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 = $iLimit + min($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 acurate 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         $sQuery = $oParams->getString('q');
 261         if (!$sQuery) {
 262             $this->setStructuredQuery(
 263                 $oParams->getString('amenity'),
 264                 $oParams->getString('street'),
 265                 $oParams->getString('city'),
 266                 $oParams->getString('county'),
 267                 $oParams->getString('state'),
 268                 $oParams->getString('country'),
 269                 $oParams->getString('postalcode')
 270             );
 271         } else {
 272             $this->setQuery($sQuery);
 273         }
 274     }
 275
 276     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 277     {
 278         $sValue = trim($sValue);
 279         if (!$sValue) {
 280             return false;
 281         }
 282         $this->aStructuredQuery[$sKey] = $sValue;
 283         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 284             $this->iMinAddressRank = $iNewMinAddressRank;
 285             $this->iMaxAddressRank = $iNewMaxAddressRank;
 286         }
 287         if ($aItemListValues) {
 288             $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 289         }
 290         return true;
 291     }
 292
 293     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 294     {
 295         $this->sQuery = false;
 296
 297         // Reset
 298         $this->iMinAddressRank = 0;
 299         $this->iMaxAddressRank = 30;
 300         $this->aAddressRankList = array();
 301
 302         $this->aStructuredQuery = array();
 303         $this->sAllowedTypesSQLList = false;
 304
 305         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 306         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 307         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 308         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 309         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 310         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 311         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 312
 313         if (!empty($this->aStructuredQuery)) {
 314             $this->sQuery = join(', ', $this->aStructuredQuery);
 315             if ($this->iMaxAddressRank < 30) {
 316                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 317             }
 318         }
 319     }
 320
 321     public function fallbackStructuredQuery()
 322     {
 323         $aParams = $this->aStructuredQuery;
 324
 325         if (!$aParams || count($aParams) == 1) {
 326             return false;
 327         }
 328
 329         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 330
 331         foreach ($aOrderToFallback as $sType) {
 332             if (isset($aParams[$sType])) {
 333                 unset($aParams[$sType]);
 334                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 335                 return true;
 336             }
 337         }
 338
 339         return false;
 340     }
 341
 342     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens)
 343     {
 344         /*
 345              Calculate all searches using oValidTokens i.e.
 346              'Wodsworth Road, Sheffield' =>
 347
 348              Phrase Wordset
 349              0      0       (wodsworth road)
 350              0      1       (wodsworth)(road)
 351              1      0       (sheffield)
 352
 353              Score how good the search is so they can be ordered
 354          */
 355         foreach ($aPhrases as $iPhrase => $oPhrase) {
 356             $aNewPhraseSearches = array();
 357             $oPosition = new SearchPosition(
 358                 $oPhrase->getPhraseType(),
 359                 $iPhrase,
 360                 count($aPhrases)
 361             );
 362
 363             foreach ($oPhrase->getWordSets() as $aWordset) {
 364                 $aWordsetSearches = $aSearches;
 365
 366                 // Add all words from this wordset
 367                 foreach ($aWordset as $iToken => $sToken) {
 368                     $aNewWordsetSearches = array();
 369                     $oPosition->setTokenPosition($iToken, count($aWordset));
 370
 371                     foreach ($aWordsetSearches as $oCurrentSearch) {
 372                         foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 373                             if ($oSearchTerm->isExtendable($oCurrentSearch, $oPosition)) {
 374                                 $aNewSearches = $oSearchTerm->extendSearch(
 375                                     $oCurrentSearch,
 376                                     $oPosition
 377                                 );
 378
 379                                 foreach ($aNewSearches as $oSearch) {
 380                                     if ($oSearch->getRank() < $this->iMaxRank) {
 381                                         $aNewWordsetSearches[] = $oSearch;
 382                                     }
 383                                 }
 384                             }
 385                         }
 386                     }
 387                     // Sort and cut
 388                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 389                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 390                 }
