lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10 require_once(CONST_BasePath.'/lib/TokenList.php');
  11
  12 class Geocode
  13 {
  14     protected $oDB;
  15
  16     protected $oPlaceLookup;
  17
  18     protected $aLangPrefOrder = array();
  19
  20     protected $bIncludeAddressDetails = false;
  21
  22     protected $aExcludePlaceIDs = array();
  23     protected $bReverseInPlan = false;
  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     protected $oNormalizer = null;
  48
  49
  50     public function __construct(&$oDB)
  51     {
  52         $this->oDB =& $oDB;
  53         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  54         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  55     }
  56
  57     private function normTerm($sTerm)
  58     {
  59         if ($this->oNormalizer === null) {
  60             return $sTerm;
  61         }
  62
  63         return $this->oNormalizer->transliterate($sTerm);
  64     }
  65
  66     public function setReverseInPlan($bReverse)
  67     {
  68         $this->bReverseInPlan = $bReverse;
  69     }
  70
  71     public function setLanguagePreference($aLangPref)
  72     {
  73         $this->aLangPrefOrder = $aLangPref;
  74     }
  75
  76     public function getMoreUrlParams()
  77     {
  78         if ($this->aStructuredQuery) {
  79             $aParams = $this->aStructuredQuery;
  80         } else {
  81             $aParams = array('q' => $this->sQuery);
  82         }
  83
  84         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  85
  86         if ($this->aExcludePlaceIDs) {
  87             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  88         }
  89
  90         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  91         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  92
  93         if ($this->aCountryCodes) {
  94             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  95         }
  96
  97         if ($this->aViewBox) {
  98             $aParams['viewbox'] = join(',', $this->aViewBox);
  99         }
 100
 101         return $aParams;
 102     }
 103
 104     public function setLimit($iLimit = 10)
 105     {
 106         if ($iLimit > 50) $iLimit = 50;
 107         if ($iLimit < 1) $iLimit = 1;
 108
 109         $this->iFinalLimit = $iLimit;
 110         $this->iLimit = $iLimit + min($iLimit, 10);
 111     }
 112
 113     public function setFeatureType($sFeatureType)
 114     {
 115         switch ($sFeatureType) {
 116             case 'country':
 117                 $this->setRankRange(4, 4);
 118                 break;
 119             case 'state':
 120                 $this->setRankRange(8, 8);
 121                 break;
 122             case 'city':
 123                 $this->setRankRange(14, 16);
 124                 break;
 125             case 'settlement':
 126                 $this->setRankRange(8, 20);
 127                 break;
 128         }
 129     }
 130
 131     public function setRankRange($iMin, $iMax)
 132     {
 133         $this->iMinAddressRank = $iMin;
 134         $this->iMaxAddressRank = $iMax;
 135     }
 136
 137     public function setViewbox($aViewbox)
 138     {
 139         $aBox = array_map('floatval', $aViewbox);
 140
 141         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 142         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 143         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 144         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 145
 146         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 147             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 148         ) {
 149             userError("Bad parameter 'viewbox'. Not a box.");
 150         }
 151     }
 152
 153     private function viewboxImportanceFactor($fX, $fY)
 154     {
 155         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 156         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 157
 158         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 159         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 160
 161         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 162             return 1;
 163         }
 164
 165         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 166             return 0.5;
 167         }
 168
 169         return 0.25;
 170     }
 171
 172     public function setQuery($sQueryString)
 173     {
 174         $this->sQuery = $sQueryString;
 175         $this->aStructuredQuery = false;
 176     }
 177
 178     public function getQueryString()
 179     {
 180         return $this->sQuery;
 181     }
 182
 183
 184     public function loadParamArray($oParams, $sForceGeometryType = null)
 185     {
 186         $this->bIncludeAddressDetails
 187          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 188
 189         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 190
 191         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 192         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 193
 194         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 195
 196         // List of excluded Place IDs - used for more acurate pageing
 197         $sExcluded = $oParams->getStringList('exclude_place_ids');
 198         if ($sExcluded) {
 199             foreach ($sExcluded as $iExcludedPlaceID) {
 200                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 201                 if ($iExcludedPlaceID)
