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