lib-php/Geocode.php

   1 <?php
   2 /**
   3  * SPDX-License-Identifier: GPL-2.0-only
   4  *
   5  * This file is part of Nominatim. (https://nominatim.org)
   6  *
   7  * Copyright (C) 2022 by the Nominatim developer community.
   8  * For a full list of authors see the git log.
   9  */
  10
  11 namespace Nominatim;
  12
  13 require_once(CONST_LibDir.'/PlaceLookup.php');
  14 require_once(CONST_LibDir.'/Phrase.php');
  15 require_once(CONST_LibDir.'/ReverseGeocode.php');
  16 require_once(CONST_LibDir.'/SearchDescription.php');
  17 require_once(CONST_LibDir.'/SearchContext.php');
  18 require_once(CONST_LibDir.'/SearchPosition.php');
  19 require_once(CONST_LibDir.'/TokenList.php');
  20 require_once(CONST_TokenizerDir.'/tokenizer.php');
  21
  22 class Geocode
  23 {
  24     protected $oDB;
  25
  26     protected $oPlaceLookup;
  27     protected $oTokenizer;
  28
  29     protected $aLangPrefOrder = array();
  30
  31     protected $aExcludePlaceIDs = array();
  32
  33     protected $iLimit = 20;
  34     protected $iFinalLimit = 10;
  35     protected $iOffset = 0;
  36     protected $bFallback = false;
  37
  38     protected $aCountryCodes = false;
  39
  40     protected $bBoundedSearch = false;
  41     protected $aViewBox = false;
  42     protected $aRoutePoints = false;
  43     protected $aRouteWidth = false;
  44
  45     protected $iMaxRank = 20;
  46     protected $iMinAddressRank = 0;
  47     protected $iMaxAddressRank = 30;
  48     protected $aAddressRankList = array();
  49
  50     protected $sAllowedTypesSQLList = false;
  51
  52     protected $sQuery = false;
  53     protected $aStructuredQuery = false;
  54
  55
  56     public function __construct(&$oDB)
  57     {
  58         $this->oDB =& $oDB;
  59         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  60         $this->oTokenizer = new \Nominatim\Tokenizer($this->oDB);
  61     }
  62
  63     public function setLanguagePreference($aLangPref)
  64     {
  65         $this->aLangPrefOrder = $aLangPref;
  66     }
  67
  68     public function getMoreUrlParams()
  69     {
  70         if ($this->aStructuredQuery) {
  71             $aParams = $this->aStructuredQuery;
  72         } else {
  73             $aParams = array('q' => $this->sQuery);
  74         }
  75
  76         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  77
  78         if ($this->aExcludePlaceIDs) {
  79             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  80         }
  81
  82         if ($this->bBoundedSearch) {
  83             $aParams['bounded'] = '1';
  84         }
  85
  86         if ($this->aCountryCodes) {
  87             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  88         }
  89
  90         if ($this->aViewBox) {
  91             $aParams['viewbox'] = join(',', $this->aViewBox);
  92         }
  93
  94         return $aParams;
  95     }
  96
  97     public function setLimit($iLimit = 10)
  98     {
  99         if ($iLimit > 50) {
 100             $iLimit = 50;
 101         } elseif ($iLimit < 1) {
 102             $iLimit = 1;
 103         }
 104
 105         $this->iFinalLimit = $iLimit;
 106         $this->iLimit = $iLimit + max($iLimit, 10);
 107     }
 108
 109     public function setFeatureType($sFeatureType)
 110     {
 111         switch ($sFeatureType) {
 112             case 'country':
 113                 $this->setRankRange(4, 4);
 114                 break;
 115             case 'state':
 116                 $this->setRankRange(8, 8);
 117                 break;
 118             case 'city':
 119                 $this->setRankRange(14, 16);
 120                 break;
 121             case 'settlement':
 122                 $this->setRankRange(8, 20);
 123                 break;
 124         }
 125     }
 126
 127     public function setRankRange($iMin, $iMax)
 128     {
 129         $this->iMinAddressRank = $iMin;
 130         $this->iMaxAddressRank = $iMax;
 131     }
 132
 133     public function setViewbox($aViewbox)
 134     {
 135         $aBox = array_map('floatval', $aViewbox);
 136
 137         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 138         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 139         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 140         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 141
 142         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 143             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 144         ) {
 145             userError("Bad parameter 'viewbox'. Not a box.");
 146         }
 147     }
 148
 149     private function viewboxImportanceFactor($fX, $fY)
 150     {
 151         if (!$this->aViewBox) {
 152             return 1;
 153         }
 154
 155         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 156         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 157
 158         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 159         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 160
 161         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 162             return 1;
 163         }
 164
 165         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 166             return 0.5;
 167         }
 168
 169         return 0.25;
 170     }
 171
 172     public function setQuery($sQueryString)
 173     {
 174         $this->sQuery = $sQueryString;
 175         $this->aStructuredQuery = false;
 176     }
 177
 178     public function getQueryString()
