lib-php/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_LibDir.'/PlaceLookup.php');
   6 require_once(CONST_LibDir.'/Phrase.php');
   7 require_once(CONST_LibDir.'/ReverseGeocode.php');
   8 require_once(CONST_LibDir.'/SearchDescription.php');
   9 require_once(CONST_LibDir.'/SearchContext.php');
  10 require_once(CONST_LibDir.'/TokenList.php');
  11 require_once(CONST_TokenizerDir.'/tokenizer.php');
  12
  13 class Geocode
  14 {
  15     protected $oDB;
  16
  17     protected $oPlaceLookup;
  18     protected $oTokenizer;
  19
  20     protected $aLangPrefOrder = array();
  21
  22     protected $aExcludePlaceIDs = array();
  23
  24     protected $iLimit = 20;
  25     protected $iFinalLimit = 10;
  26     protected $iOffset = 0;
  27     protected $bFallback = false;
  28
  29     protected $aCountryCodes = false;
  30
  31     protected $bBoundedSearch = false;
  32     protected $aViewBox = false;
  33     protected $aRoutePoints = false;
  34     protected $aRouteWidth = false;
  35
  36     protected $iMaxRank = 20;
  37     protected $iMinAddressRank = 0;
  38     protected $iMaxAddressRank = 30;
  39     protected $aAddressRankList = array();
  40
  41     protected $sAllowedTypesSQLList = false;
  42
  43     protected $sQuery = false;
  44     protected $aStructuredQuery = false;
  45
  46
  47     public function __construct(&$oDB)
  48     {
  49         $this->oDB =& $oDB;
  50         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  51         $this->oTokenizer = new \Nominatim\Tokenizer($this->oDB);
  52     }
  53
  54     public function setLanguagePreference($aLangPref)
  55     {
  56         $this->aLangPrefOrder = $aLangPref;
  57     }
  58
  59     public function getMoreUrlParams()
  60     {
  61         if ($this->aStructuredQuery) {
  62             $aParams = $this->aStructuredQuery;
  63         } else {
  64             $aParams = array('q' => $this->sQuery);
  65         }
  66
  67         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  68
  69         if ($this->aExcludePlaceIDs) {
  70             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  71         }
  72
  73         if ($this->bBoundedSearch) {
  74             $aParams['bounded'] = '1';
  75         }
  76
  77         if ($this->aCountryCodes) {
  78             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  79         }
  80
  81         if ($this->aViewBox) {
  82             $aParams['viewbox'] = join(',', $this->aViewBox);
  83         }
  84
  85         return $aParams;
  86     }
  87
  88     public function setLimit($iLimit = 10)
  89     {
  90         if ($iLimit > 50) {
  91             $iLimit = 50;
  92         } elseif ($iLimit < 1) {
  93             $iLimit = 1;
  94         }
  95
  96         $this->iFinalLimit = $iLimit;
  97         $this->iLimit = $iLimit + min($iLimit, 10);
  98     }
  99
 100     public function setFeatureType($sFeatureType)
 101     {
 102         switch ($sFeatureType) {
 103             case 'country':
 104                 $this->setRankRange(4, 4);
 105                 break;
 106             case 'state':
 107                 $this->setRankRange(8, 8);
 108                 break;
 109             case 'city':
 110                 $this->setRankRange(14, 16);
 111                 break;
 112             case 'settlement':
 113                 $this->setRankRange(8, 20);
 114                 break;
 115         }
 116     }
 117
 118     public function setRankRange($iMin, $iMax)
 119     {
 120         $this->iMinAddressRank = $iMin;
 121         $this->iMaxAddressRank = $iMax;
 122     }
 123
 124     public function setViewbox($aViewbox)
 125     {
 126         $aBox = array_map('floatval', $aViewbox);
 127
 128         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 129         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 130         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 131         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 132
 133         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 134             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 135         ) {
 136             userError("Bad parameter 'viewbox'. Not a box.");
 137         }
 138     }
 139
 140     private function viewboxImportanceFactor($fX, $fY)
 141     {
 142         if (!$this->aViewBox) {
 143             return 1;
 144         }
 145
 146         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 147         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 148
 149         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 150         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 151
 152         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 153             return 1;
 154         }
 155
 156         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 157             return 0.5;
 158         }
 159
 160         return 0.25;
 161     }
 162
 163     public function setQuery($sQueryString)
 164     {
 165         $this->sQuery = $sQueryString;
 166         $this->aStructuredQuery = false;
 167     }
 168
 169     public function getQueryString()
 170     {
 171         return $this->sQuery;
 172     }
 173
 174
 175     public function loadParamArray($oParams, $sForceGeometryType = null)
 176     {
 177         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 178
 179         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 180         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 181
 182         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 183
 184         // List of excluded Place IDs - used for more acurate pageing
