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 $oSearchTerm) {
 365                             if ($oSearchTerm->isExtendable($oCurrentSearch, $oPosition)) {
 366                                 $aNewSearches = $oSearchTerm->extendSearch(
 367                                     $oCurrentSearch,
 368                                     $oPosition
 369                                 );
 370
 371                                 foreach ($aNewSearches as $oSearch) {
 372                                     if ($oSearch->getRank() < $this->iMaxRank) {
 373                                         $aNewWordsetSearches[] = $oSearch;
 374                                     }
 375                                 }
 376                             }
 377                         }
 378                     }
 379                     // Sort and cut
 380                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 381                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 382                 }
 383
 384                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 385                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 386
 387                 $aSearchHash = array();
 388                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 389                     $sHash = serialize($aSearch);
 390                     if (isset($aSearchHash[$sHash])) {
 391                         unset($aNewPhraseSearches[$iSearch]);
 392                     } else {
 393                         $aSearchHash[$sHash] = 1;
 394                     }
 395                 }
 396
 397                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 398             }
 399
 400             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 401             $aGroupedSearches = array();
 402             foreach ($aNewPhraseSearches as $aSearch) {
 403                 $iRank = $aSearch->getRank();
 404                 if ($iRank < $this->iMaxRank) {
 405                     if (!isset($aGroupedSearches[$iRank])) {
 406                         $aGroupedSearches[$iRank] = array();
 407                     }
 408                     $aGroupedSearches[$iRank][] = $aSearch;
 409                 }
 410             }
 411             ksort($aGroupedSearches);
 412
 413             $iSearchCount = 0;
 414             $aSearches = array();
 415             foreach ($aGroupedSearches as $aNewSearches) {
 416                 $iSearchCount += count($aNewSearches);
 417                 $aSearches = array_merge($aSearches, $aNewSearches);
 418                 if ($iSearchCount > 50) {
 419                     break;
 420                 }
 421             }
 422         }
 423
 424         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 425         $aGroupedSearches = array();
 426         foreach ($aSearches as $oSearch) {
 427             if (!$oSearch->isValidSearch()) {
 428                 continue;
 429             }
 430
 431             $iRank = $oSearch->getRank();
 432             if (!isset($aGroupedSearches[$iRank])) {
 433                 $aGroupedSearches[$iRank] = array();
 434             }
 435             $aGroupedSearches[$iRank][] = $oSearch;
 436         }
 437         ksort($aGroupedSearches);
 438
 439         return $aGroupedSearches;
 440     }
 441
 442     /* Perform the actual query lookup.
 443
 444         Returns an ordered list of results, each with the following fields:
 445             osm_type: type of corresponding OSM object
 446                         N - node
 447                         W - way
 448                         R - relation
 449                         P - postcode (internally computed)
 450             osm_id: id of corresponding OSM object
 451             class: general object class (corresponds to tag key of primary OSM tag)
 452             type: subclass of object (corresponds to tag value of primary OSM tag)
 453             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 454             rank_search: rank in search hierarchy
 455                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 456             rank_address: rank in address hierarchy (determines orer in address)
 457             place_id: internal key (may differ between different instances)
 458             country_code: ISO country code
 459             langaddress: localized full address
 460             placename: localized name of object
 461             ref: content of ref tag (if available)
 462             lon: longitude
 463             lat: latitude
 464             importance: importance of place based on Wikipedia link count
 465             addressimportance: cumulated importance of address elements
 466             extra_place: type of place (for admin boundaries, if there is a place tag)
 467             aBoundingBox: bounding Box
 468             label: short description of the object class/type (English only)
 469             name: full name (currently the same as langaddress)
 470             foundorder: secondary ordering for places with same importance
 471     */
 472
 473
 474     public function lookup()
 475     {
 476         Debug::newFunction('Geocode::lookup');
 477         if (!$this->sQuery && !$this->aStructuredQuery) {
 478             return array();
 479         }
 480
 481         Debug::printDebugArray('Geocode', $this);