 202                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 203             }
 204
 205             if (isset($aExcludePlaceIDs))
 206                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 207         }
 208
 209         // Only certain ranks of feature
 210         $sFeatureType = $oParams->getString('featureType');
 211         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 212         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 213
 214         // Country code list
 215         $sCountries = $oParams->getStringList('countrycodes');
 216         if ($sCountries) {
 217             foreach ($sCountries as $sCountryCode) {
 218                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 219                     $aCountries[] = strtolower($sCountryCode);
 220                 }
 221             }
 222             if (isset($aCountries))
 223                 $this->aCountryCodes = $aCountries;
 224         }
 225
 226         $aViewbox = $oParams->getStringList('viewboxlbrt');
 227         if ($aViewbox) {
 228             if (count($aViewbox) != 4) {
 229                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 230             }
 231             $this->setViewbox($aViewbox);
 232         } else {
 233             $aViewbox = $oParams->getStringList('viewbox');
 234             if ($aViewbox) {
 235                 if (count($aViewbox) != 4) {
 236                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 237                 }
 238                 $this->setViewBox($aViewbox);
 239             } else {
 240                 $aRoute = $oParams->getStringList('route');
 241                 $fRouteWidth = $oParams->getFloat('routewidth');
 242                 if ($aRoute && $fRouteWidth) {
 243                     $this->aRoutePoints = $aRoute;
 244                     $this->aRouteWidth = $fRouteWidth;
 245                 }
 246             }
 247         }
 248
 249         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 250         $this->oPlaceLookup->setIncludeAddressDetails(false);
 251         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 252
 253         if ($this->bIncludeAddressDetails
 254             && $oParams->getString('format', '') == 'geocodejson'
 255            ) {
 256             $this->oPlaceLookup->setAddressAdminLevels(true);
 257         }
 258     }
 259
 260     public function setQueryFromParams($oParams)
 261     {
 262         // Search query
 263         $sQuery = $oParams->getString('q');
 264         if (!$sQuery) {
 265             $this->setStructuredQuery(
 266                 $oParams->getString('amenity'),
 267                 $oParams->getString('street'),
 268                 $oParams->getString('city'),
 269                 $oParams->getString('county'),
 270                 $oParams->getString('state'),
 271                 $oParams->getString('country'),
 272                 $oParams->getString('postalcode')
 273             );
 274             $this->setReverseInPlan(false);
 275         } else {
 276             $this->setQuery($sQuery);
 277         }
 278     }
 279
 280     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 281     {
 282         $sValue = trim($sValue);
 283         if (!$sValue) return false;
 284         $this->aStructuredQuery[$sKey] = $sValue;
 285         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 286             $this->iMinAddressRank = $iNewMinAddressRank;
 287             $this->iMaxAddressRank = $iNewMaxAddressRank;
 288         }
 289         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 290         return true;
 291     }
 292
 293     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 294     {
 295         $this->sQuery = false;
 296
 297         // Reset
 298         $this->iMinAddressRank = 0;
 299         $this->iMaxAddressRank = 30;
 300         $this->aAddressRankList = array();
 301
 302         $this->aStructuredQuery = array();
 303         $this->sAllowedTypesSQLList = false;
 304
 305         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 306         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 307         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 308         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 309         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 310         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 311         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 312
 313         if (!empty($this->aStructuredQuery)) {
 314             $this->sQuery = join(', ', $this->aStructuredQuery);
 315             if ($this->iMaxAddressRank < 30) {
 316                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 317             }
 318         }
 319     }
 320
 321     public function fallbackStructuredQuery()
 322     {
 323         if (!$this->aStructuredQuery) return false;
 324
 325         $aParams = $this->aStructuredQuery;
 326
 327         if (count($aParams) == 1) return false;
 328
 329         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 330
 331         foreach ($aOrderToFallback as $sType) {
 332             if (isset($aParams[$sType])) {
 333                 unset($aParams[$sType]);
 334                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 335                 return true;
 336             }
 337         }
 338
 339         return false;
