSQL .= "     place_id AS osm_id, ";
 601                 $sSQL .= "     'place' AS class,";
 602                 $sSQL .= "     'house' AS type, ";
 603                 $sSQL .= "     null AS admin_level, ";
 604                 $sSQL .= "     0 AS rank_search,";
 605                 $sSQL .= "     0 AS rank_address, ";
 606                 $sSQL .= "     min(place_id) AS place_id,";
 607                 $sSQL .= "     min(parent_place_id) AS parent_place_id, ";
 608                 $sSQL .= "     'us' AS country_code, ";
 609                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
 610                 $sSQL .= "     null AS placename, ";
 611                 $sSQL .= "     null AS ref, ";
 612                 if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
 613                 if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
 614                 $sSQL .= "     avg(ST_X(centroid)) AS lon, ";
 615                 $sSQL .= "     avg(ST_Y(centroid)) AS lat, ";
 616                 $sSQL .= "     -1.10".$sImportanceSQL." AS importance, ";
 617                 $sSQL .= "     ( ";
 618                 $sSQL .= "       SELECT max(p.importance*(p.rank_address+2))";
 619                 $sSQL .= "       FROM ";
 620                 $sSQL .= "          place_addressline s, ";
 621                 $sSQL .= "          placex p";
 622                 $sSQL .= "       WHERE s.place_id = min(location_property_aux.parent_place_id)";
 623                 $sSQL .= "         AND p.place_id = s.address_place_id ";
 624                 $sSQL .= "         AND s.isaddress";
 625                 $sSQL .= "         AND p.importance is not null";
 626                 $sSQL .= "     ) AS addressimportance, ";
 627                 $sSQL .= "     null AS extra_place ";
 628                 $sSQL .= "  FROM location_property_aux ";
 629                 $sSQL .= "  WHERE place_id in ($sPlaceIDs) ";
 630                 $sSQL .= "    AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
 631                 $sSQL .= "  GROUP BY ";
 632                 $sSQL .= "     place_id, ";
 633                 if (!$this->bDeDupe) $sSQL .= "place_id, ";
 634                 $sSQL .= "     get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
 635             }
 636         }
 637
 638         $sSQL .= " order by importance desc";
 639         if (CONST_Debug) {
 640             echo "<hr>";
 641             var_dump($sSQL);
 642         }
 643         $aSearchResults = chksql(
 644             $this->oDB->getAll($sSQL),
 645             "Could not get details for place."
 646         );
 647
 648         return $aSearchResults;
 649     }
 650
 651     public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
 652     {
 653         /*
 654              Calculate all searches using aValidTokens i.e.
 655              'Wodsworth Road, Sheffield' =>
 656
 657              Phrase Wordset
 658              0      0       (wodsworth road)
 659              0      1       (wodsworth)(road)
 660              1      0       (sheffield)
 661
 662              Score how good the search is so they can be ordered
 663          */
 664         $iGlobalRank = 0;
 665
 666         foreach ($aPhrases as $iPhrase => $aPhrase) {
 667             $aNewPhraseSearches = array();
 668             if ($bStructuredPhrases) {
 669                 $sPhraseType = $aPhraseTypes[$iPhrase];
 670             } else {
 671                 $sPhraseType = '';
 672             }
 673
 674             foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) {
 675                 // Too many permutations - too expensive
 676                 if ($iWordSet > 120) break;
 677
 678                 $aWordsetSearches = $aSearches;
 679
 680                 // Add all words from this wordset
 681                 foreach ($aWordset as $iToken => $sToken) {
 682                     //echo "<br><b>$sToken</b>";
 683                     $aNewWordsetSearches = array();
 684
 685                     foreach ($aWordsetSearches as $oCurrentSearch) {
 686                         //echo "<i>";
 687                         //var_dump($oCurrentSearch);
 688                         //echo "</i>";
 689
 690                         // If the token is valid
 691                         if (isset($aValidTokens[' '.$sToken])) {
 692                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 693                                 // Recheck if the original word shows up in the query.