 185         $sExcluded = $oParams->getStringList('exclude_place_ids');
 186         if ($sExcluded) {
 187             foreach ($sExcluded as $iExcludedPlaceID) {
 188                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 189                 if ($iExcludedPlaceID) {
 190                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 191                 }
 192             }
 193
 194             if (isset($aExcludePlaceIDs)) {
 195                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 196             }
 197         }
 198
 199         // Only certain ranks of feature
 200         $sFeatureType = $oParams->getString('featureType');
 201         if (!$sFeatureType) {
 202             $sFeatureType = $oParams->getString('featuretype');
 203         }
 204         if ($sFeatureType) {
 205             $this->setFeatureType($sFeatureType);
 206         }
 207
 208         // Country code list
 209         $sCountries = $oParams->getStringList('countrycodes');
 210         if ($sCountries) {
 211             foreach ($sCountries as $sCountryCode) {
 212                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 213                     $aCountries[] = strtolower($sCountryCode);
 214                 }
 215             }
 216             if (isset($aCountries)) {
 217                 $this->aCountryCodes = $aCountries;
 218             }
 219         }
 220
 221         $aViewbox = $oParams->getStringList('viewboxlbrt');
 222         if ($aViewbox) {
 223             if (count($aViewbox) != 4) {
 224                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 225             }
 226             $this->setViewbox($aViewbox);
 227         } else {
 228             $aViewbox = $oParams->getStringList('viewbox');
 229             if ($aViewbox) {
 230                 if (count($aViewbox) != 4) {
 231                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 232                 }
 233                 $this->setViewBox($aViewbox);
 234             } else {
 235                 $aRoute = $oParams->getStringList('route');
 236                 $fRouteWidth = $oParams->getFloat('routewidth');
 237                 if ($aRoute && $fRouteWidth) {
 238                     $this->aRoutePoints = $aRoute;
 239                     $this->aRouteWidth = $fRouteWidth;
 240                 }
 241             }
 242         }
 243
 244         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 245         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 246     }
 247
 248     public function setQueryFromParams($oParams)
 249     {
 250         // Search query
 251         $sQuery = $oParams->getString('q');
 252         if (!$sQuery) {
 253             $this->setStructuredQuery(
 254                 $oParams->getString('amenity'),
 255                 $oParams->getString('street'),
 256                 $oParams->getString('city'),
 257                 $oParams->getString('county'),
 258                 $oParams->getString('state'),
 259                 $oParams->getString('country'),
 260                 $oParams->getString('postalcode')
 261             );
 262         } else {
 263             $this->setQuery($sQuery);
 264         }
 265     }
 266
 267     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 268     {
 269         $sValue = trim($sValue);
 270         if (!$sValue) {
 271             return false;
 272         }
 273         $this->aStructuredQuery[$sKey] = $sValue;
 274         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 275             $this->iMinAddressRank = $iNewMinAddressRank;
 276             $this->iMaxAddressRank = $iNewMaxAddressRank;
 277         }
 278         if ($aItemListValues) {
 279             $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 280         }
 281         return true;
 282     }
 283
 284     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 285     {
 286         $this->sQuery = false;
 287
 288         // Reset
 289         $this->iMinAddressRank = 0;
 290         $this->iMaxAddressRank = 30;
 291         $this->aAddressRankList = array();
 292
 293         $this->aStructuredQuery = array();
 294         $this->sAllowedTypesSQLList = false;
 295
 296         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 297         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 298         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 299         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 300         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 301         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 302         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 303
 304         if (!empty($this->aStructuredQuery)) {
 305             $this->sQuery = join(', ', $this->aStructuredQuery);
 306             if ($this->iMaxAddressRank < 30) {
 307                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 308             }
 309         }
 310     }
 311
 312     public function fallbackStructuredQuery()
 313     {
 314         $aParams = $this->aStructuredQuery;
 315
 316         if (!$aParams || count($aParams) == 1) {
 317             return false;
 318         }
 319
 320         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 321
 322         foreach ($aOrderToFallback as $sType) {
 323             if (isset($aParams[$sType])) {
 324                 unset($aParams[$sType]);
 325                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 326                 return true;
 327             }
 328         }
 329
 330         return false;
 331     }
 332
 333     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens)