 340     }
 341
 342     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
 343     {
 344         /*
 345              Calculate all searches using oValidTokens i.e.
 346              'Wodsworth Road, Sheffield' =>
 347
 348              Phrase Wordset
 349              0      0       (wodsworth road)
 350              0      1       (wodsworth)(road)
 351              1      0       (sheffield)
 352
 353              Score how good the search is so they can be ordered
 354          */
 355         foreach ($aPhrases as $iPhrase => $oPhrase) {
 356             $aNewPhraseSearches = array();
 357             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 358
 359             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 360                 // Too many permutations - too expensive
 361                 if ($iWordSet > 120) break;
 362
 363                 $aWordsetSearches = $aSearches;
 364
 365                 // Add all words from this wordset
 366                 foreach ($aWordset as $iToken => $sToken) {
 367                     //echo "<br><b>$sToken</b>";
 368                     $aNewWordsetSearches = array();
 369
 370                     foreach ($aWordsetSearches as $oCurrentSearch) {
 371                         //echo "<i>";
 372                         //var_dump($oCurrentSearch);
 373                         //echo "</i>";
 374
 375                         // Tokens with full name matches.
 376                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 377                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 378                                 $oSearchTerm,
 379                                 $oValidTokens->contains($sToken)
 380                                   && strpos($sToken, ' ') === false,
 381                                 $sPhraseType,
 382                                 $iToken == 0 && $iPhrase == 0,
 383                                 $iPhrase == 0,
 384                                 $iToken + 1 == count($aWordset)
 385                                   && $iPhrase + 1 == count($aPhrases)
 386                             );
 387
 388                             foreach ($aNewSearches as $oSearch) {
 389                                 if ($oSearch->getRank() < $this->iMaxRank) {
 390                                     $aNewWordsetSearches[] = $oSearch;
 391                                 }
 392                             }
 393                         }
 394                         // Look for partial matches.
 395                         // Note that there is no point in adding country terms here
 396                         // because country is omitted in the address.
 397                         if ($sPhraseType != 'country') {
 398                             // Allow searching for a word - but at extra cost
 399                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 400                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 401                                     $sToken,
 402                                     $oSearchTerm,
 403                                     $bIsStructured,
 404                                     $iPhrase,
 405                                     $oValidTokens->get(' '.$sToken)
 406                                 );
 407
 408                                 foreach ($aNewSearches as $oSearch) {
 409                                     if ($oSearch->getRank() < $this->iMaxRank) {
 410                                         $aNewWordsetSearches[] = $oSearch;
 411                                     }
 412                                 }
 413                             }
 414                         }
 415                     }
 416                     // Sort and cut
 417                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 418                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 419                 }
 420                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 421
 422                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 423                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 424
 425                 $aSearchHash = array();
 426                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 427                     $sHash = serialize($aSearch);
 428                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 429                     else $aSearchHash[$sHash] = 1;
 430                 }
 431
 432                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 433             }
 434
 435             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 436             $aGroupedSearches = array();
 437             foreach ($aNewPhraseSearches as $aSearch) {
 438                 $iRank = $aSearch->getRank();
 439                 if ($iRank < $this->iMaxRank) {
 440                     if (!isset($aGroupedSearches[$iRank])) {
 441                         $aGroupedSearches[$iRank] = array();
 442                     }
 443                     $aGroupedSearches[$iRank][] = $aSearch;
 444                 }
 445             }
 446             ksort($aGroupedSearches);
 447
 448             $iSearchCount = 0;
 449             $aSearches = array();
 450             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 451                 $iSearchCount += count($aNewSearches);
 452                 $aSearches = array_merge($aSearches, $aNewSearches);
 453                 if ($iSearchCount > 50) break;
 454             }
 455         }
 456
 457         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 458         $aGroupedSearches = array();
 459         foreach ($aSearches as $oSearch) {
 460             if (!$oSearch->isValidSearch()) {
 461                 continue;
 462             }
 463
 464             $iRank = $oSearch->getRank();
 465             if (!isset($aGroupedSearches[$iRank])) {
 466                 $aGroupedSearches[$iRank] = array();
 467             }
 468             $aGroupedSearches[$iRank][] = $oSearch;
 469         }
 470         ksort($aGroupedSearches);
 471
 472         return $aGroupedSearches;
 473     }
 474
 475     /* Perform the actual query lookup.