 694                                 $bWordInQuery = false;
 695                                 if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
 696                                     $bWordInQuery = strpos(
 697                                         $sNormQuery,
 698                                         $this->normTerm($aSearchTerm['word'])
 699                                     ) !== false;
 700                                 }
 701                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 702                                     $aSearchTerm,
 703                                     $bWordInQuery,
 704                                     isset($aValidTokens[$sToken])
 705                                       && strpos($sToken, ' ') === false,
 706                                     $sPhraseType,
 707                                     $iToken == 0 && $iPhrase == 0,
 708                                     $iPhrase == 0,
 709                                     $iToken + 1 == sizeof($aWordset)
 710                                       && $iPhrase + 1 == sizeof($aPhrases),
 711                                     $iGlobalRank
 712                                 );
 713
 714                                 foreach ($aNewSearches as $oSearch) {
 715                                     if ($oSearch->getRank() < $this->iMaxRank) {
 716                                         $aNewWordsetSearches[] = $oSearch;
 717                                     }
 718                                 }
 719                             }
 720                         }
 721                         // Look for partial matches.
 722                         // Note that there is no point in adding country terms here
 723                         // because country is omitted in the address.
 724                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 725                             // Allow searching for a word - but at extra cost
 726                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 727                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 728                                     $aSearchTerm,
 729                                     $bStructuredPhrases,
 730                                     $iPhrase,
 731                                     $aWordFrequencyScores,
 732                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 733                                 );
 734
 735                                 foreach ($aNewSearches as $oSearch) {
 736                                     if ($oSearch->getRank() < $this->iMaxRank) {
 737                                         $aNewWordsetSearches[] = $oSearch;
 738                                     }
 739                                 }
 740                             }
 741                         }
 742                     }
 743                     // Sort and cut
 744                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 745                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 746                 }
 747                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 748
 749                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 750                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 751
 752                 $aSearchHash = array();
 753                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 754                     $sHash = serialize($aSearch);
 755                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 756                     else $aSearchHash[$sHash] = 1;
 757                 }
 758
 759                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 760             }
 761
 762             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 763             $aGroupedSearches = array();
 764             foreach ($aNewPhraseSearches as $aSearch) {
 765                 $iRank = $aSearch->getRank();
 766                 if ($iRank < $this->iMaxRank) {
 767                     if (!isset($aGroupedSearches[$iRank])) {
 768                         $aGroupedSearches[$iRank] = array();
 769                     }
 770                     $aGroupedSearches[$iRank][] = $aSearch;
 771                 }
 772             }
 773             ksort($aGroupedSearches);
 774
 775             $iSearchCount = 0;
 776             $aSearches = array();
 777             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 778                 $iSearchCount += sizeof($aNewSearches);
 779                 $aSearches = array_merge($aSearches, $aNewSearches);
 780                 if ($iSearchCount > 50) break;
 781             }
 782
 783             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 784         }
 785
 786         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 787         $aGroupedSearches = array();
 788         foreach ($aSearches as $oSearch) {
 789             if (!$oSearch->isValidSearch($this->aCountryCodes)) {
 790                 continue;
 791             }
 792
 793             $iRank = $oSearch->addToRank($iGlobalRank);
 794             if (!isset($aGroupedSearches[$iRank])) {
 795                 $aGroupedSearches[$iRank] = array();
 796             }
 797             $aGroupedSearches[$iRank][] = $oSearch;
 798         }
 799         ksort($aGroupedSearches);
 800
 801         return $aGroupedSearches;
 802     }
 803
 804     /* Perform the actual query lookup.