 334     {
 335         /*
 336              Calculate all searches using oValidTokens i.e.
 337              'Wodsworth Road, Sheffield' =>
 338
 339              Phrase Wordset
 340              0      0       (wodsworth road)
 341              0      1       (wodsworth)(road)
 342              1      0       (sheffield)
 343
 344              Score how good the search is so they can be ordered
 345          */
 346         foreach ($aPhrases as $iPhrase => $oPhrase) {
 347             $aNewPhraseSearches = array();
 348             $sPhraseType = $oPhrase->getPhraseType();
 349
 350             foreach ($oPhrase->getWordSets() as $aWordset) {
 351                 $aWordsetSearches = $aSearches;
 352
 353                 // Add all words from this wordset
 354                 foreach ($aWordset as $iToken => $sToken) {
 355                     $aNewWordsetSearches = array();
 356
 357                     foreach ($aWordsetSearches as $oCurrentSearch) {
 358                         // Tokens with full name matches.
 359                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 360                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 361                                 $oSearchTerm,
 362                                 $sPhraseType,
 363                                 $iToken == 0 && $iPhrase == 0,
 364                                 $iPhrase == 0,
 365                                 $iToken + 1 == count($aWordset)
 366                                   && $iPhrase + 1 == count($aPhrases)
 367                             );
 368
 369                             foreach ($aNewSearches as $oSearch) {
 370                                 if ($oSearch->getRank() < $this->iMaxRank) {
 371                                     $aNewWordsetSearches[] = $oSearch;
 372                                 }
 373                             }
 374                         }
 375                         // Look for partial matches.
 376                         // Note that there is no point in adding country terms here
 377                         // because country is omitted in the address.
 378                         if ($sPhraseType != 'country') {
 379                             // Allow searching for a word - but at extra cost
 380                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 381                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 382                                     $sToken,
 383                                     $oSearchTerm,
 384                                     (bool) $sPhraseType,
 385                                     $iPhrase,
 386                                     $oValidTokens->get(' '.$sToken)
 387                                 );
 388
 389                                 foreach ($aNewSearches as $oSearch) {
 390                                     if ($oSearch->getRank() < $this->iMaxRank) {
 391                                         $aNewWordsetSearches[] = $oSearch;
 392                                     }
 393                                 }
 394                             }
 395                         }
 396                     }
 397                     // Sort and cut
 398                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 399                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 400                 }
 401
 402                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 403                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 404
 405                 $aSearchHash = array();
 406                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 407                     $sHash = serialize($aSearch);
 408                     if (isset($aSearchHash[$sHash])) {
 409                         unset($aNewPhraseSearches[$iSearch]);
 410                     } else {
 411                         $aSearchHash[$sHash] = 1;
 412                     }
 413                 }
 414
 415                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 416             }
 417
 418             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 419             $aGroupedSearches = array();
 420             foreach ($aNewPhraseSearches as $aSearch) {
 421                 $iRank = $aSearch->getRank();
 422                 if ($iRank < $this->iMaxRank) {
 423                     if (!isset($aGroupedSearches[$iRank])) {
 424                         $aGroupedSearches[$iRank] = array();
 425                     }
 426                     $aGroupedSearches[$iRank][] = $aSearch;
 427                 }
 428             }
 429             ksort($aGroupedSearches);
 430
 431             $iSearchCount = 0;
 432             $aSearches = array();
 433             foreach ($aGroupedSearches as $aNewSearches) {
 434                 $iSearchCount += count($aNewSearches);
 435                 $aSearches = array_merge($aSearches, $aNewSearches);
 436                 if ($iSearchCount > 50) {
 437                     break;
 438                 }
 439             }
 440         }
 441
 442         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 443         $aGroupedSearches = array();
 444         foreach ($aSearches as $oSearch) {
 445             if (!$oSearch->isValidSearch()) {
 446                 continue;
 447             }
 448
 449             $iRank = $oSearch->getRank();
 450             if (!isset($aGroupedSearches[$iRank])) {
 451                 $aGroupedSearches[$iRank] = array();
 452             }
 453             $aGroupedSearches[$iRank][] = $oSearch;
 454         }
 455         ksort($aGroupedSearches);
 456
 457         return $aGroupedSearches;
 458     }
 459
 460     /* Perform the actual query lookup.