 179     {
 180         return $this->sQuery;
 181     }
 182
 183
 184     public function loadParamArray($oParams, $sForceGeometryType = null)
 185     {
 186         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 187
 188         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 189         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 190
 191         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 192
 193         // List of excluded Place IDs - used for more accurate pageing
 194         $sExcluded = $oParams->getStringList('exclude_place_ids');
 195         if ($sExcluded) {
 196             foreach ($sExcluded as $iExcludedPlaceID) {
 197                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 198                 if ($iExcludedPlaceID) {
 199                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 200                 }
 201             }
 202
 203             if (isset($aExcludePlaceIDs)) {
 204                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 205             }
 206         }
 207
 208         // Only certain ranks of feature
 209         $sFeatureType = $oParams->getString('featureType');
 210         if (!$sFeatureType) {
 211             $sFeatureType = $oParams->getString('featuretype');
 212         }
 213         if ($sFeatureType) {
 214             $this->setFeatureType($sFeatureType);
 215         }
 216
 217         // Country code list
 218         $sCountries = $oParams->getStringList('countrycodes');
 219         if ($sCountries) {
 220             foreach ($sCountries as $sCountryCode) {
 221                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 222                     $aCountries[] = strtolower($sCountryCode);
 223                 }
 224             }
 225             if (isset($aCountries)) {
 226                 $this->aCountryCodes = $aCountries;
 227             }
 228         }
 229
 230         $aViewbox = $oParams->getStringList('viewboxlbrt');
 231         if ($aViewbox) {
 232             if (count($aViewbox) != 4) {
 233                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 234             }
 235             $this->setViewbox($aViewbox);
 236         } else {
 237             $aViewbox = $oParams->getStringList('viewbox');
 238             if ($aViewbox) {
 239                 if (count($aViewbox) != 4) {
 240                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 241                 }
 242                 $this->setViewBox($aViewbox);
 243             } else {
 244                 $aRoute = $oParams->getStringList('route');
 245                 $fRouteWidth = $oParams->getFloat('routewidth');
 246                 if ($aRoute && $fRouteWidth) {
 247                     $this->aRoutePoints = $aRoute;
 248                     $this->aRouteWidth = $fRouteWidth;
 249                 }
 250             }
 251         }
 252
 253         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 254         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 255     }
 256
 257     public function setQueryFromParams($oParams)
 258     {
 259         // Search query
 260         $this->setStructuredQuery(
 261             $oParams->getString('amenity'),
 262             $oParams->getString('street'),
 263             $oParams->getString('city'),
 264             $oParams->getString('county'),
 265             $oParams->getString('state'),
 266             $oParams->getString('country'),
 267             $oParams->getString('postalcode')
 268         );
 269         if (!$this->sQuery) {
 270             $sQuery = $oParams->getString('q');
 271
 272             if ($sQuery) {
 273                 $this->setQuery($sQuery);
 274             }
 275         }
 276     }
 277
 278     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 279     {
 280         $sValue = trim($sValue);
 281         if (!$sValue) {
 282             return false;
 283         }
 284         $this->aStructuredQuery[$sKey] = $sValue;
 285         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 286             $this->iMinAddressRank = $iNewMinAddressRank;
 287             $this->iMaxAddressRank = $iNewMaxAddressRank;
 288         }
 289         if ($aItemListValues) {
 290             $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 291         }
 292         return true;
 293     }
 294
 295     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 296     {
 297         $this->sQuery = false;
 298
 299         if ($sAmenity || $sStreet || $sCity || $sCounty || $sState || $sCountry || $sPostalCode) {
 300             // Reset
 301             $this->iMinAddressRank = 0;
 302             $this->iMaxAddressRank = 30;
 303             $this->aAddressRankList = array();
 304
 305             $this->aStructuredQuery = array();
 306             $this->sAllowedTypesSQLList = false;
 307
 308             $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 309             $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 310             $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 311             $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 312             $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 313             $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 314             $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 315
 316             if (!empty($this->aStructuredQuery)) {
 317                 $this->sQuery = join(', ', $this->aStructuredQuery);