 805
 806         Returns an ordered list of results, each with the following fields:
 807             osm_type: type of corresponding OSM object
 808                         N - node
 809                         W - way
 810                         R - relation
 811                         P - postcode (internally computed)
 812             osm_id: id of corresponding OSM object
 813             class: general object class (corresponds to tag key of primary OSM tag)
 814             type: subclass of object (corresponds to tag value of primary OSM tag)
 815             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 816             rank_search: rank in search hierarchy
 817                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 818             rank_address: rank in address hierarchy (determines orer in address)
 819             place_id: internal key (may differ between different instances)
 820             country_code: ISO country code
 821             langaddress: localized full address
 822             placename: localized name of object
 823             ref: content of ref tag (if available)
 824             lon: longitude
 825             lat: latitude
 826             importance: importance of place based on Wikipedia link count
 827             addressimportance: cumulated importance of address elements
 828             extra_place: type of place (for admin boundaries, if there is a place tag)
 829             aBoundingBox: bounding Box
 830             label: short description of the object class/type (English only)
 831             name: full name (currently the same as langaddress)
 832             foundorder: secondary ordering for places with same importance
 833     */
 834
 835
 836     public function lookup()
 837     {
 838         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 839
 840         $oCtx = new SearchContext();
 841
 842         if ($this->aRoutePoints) {
 843             $oCtx->setViewboxFromRoute(
 844                 $this->oDB,
 845                 $this->aRoutePoints,
 846                 $this->aRouteWidth,
 847                 $this->bBoundedSearch
 848             );
 849         } else if ($this->aViewBox) {
 850             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 851         }
 852
 853         $sNormQuery = $this->normTerm($this->sQuery);
 854         $sLanguagePrefArraySQL = getArraySQL(
 855             array_map("getDBQuoted", $this->aLangPrefOrder)
 856         );
 857         $sCountryCodesSQL = false;
 858         if ($this->aCountryCodes) {
 859             $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes));
 860         }
 861
 862         $sQuery = $this->sQuery;
 863         if (!preg_match('//u', $sQuery)) {
 864             userError("Query string is not UTF-8 encoded.");
 865         }
 866
 867         // Conflicts between US state abreviations and various words for 'the' in different languages
 868         if (isset($this->aLangPrefOrder['name:en'])) {
 869             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 870             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 871             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 872         }
 873
 874         // Do we have anything that looks like a lat/lon pair?
 875         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 876
 877         $aSearchResults = array();
 878         if ($sQuery || $this->aStructuredQuery) {
 879             // Start with a single blank search
 880             $aSearches = array(new SearchDescription($oCtx));
 881
 882             if ($sQuery) {
 883                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 884             }
 885
 886             $sSpecialTerm = '';
 887             if ($sQuery) {
 888                 preg_match_all(
 889                     '/\\[([\\w ]*)\\]/u',
 890                     $sQuery,
 891                     $aSpecialTermsRaw,
 892                     PREG_SET_ORDER
 893                 );
 894                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 895                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 896                     if (!$sSpecialTerm) {
 897                         $sSpecialTerm = $aSpecialTerm[1];
 898                     }
 899                 }
 900             }
 901             if (!$sSpecialTerm && $this->aStructuredQuery
 902                 && isset($this->aStructuredQuery['amenity'])) {
 903                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 904                 unset($this->aStructuredQuery['amenity']);
 905             }
 906
 907             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 908                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 909                 $sToken = chksql(
 910                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 911                     "Cannot decode query. Wrong encoding?"
 912                 );
 913                 $sSQL = 'SELECT class, type FROM word ';
 914                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 915                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 916                 if (CONST_Debug) var_Dump($sSQL);
 917                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 918                 $aNewSearches = array();
 919                 foreach ($aSearches as $oSearch) {
 920                     foreach ($aSearchWords as $aSearchTerm) {
 921                         $oNewSearch = clone $oSearch;
 922                         $oNewSearch->setPoiSearch(
 923                             Operator::TYPE,
 924                             $aSearchTerm['class'],
 925                             $aSearchTerm['type']
 926                         );
 927                         $aNewSearches[] = $oNewSearch;
 928                     }
 929                 }
 930                 $aSearches = $aNewSearches;
 931             }
 932
 933             // Split query into phrases
 934             // Commas are used to reduce the search space by indicating where phrases split
 935             if ($this->aStructuredQuery) {
 936                 $aPhrases = $this->aStructuredQuery;
 937                 $bStructuredPhrases = true;
 938             } else {
 939                 $aPhrases = explode(',', $sQuery);
 940                 $bStructuredPhrases = false;
 941             }
 942
 943             // Convert each phrase to standard form
 944             // Create a list of standard words
 945             // Get all 'sets' of words
 946             // Generate a complete list of all
 947             $aTokens = array();
 948             foreach ($aPhrases as $iPhrase => $sPhrase) {
 949                 $aPhrase = chksql(
 950                     $this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"),
 951                     "Cannot normalize query string (is it a UTF-8 string?)"