 391
 392                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 393                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 394
 395                 $aSearchHash = array();
 396                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 397                     $sHash = serialize($aSearch);
 398                     if (isset($aSearchHash[$sHash])) {
 399                         unset($aNewPhraseSearches[$iSearch]);
 400                     } else {
 401                         $aSearchHash[$sHash] = 1;
 402                     }
 403                 }
 404
 405                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 406             }
 407
 408             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 409             $aGroupedSearches = array();
 410             foreach ($aNewPhraseSearches as $aSearch) {
 411                 $iRank = $aSearch->getRank();
 412                 if ($iRank < $this->iMaxRank) {
 413                     if (!isset($aGroupedSearches[$iRank])) {
 414                         $aGroupedSearches[$iRank] = array();
 415                     }
 416                     $aGroupedSearches[$iRank][] = $aSearch;
 417                 }
 418             }
 419             ksort($aGroupedSearches);
 420
 421             $iSearchCount = 0;
 422             $aSearches = array();
 423             foreach ($aGroupedSearches as $aNewSearches) {
 424                 $iSearchCount += count($aNewSearches);
 425                 $aSearches = array_merge($aSearches, $aNewSearches);
 426                 if ($iSearchCount > 50) {
 427                     break;
 428                 }
 429             }
 430         }
 431
 432         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 433         $aGroupedSearches = array();
 434         foreach ($aSearches as $oSearch) {
 435             if (!$oSearch->isValidSearch()) {
 436                 continue;
 437             }
 438
 439             $iRank = $oSearch->getRank();
 440             if (!isset($aGroupedSearches[$iRank])) {
 441                 $aGroupedSearches[$iRank] = array();
 442             }
 443             $aGroupedSearches[$iRank][] = $oSearch;
 444         }
 445         ksort($aGroupedSearches);
 446
 447         return $aGroupedSearches;
 448     }
 449
 450     /* Perform the actual query lookup.
 451
 452         Returns an ordered list of results, each with the following fields:
 453             osm_type: type of corresponding OSM object
 454                         N - node
 455                         W - way
 456                         R - relation
 457                         P - postcode (internally computed)
 458             osm_id: id of corresponding OSM object
 459             class: general object class (corresponds to tag key of primary OSM tag)
 460             type: subclass of object (corresponds to tag value of primary OSM tag)
 461             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 462             rank_search: rank in search hierarchy
 463                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 464             rank_address: rank in address hierarchy (determines orer in address)
 465             place_id: internal key (may differ between different instances)
 466             country_code: ISO country code
 467             langaddress: localized full address
 468             placename: localized name of object
 469             ref: content of ref tag (if available)
 470             lon: longitude
 471             lat: latitude
 472             importance: importance of place based on Wikipedia link count
 473             addressimportance: cumulated importance of address elements
 474             extra_place: type of place (for admin boundaries, if there is a place tag)
 475             aBoundingBox: bounding Box
 476             label: short description of the object class/type (English only)
 477             name: full name (currently the same as langaddress)
 478             foundorder: secondary ordering for places with same importance
 479     */
 480
 481
 482     public function lookup()
 483     {
 484         Debug::newFunction('Geocode::lookup');
 485         if (!$this->sQuery && !$this->aStructuredQuery) {
 486             return array();
 487         }
 488
 489         Debug::printDebugArray('Geocode', $this);
 490
 491         $oCtx = new SearchContext();
 492
 493         if ($this->aRoutePoints) {
 494             $oCtx->setViewboxFromRoute(
 495                 $this->oDB,
 496                 $this->aRoutePoints,
 497                 $this->aRouteWidth,
 498                 $this->bBoundedSearch
 499             );
 500         } elseif ($this->aViewBox) {
 501             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 502         }
 503         if ($this->aExcludePlaceIDs) {
 504             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 505         }
 506         if ($this->aCountryCodes) {
 507             $oCtx->setCountryList($this->aCountryCodes);
 508         }
 509
 510         Debug::newSection('Query Preprocessing');
 511
 512         $sQuery = $this->sQuery;
 513         if (!preg_match('//u', $sQuery)) {
 514             userError('Query string is not UTF-8 encoded.');
 515         }
 516
 517         // Do we have anything that looks like a lat/lon pair?
