lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10
  11 class Geocode
  12 {
  13     protected $oDB;
  14
  15     protected $aLangPrefOrder = array();
  16
  17     protected $bIncludeAddressDetails = false;
  18     protected $bIncludeExtraTags = false;
  19     protected $bIncludeNameDetails = false;
  20
  21     protected $bIncludePolygonAsPoints = false;
  22     protected $bIncludePolygonAsText = false;
  23     protected $bIncludePolygonAsGeoJSON = false;
  24     protected $bIncludePolygonAsKML = false;
  25     protected $bIncludePolygonAsSVG = false;
  26     protected $fPolygonSimplificationThreshold = 0.0;
  27
  28     protected $aExcludePlaceIDs = array();
  29     protected $bDeDupe = true;
  30     protected $bReverseInPlan = false;
  31
  32     protected $iLimit = 20;
  33     protected $iFinalLimit = 10;
  34     protected $iOffset = 0;
  35     protected $bFallback = false;
  36
  37     protected $aCountryCodes = false;
  38
  39     protected $bBoundedSearch = false;
  40     protected $aViewBox = false;
  41     protected $aRoutePoints = false;
  42     protected $aRouteWidth = false;
  43
  44     protected $iMaxRank = 20;
  45     protected $iMinAddressRank = 0;
  46     protected $iMaxAddressRank = 30;
  47     protected $aAddressRankList = array();
  48
  49     protected $sAllowedTypesSQLList = false;
  50
  51     protected $sQuery = false;
  52     protected $aStructuredQuery = false;
  53
  54     protected $oNormalizer = null;
  55
  56
  57     public function __construct(&$oDB)
  58     {
  59         $this->oDB =& $oDB;
  60         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  61     }
  62
  63     private function normTerm($sTerm)
  64     {
  65         if ($this->oNormalizer === null) {
  66             return $sTerm;
  67         }
  68
  69         return $this->oNormalizer->transliterate($sTerm);
  70     }
  71
  72     public function setReverseInPlan($bReverse)
  73     {
  74         $this->bReverseInPlan = $bReverse;
  75     }
  76
  77     public function setLanguagePreference($aLangPref)
  78     {
  79         $this->aLangPrefOrder = $aLangPref;
  80     }
  81
  82     public function getMoreUrlParams()
  83     {
  84         if ($this->aStructuredQuery) {
  85             $aParams = $this->aStructuredQuery;
  86         } else {
  87             $aParams = array('q' => $this->sQuery);
  88         }
  89
  90         if ($this->aExcludePlaceIDs) {
  91             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  92         }
  93
  94         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  95         if ($this->bIncludeExtraTags) $aParams['extratags'] = '1';
  96         if ($this->bIncludeNameDetails) $aParams['namedetails'] = '1';
  97
  98         if ($this->bIncludePolygonAsPoints) $aParams['polygon'] = '1';
  99         if ($this->bIncludePolygonAsText) $aParams['polygon_text'] = '1';
 100         if ($this->bIncludePolygonAsGeoJSON) $aParams['polygon_geojson'] = '1';
 101         if ($this->bIncludePolygonAsKML) $aParams['polygon_kml'] = '1';
 102         if ($this->bIncludePolygonAsSVG) $aParams['polygon_svg'] = '1';
 103
 104         if ($this->fPolygonSimplificationThreshold > 0.0) {
 105             $aParams['polygon_threshold'] = $this->fPolygonSimplificationThreshold;
 106         }
 107
 108         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
 109         if (!$this->bDeDupe) $aParams['dedupe'] = '0';
 110
 111         if ($this->aCountryCodes) {
 112             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
 113         }
 114
 115         if ($this->aViewBox) {
 116             $aParams['viewbox'] = join(',', $this->aViewBox);
 117         }
 118
 119         return $aParams;
 120     }
 121
 122     public function setIncludePolygonAsPoints($b = true)
 123     {
 124         $this->bIncludePolygonAsPoints = $b;
 125     }
 126
 127     public function setIncludePolygonAsText($b = true)
 128     {
 129         $this->bIncludePolygonAsText = $b;
 130     }
 131
 132     public function setIncludePolygonAsGeoJSON($b = true)
 133     {
 134         $this->bIncludePolygonAsGeoJSON = $b;
 135     }
 136
 137     public function setIncludePolygonAsKML($b = true)
 138     {
 139         $this->bIncludePolygonAsKML = $b;
 140     }
 141
 142     public function setIncludePolygonAsSVG($b = true)
 143     {
 144         $this->bIncludePolygonAsSVG = $b;