 318                 if ($this->iMaxAddressRank < 30) {
 319                     $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 320                 }
 321             }
 322         }
 323     }
 324
 325     public function fallbackStructuredQuery()
 326     {
 327         $aParams = $this->aStructuredQuery;
 328
 329         if (!$aParams || count($aParams) == 1) {
 330             return false;
 331         }
 332
 333         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 334
 335         foreach ($aOrderToFallback as $sType) {
 336             if (isset($aParams[$sType])) {
 337                 unset($aParams[$sType]);
 338                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 339                 return true;
 340             }
 341         }
 342
 343         return false;
 344     }
 345
 346     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens)
 347     {
 348         /*
 349              Calculate all searches using oValidTokens i.e.
 350              'Wodsworth Road, Sheffield' =>
 351
 352              Phrase Wordset
 353              0      0       (wodsworth road)
 354              0      1       (wodsworth)(road)
 355              1      0       (sheffield)
 356
 357              Score how good the search is so they can be ordered
 358          */
 359         foreach ($aPhrases as $iPhrase => $oPhrase) {
 360             $aNewPhraseSearches = array();
 361             $oPosition = new SearchPosition(
 362                 $oPhrase->getPhraseType(),
 363                 $iPhrase,
 364                 count($aPhrases)
 365             );
 366
 367             foreach ($oPhrase->getWordSets() as $aWordset) {
 368                 $aWordsetSearches = $aSearches;
 369
 370                 // Add all words from this wordset
 371                 foreach ($aWordset as $iToken => $sToken) {
 372                     $aNewWordsetSearches = array();
 373                     $oPosition->setTokenPosition($iToken, count($aWordset));
 374
 375                     foreach ($aWordsetSearches as $oCurrentSearch) {
 376                         foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 377                             if ($oSearchTerm->isExtendable($oCurrentSearch, $oPosition)) {
 378                                 $aNewSearches = $oSearchTerm->extendSearch(
 379                                     $oCurrentSearch,
 380                                     $oPosition
 381                                 );
 382
 383                                 foreach ($aNewSearches as $oSearch) {
 384                                     if ($oSearch->getRank() < $this->iMaxRank) {
 385                                         $aNewWordsetSearches[] = $oSearch;
 386                                     }
 387                                 }
 388                             }
 389                         }
 390                     }
 391                     // Sort and cut
 392                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 393                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 394                 }
 395
 396                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 397                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 398
 399                 $aSearchHash = array();
 400                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 401                     $sHash = serialize($aSearch);
 402                     if (isset($aSearchHash[$sHash])) {
 403                         unset($aNewPhraseSearches[$iSearch]);
 404                     } else {
 405                         $aSearchHash[$sHash] = 1;
 406                     }
 407                 }
 408
 409                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 410             }
 411
 412             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 413             $aGroupedSearches = array();
 414             foreach ($aNewPhraseSearches as $aSearch) {
 415                 $iRank = $aSearch->getRank();
 416                 if ($iRank < $this->iMaxRank) {
 417                     if (!isset($aGroupedSearches[$iRank])) {
 418                         $aGroupedSearches[$iRank] = array();
 419                     }
 420                     $aGroupedSearches[$iRank][] = $aSearch;
 421                 }
 422             }
 423             ksort($aGroupedSearches);
 424
 425             $iSearchCount = 0;
 426             $aSearches = array();
 427             foreach ($aGroupedSearches as $aNewSearches) {
 428                 $iSearchCount += count($aNewSearches);
 429                 $aSearches = array_merge($aSearches, $aNewSearches);
 430                 if ($iSearchCount > 50) {
 431                     break;
 432                 }
 433             }
 434         }
 435
 436         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 437         $aGroupedSearches = array();
 438         foreach ($aSearches as $oSearch) {
 439             if (!$oSearch->isValidSearch()) {
 440                 continue;
 441             }
 442
 443             $iRank = $oSearch->getRank();
 444             if (!isset($aGroupedSearches[$iRank])) {
 445                 $aGroupedSearches[$iRank] = array();
 446             }
 447             $aGroupedSearches[$iRank][] = $oSearch;
 448         }
 449         ksort($aGroupedSearches);
 450
 451         return $aGroupedSearches;
 452     }
 453
 454     /* Perform the actual query lookup.