 518         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 519
 520         if ($sQuery || $this->aStructuredQuery) {
 521             // Start with a single blank search
 522             $aSearches = array(new SearchDescription($oCtx));
 523
 524             if ($sQuery) {
 525                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 526             }
 527
 528             $sSpecialTerm = '';
 529             if ($sQuery) {
 530                 preg_match_all(
 531                     '/\\[([\\w ]*)\\]/u',
 532                     $sQuery,
 533                     $aSpecialTermsRaw,
 534                     PREG_SET_ORDER
 535                 );
 536                 if (!empty($aSpecialTermsRaw)) {
 537                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 538                 }
 539
 540                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 541                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 542                     if (!$sSpecialTerm) {
 543                         $sSpecialTerm = $aSpecialTerm[1];
 544                     }
 545                 }
 546             }
 547             if (!$sSpecialTerm && $this->aStructuredQuery
 548                 && isset($this->aStructuredQuery['amenity'])) {
 549                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 550                 unset($this->aStructuredQuery['amenity']);
 551             }
 552
 553             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 554                 $aTokens = $this->oTokenizer->tokensForSpecialTerm($sSpecialTerm);
 555
 556                 if (!empty($aTokens)) {
 557                     $aNewSearches = array();
 558                     $oPosition = new SearchPosition('', 0, 1);
 559                     $oPosition->setTokenPosition(0, 1);
 560
 561                     foreach ($aSearches as $oSearch) {
 562                         foreach ($aTokens as $oToken) {
 563                             $aNewSearches = array_merge(
 564                                 $aNewSearches,
 565                                 $oToken->extendSearch($oSearch, $oPosition)
 566                             );
 567                         }
 568                     }
 569                     $aSearches = $aNewSearches;
 570                 }
 571             }
 572
 573             // Split query into phrases
 574             // Commas are used to reduce the search space by indicating where phrases split
 575             $aPhrases = array();
 576             if ($this->aStructuredQuery) {
 577                 foreach ($this->aStructuredQuery as $iPhrase => $sPhrase) {
 578                     $aPhrases[] = new Phrase($sPhrase, $iPhrase);
 579                 }
 580             } else {
 581                 foreach (explode(',', $sQuery) as $sPhrase) {
 582                     $aPhrases[] = new Phrase($sPhrase, '');
 583                 }
 584             }
 585
 586             Debug::printDebugArray('Search context', $oCtx);
 587             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 588
 589             Debug::newSection('Tokenization');
 590             $oValidTokens = $this->oTokenizer->extractTokensFromPhrases($aPhrases);
 591
 592             if ($oValidTokens->count() > 0) {
 593                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 594
 595                 $aPhrases = array_filter($aPhrases, function ($oPhrase) {
 596                     return $oPhrase->getWordSets() !== null;
 597                 });
 598
 599                 // Any words that have failed completely?
 600                 // TODO: suggestions
 601
 602                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 603                 Debug::printDebugTable('Phrases', $aPhrases);
 604
 605                 Debug::newSection('Search candidates');
 606
 607                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 608
 609                 if (!$this->aStructuredQuery) {
 610                     // Reverse phrase array and also reverse the order of the wordsets in
 611                     // the first and final phrase. Don't bother about phrases in the middle
 612                     // because order in the address doesn't matter.
 613                     $aPhrases = array_reverse($aPhrases);
 614                     $aPhrases[0]->invertWordSets();
 615                     if (count($aPhrases) > 1) {
 616                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 617                     }
 618                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 619
 620                     foreach ($aReverseGroupedSearches as $aSearches) {
 621                         foreach ($aSearches as $aSearch) {
 622                             if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 623                                 $aGroupedSearches[$aSearch->getRank()] = array();
 624                             }
 625                             $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 626                         }
 627                     }
 628
 629                     ksort($aGroupedSearches);
 630                 }
 631             } else {
 632                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 633                 $aGroupedSearches = array();
 634                 foreach ($aSearches as $aSearch) {
 635                     if ($aSearch->getRank() < $this->iMaxRank) {
 636                         if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 637                             $aGroupedSearches[$aSearch->getRank()] = array();
 638                         }
 639                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 640                     }
 641                 }
 642                 ksort($aGroupedSearches);
 643             }
 644
 645             // Filter out duplicate searches
 646             $aSearchHash = array();
 647             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 648                 foreach ($aSearches as $iSearch => $aSearch) {
 649                     $sHash = serialize($aSearch);
 650                     if (isset($aSearchHash[$sHash])) {
 651                         unset($aGroupedSearches[$iGroup][$iSearch]);
 652                         if (empty($aGroupedSearches[$iGroup])) {
 653                             unset($aGroupedSearches[$iGroup]);
 654                         }
 655                     } else {
 656                         $aSearchHash[$sHash] = 1;
 657                     }
 658                 }
 659             }
 660
 661             Debug::printGroupedSearch(
 662                 $aGroupedSearches,
 663                 $oValidTokens->debugTokenByWordIdList()
 664             );
 665
 666             // Start the search process
 667             $iGroupLoop = 0;
 668             $iQueryLoop = 0;
 669             $aNextResults = array();
 670             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 671                 $iGroupLoop++;
 672                 $aResults = $aNextResults;
 673                 foreach ($aSearches as $oSearch) {
 674                     $iQueryLoop++;
 675
 676                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 677                     Debug::printGroupedSearch(
 678                         array($iGroupedRank => array($oSearch)),
 679                         $oValidTokens->debugTokenByWordIdList()
 680                     );
 681
 682                     $aNewResults = $oSearch->query(
 683                         $this->oDB,
 684                         $this->iMinAddressRank,
 685                         $this->iMaxAddressRank,
 686                         $this->iLimit
 687                     );
 688
 689                     // The same result may appear in different rounds, only
 690                     // use the one with minimal rank.