 145     }
 146
 147     public function setPolygonSimplificationThreshold($f)
 148     {
 149         $this->fPolygonSimplificationThreshold = $f;
 150     }
 151
 152     public function setLimit($iLimit = 10)
 153     {
 154         if ($iLimit > 50) $iLimit = 50;
 155         if ($iLimit < 1) $iLimit = 1;
 156
 157         $this->iFinalLimit = $iLimit;
 158         $this->iLimit = $iLimit + min($iLimit, 10);
 159     }
 160
 161     public function setFeatureType($sFeatureType)
 162     {
 163         switch ($sFeatureType) {
 164             case 'country':
 165                 $this->setRankRange(4, 4);
 166                 break;
 167             case 'state':
 168                 $this->setRankRange(8, 8);
 169                 break;
 170             case 'city':
 171                 $this->setRankRange(14, 16);
 172                 break;
 173             case 'settlement':
 174                 $this->setRankRange(8, 20);
 175                 break;
 176         }
 177     }
 178
 179     public function setRankRange($iMin, $iMax)
 180     {
 181         $this->iMinAddressRank = $iMin;
 182         $this->iMaxAddressRank = $iMax;
 183     }
 184
 185     public function setViewbox($aViewbox)
 186     {
 187         $aBox = array_map('floatval', $aViewbox);
 188
 189         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 190         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 191         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 192         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 193
 194         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 195             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 196         ) {
 197             userError("Bad parameter 'viewbox'. Not a box.");
 198         }
 199     }
 200
 201     private function viewboxImportanceFactor($fX, $fY)
 202     {
 203         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 204         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 205
 206         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 207         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 208
 209         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 210             return 1;
 211         }
 212
 213         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 214             return 0.5;
 215         }
 216
 217         return 0.25;
 218     }
 219
 220     public function setQuery($sQueryString)
 221     {
 222         $this->sQuery = $sQueryString;
 223         $this->aStructuredQuery = false;
 224     }
 225
 226     public function getQueryString()
 227     {
 228         return $this->sQuery;
 229     }
 230
 231
 232     public function loadParamArray($oParams)
 233     {
 234         $this->bIncludeAddressDetails
 235          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 236         $this->bIncludeExtraTags
 237          = $oParams->getBool('extratags', $this->bIncludeExtraTags);
 238         $this->bIncludeNameDetails
 239          = $oParams->getBool('namedetails', $this->bIncludeNameDetails);
 240
 241         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 242         $this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe);
 243
 244         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 245         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 246
 247         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 248
 249         // List of excluded Place IDs - used for more acurate pageing
 250         $sExcluded = $oParams->getStringList('exclude_place_ids');
 251         if ($sExcluded) {
 252             foreach ($sExcluded as $iExcludedPlaceID) {
 253                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 254                 if ($iExcludedPlaceID)
 255                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 256             }
 257
 258             if (isset($aExcludePlaceIDs))
 259                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 260         }
 261
 262         // Only certain ranks of feature
 263         $sFeatureType = $oParams->getString('featureType');
 264         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 265         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 266
 267         // Country code list
 268         $sCountries = $oParams->getStringList('countrycodes');