 455
 456         Returns an ordered list of results, each with the following fields:
 457             osm_type: type of corresponding OSM object
 458                         N - node
 459                         W - way
 460                         R - relation
 461                         P - postcode (internally computed)
 462             osm_id: id of corresponding OSM object
 463             class: general object class (corresponds to tag key of primary OSM tag)
 464             type: subclass of object (corresponds to tag value of primary OSM tag)
 465             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 466             rank_search: rank in search hierarchy
 467                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 468             rank_address: rank in address hierarchy (determines orer in address)
 469             place_id: internal key (may differ between different instances)
 470             country_code: ISO country code
 471             langaddress: localized full address
 472             placename: localized name of object
 473             ref: content of ref tag (if available)
 474             lon: longitude
 475             lat: latitude
 476             importance: importance of place based on Wikipedia link count
 477             addressimportance: cumulated importance of address elements
 478             extra_place: type of place (for admin boundaries, if there is a place tag)
 479             aBoundingBox: bounding Box
 480             label: short description of the object class/type (English only)
 481             name: full name (currently the same as langaddress)
 482             foundorder: secondary ordering for places with same importance
 483     */
 484
 485
 486     public function lookup()
 487     {
 488         Debug::newFunction('Geocode::lookup');
 489         if (!$this->sQuery && !$this->aStructuredQuery) {
 490             return array();
 491         }
 492
 493         Debug::printDebugArray('Geocode', $this);
 494
 495         $oCtx = new SearchContext();
 496
 497         if ($this->aRoutePoints) {
 498             $oCtx->setViewboxFromRoute(
 499                 $this->oDB,
 500                 $this->aRoutePoints,
 501                 $this->aRouteWidth,
 502                 $this->bBoundedSearch
 503             );
 504         } elseif ($this->aViewBox) {
 505             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 506         }
 507         if ($this->aExcludePlaceIDs) {
 508             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 509         }
 510         if ($this->aCountryCodes) {
 511             $oCtx->setCountryList($this->aCountryCodes);
 512         }
 513
 514         Debug::newSection('Query Preprocessing');
 515
 516         $sQuery = $this->sQuery;
 517         if (!preg_match('//u', $sQuery)) {
 518             userError('Query string is not UTF-8 encoded.');
 519         }
 520
 521         // Do we have anything that looks like a lat/lon pair?