 461
 462         Returns an ordered list of results, each with the following fields:
 463             osm_type: type of corresponding OSM object
 464                         N - node
 465                         W - way
 466                         R - relation
 467                         P - postcode (internally computed)
 468             osm_id: id of corresponding OSM object
 469             class: general object class (corresponds to tag key of primary OSM tag)
 470             type: subclass of object (corresponds to tag value of primary OSM tag)
 471             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 472             rank_search: rank in search hierarchy
 473                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 474             rank_address: rank in address hierarchy (determines orer in address)
 475             place_id: internal key (may differ between different instances)
 476             country_code: ISO country code
 477             langaddress: localized full address
 478             placename: localized name of object
 479             ref: content of ref tag (if available)
 480             lon: longitude
 481             lat: latitude
 482             importance: importance of place based on Wikipedia link count
 483             addressimportance: cumulated importance of address elements
 484             extra_place: type of place (for admin boundaries, if there is a place tag)
 485             aBoundingBox: bounding Box
 486             label: short description of the object class/type (English only)
 487             name: full name (currently the same as langaddress)
 488             foundorder: secondary ordering for places with same importance
 489     */
 490
 491
 492     public function lookup()
 493     {
 494         Debug::newFunction('Geocode::lookup');
 495         if (!$this->sQuery && !$this->aStructuredQuery) {
 496             return array();
 497         }
 498
 499         Debug::printDebugArray('Geocode', $this);
 500
 501         $oCtx = new SearchContext();
 502
 503         if ($this->aRoutePoints) {
 504             $oCtx->setViewboxFromRoute(
 505                 $this->oDB,
 506                 $this->aRoutePoints,
 507                 $this->aRouteWidth,
 508                 $this->bBoundedSearch
 509             );
 510         } elseif ($this->aViewBox) {
 511             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 512         }
 513         if ($this->aExcludePlaceIDs) {
 514             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 515         }
 516         if ($this->aCountryCodes) {
 517             $oCtx->setCountryList($this->aCountryCodes);
 518         }
 519         $this->oTokenizer->setCountryRestriction($this->aCountryCodes);
 520
 521         Debug::newSection('Query Preprocessing');
 522
 523         $sQuery = $this->sQuery;
 524         if (!preg_match('//u', $sQuery)) {
 525             userError('Query string is not UTF-8 encoded.');
 526         }
 527
 528         // Conflicts between US state abreviations and various words for 'the' in different languages
 529         if (isset($this->aLangPrefOrder['name:en'])) {
 530             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 531             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 532             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 533         }
 534
 535         // Do we have anything that looks like a lat/lon pair?