 476
 477         Returns an ordered list of results, each with the following fields:
 478             osm_type: type of corresponding OSM object
 479                         N - node
 480                         W - way
 481                         R - relation
 482                         P - postcode (internally computed)
 483             osm_id: id of corresponding OSM object
 484             class: general object class (corresponds to tag key of primary OSM tag)
 485             type: subclass of object (corresponds to tag value of primary OSM tag)
 486             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 487             rank_search: rank in search hierarchy
 488                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 489             rank_address: rank in address hierarchy (determines orer in address)
 490             place_id: internal key (may differ between different instances)
 491             country_code: ISO country code
 492             langaddress: localized full address
 493             placename: localized name of object
 494             ref: content of ref tag (if available)
 495             lon: longitude
 496             lat: latitude
 497             importance: importance of place based on Wikipedia link count
 498             addressimportance: cumulated importance of address elements
 499             extra_place: type of place (for admin boundaries, if there is a place tag)
 500             aBoundingBox: bounding Box
 501             label: short description of the object class/type (English only)
 502             name: full name (currently the same as langaddress)
 503             foundorder: secondary ordering for places with same importance
 504     */
 505
 506
 507     public function lookup()
 508     {
 509         Debug::newFunction('Geocode::lookup');
 510         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 511
 512         Debug::printDebugArray('Geocode', $this);
 513
 514         $oCtx = new SearchContext();
 515
 516         if ($this->aRoutePoints) {
 517             $oCtx->setViewboxFromRoute(
 518                 $this->oDB,
 519                 $this->aRoutePoints,
 520                 $this->aRouteWidth,
 521                 $this->bBoundedSearch
 522             );
 523         } elseif ($this->aViewBox) {
 524             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 525         }
 526         if ($this->aExcludePlaceIDs) {
 527             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 528         }
 529         if ($this->aCountryCodes) {
 530             $oCtx->setCountryList($this->aCountryCodes);
 531         }
 532
 533         Debug::newSection('Query Preprocessing');
 534
 535         $sNormQuery = $this->normTerm($this->sQuery);
 536         Debug::printVar('Normalized query', $sNormQuery);
 537
 538         $sLanguagePrefArraySQL = getArraySQL(
 539             array_map('getDBQuoted', $this->aLangPrefOrder)
 540         );
 541
 542         $sQuery = $this->sQuery;
 543         if (!preg_match('//u', $sQuery)) {
 544             userError('Query string is not UTF-8 encoded.');
 545         }
 546
 547         // Conflicts between US state abreviations and various words for 'the' in different languages
 548         if (isset($this->aLangPrefOrder['name:en'])) {
 549             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 550             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 551             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 552         }
 553
 554         // Do we have anything that looks like a lat/lon pair?
 555         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 556
 557         $aResults = array();
 558         if ($sQuery || $this->aStructuredQuery) {
 559             // Start with a single blank search
 560             $aSearches = array(new SearchDescription($oCtx));
 561
 562             if ($sQuery) {
 563                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 564             }
 565
 566             $sSpecialTerm = '';
 567             if ($sQuery) {
 568                 preg_match_all(
 569                     '/\\[([\\w ]*)\\]/u',
 570                     $sQuery,
 571                     $aSpecialTermsRaw,
 572                     PREG_SET_ORDER
 573                 );
 574                 if (!empty($aSpecialTermsRaw)) {
 575                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 576                 }
 577
 578                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 579                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 580                     if (!$sSpecialTerm) {
 581                         $sSpecialTerm = $aSpecialTerm[1];
 582                     }
 583                 }
 584             }
 585             if (!$sSpecialTerm && $this->aStructuredQuery
 586                 && isset($this->aStructuredQuery['amenity'])) {
 587                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 588                 unset($this->aStructuredQuery['amenity']);
 589             }
 590
 591             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 592                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 593                 $sToken = chksql(
 594                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 595                     'Cannot decode query. Wrong encoding?'