 482
 483         $oCtx = new SearchContext();
 484
 485         if ($this->aRoutePoints) {
 486             $oCtx->setViewboxFromRoute(
 487                 $this->oDB,
 488                 $this->aRoutePoints,
 489                 $this->aRouteWidth,
 490                 $this->bBoundedSearch
 491             );
 492         } elseif ($this->aViewBox) {
 493             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 494         }
 495         if ($this->aExcludePlaceIDs) {
 496             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 497         }
 498         if ($this->aCountryCodes) {
 499             $oCtx->setCountryList($this->aCountryCodes);
 500         }
 501         $this->oTokenizer->setCountryRestriction($this->aCountryCodes);
 502
 503         Debug::newSection('Query Preprocessing');
 504
 505         $sQuery = $this->sQuery;
 506         if (!preg_match('//u', $sQuery)) {
 507             userError('Query string is not UTF-8 encoded.');
 508         }
 509
 510         // Conflicts between US state abreviations and various words for 'the' in different languages
 511         if (isset($this->aLangPrefOrder['name:en'])) {
 512             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 513             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 514             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 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                     $aNewSearches = array();
 558                     foreach ($aSearches as $oSearch) {
 559                         foreach ($aTokens as $oToken) {
 560                             $oNewSearch = clone $oSearch;
 561                             $oNewSearch->setPoiSearch(
 562                                 $oToken->iOperator,
 563                                 $oToken->sClass,
 564                                 $oToken->sType
 565                             );
 566                             $aNewSearches[] = $oNewSearch;
 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, $oValidTokens);
 619
 620                     foreach ($aGroupedSearches as $aSearches) {
 621                         foreach ($aSearches as $aSearch) {
 622                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 623                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 624                             }
 625                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 626                         }
 627                     }
 628
 629                     $aGroupedSearches = $aReverseGroupedSearches;
 630                     ksort($aGroupedSearches);
 631                 }
 632             } else {
 633                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 634                 $aGroupedSearches = array();
 635                 foreach ($aSearches as $aSearch) {
 636                     if ($aSearch->getRank() < $this->iMaxRank) {
 637                         if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 638                             $aGroupedSearches[$aSearch->getRank()] = array();
 639                         }
 640                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 641                     }
 642                 }
 643                 ksort($aGroupedSearches);
 644             }
 645
 646             // Filter out duplicate searches
 647             $aSearchHash = array();
 648             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 649                 foreach ($aSearches as $iSearch => $aSearch) {
 650                     $sHash = serialize($aSearch);
 651                     if (isset($aSearchHash[$sHash])) {
 652                         unset($aGroupedSearches[$iGroup][$iSearch]);
 653                         if (empty($aGroupedSearches[$iGroup])) {
 654                             unset($aGroupedSearches[$iGroup]);
 655                         }
 656                     } else {
 657                         $aSearchHash[$sHash] = 1;
 658                     }
 659                 }
 660             }
 661
 662             Debug::printGroupedSearch(
 663                 $aGroupedSearches,
 664                 $oValidTokens->debugTokenByWordIdList()
 665             );
 666
 667             // Start the search process
 668             $iGroupLoop = 0;
 669             $iQueryLoop = 0;
 670             $aNextResults = array();
 671             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 672                 $iGroupLoop++;
 673                 $aResults = $aNextResults;
 674                 foreach ($aSearches as $oSearch) {
 675                     $iQueryLoop++;
 676
 677                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 678                     Debug::printGroupedSearch(
 679                         array($iGroupedRank => array($oSearch)),
 680                         $oValidTokens->debugTokenByWordIdList()
 681                     );
 682
 683                     $aNewResults = $oSearch->query(
 684                         $this->oDB,
 685                         $this->iMinAddressRank,
 686                         $this->iMaxAddressRank,
 687                         $this->iLimit
 688                     );
 689
 690                     // The same result may appear in different rounds, only
 691                     // use the one with minimal rank.