 522         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 523
 524         if ($sQuery || $this->aStructuredQuery) {
 525             // Start with a single blank search
 526             $aSearches = array(new SearchDescription($oCtx));
 527
 528             if ($sQuery) {
 529                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 530             }
 531
 532             $sSpecialTerm = '';
 533             if ($sQuery) {
 534                 preg_match_all(
 535                     '/\\[([\\w ]*)\\]/u',
 536                     $sQuery,
 537                     $aSpecialTermsRaw,
 538                     PREG_SET_ORDER
 539                 );
 540                 if (!empty($aSpecialTermsRaw)) {
 541                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 542                 }
 543
 544                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 545                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 546                     if (!$sSpecialTerm) {
 547                         $sSpecialTerm = $aSpecialTerm[1];
 548                     }
 549                 }
 550             }
 551             if (!$sSpecialTerm && $this->aStructuredQuery
 552                 && isset($this->aStructuredQuery['amenity'])) {
 553                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 554                 unset($this->aStructuredQuery['amenity']);
 555             }
 556
 557             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 558                 $aTokens = $this->oTokenizer->tokensForSpecialTerm($sSpecialTerm);
 559
 560                 if (!empty($aTokens)) {
 561                     $aNewSearches = array();
 562                     $oPosition = new SearchPosition('', 0, 1);
 563                     $oPosition->setTokenPosition(0, 1);
 564
 565                     foreach ($aSearches as $oSearch) {
 566                         foreach ($aTokens as $oToken) {
 567                             $aNewSearches = array_merge(
 568                                 $aNewSearches,
 569                                 $oToken->extendSearch($oSearch, $oPosition)
 570                             );
 571                         }
 572                     }
 573                     $aSearches = $aNewSearches;
 574                 }
 575             }
 576
 577             // Split query into phrases
 578             // Commas are used to reduce the search space by indicating where phrases split
 579             $aPhrases = array();
 580             if ($this->aStructuredQuery) {
 581                 foreach ($this->aStructuredQuery as $iPhrase => $sPhrase) {
 582                     $aPhrases[] = new Phrase($sPhrase, $iPhrase);
 583                 }
 584             } else {
 585                 foreach (explode(',', $sQuery) as $sPhrase) {
 586                     $aPhrases[] = new Phrase($sPhrase, '');
 587                 }
 588             }
 589
 590             Debug::printDebugArray('Search context', $oCtx);
 591             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 592
 593             Debug::newSection('Tokenization');
 594             $oValidTokens = $this->oTokenizer->extractTokensFromPhrases($aPhrases);
 595
 596             if ($oValidTokens->count() > 0) {
 597                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 598
 599                 $aPhrases = array_filter($aPhrases, function ($oPhrase) {
 600                     return $oPhrase->getWordSets() !== null;
 601                 });
 602
 603                 // Any words that have failed completely?
 604                 // TODO: suggestions
 605
 606                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 607                 Debug::printDebugTable('Phrases', $aPhrases);
 608
 609                 Debug::newSection('Search candidates');
 610
 611                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 612
 613                 if (!$this->aStructuredQuery) {
 614                     // Reverse phrase array and also reverse the order of the wordsets in
 615                     // the first and final phrase. Don't bother about phrases in the middle
 616                     // because order in the address doesn't matter.
 617                     $aPhrases = array_reverse($aPhrases);
 618                     $aPhrases[0]->invertWordSets();
 619                     if (count($aPhrases) > 1) {
 620                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 621                     }
 622                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 623
 624                     foreach ($aReverseGroupedSearches as $aSearches) {
 625                         foreach ($aSearches as $aSearch) {
 626                             if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 627                                 $aGroupedSearches[$aSearch->getRank()] = array();
 628                             }
 629                             $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 630                         }
 631                     }
 632
 633                     ksort($aGroupedSearches);
 634                 }
 635             } else {
 636                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 637                 $aGroupedSearches = array();
 638                 foreach ($aSearches as $aSearch) {
 639                     if ($aSearch->getRank() < $this->iMaxRank) {
 640                         if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 641                             $aGroupedSearches[$aSearch->getRank()] = array();
 642                         }
 643                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 644                     }
 645                 }
 646                 ksort($aGroupedSearches);
 647             }
 648
 649             // Filter out duplicate searches
 650             $aSearchHash = array();
 651             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 652                 foreach ($aSearches as $iSearch => $aSearch) {
 653                     $sHash = serialize($aSearch);
 654                     if (isset($aSearchHash[$sHash])) {
 655                         unset($aGroupedSearches[$iGroup][$iSearch]);
 656                         if (empty($aGroupedSearches[$iGroup])) {
 657                             unset($aGroupedSearches[$iGroup]);
 658                         }
 659                     } else {
 660                         $aSearchHash[$sHash] = 1;
 661                     }
 662                 }
 663             }
 664
 665             Debug::printGroupedSearch(
 666                 $aGroupedSearches,
 667                 $oValidTokens->debugTokenByWordIdList()
 668             );
 669
 670             // Start the search process
 671             $iGroupLoop = 0;
 672             $iQueryLoop = 0;
 673             $aNextResults = array();
 674             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 675                 $iGroupLoop++;
 676                 $aResults = $aNextResults;
 677                 foreach ($aSearches as $oSearch) {
 678                     $iQueryLoop++;
 679
 680                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 681                     Debug::printGroupedSearch(
 682                         array($iGroupedRank => array($oSearch)),
 683                         $oValidTokens->debugTokenByWordIdList()
 684                     );
 685
 686                     $aNewResults = $oSearch->query(
 687                         $this->oDB,
 688                         $this->iMinAddressRank,
 689                         $this->iMaxAddressRank,
 690                         $this->iLimit
 691                     );
 692
 693                     // The same result may appear in different rounds, only
 694                     // use the one with minimal rank.