 596                 );
 597                 $sSQL = 'SELECT class, type FROM word ';
 598                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 599                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 600
 601                 Debug::printSQL($sSQL);
 602                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 603                 $aNewSearches = array();
 604                 foreach ($aSearches as $oSearch) {
 605                     foreach ($aSearchWords as $aSearchTerm) {
 606                         $oNewSearch = clone $oSearch;
 607                         $oNewSearch->setPoiSearch(
 608                             Operator::TYPE,
 609                             $aSearchTerm['class'],
 610                             $aSearchTerm['type']
 611                         );
 612                         $aNewSearches[] = $oNewSearch;
 613                     }
 614                 }
 615                 $aSearches = $aNewSearches;
 616             }
 617
 618             // Split query into phrases
 619             // Commas are used to reduce the search space by indicating where phrases split
 620             if ($this->aStructuredQuery) {
 621                 $aInPhrases = $this->aStructuredQuery;
 622                 $bStructuredPhrases = true;
 623             } else {
 624                 $aInPhrases = explode(',', $sQuery);
 625                 $bStructuredPhrases = false;
 626             }
 627
 628             Debug::printDebugArray('Search context', $oCtx);
 629             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 630             Debug::printVar('Final query phrases', $aInPhrases);
 631
 632             // Convert each phrase to standard form
 633             // Create a list of standard words
 634             // Get all 'sets' of words
 635             // Generate a complete list of all
 636             Debug::newSection('Tokenization');
 637             $aTokens = array();
 638             $aPhrases = array();
 639             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 640                 $sPhrase = chksql(
 641                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 642                     'Cannot normalize query string (is it a UTF-8 string?)'
 643                 );
 644                 if (trim($sPhrase)) {
 645                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 646                     $oPhrase->addTokens($aTokens);
 647                     $aPhrases[] = $oPhrase;
 648                 }
 649             }
 650
 651             Debug::printDebugTable('Phrases', $aPhrases);
 652             Debug::printVar('Tokens', $aTokens);
 653
 654             $oValidTokens = new TokenList();
 655
 656             if (!empty($aTokens)) {
 657                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 658                 $sSQL .= ' FROM word ';
 659                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 660
 661                 Debug::printSQL($sSQL);
 662
 663                 $oValidTokens->addTokensFromDB(
 664                     $this->oDB,
 665                     $aTokens,
 666                     $this->aCountryCodes,
 667                     $sNormQuery,
 668                     $this->oNormalizer
 669                 );
 670
 671                 // Try more interpretations for Tokens that could not be matched.
 672                 foreach ($aTokens as $sToken) {
 673                     if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
 674                         if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 675                             // US ZIP+4 codes - merge in the 5-digit ZIP code
 676                             $oValidTokens->addToken(
 677                                 $sToken,
 678                                 new Token\Postcode(null, $aData[1], 'us')
 679                             );
 680                         } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
 681                             // Unknown single word token with a number.
 682                             // Assume it is a house number.
 683                             $oValidTokens->addToken(
 684                                 $sToken,
 685                                 new Token\HouseNumber(null, trim($sToken))
 686                             );
 687                         }
 688                     }
 689                 }
 690
 691                 // Any words that have failed completely?
 692                 // TODO: suggestions
 693
 694                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 695
 696                 Debug::newSection('Search candidates');
 697
 698                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
 699
 700                 if ($this->bReverseInPlan) {
 701                     // Reverse phrase array and also reverse the order of the wordsets in
 702                     // the first and final phrase. Don't bother about phrases in the middle
 703                     // because order in the address doesn't matter.