 269         if ($sCountries) {
 270             foreach ($sCountries as $sCountryCode) {
 271                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 272                     $aCountries[] = strtolower($sCountryCode);
 273                 }
 274             }
 275             if (isset($aCountries))
 276                 $this->aCountryCodes = $aCountries;
 277         }
 278
 279         $aViewbox = $oParams->getStringList('viewboxlbrt');
 280         if ($aViewbox) {
 281             if (count($aViewbox) != 4) {
 282                 userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
 283             }
 284             $this->setViewbox($aViewbox);
 285         } else {
 286             $aViewbox = $oParams->getStringList('viewbox');
 287             if ($aViewbox) {
 288                 if (count($aViewbox) != 4) {
 289                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
 290                 }
 291                 $this->setViewBox($aViewbox);
 292             } else {
 293                 $aRoute = $oParams->getStringList('route');
 294                 $fRouteWidth = $oParams->getFloat('routewidth');
 295                 if ($aRoute && $fRouteWidth) {
 296                     $this->aRoutePoints = $aRoute;
 297                     $this->aRouteWidth = $fRouteWidth;
 298                 }
 299             }
 300         }
 301     }
 302
 303     public function setQueryFromParams($oParams)
 304     {
 305         // Search query
 306         $sQuery = $oParams->getString('q');
 307         if (!$sQuery) {
 308             $this->setStructuredQuery(
 309                 $oParams->getString('amenity'),
 310                 $oParams->getString('street'),
 311                 $oParams->getString('city'),
 312                 $oParams->getString('county'),
 313                 $oParams->getString('state'),
 314                 $oParams->getString('country'),
 315                 $oParams->getString('postalcode')
 316             );
 317             $this->setReverseInPlan(false);
 318         } else {
 319             $this->setQuery($sQuery);
 320         }
 321     }
 322
 323     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 324     {
 325         $sValue = trim($sValue);
 326         if (!$sValue) return false;
 327         $this->aStructuredQuery[$sKey] = $sValue;
 328         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 329             $this->iMinAddressRank = $iNewMinAddressRank;
 330             $this->iMaxAddressRank = $iNewMaxAddressRank;
 331         }
 332         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 333         return true;
 334     }
 335
 336     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 337     {
 338         $this->sQuery = false;
 339
 340         // Reset
 341         $this->iMinAddressRank = 0;
 342         $this->iMaxAddressRank = 30;
 343         $this->aAddressRankList = array();
 344
 345         $this->aStructuredQuery = array();
 346         $this->sAllowedTypesSQLList = false;
 347
 348         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 349         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 350         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 351         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 352         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 353         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 354         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 355
 356         if (sizeof($this->aStructuredQuery) > 0) {
 357             $this->sQuery = join(', ', $this->aStructuredQuery);
 358             if ($this->iMaxAddressRank < 30) {
 359                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 360             }
 361         }
 362     }
 363
 364     public function fallbackStructuredQuery()
 365     {
 366         if (!$this->aStructuredQuery) return false;
 367
 368         $aParams = $this->aStructuredQuery;
 369
 370         if (sizeof($aParams) == 1) return false;
 371
 372         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 373
 374         foreach ($aOrderToFallback as $sType) {
 375             if (isset($aParams[$sType])) {
 376                 unset($aParams[$sType]);
 377                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 378                 return true;
 379             }
 380         }
 381
 382         return false;