 692                     foreach ($aNewResults as $iPlace => $oRes) {
 693                         if (!isset($aResults[$iPlace])
 694                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 695                             $aResults[$iPlace] = $oRes;
 696                         }
 697                     }
 698
 699                     if ($iQueryLoop > 20) {
 700                         break;
 701                     }
 702                 }
 703
 704                 if (!empty($aResults)) {
 705                     $aSplitResults = Result::splitResults($aResults);
 706                     Debug::printVar('Split results', $aSplitResults);
 707                     if ($iGroupLoop <= 4
 708                         && reset($aSplitResults['head'])->iResultRank > 0
 709                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 710                         // Haven't found an exact match for the query yet.
 711                         // Therefore add result from the next group level.
 712                         $aNextResults = $aSplitResults['head'];
 713                         foreach ($aNextResults as $oRes) {
 714                             $oRes->iResultRank--;
 715                         }
 716                         foreach ($aSplitResults['tail'] as $oRes) {
 717                             $oRes->iResultRank--;
 718                             $aNextResults[$oRes->iId] = $oRes;
 719                         }
 720                         $aResults = array();
 721                     } else {
 722                         $aResults = $aSplitResults['head'];
 723                     }
 724                 }
 725
 726                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 727                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 728                     // reduces the number of place ids, like a filter
 729                     // rank_address is 30 for interpolated housenumbers
 730                     $aFilterSql = array();
 731                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 732                     if ($sPlaceIds) {
 733                         $sSQL = 'SELECT place_id FROM placex ';
 734                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 735                         $sSQL .= '  AND (';
 736                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 737                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 738                         if ($this->aAddressRankList) {
 739                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 740                         }
 741                         $sSQL .= ')';
 742                         $aFilterSql[] = $sSQL;
 743                     }
 744                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 745                     if ($sPlaceIds) {
 746                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 747                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 748                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 749                         if ($this->aAddressRankList) {
 750                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 751                         }
 752                         $sSQL .= ') ';
 753                         $aFilterSql[] = $sSQL;
 754                     }
 755
 756                     $aFilteredIDs = array();
 757                     if ($aFilterSql) {
 758                         $sSQL = join(' UNION ', $aFilterSql);
 759                         Debug::printSQL($sSQL);
 760                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 761                     }
 762
 763                     $tempIDs = array();
 764                     foreach ($aResults as $oResult) {
 765                         if (($this->iMaxAddressRank == 30 &&
 766                              ($oResult->iTable == Result::TABLE_OSMLINE
 767                               || $oResult->iTable == Result::TABLE_TIGER))
 768                             || in_array($oResult->iId, $aFilteredIDs)
 769                         ) {
 770                             $tempIDs[$oResult->iId] = $oResult;
 771                         }
 772                     }
 773                     $aResults = $tempIDs;
 774                 }
 775
 776                 if (!empty($aResults) || $iGroupLoop > 4 || $iQueryLoop > 30) {
 777                     break;
 778                 }
 779             }
 780         } else {
 781             // Just interpret as a reverse geocode
 782             $oReverse = new ReverseGeocode($this->oDB);
 783             $oReverse->setZoom(18);
 784
 785             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 786
 787             Debug::printVar('Reverse search', $oLookup);
 788
 789             if ($oLookup) {
 790                 $aResults = array($oLookup->iId => $oLookup);
 791             }
 792         }
 793
 794         // No results? Done
 795         if (empty($aResults)) {
 796             if ($this->bFallback && $this->fallbackStructuredQuery()) {
 797                 return $this->lookup();
 798             }
 799
 800             return array();
 801         }
 802
 803         if ($this->aAddressRankList) {
 804             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 805         }
 806         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 807         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 808         if ($oCtx->hasNearPoint()) {
 809             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 810         }
 811
 812         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 813
 814         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 815         foreach ($aRecheckWords as $i => $sWord) {
 816             if (!preg_match('/[\pL\pN]/', $sWord)) {
 817                 unset($aRecheckWords[$i]);
 818             }
 819         }
 820
 821         Debug::printVar('Recheck words', $aRecheckWords);
 822
 823         foreach ($aSearchResults as $iIdx => $aResult) {
 824             $fRadius = ClassTypes\getDefRadius($aResult);
 825
 826             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 827             if ($aOutlineResult) {
 828                 $aResult = array_merge($aResult, $aOutlineResult);
 829             }
 830
 831             // Is there an icon set for this type of result?