 691                     foreach ($aNewResults as $iPlace => $oRes) {
 692                         if (!isset($aResults[$iPlace])
 693                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 694                             $aResults[$iPlace] = $oRes;
 695                         }
 696                     }
 697
 698                     if ($iQueryLoop > 20) {
 699                         break;
 700                     }
 701                 }
 702
 703                 if (!empty($aResults)) {
 704                     $aSplitResults = Result::splitResults($aResults);
 705                     Debug::printVar('Split results', $aSplitResults);
 706                     if ($iGroupLoop <= 4
 707                         && reset($aSplitResults['head'])->iResultRank > 0
 708                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 709                         // Haven't found an exact match for the query yet.
 710                         // Therefore add result from the next group level.
 711                         $aNextResults = $aSplitResults['head'];
 712                         foreach ($aNextResults as $oRes) {
 713                             $oRes->iResultRank--;
 714                         }
 715                         foreach ($aSplitResults['tail'] as $oRes) {
 716                             $oRes->iResultRank--;
 717                             $aNextResults[$oRes->iId] = $oRes;
 718                         }
 719                         $aResults = array();
 720                     } else {
 721                         $aResults = $aSplitResults['head'];
 722                     }
 723                 }
 724
 725                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 726                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 727                     // reduces the number of place ids, like a filter
 728                     // rank_address is 30 for interpolated housenumbers
 729                     $aFilterSql = array();
 730                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 731                     if ($sPlaceIds) {
 732                         $sSQL = 'SELECT place_id FROM placex ';
 733                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 734                         $sSQL .= '  AND (';
 735                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 736                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 737                         if ($this->aAddressRankList) {
 738                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 739                         }
 740                         $sSQL .= ')';
 741                         $aFilterSql[] = $sSQL;
 742                     }
 743                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 744                     if ($sPlaceIds) {
 745                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 746                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 747                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 748                         if ($this->aAddressRankList) {
 749                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 750                         }
 751                         $sSQL .= ') ';
 752                         $aFilterSql[] = $sSQL;
 753                     }
 754
 755                     $aFilteredIDs = array();
 756                     if ($aFilterSql) {
 757                         $sSQL = join(' UNION ', $aFilterSql);
 758                         Debug::printSQL($sSQL);
 759                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 760                     }
 761
 762                     $tempIDs = array();
 763                     foreach ($aResults as $oResult) {
 764                         if (($this->iMaxAddressRank == 30 &&
 765                              ($oResult->iTable == Result::TABLE_OSMLINE
 766                               || $oResult->iTable == Result::TABLE_TIGER))
 767                             || in_array($oResult->iId, $aFilteredIDs)
 768                         ) {
 769                             $tempIDs[$oResult->iId] = $oResult;
 770                         }
 771                     }
 772                     $aResults = $tempIDs;
 773                 }
 774
 775                 if (!empty($aResults) || $iGroupLoop > 4 || $iQueryLoop > 30) {
 776                     break;
 777                 }
 778             }
 779         } else {
 780             // Just interpret as a reverse geocode
 781             $oReverse = new ReverseGeocode($this->oDB);
 782             $oReverse->setZoom(18);
 783
 784             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 785
 786             Debug::printVar('Reverse search', $oLookup);
 787
 788             if ($oLookup) {
 789                 $aResults = array($oLookup->iId => $oLookup);
 790             }
 791         }
 792
 793         // No results? Done
 794         if (empty($aResults)) {
 795             if ($this->bFallback && $this->fallbackStructuredQuery()) {
 796                 return $this->lookup();
 797             }
 798
 799             return array();
 800         }
 801
 802         if ($this->aAddressRankList) {
 803             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 804         }
 805         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 806         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 807         if ($oCtx->hasNearPoint()) {
 808             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 809         }
 810
 811         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 812
 813         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 814         foreach ($aRecheckWords as $i => $sWord) {
 815             if (!preg_match('/[\pL\pN]/', $sWord)) {
 816                 unset($aRecheckWords[$i]);
 817             }
 818         }
 819
 820         Debug::printVar('Recheck words', $aRecheckWords);
 821
 822         foreach ($aSearchResults as $iIdx => $aResult) {
 823             $fRadius = ClassTypes\getDefRadius($aResult);
 824
 825             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 826             if ($aOutlineResult) {
 827                 $aResult = array_merge($aResult, $aOutlineResult);
 828             }
 829
 830             // Is there an icon set for this type of result?