 383     }
 384
 385     public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
 386     {
 387         /*
 388              Calculate all searches using aValidTokens i.e.
 389              'Wodsworth Road, Sheffield' =>
 390
 391              Phrase Wordset
 392              0      0       (wodsworth road)
 393              0      1       (wodsworth)(road)
 394              1      0       (sheffield)
 395
 396              Score how good the search is so they can be ordered
 397          */
 398         $iGlobalRank = 0;
 399
 400         foreach ($aPhrases as $iPhrase => $oPhrase) {
 401             $aNewPhraseSearches = array();
 402             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 403
 404             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 405                 // Too many permutations - too expensive
 406                 if ($iWordSet > 120) break;
 407
 408                 $aWordsetSearches = $aSearches;
 409
 410                 // Add all words from this wordset
 411                 foreach ($aWordset as $iToken => $sToken) {
 412                     //echo "<br><b>$sToken</b>";
 413                     $aNewWordsetSearches = array();
 414
 415                     foreach ($aWordsetSearches as $oCurrentSearch) {
 416                         //echo "<i>";
 417                         //var_dump($oCurrentSearch);
 418                         //echo "</i>";
 419
 420                         // If the token is valid
 421                         if (isset($aValidTokens[' '.$sToken])) {
 422                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 423                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 424                                     $aSearchTerm,
 425                                     isset($aValidTokens[$sToken])
 426                                       && strpos($sToken, ' ') === false,
 427                                     $sPhraseType,
 428                                     $iToken == 0 && $iPhrase == 0,
 429                                     $iPhrase == 0,
 430                                     $iToken + 1 == sizeof($aWordset)
 431                                       && $iPhrase + 1 == sizeof($aPhrases),
 432                                     $iGlobalRank
 433                                 );
 434
 435                                 foreach ($aNewSearches as $oSearch) {
 436                                     if ($oSearch->getRank() < $this->iMaxRank) {
 437                                         $aNewWordsetSearches[] = $oSearch;
 438                                     }
 439                                 }
 440                             }
 441                         }
 442                         // Look for partial matches.
 443                         // Note that there is no point in adding country terms here
 444                         // because country is omitted in the address.
 445                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 446                             // Allow searching for a word - but at extra cost
 447                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 448                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 449                                     $aSearchTerm,
 450                                     $bIsStructured,
 451                                     $iPhrase,
 452                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 453                                 );
 454
 455                                 foreach ($aNewSearches as $oSearch) {
 456                                     if ($oSearch->getRank() < $this->iMaxRank) {
 457                                         $aNewWordsetSearches[] = $oSearch;
 458                                     }
 459                                 }
 460                             }
 461                         }
 462                     }
 463                     // Sort and cut
 464                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 465                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 466                 }
 467                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 468
 469                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 470                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 471
 472                 $aSearchHash = array();
 473                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 474                     $sHash = serialize($aSearch);
 475                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 476                     else $aSearchHash[$sHash] = 1;
 477                 }
 478
 479                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 480             }
 481
 482             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 483             $aGroupedSearches = array();
 484             foreach ($aNewPhraseSearches as $aSearch) {
 485                 $iRank = $aSearch->getRank();
 486                 if ($iRank < $this->iMaxRank) {
 487                     if (!isset($aGroupedSearches[$iRank])) {
 488                         $aGroupedSearches[$iRank] = array();
 489                     }
 490                     $aGroupedSearches[$iRank][] = $aSearch;
 491                 }
 492             }
 493             ksort($aGroupedSearches);
 494
 495             $iSearchCount = 0;
 496             $aSearches = array();
 497             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 498                 $iSearchCount += sizeof($aNewSearches);
 499                 $aSearches = array_merge($aSearches, $aNewSearches);
 500                 if ($iSearchCount > 50) break;
 501             }
 502
 503             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 504         }
 505
 506         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 507         $aGroupedSearches = array();
 508         foreach ($aSearches as $oSearch) {
 509             if (!$oSearch->isValidSearch()) {
 510                 continue;
 511             }
 512
 513             $iRank = $oSearch->addToRank($iGlobalRank);
 514             if (!isset($aGroupedSearches[$iRank])) {
 515                 $aGroupedSearches[$iRank] = array();
 516             }
 517             $aGroupedSearches[$iRank][] = $oSearch;
 518         }
 519         ksort($aGroupedSearches);
 520
 521         return $aGroupedSearches;
 522     }
 523
 524     /* Perform the actual query lookup.