 832             $sIcon = ClassTypes\getIconFile($aResult);
 833             if (isset($sIcon)) {
 834                 $aResult['icon'] = $sIcon;
 835             }
 836
 837             $sLabel = ClassTypes\getLabel($aResult);
 838             if (isset($sLabel)) {
 839                 $aResult['label'] = $sLabel;
 840             }
 841             $aResult['name'] = $aResult['langaddress'];
 842
 843             if ($oCtx->hasNearPoint()) {
 844                 $aResult['importance'] = 0.001;
 845                 $aResult['foundorder'] = $aResult['addressimportance'];
 846             } else {
 847                 $aResult['importance'] = max(0.001, $aResult['importance']);
 848                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 849                     $aResult['lon'],
 850                     $aResult['lat']
 851                 );
 852
 853                 // secondary ordering (for results with same importance (the smaller the better):
 854                 // - approximate importance of address parts
 855                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 856                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 857                 } else {
 858                     $aResult['foundorder'] = -$aResult['importance'];
 859                 }
 860                 // - number of exact matches from the query
 861                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 862                 // - importance of the class/type
 863                 $iClassImportance = ClassTypes\getImportance($aResult);
 864                 if (isset($iClassImportance)) {
 865                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 866                 } else {
 867                     $aResult['foundorder'] += 0.01;
 868                 }
 869                 // - rank
 870                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 871
 872                 // Adjust importance for the number of exact string matches in the result
 873                 $iCountWords = 0;
 874                 $sAddress = $aResult['langaddress'];
 875                 foreach ($aRecheckWords as $i => $sWord) {
 876                     if (stripos($sAddress, $sWord)!==false) {
 877                         $iCountWords++;
 878                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 879                             $iCountWords += 0.1;
 880                         }
 881                     }
 882                 }
 883
 884                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 885                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 886             }
 887             $aSearchResults[$iIdx] = $aResult;
 888         }
 889         uasort($aSearchResults, 'byImportance');
 890         Debug::printVar('Pre-filter results', $aSearchResults);
 891
 892         $aOSMIDDone = array();
 893         $aClassTypeNameDone = array();
 894         $aToFilter = $aSearchResults;
 895         $aSearchResults = array();
 896
 897         foreach ($aToFilter as $aResult) {
 898             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 899             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 900                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 901             ) {
 902                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 903                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 904                 $aSearchResults[] = $aResult;
 905             }
 906
 907             // Absolute limit on number of results
 908             if (count($aSearchResults) >= $this->iFinalLimit) {
 909                 break;
 910             }
 911         }
 912
 913         Debug::printVar('Post-filter results', $aSearchResults);
 914         return $aSearchResults;
 915     } // end lookup()
 916
 917     public function debugInfo()
 918     {
 919         return array(
 920                 'Query' => $this->sQuery,
 921                 'Structured query' => $this->aStructuredQuery,
 922                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 923                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 924                 'Limit (for searches)' => $this->iLimit,
 925                 'Limit (for results)'=> $this->iFinalLimit,
 926                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 927                 'Bounded search' => $this->bBoundedSearch,
 928                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 929                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 930                 'Route width' => $this->aRouteWidth,
 931                 'Max rank' => $this->iMaxRank,
 932                 'Min address rank' => $this->iMinAddressRank,
 933                 'Max address rank' => $this->iMaxAddressRank,
 934                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 935                );
 936     }
 937 } // end class