 536         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 537
 538         if ($sQuery || $this->aStructuredQuery) {
 539             // Start with a single blank search
 540             $aSearches = array(new SearchDescription($oCtx));
 541
 542             if ($sQuery) {
 543                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 544             }
 545
 546             $sSpecialTerm = '';
 547             if ($sQuery) {
 548                 preg_match_all(
 549                     '/\\[([\\w ]*)\\]/u',
 550                     $sQuery,
 551                     $aSpecialTermsRaw,
 552                     PREG_SET_ORDER
 553                 );
 554                 if (!empty($aSpecialTermsRaw)) {
 555                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 556                 }
 557
 558                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 559                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 560                     if (!$sSpecialTerm) {
 561                         $sSpecialTerm = $aSpecialTerm[1];
 562                     }
 563                 }
 564             }
 565             if (!$sSpecialTerm && $this->aStructuredQuery
 566                 && isset($this->aStructuredQuery['amenity'])) {
 567                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 568                 unset($this->aStructuredQuery['amenity']);
 569             }
 570
 571             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 572                 $aTokens = $this->oTokenizer->tokensForSpecialTerm($sSpecialTerm);
 573
 574                 if (!empty($aTokens)) {
 575                     $aNewSearches = array();
 576                     foreach ($aSearches as $oSearch) {
 577                         foreach ($aTokens as $oToken) {
 578                             $oNewSearch = clone $oSearch;
 579                             $oNewSearch->setPoiSearch(
 580                                 $oToken->iOperator,
 581                                 $oToken->sClass,
 582                                 $oToken->sType
 583                             );
 584                             $aNewSearches[] = $oNewSearch;
 585                         }
 586                     }
 587                     $aSearches = $aNewSearches;
 588                 }
 589             }
 590
 591             // Split query into phrases
 592             // Commas are used to reduce the search space by indicating where phrases split
 593             $aPhrases = array();
 594             if ($this->aStructuredQuery) {
 595                 foreach ($this->aStructuredQuery as $iPhrase => $sPhrase) {
 596                     $aPhrases[] = new Phrase($sPhrase, $iPhrase);
 597                 }
 598             } else {
 599                 foreach (explode(',', $sQuery) as $sPhrase) {
 600                     $aPhrases[] = new Phrase($sPhrase, '');
 601                 }
 602             }
 603
 604             Debug::printDebugArray('Search context', $oCtx);
 605             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 606
 607             Debug::newSection('Tokenization');
 608             $oValidTokens = $this->oTokenizer->extractTokensFromPhrases($aPhrases);
 609
 610             if ($oValidTokens->count() > 0) {
 611                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 612
 613                 $aPhrases = array_filter($aPhrases, function ($oPhrase) {
 614                     return $oPhrase->getWordSets() !== null;
 615                 });
 616
 617                 // Any words that have failed completely?
 618                 // TODO: suggestions
 619
 620                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 621                 Debug::printDebugTable('Phrases', $aPhrases);
 622
 623                 Debug::newSection('Search candidates');
 624
 625                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 626
 627                 if (!$this->aStructuredQuery) {
 628                     // Reverse phrase array and also reverse the order of the wordsets in
 629                     // the first and final phrase. Don't bother about phrases in the middle
 630                     // because order in the address doesn't matter.
 631                     $aPhrases = array_reverse($aPhrases);
 632                     $aPhrases[0]->invertWordSets();
 633                     if (count($aPhrases) > 1) {
 634                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 635                     }
 636                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 637
 638                     foreach ($aGroupedSearches as $aSearches) {
 639                         foreach ($aSearches as $aSearch) {
 640                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 641                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 642                             }
 643                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 644                         }
 645                     }
 646
 647                     $aGroupedSearches = $aReverseGroupedSearches;
 648                     ksort($aGroupedSearches);
 649                 }
 650             } else {
 651                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 652                 $aGroupedSearches = array();
 653                 foreach ($aSearches as $aSearch) {
 654                     if ($aSearch->getRank() < $this->iMaxRank) {
 655                         if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 656                             $aGroupedSearches[$aSearch->getRank()] = array();
 657                         }
 658                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 659                     }
 660                 }
 661                 ksort($aGroupedSearches);
 662             }
 663
 664             // Filter out duplicate searches
 665             $aSearchHash = array();
 666             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 667                 foreach ($aSearches as $iSearch => $aSearch) {
 668                     $sHash = serialize($aSearch);
 669                     if (isset($aSearchHash[$sHash])) {
 670                         unset($aGroupedSearches[$iGroup][$iSearch]);
 671                         if (empty($aGroupedSearches[$iGroup])) {
 672                             unset($aGroupedSearches[$iGroup]);
 673                         }
 674                     } else {
 675                         $aSearchHash[$sHash] = 1;
 676                     }
 677                 }
 678             }
 679
 680             Debug::printGroupedSearch(
 681                 $aGroupedSearches,
 682                 $oValidTokens->debugTokenByWordIdList()
 683             );
 684
 685             // Start the search process
 686             $iGroupLoop = 0;
 687             $iQueryLoop = 0;
 688             $aNextResults = array();
 689             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 690                 $iGroupLoop++;
 691                 $aResults = $aNextResults;
 692                 foreach ($aSearches as $oSearch) {
 693                     $iQueryLoop++;
 694
 695                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 696                     Debug::printGroupedSearch(
 697                         array($iGroupedRank => array($oSearch)),
 698                         $oValidTokens->debugTokenByWordIdList()
 699                     );
 700
 701                     $aNewResults = $oSearch->query(
 702                         $this->oDB,
 703                         $this->iMinAddressRank,
 704                         $this->iMaxAddressRank,
 705                         $this->iLimit
 706                     );
 707
 708                     // The same result may appear in different rounds, only
 709                     // use the one with minimal rank.
