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