 952                 );
 953                 if (trim($aPhrase['string'])) {
 954                     $aPhrases[$iPhrase] = $aPhrase;
 955                     $aPhrases[$iPhrase]['words'] = explode(' ', $aPhrases[$iPhrase]['string']);
 956                     $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0);
 957                     $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets']));
 958                 } else {
 959                     unset($aPhrases[$iPhrase]);
 960                 }
 961             }
 962
 963             // Reindex phrases - we make assumptions later on that they are numerically keyed in order
 964             $aPhraseTypes = array_keys($aPhrases);
 965             $aPhrases = array_values($aPhrases);
 966
 967             if (sizeof($aTokens)) {
 968                 // Check which tokens we have, get the ID numbers
 969                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 970                 $sSQL .= ' FROM word ';
 971                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
 972
 973                 if (CONST_Debug) var_Dump($sSQL);
 974
 975                 $aValidTokens = array();
 976                 $aDatabaseWords = chksql(
 977                     $this->oDB->getAll($sSQL),
 978                     "Could not get word tokens."
 979                 );
 980                 $aPossibleMainWordIDs = array();
 981                 $aWordFrequencyScores = array();
 982                 foreach ($aDatabaseWords as $aToken) {
 983                     // Very special case - require 2 letter country param to match the country code found
 984                     if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
 985                         && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
 986                     ) {
 987                         continue;
 988                     }
 989
 990                     if (isset($aValidTokens[$aToken['word_token']])) {
 991                         $aValidTokens[$aToken['word_token']][] = $aToken;
 992                     } else {
 993                         $aValidTokens[$aToken['word_token']] = array($aToken);
 994                     }
 995                     if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
 996                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
 997                 }
 998                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 999
1000                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
1001                 foreach ($aTokens as $sToken) {
1002                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
1003                         if (isset($aValidTokens[$aData[1]])) {
1004                             foreach ($aValidTokens[$aData[1]] as $aToken) {
1005                                 if (!$aToken['class']) {
1006                                     if (isset($aValidTokens[$sToken])) {
1007                                         $aValidTokens[$sToken][] = $aToken;
1008                                     } else {
1009                                         $aValidTokens[$sToken] = array($aToken);
1010                                     }
1011                                 }
1012                             }
1013                         }
1014                     }
1015                 }
1016
1017                 foreach ($aTokens as $sToken) {
1018                     // Unknown single word token with a number - assume it is a house number
1019                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
1020                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
1021                     }
1022                 }
1023
1024                 // Any words that have failed completely?
1025                 // TODO: suggestions
1026
1027                 // Start the search process
1028                 // array with: placeid => -1 | tiger-housenumber
1029                 $aResultPlaceIDs = array();
1030
1031                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
1032
1033                 if ($this->bReverseInPlan) {
1034                     // Reverse phrase array and also reverse the order of the wordsets in
1035                     // the first and final phrase. Don't bother about phrases in the middle
1036                     // because order in the address doesn't matter.