 525
 526         Returns an ordered list of results, each with the following fields:
 527             osm_type: type of corresponding OSM object
 528                         N - node
 529                         W - way
 530                         R - relation
 531                         P - postcode (internally computed)
 532             osm_id: id of corresponding OSM object
 533             class: general object class (corresponds to tag key of primary OSM tag)
 534             type: subclass of object (corresponds to tag value of primary OSM tag)
 535             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 536             rank_search: rank in search hierarchy
 537                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 538             rank_address: rank in address hierarchy (determines orer in address)
 539             place_id: internal key (may differ between different instances)
 540             country_code: ISO country code
 541             langaddress: localized full address
 542             placename: localized name of object
 543             ref: content of ref tag (if available)
 544             lon: longitude
 545             lat: latitude
 546             importance: importance of place based on Wikipedia link count
 547             addressimportance: cumulated importance of address elements
 548             extra_place: type of place (for admin boundaries, if there is a place tag)
 549             aBoundingBox: bounding Box
 550             label: short description of the object class/type (English only)
 551             name: full name (currently the same as langaddress)
 552             foundorder: secondary ordering for places with same importance
 553     */
 554
 555
 556     public function lookup()
 557     {
 558         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 559
 560         $oCtx = new SearchContext();
 561
 562         if ($this->aRoutePoints) {
 563             $oCtx->setViewboxFromRoute(
 564                 $this->oDB,
 565                 $this->aRoutePoints,
 566                 $this->aRouteWidth,
 567                 $this->bBoundedSearch
 568             );
 569         } elseif ($this->aViewBox) {
 570             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 571         }
 572         if ($this->aExcludePlaceIDs) {
 573             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 574         }
 575         if ($this->aCountryCodes) {
 576             $oCtx->setCountryList($this->aCountryCodes);
 577         }
 578
 579         $sNormQuery = $this->normTerm($this->sQuery);
 580         $sLanguagePrefArraySQL = getArraySQL(
 581             array_map("getDBQuoted", $this->aLangPrefOrder)
 582         );
 583
 584         $sQuery = $this->sQuery;
 585         if (!preg_match('//u', $sQuery)) {
 586             userError("Query string is not UTF-8 encoded.");
 587         }
 588
 589         // Conflicts between US state abreviations and various words for 'the' in different languages
 590         if (isset($this->aLangPrefOrder['name:en'])) {
 591             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 592             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 593             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 594         }
 595
 596         // Do we have anything that looks like a lat/lon pair?
 597         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 598
 599         $aResults = array();
 600         if ($sQuery || $this->aStructuredQuery) {
 601             // Start with a single blank search
 602             $aSearches = array(new SearchDescription($oCtx));
 603
 604             if ($sQuery) {
 605                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 606             }
 607
 608             $sSpecialTerm = '';
 609             if ($sQuery) {
 610                 preg_match_all(
 611                     '/\\[([\\w ]*)\\]/u',
 612                     $sQuery,
 613                     $aSpecialTermsRaw,
 614                     PREG_SET_ORDER
 615                 );
 616                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 617                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 618                     if (!$sSpecialTerm) {
 619                         $sSpecialTerm = $aSpecialTerm[1];
 620                     }
 621                 }
 622             }
 623             if (!$sSpecialTerm && $this->aStructuredQuery
 624                 && isset($this->aStructuredQuery['amenity'])) {
 625                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 626                 unset($this->aStructuredQuery['amenity']);
 627             }
 628
 629             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 630                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 631                 $sToken = chksql(
 632                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 633                     "Cannot decode query. Wrong encoding?"
 634                 );
 635                 $sSQL = 'SELECT class, type FROM word ';
 636                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 637                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 638                 if (CONST_Debug) var_Dump($sSQL);
 639                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 640                 $aNewSearches = array();
 641                 foreach ($aSearches as $oSearch) {
 642                     foreach ($aSearchWords as $aSearchTerm) {
 643                         $oNewSearch = clone $oSearch;
 644                         $oNewSearch->setPoiSearch(
 645                             Operator::TYPE,
 646                             $aSearchTerm['class'],
 647                             $aSearchTerm['type']
 648                         );
 649                         $aNewSearches[] = $oNewSearch;
 650                     }
 651                 }
 652                 $aSearches = $aNewSearches;
 653             }
 654
 655             // Split query into phrases
 656             // Commas are used to reduce the search space by indicating where phrases split
 657             if ($this->aStructuredQuery) {
 658                 $aInPhrases = $this->aStructuredQuery;
 659                 $bStructuredPhrases = true;
 660             } else {
 661                 $aInPhrases = explode(',', $sQuery);
 662                 $bStructuredPhrases = false;
 663             }
 664
 665             // Convert each phrase to standard form
 666             // Create a list of standard words
 667             // Get all 'sets' of words
 668             // Generate a complete list of all
 669             $aTokens = array();
 670             $aPhrases = array();
 671             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 672                 $sPhrase = chksql(
 673                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 674                     "Cannot normalize query string (is it a UTF-8 string?)"