 710                     foreach ($aNewResults as $iPlace => $oRes) {
 711                         if (!isset($aResults[$iPlace])
 712                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 713                             $aResults[$iPlace] = $oRes;
 714                         }
 715                     }
 716
 717                     if ($iQueryLoop > 20) {
 718                         break;
 719                     }
 720                 }
 721
 722                 if (!empty($aResults)) {
 723                     $aSplitResults = Result::splitResults($aResults);
 724                     Debug::printVar('Split results', $aSplitResults);
 725                     if ($iGroupLoop <= 4
 726                         && reset($aSplitResults['head'])->iResultRank > 0
 727                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 728                         // Haven't found an exact match for the query yet.
 729                         // Therefore add result from the next group level.
 730                         $aNextResults = $aSplitResults['head'];
 731                         foreach ($aNextResults as $oRes) {
 732                             $oRes->iResultRank--;
 733                         }
 734                         foreach ($aSplitResults['tail'] as $oRes) {
 735                             $oRes->iResultRank--;
 736                             $aNextResults[$oRes->iId] = $oRes;
 737                         }
 738                         $aResults = array();
 739                     } else {
 740                         $aResults = $aSplitResults['head'];
 741                     }
 742                 }
 743
 744                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 745                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 746                     // reduces the number of place ids, like a filter
 747                     // rank_address is 30 for interpolated housenumbers
 748                     $aFilterSql = array();
 749                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 750                     if ($sPlaceIds) {
 751                         $sSQL = 'SELECT place_id FROM placex ';
 752                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 753                         $sSQL .= '  AND (';
 754                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 755                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 756                         if ($this->aAddressRankList) {
 757                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 758                         }
 759                         $sSQL .= ')';
 760                         $aFilterSql[] = $sSQL;
 761                     }
 762                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 763                     if ($sPlaceIds) {
 764                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 765                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 766                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 767                         if ($this->aAddressRankList) {
 768                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 769                         }
 770                         $sSQL .= ') ';
 771                         $aFilterSql[] = $sSQL;
 772                     }
 773
 774                     $aFilteredIDs = array();
 775                     if ($aFilterSql) {
 776                         $sSQL = join(' UNION ', $aFilterSql);
 777                         Debug::printSQL($sSQL);
 778                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 779                     }
 780
 781                     $tempIDs = array();
 782                     foreach ($aResults as $oResult) {
 783                         if (($this->iMaxAddressRank == 30 &&
 784                              ($oResult->iTable == Result::TABLE_OSMLINE
 785                               || $oResult->iTable == Result::TABLE_TIGER))
 786                             || in_array($oResult->iId, $aFilteredIDs)
 787                         ) {
 788                             $tempIDs[$oResult->iId] = $oResult;
 789                         }
 790                     }
 791                     $aResults = $tempIDs;
 792                 }
 793
 794                 if (!empty($aResults) || $iGroupLoop > 4 || $iQueryLoop > 30) {
 795                     break;
 796                 }
 797             }
 798         } else {
 799             // Just interpret as a reverse geocode
 800             $oReverse = new ReverseGeocode($this->oDB);
 801             $oReverse->setZoom(18);
 802
 803             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 804
 805             Debug::printVar('Reverse search', $oLookup);
 806
 807             if ($oLookup) {
 808                 $aResults = array($oLookup->iId => $oLookup);
 809             }
 810         }
 811
 812         // No results? Done
 813         if (empty($aResults)) {
 814             if ($this->bFallback && $this->fallbackStructuredQuery()) {
 815                 return $this->lookup();
 816             }
 817
 818             return array();
 819         }
 820
 821         if ($this->aAddressRankList) {
 822             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 823         }
 824         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 825         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 826         if ($oCtx->hasNearPoint()) {
 827             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 828         }
 829
 830         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 831
 832         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 833         foreach ($aRecheckWords as $i => $sWord) {
 834             if (!preg_match('/[\pL\pN]/', $sWord)) {
 835                 unset($aRecheckWords[$i]);
 836             }
 837         }
 838
 839         Debug::printVar('Recheck words', $aRecheckWords);
 840
 841         foreach ($aSearchResults as $iIdx => $aResult) {
 842             $fRadius = ClassTypes\getDefRadius($aResult);
 843
 844             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 845             if ($aOutlineResult) {
 846                 $aResult = array_merge($aResult, $aOutlineResult);
 847             }
 848
 849             // Is there an icon set for this type of result?