 695                     foreach ($aNewResults as $iPlace => $oRes) {
 696                         if (!isset($aResults[$iPlace])
 697                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 698                             $aResults[$iPlace] = $oRes;
 699                         }
 700                     }
 701
 702                     if ($iQueryLoop > 30) {
 703                         break;
 704                     }
 705                 }
 706
 707                 if (!empty($aResults)) {
 708                     $aSplitResults = Result::splitResults($aResults);
 709                     Debug::printVar('Split results', $aSplitResults);
 710                     if ($iGroupLoop <= 4
 711                         && reset($aSplitResults['head'])->iResultRank > 0
 712                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 713                         // Haven't found an exact match for the query yet.
 714                         // Therefore add result from the next group level.
 715                         $aNextResults = $aSplitResults['head'];
 716                         foreach ($aNextResults as $oRes) {
 717                             $oRes->iResultRank--;
 718                         }
 719                         foreach ($aSplitResults['tail'] as $oRes) {
 720                             $oRes->iResultRank--;
 721                             $aNextResults[$oRes->iId] = $oRes;
 722                         }
 723                         $aResults = array();
 724                     } else {
 725                         $aResults = $aSplitResults['head'];
 726                     }
 727                 }
 728
 729                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 730                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 731                     // reduces the number of place ids, like a filter
 732                     // rank_address is 30 for interpolated housenumbers
 733                     $aFilterSql = array();
 734                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 735                     if ($sPlaceIds) {
 736                         $sSQL = 'SELECT place_id FROM placex ';
 737                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 738                         $sSQL .= '  AND (';
 739                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 740                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 741                         if ($this->aAddressRankList) {
 742                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 743                         }
 744                         $sSQL .= ')';
 745                         $aFilterSql[] = $sSQL;
 746                     }
 747                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 748                     if ($sPlaceIds) {
 749                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 750                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 751                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 752                         if ($this->aAddressRankList) {
 753                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 754                         }
 755                         $sSQL .= ') ';
 756                         $aFilterSql[] = $sSQL;
 757                     }
 758
 759                     $aFilteredIDs = array();
 760                     if ($aFilterSql) {
 761                         $sSQL = join(' UNION ', $aFilterSql);
 762                         Debug::printSQL($sSQL);
 763                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 764                     }
 765
 766                     $tempIDs = array();
 767                     foreach ($aResults as $oResult) {
 768                         if (($this->iMaxAddressRank == 30 &&
 769                              ($oResult->iTable == Result::TABLE_OSMLINE
 770                               || $oResult->iTable == Result::TABLE_TIGER))
 771                             || in_array($oResult->iId, $aFilteredIDs)
 772                         ) {
 773                             $tempIDs[$oResult->iId] = $oResult;
 774                         }
 775                     }
 776                     $aResults = $tempIDs;
 777                 }
 778
 779                 if (!empty($aResults) || $iGroupLoop > 6 || $iQueryLoop > 40) {
 780                     break;
 781                 }
 782             }
 783         } else {
 784             // Just interpret as a reverse geocode
 785             $oReverse = new ReverseGeocode($this->oDB);
 786             $oReverse->setZoom(18);
 787
 788             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 789
 790             Debug::printVar('Reverse search', $oLookup);
 791
 792             if ($oLookup) {
 793                 $aResults = array($oLookup->iId => $oLookup);
 794             }
 795         }
 796
 797         // No results? Done
 798         if (empty($aResults)) {
 799             if ($this->bFallback && $this->fallbackStructuredQuery()) {
 800                 return $this->lookup();
 801             }
 802
 803             return array();
 804         }
 805
 806         if ($this->aAddressRankList) {
 807             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 808         }
 809         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 810         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 811         if ($oCtx->hasNearPoint()) {
 812             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 813         }
 814
 815         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 816
 817         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 818         foreach ($aRecheckWords as $i => $sWord) {
 819             if (!preg_match('/[\pL\pN]/', $sWord)) {
 820                 unset($aRecheckWords[$i]);
 821             }
 822         }
 823
 824         Debug::printVar('Recheck words', $aRecheckWords);
 825
 826         foreach ($aSearchResults as $iIdx => $aResult) {
 827             $fRadius = ClassTypes\getDefRadius($aResult);
 828
 829             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 830             if ($aOutlineResult) {
 831                 $aResult = array_merge($aResult, $aOutlineResult);
 832             }
 833
 834             // Is there an icon set for this type of result?