 675                 );
 676                 if (trim($sPhrase)) {
 677                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 678                     $oPhrase->addTokens($aTokens);
 679                     $aPhrases[] = $oPhrase;
 680                 }
 681             }
 682
 683             if (sizeof($aTokens)) {
 684                 // Check which tokens we have, get the ID numbers
 685                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 686                 $sSQL .= ' FROM word ';
 687                 $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')';
 688
 689                 if (CONST_Debug) var_Dump($sSQL);
 690
 691                 $aValidTokens = array();
 692                 $aDatabaseWords = chksql(
 693                     $this->oDB->getAll($sSQL),
 694                     "Could not get word tokens."
 695                 );
 696                 $aWordFrequencyScores = array();
 697                 foreach ($aDatabaseWords as $aToken) {
 698                     // Filter country tokens that do not match restricted countries.
 699                     if ($this->aCountryCodes
 700                         && $aToken['country_code']
 701                         && !in_array($aToken['country_code'], $this->aCountryCodes)
 702                     ) {
 703                         continue;
 704                     }
 705
 706                     // Special terms need to appear in their normalized form.
 707                     if ($aToken['word'] && $aToken['class']) {
 708                         $sNormWord = $this->normTerm($aToken['word']);
 709                         if (strpos($sNormQuery, $sNormWord) === false) {
 710                             continue;
 711                         }
 712                     }
 713
 714                     if (isset($aValidTokens[$aToken['word_token']])) {
 715                         $aValidTokens[$aToken['word_token']][] = $aToken;
 716                     } else {
 717                         $aValidTokens[$aToken['word_token']] = array($aToken);
 718                     }
 719                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
 720                 }
 721                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 722
 723                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
 724                 foreach ($aTokens as $sToken) {
 725                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 726                         if (isset($aValidTokens[$aData[1]])) {
 727                             foreach ($aValidTokens[$aData[1]] as $aToken) {
 728                                 if (!$aToken['class']) {
 729                                     if (isset($aValidTokens[$sToken])) {
 730                                         $aValidTokens[$sToken][] = $aToken;
 731                                     } else {
 732                                         $aValidTokens[$sToken] = array($aToken);
 733                                     }
 734                                 }
 735                             }
 736                         }
 737                     }
 738                 }
 739
 740                 foreach ($aTokens as $sToken) {
 741                     // Unknown single word token with a number - assume it is a house number
 742                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
 743                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
 744                     }
 745                 }
 746
 747                 // Any words that have failed completely?
 748                 // TODO: suggestions
 749
 750                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 751
 752                 if ($this->bReverseInPlan) {
 753                     // Reverse phrase array and also reverse the order of the wordsets in
 754                     // the first and final phrase. Don't bother about phrases in the middle
 755                     // because order in the address doesn't matter.