 704                     $aPhrases = array_reverse($aPhrases);
 705                     $aPhrases[0]->invertWordSets();
 706                     if (count($aPhrases) > 1) {
 707                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 708                     }
 709                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
 710
 711                     foreach ($aGroupedSearches as $aSearches) {
 712                         foreach ($aSearches as $aSearch) {
 713                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 714                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 715                             }
 716                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 717                         }
 718                     }
 719
 720                     $aGroupedSearches = $aReverseGroupedSearches;
 721                     ksort($aGroupedSearches);
 722                 }
 723             } else {
 724                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 725                 $aGroupedSearches = array();
 726                 foreach ($aSearches as $aSearch) {
 727                     if ($aSearch->getRank() < $this->iMaxRank) {
 728                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 729                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 730                     }
 731                 }
 732                 ksort($aGroupedSearches);
 733             }
 734
 735             // Filter out duplicate searches
 736             $aSearchHash = array();
 737             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 738                 foreach ($aSearches as $iSearch => $aSearch) {
 739                     $sHash = serialize($aSearch);
 740                     if (isset($aSearchHash[$sHash])) {
 741                         unset($aGroupedSearches[$iGroup][$iSearch]);
 742                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 743                     } else {
 744                         $aSearchHash[$sHash] = 1;
 745                     }
 746                 }
 747             }
 748
 749             Debug::printGroupedSearch(
 750                 $aGroupedSearches,
 751                 $oValidTokens->debugTokenByWordIdList()
 752             );
 753
 754             // Start the search process
 755             $iGroupLoop = 0;
 756             $iQueryLoop = 0;
 757             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 758                 $iGroupLoop++;
 759                 foreach ($aSearches as $oSearch) {
 760                     $iQueryLoop++;
 761
 762                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 763                     Debug::printGroupedSearch(
 764                         array($iGroupedRank => array($oSearch)),
 765                         $oValidTokens->debugTokenByWordIdList()
 766                     );
 767
 768                     $aResults += $oSearch->query(
 769                         $this->oDB,
 770                         $this->iMinAddressRank,
 771                         $this->iMaxAddressRank,
 772                         $this->iLimit
 773                     );
 774
 775                     if ($iQueryLoop > 20) break;
 776                 }
 777
 778                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 779                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 780                     // reduces the number of place ids, like a filter
 781                     // rank_address is 30 for interpolated housenumbers
 782                     $aFilterSql = array();
 783                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 784                     if ($sPlaceIds) {
 785                         $sSQL = 'SELECT place_id FROM placex ';
 786                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 787                         $sSQL .= '  AND (';
 788                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 789                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 790                             $sSQL .= "     OR (extratags->'place') = 'city'";
 791                         }
 792                         if ($this->aAddressRankList) {
 793                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 794                         }
 795                         $sSQL .= ')';
 796                         $aFilterSql[] = $sSQL;
 797                     }
 798                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 799                     if ($sPlaceIds) {
 800                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 801                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 802                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 803                         if ($this->aAddressRankList) {
 804                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 805                         }
 806                         $sSQL .= ') ';
 807                         $aFilterSql[] = $sSQL;
 808                     }
 809
 810                     $aFilteredIDs = array();
 811                     if ($aFilterSql) {
 812                         $sSQL = join(' UNION ', $aFilterSql);
 813                         Debug::printSQL($sSQL);
 814                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 815                     }
 816
 817                     $tempIDs = array();
 818                     foreach ($aResults as $oResult) {
 819                         if (($this->iMaxAddressRank == 30 &&
 820                              ($oResult->iTable == Result::TABLE_OSMLINE
 821                               || $oResult->iTable == Result::TABLE_AUX
 822                               || $oResult->iTable == Result::TABLE_TIGER))
 823                             || in_array($oResult->iId, $aFilteredIDs)
 824                         ) {
 825                             $tempIDs[$oResult->iId] = $oResult;
 826                         }
 827                     }
 828                     $aResults = $tempIDs;
 829                 }
 830
 831                 if (!empty($aResults)) break;
 832                 if ($iGroupLoop > 4) break;
 833                 if ($iQueryLoop > 30) break;
 834             }
 835         } else {
 836             // Just interpret as a reverse geocode
 837             $oReverse = new ReverseGeocode($this->oDB);
 838             $oReverse->setZoom(18);
 839
 840             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 841
 842             Debug::printVar('Reverse search', $oLookup);
 843
 844             if ($oLookup) {
 845                 $aResults = array($oLookup->iId => $oLookup);
 846             }
 847         }
 848
 849         // No results? Done
 850         if (empty($aResults)) {
 851             if ($this->bFallback) {
 852                 if ($this->fallbackStructuredQuery()) {
 853                     return $this->lookup();
 854                 }
 855             }
 856
 857             return array();
 858         }
 859
 860         if ($this->aAddressRankList) {
 861             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 862         }
 863         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 864         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 865         if ($oCtx->hasNearPoint()) {
 866             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 867         }
 868
 869         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 870
 871         $aClassType = getClassTypesWithImportance();
 872         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 873         foreach ($aRecheckWords as $i => $sWord) {
 874             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 875         }
 876
 877         Debug::printVar('Recheck words', $aRecheckWords);
 878
 879         foreach ($aSearchResults as $iIdx => $aResult) {
 880             // Default
 881             $fDiameter = getResultDiameter($aResult);
 882
 883             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 884             if ($aOutlineResult) {
 885                 $aResult = array_merge($aResult, $aOutlineResult);
 886             }
 887
 888             if ($aResult['extra_place'] == 'city') {
 889                 $aResult['class'] = 'place';
 890                 $aResult['type'] = 'city';
 891                 $aResult['rank_search'] = 16;
 892             }
 893
 894             // Is there an icon set for this type of result?