1037                     $aPhrases = array_reverse($aPhrases);
1038                     $aPhrases[0]['wordsets'] = getInverseWordSets($aPhrases[0]['words'], 0);
1039                     if (sizeof($aPhrases) > 1) {
1040                         $aFinalPhrase = end($aPhrases);
1041                         $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
1042                     }
1043                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
1044
1045                     foreach ($aGroupedSearches as $aSearches) {
1046                         foreach ($aSearches as $aSearch) {
1047                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
1048                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
1049                             }
1050                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
1051                         }
1052                     }
1053
1054                     $aGroupedSearches = $aReverseGroupedSearches;
1055                     ksort($aGroupedSearches);
1056                 }
1057             } else {
1058                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
1059                 $aGroupedSearches = array();
1060                 foreach ($aSearches as $aSearch) {
1061                     if ($aSearch->getRank() < $this->iMaxRank) {
1062                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
1063                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
1064                     }
1065                 }
1066                 ksort($aGroupedSearches);
1067             }
1068
1069             // Filter out duplicate searches
1070             $aSearchHash = array();
1071             foreach ($aGroupedSearches as $iGroup => $aSearches) {
1072                 foreach ($aSearches as $iSearch => $aSearch) {
1073                     $sHash = serialize($aSearch);
1074                     if (isset($aSearchHash[$sHash])) {
1075                         unset($aGroupedSearches[$iGroup][$iSearch]);
1076                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
1077                     } else {
1078                         $aSearchHash[$sHash] = 1;
1079                     }
1080                 }
1081             }
1082
1083             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
1084
1085             $iGroupLoop = 0;
1086             $iQueryLoop = 0;
1087             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
1088                 $iGroupLoop++;
1089                 foreach ($aSearches as $oSearch) {
1090                     $iQueryLoop++;
1091                     $searchedHousenumber = -1;
1092
1093                     if (CONST_Debug) echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
1094                     if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
1095
1096                     $aPlaceIDs = array();
1097                     if ($oSearch->isCountrySearch()) {
1098                         // Just looking for a country - look it up
1099                         if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
1100                             $aPlaceIDs = $oSearch->queryCountry($this->oDB);
1101                         }
1102                     } elseif (!$oSearch->isNamedSearch()) {
1103                         // looking for a POI in a geographic area
1104                         if (!$oCtx->isBoundedSearch()) {
1105                             continue;
1106                         }
1107
1108                         $aPlaceIDs = $oSearch->queryNearbyPoi(
1109                             $this->oDB,
1110                             $sCountryCodesSQL,
1111                             $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1112                             $this->iLimit
1113                         );
1114                     } elseif ($oSearch->isOperator(Operator::POSTCODE)) {
1115                         $aPlaceIDs = $oSearch->queryPostcode(
1116                             $this->oDB,
1117                             $sCountryCodesSQL,
1118                             $this->iLimit
1119                         );
1120                     } else {
1121                         // Ordinary search:
1122                         // First search for places according to name and address.
1123                         $aNamedPlaceIDs = $oSearch->queryNamedPlace(
1124                             $this->oDB,
1125                             $aWordFrequencyScores,
1126                             $sCountryCodesSQL,
1127                             $this->iMinAddressRank,
1128                             $this->iMaxAddressRank,
1129                             $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1130                             $this->iLimit
1131                         );
1132
1133                         if (sizeof($aNamedPlaceIDs)) {
1134                             foreach ($aNamedPlaceIDs as $aRow) {
1135                                 $aPlaceIDs[] = $aRow['place_id'];
1136                                 $this->exactMatchCache[$aRow['place_id']] = $aRow['exactmatch'];
1137                             }
1138                         }
1139
1140                         //now search for housenumber, if housenumber provided
1141                         if ($oSearch->hasHouseNumber() && sizeof($aPlaceIDs)) {
1142                             $aResult = $oSearch->queryHouseNumber(
1143                                 $this->oDB,
1144                                 $aPlaceIDs,
1145                                 $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1146                                 $this->iLimit
1147                             );
1148
1149                             if (sizeof($aResult)) {
1150                                 $searchedHousenumber = $aResult['iHouseNumber'];
1151                                 $aPlaceIDs = $aResult['aPlaceIDs'];
1152                             } elseif (!$oSearch->looksLikeFullAddress()) {
1153                                 $aPlaceIDs = array();
1154                             }
1155                         }
1156
1157                         // finally get POIs if requested
1158                         if ($oSearch->isPoiSearch() && sizeof($aPlaceIDs)) {
1159                             $aPlaceIDs = $oSearch->queryPoiByOperator(
1160                                 $this->oDB,
1161                                 $aPlaceIDs,
1162                                 $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '',
1163                                 $this->iLimit
1164                             );
1165                         }
1166                     }
1167
1168                     if (CONST_Debug) {