 756                     $aPhrases = array_reverse($aPhrases);
 757                     $aPhrases[0]->invertWordSets();
 758                     if (sizeof($aPhrases) > 1) {
 759                         $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
 760                     }
 761                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 762
 763                     foreach ($aGroupedSearches as $aSearches) {
 764                         foreach ($aSearches as $aSearch) {
 765                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 766                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 767                             }
 768                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 769                         }
 770                     }
 771
 772                     $aGroupedSearches = $aReverseGroupedSearches;
 773                     ksort($aGroupedSearches);
 774                 }
 775             } else {
 776                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 777                 $aGroupedSearches = array();
 778                 foreach ($aSearches as $aSearch) {
 779                     if ($aSearch->getRank() < $this->iMaxRank) {
 780                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 781                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 782                     }
 783                 }
 784                 ksort($aGroupedSearches);
 785             }
 786
 787             // Filter out duplicate searches
 788             $aSearchHash = array();
 789             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 790                 foreach ($aSearches as $iSearch => $aSearch) {
 791                     $sHash = serialize($aSearch);
 792                     if (isset($aSearchHash[$sHash])) {
 793                         unset($aGroupedSearches[$iGroup][$iSearch]);
 794                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
 795                     } else {
 796                         $aSearchHash[$sHash] = 1;
 797                     }
 798                 }
 799             }
 800
 801             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 802
 803             // Start the search process
 804             $iGroupLoop = 0;
 805             $iQueryLoop = 0;
 806             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 807                 $iGroupLoop++;
 808                 foreach ($aSearches as $oSearch) {
 809                     $iQueryLoop++;
 810
 811                     if (CONST_Debug) {
 812                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
 813                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
 814                     }
 815
 816                     $aResults += $oSearch->query(
 817                         $this->oDB,
 818                         $aWordFrequencyScores,
 819                         $this->iMinAddressRank,
 820                         $this->iMaxAddressRank,
 821                         $this->iLimit
 822                     );
 823
 824                     if ($iQueryLoop > 20) break;
 825                 }
 826
 827                 if (sizeof($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 828                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 829                     // reduces the number of place ids, like a filter
 830                     // rank_address is 30 for interpolated housenumbers
 831                     $aFilterSql = array();
 832                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 833                     if ($sPlaceIds) {
 834                         $sSQL = 'SELECT place_id FROM placex ';
 835                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 836                         $sSQL .= "  AND (";
 837                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 838                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 839                             $sSQL .= "     OR (extratags->'place') = 'city'";
 840                         }
 841                         if ($this->aAddressRankList) {
 842                             $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
 843                         }
 844                         $sSQL .= ")";
 845                         $aFilterSql[] = $sSQL;
 846                     }
 847                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 848                     if ($sPlaceIds) {
 849                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 850                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 851                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 852                         if ($this->aAddressRankList) {
 853                             $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
 854                         }
 855                         $sSQL .= ") ";
 856                         $aFilterSql[] = $sSQL;
 857                     }
 858
 859                     $aFilteredIDs = array();
 860                     if ($aFilterSql) {
 861                         $sSQL = join(' UNION ', $aFilterSql);
 862                         if (CONST_Debug) var_dump($sSQL);
 863                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 864                     }
 865
 866                     $tempIDs = array();
 867                     foreach ($aResults as $oResult) {
 868                         if (($this->iMaxAddressRank == 30 &&
 869                              ($oResult->iTable == Result::TABLE_OSMLINE
 870                               || $oResult->iTable == Result::TABLE_AUX
 871                               || $oResult->iTable == Result::TABLE_TIGER))
 872                             || in_array($oResult->iId, $aFilteredIDs)
 873                         ) {
 874                             $tempIDs[$oResult->iId] = $oResult;
 875                         }
 876                     }
 877                     $aResults = $tempIDs;
 878                 }
 879
 880                 if (sizeof($aResults)) break;
 881                 if ($iGroupLoop > 4) break;
 882                 if ($iQueryLoop > 30) break;
 883             }
 884         } else {
 885             // Just interpret as a reverse geocode
 886             $oReverse = new ReverseGeocode($this->oDB);
 887             $oReverse->setZoom(18);
 888
 889             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 890
 891             if (CONST_Debug) var_dump("Reverse search", $aLookup);
 892
 893             if ($oLookup) {
 894                 $aResults = array($oLookup->iId => $oLookup);
 895             }
 896         }
 897
 898         // No results? Done
 899         if (!sizeof($aResults)) {
 900             if ($this->bFallback) {
 901                 if ($this->fallbackStructuredQuery()) {
 902                     return $this->lookup();