 895             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
 896                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
 897             ) {
 898                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
 899             }
 900
 901             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
 902                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
 903             ) {
 904                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
 905             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
 906                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
 907             ) {
 908                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
 909             }
 910             // if tag '&addressdetails=1' is set in query
 911             if ($this->bIncludeAddressDetails) {
 912                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
 913                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber);
 914                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
 915                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
 916                 }
 917             }
 918
 919             $aResult['name'] = $aResult['langaddress'];
 920
 921             if ($oCtx->hasNearPoint()) {
 922                 $aResult['importance'] = 0.001;
 923                 $aResult['foundorder'] = $aResult['addressimportance'];
 924             } else {
 925                 $aResult['importance'] = max(0.001, $aResult['importance']);
 926                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 927                     $aResult['lon'],
 928                     $aResult['lat']
 929                 );
 930                 // Adjust importance for the number of exact string matches in the result
 931                 $iCountWords = 0;
 932                 $sAddress = $aResult['langaddress'];
 933                 foreach ($aRecheckWords as $i => $sWord) {
 934                     if (stripos($sAddress, $sWord)!==false) {
 935                         $iCountWords++;
 936                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 937                     }
 938                 }
 939
 940                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 941
 942                 // secondary ordering (for results with same importance (the smaller the better):
 943                 // - approximate importance of address parts
 944                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
 945                 // - number of exact matches from the query
 946                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 947                 // - importance of the class/type
 948                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
 949                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
 950                 ) {
 951                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
 952                 } else {
 953                     $aResult['foundorder'] += 0.01;
 954                 }
 955             }
 956             $aSearchResults[$iIdx] = $aResult;
 957         }
 958         uasort($aSearchResults, 'byImportance');
 959         Debug::printVar('Pre-filter results', $aSearchResults);
 960
 961         $aOSMIDDone = array();
 962         $aClassTypeNameDone = array();
 963         $aToFilter = $aSearchResults;
 964         $aSearchResults = array();
 965
 966         $bFirst = true;
 967         foreach ($aToFilter as $aResult) {
 968             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 969             if ($bFirst) {
 970                 $fLat = $aResult['lat'];
 971                 $fLon = $aResult['lon'];
 972                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 973                 $bFirst = false;
 974             }
 975             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 976                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 977             ) {
 978                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 979                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 980                 $aSearchResults[] = $aResult;
 981             }
 982
 983             // Absolute limit on number of results
 984             if (count($aSearchResults) >= $this->iFinalLimit) break;
 985         }
 986
 987         Debug::printVar('Post-filter results', $aSearchResults);
 988         return $aSearchResults;
 989     } // end lookup()
 990
 991     public function debugInfo()
 992     {
 993         return array(
 994                 'Query' => $this->sQuery,
 995                 'Structured query' => $this->aStructuredQuery,
 996                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 997                 'Include address' => $this->bIncludeAddressDetails,
 998                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 999                 'Try reversed query'=> $this->bReverseInPlan,
1000                 'Limit (for searches)' => $this->iLimit,
1001                 'Limit (for results)'=> $this->iFinalLimit,
1002                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
1003                 'Bounded search' => $this->bBoundedSearch,
1004                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
1005                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
1006                 'Route width' => $this->aRouteWidth,
1007                 'Max rank' => $this->iMaxRank,
1008                 'Min address rank' => $this->iMinAddressRank,
1009                 'Max address rank' => $this->iMaxAddressRank,
1010                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
1011                );
1012     }
1013 } // end class