1169                         echo "<br><b>Place IDs:</b> ";
1170                         var_Dump($aPlaceIDs);
1171                     }
1172
1173                     if (sizeof($aPlaceIDs) && $oSearch->getPostcode()) {
1174                         $sSQL = 'SELECT place_id FROM placex';
1175                         $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
1176                         $sSQL .= " AND postcode = '".$oSearch->getPostcode()."'";
1177                         if (CONST_Debug) var_dump($sSQL);
1178                         $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1179                         if ($aFilteredPlaceIDs) {
1180                             $aPlaceIDs = $aFilteredPlaceIDs;
1181                             if (CONST_Debug) {
1182                                 echo "<br><b>Place IDs after postcode filtering:</b> ";
1183                                 var_Dump($aPlaceIDs);
1184                             }
1185                         }
1186                     }
1187
1188                     foreach ($aPlaceIDs as $iPlaceID) {
1189                         // array for placeID => -1 | Tiger housenumber
1190                         $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
1191                     }
1192                     if ($iQueryLoop > 20) break;
1193                 }
1194
1195                 if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
1196                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
1197                     // reduces the number of place ids, like a filter
1198                     // rank_address is 30 for interpolated housenumbers
1199                     $sWherePlaceId = 'WHERE place_id in (';
1200                     $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
1201
1202                     $sSQL = "SELECT place_id ";
1203                     $sSQL .= "FROM placex ".$sWherePlaceId;
1204                     $sSQL .= "  AND (";
1205                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1206                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
1207                         $sSQL .= "     OR (extratags->'place') = 'city'";
1208                     }
1209                     if ($this->aAddressRankList) {
1210                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
1211                     }
1212                     $sSQL .= "  ) UNION ";
1213                     $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
1214                     $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
1215                     if ($this->aAddressRankList) {
1216                         $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
1217                     }
1218                     $sSQL .= ") ";
1219                     if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
1220                         $sSQL .= "UNION ";
1221                         $sSQL .= "  SELECT place_id ";
1222                         $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
1223                     }
1224                     if ($this->iMaxAddressRank == 30) {
1225                         $sSQL .= "UNION ";
1226                         $sSQL .= "  SELECT place_id ";
1227                         $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;
1228                     }
1229                     if (CONST_Debug) var_dump($sSQL);
1230                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
1231                     $tempIDs = array();
1232                     foreach ($aFilteredPlaceIDs as $placeID) {
1233                         $tempIDs[$placeID] = $aResultPlaceIDs[$placeID];  //assign housenumber to placeID
1234                     }
1235                     $aResultPlaceIDs = $tempIDs;
1236                 }
1237
1238                 //exit;
1239                 if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break;
1240                 if ($iGroupLoop > 4) break;
1241                 if ($iQueryLoop > 30) break;
1242             }
1243
1244             // Did we find anything?
1245             if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) {
1246                 $aSearchResults = $this->getDetails($aResultPlaceIDs, $oCtx);
1247             }
1248         } else {
1249             // Just interpret as a reverse geocode
1250             $oReverse = new ReverseGeocode($this->oDB);
1251             $oReverse->setZoom(18);
1252
1253             $aLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
1254
1255             if (CONST_Debug) var_dump("Reverse search", $aLookup);
1256
1257             if ($aLookup['place_id']) {
1258                 $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1), $oCtx);
1259                 $aResultPlaceIDs[$aLookup['place_id']] = -1;
1260             } else {
1261                 $aSearchResults = array();
1262             }
1263         }
1264
1265         // No results? Done
1266         if (!sizeof($aSearchResults)) {
1267             if ($this->bFallback) {
1268                 if ($this->fallbackStructuredQuery()) {
1269                     return $this->lookup();
1270                 }
1271             }
1272
1273             return array();
1274         }
1275
1276         $aClassType = getClassTypesWithImportance();
1277         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
1278         foreach ($aRecheckWords as $i => $sWord) {
1279             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
1280         }
1281
1282         if (CONST_Debug) {
1283             echo '<i>Recheck words:<\i>';
1284             var_dump($aRecheckWords);
1285         }
1286
1287         $oPlaceLookup = new PlaceLookup($this->oDB);
1288         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
1289         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
1290         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
1291         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
1292         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
1293         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
1294
1295         foreach ($aSearchResults as $iResNum => $aResult) {
1296             // Default
1297             $fDiameter = getResultDiameter($aResult);
1298
1299             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
1300             if ($aOutlineResult) {
1301                 $aResult = array_merge($aResult, $aOutlineResult);
1302             }
1303
1304             if ($aResult['extra_place'] == 'city') {
1305                 $aResult['class'] = 'place';
1306                 $aResult['type'] = 'city';
1307                 $aResult['rank_search'] = 16;
1308             }
1309
1310             // Is there an icon set for this type of result?