 835             $sIcon = ClassTypes\getIconFile($aResult);
 836             if (isset($sIcon)) {
 837                 $aResult['icon'] = $sIcon;
 838             }
 839
 840             $sLabel = ClassTypes\getLabel($aResult);
 841             if (isset($sLabel)) {
 842                 $aResult['label'] = $sLabel;
 843             }
 844             $aResult['name'] = $aResult['langaddress'];
 845
 846             if ($oCtx->hasNearPoint()) {
 847                 $aResult['importance'] = 0.001;
 848                 $aResult['foundorder'] = $aResult['addressimportance'];
 849             } else {
 850                 if ($aResult['importance'] == 0) {
 851                     $aResult['importance'] = 0.0001;
 852                 }
 853                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 854                     $aResult['lon'],
 855                     $aResult['lat']
 856                 );
 857
 858                 // secondary ordering (for results with same importance (the smaller the better):
 859                 // - approximate importance of address parts
 860                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 861                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 862                 } else {
 863                     $aResult['foundorder'] = -$aResult['importance'];
 864                 }
 865                 // - number of exact matches from the query
 866                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 867                 // - importance of the class/type
 868                 $iClassImportance = ClassTypes\getImportance($aResult);
 869                 if (isset($iClassImportance)) {
 870                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 871                 } else {
 872                     $aResult['foundorder'] += 0.01;
 873                 }
 874                 // - rank
 875                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 876
 877                 // Adjust importance for the number of exact string matches in the result
 878                 $iCountWords = 0;
 879                 $sAddress = $aResult['langaddress'];
 880                 foreach ($aRecheckWords as $i => $sWord) {
 881                     if (grapheme_stripos($sAddress, $sWord)!==false) {
 882                         $iCountWords++;
 883                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 884                             $iCountWords += 0.1;
 885                         }
 886                     }
 887                 }
 888
 889                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 890                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 891             }
 892             $aSearchResults[$iIdx] = $aResult;
 893         }
 894         uasort($aSearchResults, 'byImportance');
 895         Debug::printVar('Pre-filter results', $aSearchResults);
 896
 897         $aOSMIDDone = array();
 898         $aClassTypeNameDone = array();
 899         $aToFilter = $aSearchResults;
 900         $aSearchResults = array();
 901
 902         foreach ($aToFilter as $aResult) {
 903             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 904             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 905                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 906             ) {
 907                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 908                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 909                 $aSearchResults[] = $aResult;
 910             }
 911
 912             // Absolute limit on number of results
 913             if (count($aSearchResults) >= $this->iFinalLimit) {
 914                 break;
 915             }
 916         }
 917
 918         Debug::printVar('Post-filter results', $aSearchResults);
 919         return $aSearchResults;
 920     } // end lookup()
 921
 922     public function debugInfo()
 923     {
 924         return array(
 925                 'Query' => $this->sQuery,
 926                 'Structured query' => $this->aStructuredQuery,
 927                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 928                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 929                 'Limit (for searches)' => $this->iLimit,
 930                 'Limit (for results)'=> $this->iFinalLimit,
 931                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 932                 'Bounded search' => $this->bBoundedSearch,
 933                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 934                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 935                 'Route width' => $this->aRouteWidth,
 936                 'Max rank' => $this->iMaxRank,
 937                 'Min address rank' => $this->iMinAddressRank,
 938                 'Max address rank' => $this->iMaxAddressRank,
 939                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 940                );
 941     }
 942 } // end class