 850             $sIcon = ClassTypes\getIconFile($aResult);
 851             if (isset($sIcon)) {
 852                 $aResult['icon'] = $sIcon;
 853             }
 854
 855             $sLabel = ClassTypes\getLabel($aResult);
 856             if (isset($sLabel)) {
 857                 $aResult['label'] = $sLabel;
 858             }
 859             $aResult['name'] = $aResult['langaddress'];
 860
 861             if ($oCtx->hasNearPoint()) {
 862                 $aResult['importance'] = 0.001;
 863                 $aResult['foundorder'] = $aResult['addressimportance'];
 864             } else {
 865                 $aResult['importance'] = max(0.001, $aResult['importance']);
 866                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 867                     $aResult['lon'],
 868                     $aResult['lat']
 869                 );
 870
 871                 // secondary ordering (for results with same importance (the smaller the better):
 872                 // - approximate importance of address parts
 873                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 874                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 875                 } else {
 876                     $aResult['foundorder'] = -$aResult['importance'];
 877                 }
 878                 // - number of exact matches from the query
 879                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 880                 // - importance of the class/type
 881                 $iClassImportance = ClassTypes\getImportance($aResult);
 882                 if (isset($iClassImportance)) {
 883                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 884                 } else {
 885                     $aResult['foundorder'] += 0.01;
 886                 }
 887                 // - rank
 888                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 889
 890                 // Adjust importance for the number of exact string matches in the result
 891                 $iCountWords = 0;
 892                 $sAddress = $aResult['langaddress'];
 893                 foreach ($aRecheckWords as $i => $sWord) {
 894                     if (stripos($sAddress, $sWord)!==false) {
 895                         $iCountWords++;
 896                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 897                             $iCountWords += 0.1;
 898                         }
 899                     }
 900                 }
 901
 902                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 903                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 904             }
 905             $aSearchResults[$iIdx] = $aResult;
 906         }
 907         uasort($aSearchResults, 'byImportance');
 908         Debug::printVar('Pre-filter results', $aSearchResults);
 909
 910         $aOSMIDDone = array();
 911         $aClassTypeNameDone = array();
 912         $aToFilter = $aSearchResults;
 913         $aSearchResults = array();
 914
 915         foreach ($aToFilter as $aResult) {
 916             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 917             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 918                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 919             ) {
 920                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 921                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 922                 $aSearchResults[] = $aResult;
 923             }
 924
 925             // Absolute limit on number of results
 926             if (count($aSearchResults) >= $this->iFinalLimit) {
 927                 break;
 928             }
 929         }
 930
 931         Debug::printVar('Post-filter results', $aSearchResults);
 932         return $aSearchResults;
 933     } // end lookup()
 934
 935     public function debugInfo()
 936     {
 937         return array(
 938                 'Query' => $this->sQuery,
 939                 'Structured query' => $this->aStructuredQuery,
 940                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 941                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 942                 'Limit (for searches)' => $this->iLimit,
 943                 'Limit (for results)'=> $this->iFinalLimit,
 944                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 945                 'Bounded search' => $this->bBoundedSearch,
 946                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 947                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 948                 'Route width' => $this->aRouteWidth,
 949                 'Max rank' => $this->iMaxRank,
 950                 'Min address rank' => $this->iMinAddressRank,
 951                 'Max address rank' => $this->iMaxAddressRank,
 952                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 953                );
 954     }
 955 } // end class