 903                 }
 904             }
 905
 906             return array();
 907         }
 908
 909         $oPlaceLookup = new PlaceLookup($this->oDB);
 910         $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
 911         $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
 912         $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
 913         $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
 914         $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
 915         $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
 916         $oPlaceLookup->setDeDupe($this->bDeDupe);
 917         if ($this->aAddressRankList) {
 918             $oPlaceLookup->setAddressRankList($this->aAddressRankList);
 919         }
 920         $oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 921         $oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 922         $oPlaceLookup->setIncludeExtraTags($this->bIncludeExtraTags);
 923         $oPlaceLookup->setIncludeNameDetails($this->bIncludeNameDetails);
 924         if ($oCtx->hasNearPoint()) {
 925             $oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 926         }
 927
 928         $aSearchResults = $oPlaceLookup->lookup($aResults);
 929
 930         $aClassType = getClassTypesWithImportance();
 931         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 932         foreach ($aRecheckWords as $i => $sWord) {
 933             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 934         }
 935
 936         if (CONST_Debug) {
 937             echo '<i>Recheck words:<\i>';
 938             var_dump($aRecheckWords);
 939         }
 940
 941         foreach ($aSearchResults as $iIdx => $aResult) {
 942             // Default
 943             $fDiameter = getResultDiameter($aResult);
 944
 945             $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 946             if ($aOutlineResult) {
 947                 $aResult = array_merge($aResult, $aOutlineResult);
 948             }
 949
 950             if ($aResult['extra_place'] == 'city') {
 951                 $aResult['class'] = 'place';
 952                 $aResult['type'] = 'city';
 953                 $aResult['rank_search'] = 16;
 954             }
 955
 956             // Is there an icon set for this type of result?
 957             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
 958                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
 959             ) {
 960                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
 961             }
 962
 963             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
 964                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
 965             ) {
 966                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
 967             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
 968                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
 969             ) {
 970                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
 971             }
 972             // if tag '&addressdetails=1' is set in query
 973             if ($this->bIncludeAddressDetails) {
 974                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
 975                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber);
 976                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
 977                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
 978                 }
 979             }
 980
 981             $aResult['name'] = $aResult['langaddress'];
 982
 983             if ($oCtx->hasNearPoint()) {
 984                 $aResult['importance'] = 0.001;
 985                 $aResult['foundorder'] = $aResult['addressimportance'];
 986             } else {
 987                 // Adjust importance for the number of exact string matches in the result
 988                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 989                     $aResult['lon'],
 990                     $aResult['lat']
 991                 );
 992                 $aResult['importance'] = max(0.001, $aResult['importance']);
 993                 $iCountWords = 0;
 994                 $sAddress = $aResult['langaddress'];
 995                 foreach ($aRecheckWords as $i => $sWord) {
 996                     if (stripos($sAddress, $sWord)!==false) {
 997                         $iCountWords++;
 998                         if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
 999                     }
1000                 }
1001
1002                 $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
1003
1004                 // secondary ordering (for results with same importance (the smaller the better):
1005                 // - approximate importance of address parts
1006                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
1007                 // - number of exact matches from the query
1008                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
1009                 // - importance of the class/type
1010                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
1011                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
1012                 ) {
1013                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
1014                 } else {
1015                     $aResult['foundorder'] += 0.01;
1016                 }
1017             }
1018             if (CONST_Debug) var_dump($aResult);
1019             $aSearchResults[$iIdx] = $aResult;
1020         }
1021         uasort($aSearchResults, 'byImportance');
1022
1023         $aOSMIDDone = array();
1024         $aClassTypeNameDone = array();
1025         $aToFilter = $aSearchResults;
1026         $aSearchResults = array();
1027
1028         if (CONST_Debug) var_dump($aToFilter);
1029
1030         $bFirst = true;
1031         foreach ($aToFilter as $aResult) {
1032             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
1033             if ($bFirst) {
1034                 $fLat = $aResult['lat'];
1035                 $fLon = $aResult['lon'];
1036                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
1037                 $bFirst = false;
1038             }
1039             if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
1040                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
1041             ) {
1042                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
1043                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
1044                 $aSearchResults[] = $aResult;
1045             }
1046
1047             // Absolute limit on number of results
1048             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
1049         }
1050
1051         if (CONST_Debug) var_dump($aSearchResults);
1052         return $aSearchResults;
1053     } // end lookup()
1054 } // end class