 831             $sIcon = ClassTypes\getIconFile($aResult);
 832             if (isset($sIcon)) {
 833                 $aResult['icon'] = $sIcon;
 834             }
 835
 836             $sLabel = ClassTypes\getLabel($aResult);
 837             if (isset($sLabel)) {
 838                 $aResult['label'] = $sLabel;
 839             }
 840             $aResult['name'] = $aResult['langaddress'];
 841
 842             if ($oCtx->hasNearPoint()) {
 843                 $aResult['importance'] = 0.001;
 844                 $aResult['foundorder'] = $aResult['addressimportance'];
 845             } else {
 846                 if ($aResult['importance'] == 0) {
 847                     $aResult['importance'] = 0.0001;
 848                 }
 849                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 850                     $aResult['lon'],
 851                     $aResult['lat']
 852                 );
 853
 854                 // secondary ordering (for results with same importance (the smaller the better):
 855                 // - approximate importance of address parts
 856                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 857                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 858                 } else {
 859                     $aResult['foundorder'] = -$aResult['importance'];
 860                 }
 861                 // - number of exact matches from the query
 862                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 863                 // - importance of the class/type
 864                 $iClassImportance = ClassTypes\getImportance($aResult);
 865                 if (isset($iClassImportance)) {
 866                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 867                 } else {
 868                     $aResult['foundorder'] += 0.01;
 869                 }
 870                 // - rank
 871                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 872
 873                 // Adjust importance for the number of exact string matches in the result
 874                 $iCountWords = 0;
 875                 $sAddress = $aResult['langaddress'];
 876                 foreach ($aRecheckWords as $i => $sWord) {
 877                     if (stripos($sAddress, $sWord)!==false) {
 878                         $iCountWords++;
 879                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 880                             $iCountWords += 0.1;
 881                         }
 882                     }
 883                 }
 884
 885                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 886                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 887             }
 888             $aSearchResults[$iIdx] = $aResult;
 889         }
 890         uasort($aSearchResults, 'byImportance');
 891         Debug::printVar('Pre-filter results', $aSearchResults);
 892
 893         $aOSMIDDone = array();
 894         $aClassTypeNameDone = array();
 895         $aToFilter = $aSearchResults;
 896         $aSearchResults = array();
 897
 898         foreach ($aToFilter as $aResult) {
 899             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 900             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 901                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 902             ) {
 903                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 904                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 905                 $aSearchResults[] = $aResult;
 906             }
 907
 908             // Absolute limit on number of results
 909             if (count($aSearchResults) >= $this->iFinalLimit) {
 910                 break;
 911             }
 912         }
 913
 914         Debug::printVar('Post-filter results', $aSearchResults);
 915         return $aSearchResults;
 916     } // end lookup()
 917
 918     public function debugInfo()
 919     {
 920         return array(
 921                 'Query' => $this->sQuery,
 922                 'Structured query' => $this->aStructuredQuery,
 923                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 924                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 925                 'Limit (for searches)' => $this->iLimit,
 926                 'Limit (for results)'=> $this->iFinalLimit,
 927                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 928                 'Bounded search' => $this->bBoundedSearch,
 929                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 930                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 931                 'Route width' => $this->aRouteWidth,
 932                 'Max rank' => $this->iMaxRank,
 933                 'Min address rank' => $this->iMinAddressRank,
 934                 'Max address rank' => $this->iMaxAddressRank,
 935                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 936                );
 937     }
 938 } // end class