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
 502         Debug::newSection('Query Preprocessing');
 503
 504         $sQuery = $this->sQuery;
 505         if (!preg_match('//u', $sQuery)) {
 506             userError('Query string is not UTF-8 encoded.');
 507         }
 508
 509         // Conflicts between US state abreviations and various words for 'the' in different languages
 510         if (isset($this->aLangPrefOrder['name:en'])) {
 511             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 512             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 513             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 514         }
 515
 516         // Do we have anything that looks like a lat/lon pair?
 517         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 518
 519         if ($sQuery || $this->aStructuredQuery) {
 520             // Start with a single blank search
 521             $aSearches = array(new SearchDescription($oCtx));
 522
 523             if ($sQuery) {
 524                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 525             }
 526
 527             $sSpecialTerm = '';
 528             if ($sQuery) {
 529                 preg_match_all(
 530                     '/\\[([\\w ]*)\\]/u',
 531                     $sQuery,
 532                     $aSpecialTermsRaw,
 533                     PREG_SET_ORDER
 534                 );
 535                 if (!empty($aSpecialTermsRaw)) {
 536                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 537                 }
 538
 539                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 540                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 541                     if (!$sSpecialTerm) {
 542                         $sSpecialTerm = $aSpecialTerm[1];
 543                     }
 544                 }
 545             }
 546             if (!$sSpecialTerm && $this->aStructuredQuery
 547                 && isset($this->aStructuredQuery['amenity'])) {
 548                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 549                 unset($this->aStructuredQuery['amenity']);
 550             }
 551
 552             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 553                 $aTokens = $this->oTokenizer->tokensForSpecialTerm($sSpecialTerm);
 554
 555                 if (!empty($aTokens)) {
 556                     $aNewSearches = array();
 557                     $oPosition = new SearchPosition('', 0, 1);
 558                     $oPosition->setTokenPosition(0, 1);
 559
 560                     foreach ($aSearches as $oSearch) {
 561                         foreach ($aTokens as $oToken) {
 562                             $aNewSearches = array_merge(
 563                                 $aNewSearches,
 564                                 $oToken->extendSearch($oSearch, $oPosition)
 565                             );
 566                         }
 567                     }
 568                     $aSearches = $aNewSearches;
 569                 }
 570             }
 571
 572             // Split query into phrases
 573             // Commas are used to reduce the search space by indicating where phrases split
 574             $aPhrases = array();
 575             if ($this->aStructuredQuery) {
 576                 foreach ($this->aStructuredQuery as $iPhrase => $sPhrase) {
 577                     $aPhrases[] = new Phrase($sPhrase, $iPhrase);
 578                 }
 579             } else {
 580                 foreach (explode(',', $sQuery) as $sPhrase) {
 581                     $aPhrases[] = new Phrase($sPhrase, '');
 582                 }
 583             }
 584
 585             Debug::printDebugArray('Search context', $oCtx);
 586             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 587
 588             Debug::newSection('Tokenization');
 589             $oValidTokens = $this->oTokenizer->extractTokensFromPhrases($aPhrases);
 590
 591             if ($oValidTokens->count() > 0) {
 592                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 593
 594                 $aPhrases = array_filter($aPhrases, function ($oPhrase) {
 595                     return $oPhrase->getWordSets() !== null;
 596                 });
 597
 598                 // Any words that have failed completely?
 599                 // TODO: suggestions
 600
 601                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 602                 Debug::printDebugTable('Phrases', $aPhrases);
 603
 604                 Debug::newSection('Search candidates');
 605
 606                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 607
 608                 if (!$this->aStructuredQuery) {
 609                     // Reverse phrase array and also reverse the order of the wordsets in
 610                     // the first and final phrase. Don't bother about phrases in the middle
 611                     // because order in the address doesn't matter.
 612                     $aPhrases = array_reverse($aPhrases);
 613                     $aPhrases[0]->invertWordSets();
 614                     if (count($aPhrases) > 1) {
 615                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 616                     }
 617                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 618
 619                     foreach ($aGroupedSearches as $aSearches) {
 620                         foreach ($aSearches as $aSearch) {
 621                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 622                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 623                             }
 624                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 625                         }
 626                     }
 627
 628                     $aGroupedSearches = $aReverseGroupedSearches;
 629                     ksort($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                 $aResult['importance'] = max(0.001, $aResult['importance']);
 847                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 848                     $aResult['lon'],
 849                     $aResult['lat']
 850                 );
 851
 852                 // secondary ordering (for results with same importance (the smaller the better):
 853                 // - approximate importance of address parts
 854                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 855                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 856                 } else {
 857                     $aResult['foundorder'] = -$aResult['importance'];
 858                 }
 859                 // - number of exact matches from the query
 860                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 861                 // - importance of the class/type
 862                 $iClassImportance = ClassTypes\getImportance($aResult);
 863                 if (isset($iClassImportance)) {
 864                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 865                 } else {
 866                     $aResult['foundorder'] += 0.01;
 867                 }
 868                 // - rank
 869                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 870
 871                 // Adjust importance for the number of exact string matches in the result
 872                 $iCountWords = 0;
 873                 $sAddress = $aResult['langaddress'];
 874                 foreach ($aRecheckWords as $i => $sWord) {
 875                     if (stripos($sAddress, $sWord)!==false) {
 876                         $iCountWords++;
 877                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 878                             $iCountWords += 0.1;
 879                         }
 880                     }
 881                 }
 882
 883                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 884                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 885             }
 886             $aSearchResults[$iIdx] = $aResult;
 887         }
 888         uasort($aSearchResults, 'byImportance');
 889         Debug::printVar('Pre-filter results', $aSearchResults);
 890
 891         $aOSMIDDone = array();
 892         $aClassTypeNameDone = array();
 893         $aToFilter = $aSearchResults;
 894         $aSearchResults = array();
 895
 896         foreach ($aToFilter as $aResult) {
 897             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 898             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 899                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 900             ) {
 901                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 902                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 903                 $aSearchResults[] = $aResult;
 904             }
 905
 906             // Absolute limit on number of results
 907             if (count($aSearchResults) >= $this->iFinalLimit) {
 908                 break;
 909             }
 910         }
 911
 912         Debug::printVar('Post-filter results', $aSearchResults);
 913         return $aSearchResults;
 914     } // end lookup()
 915
 916     public function debugInfo()
 917     {
 918         return array(
 919                 'Query' => $this->sQuery,
 920                 'Structured query' => $this->aStructuredQuery,
 921                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 922                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 923                 'Limit (for searches)' => $this->iLimit,
 924                 'Limit (for results)'=> $this->iFinalLimit,
 925                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 926                 'Bounded search' => $this->bBoundedSearch,
 927                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 928                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 929                 'Route width' => $this->aRouteWidth,
 930                 'Max rank' => $this->iMaxRank,
 931                 'Min address rank' => $this->iMinAddressRank,
 932                 'Max address rank' => $this->iMaxAddressRank,
 933                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 934                );
 935     }
 936 } // end class