1311             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
1312                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
1313             ) {
1314                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
1315             }
1316
1317             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
1318                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
1319             ) {
1320                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
1321             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
1322                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
1323             ) {
1324                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
1325             }
1326             // if tag '&addressdetails=1' is set in query
1327             if ($this->bIncludeAddressDetails) {
1328                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
1329                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);
1330                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
1331                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
1332                 }
1333             }
1334
1335             if ($this->bIncludeExtraTags) {
1336                 if ($aResult['extra']) {
1337                     $aResult['sExtraTags'] = json_decode($aResult['extra']);
1338                 } else {
1339                     $aResult['sExtraTags'] = (object) array();
1340                 }
1341             }
1342
1343             if ($this->bIncludeNameDetails) {
1344                 if ($aResult['names']) {
1345                     $aResult['sNameDetails'] = json_decode($aResult['names']);
1346                 } else {
1347                     $aResult['sNameDetails'] = (object) array();
1348                 }
1349             }
1350
1351             // Adjust importance for the number of exact string matches in the result
1352             $aResult['importance'] = max(0.001, $aResult['importance']);
1353             $iCountWords = 0;
1354             $sAddress = $aResult['langaddress'];
1355             foreach ($aRecheckWords as $i => $sWord) {
1356                 if (stripos($sAddress, $sWord)!==false) {
1357                     $iCountWords++;
1358                     if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
1359                 }
1360             }
1361
1362             $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
1363
1364             $aResult['name'] = $aResult['langaddress'];
1365             // secondary ordering (for results with same importance (the smaller the better):
1366             // - approximate importance of address parts
1367             $aResult['foundorder'] = -$aResult['addressimportance']/10;
1368             // - number of exact matches from the query
1369             if (isset($this->exactMatchCache[$aResult['place_id']])) {
1370                 $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']];
1371             } elseif (isset($this->exactMatchCache[$aResult['parent_place_id']])) {
1372                 $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']];
1373             }
1374             // - importance of the class/type
1375             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1376                 && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1377             ) {
1378                 $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1379             } else {
1380                 $aResult['foundorder'] += 0.01;
1381             }
1382             if (CONST_Debug) var_dump($aResult);
1383             $aSearchResults[$iResNum] = $aResult;
1384         }
1385         uasort($aSearchResults, 'byImportance');
1386
1387         $aOSMIDDone = array();
1388         $aClassTypeNameDone = array();
1389         $aToFilter = $aSearchResults;
1390         $aSearchResults = array();
1391
1392         $bFirst = true;
1393         foreach ($aToFilter as $iResNum => $aResult) {
1394             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1395             if ($bFirst) {
1396                 $fLat = $aResult['lat'];
1397                 $fLon = $aResult['lon'];
1398                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1399                 $bFirst = false;
1400             }
1401             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1402                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1403             ) {
1404                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1405                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1406                 $aSearchResults[] = $aResult;
1407             }
1408
1409             // Absolute limit on number of results
1410             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1411         }
1412
1413         return $aSearchResults;
1414     } // end lookup()
1415 } // end class