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
  11 class Geocode
  12 {
  13     protected $oDB;
  14
  15     protected $oPlaceLookup;
  16
  17     protected $aLangPrefOrder = array();
  18
  19     protected $bIncludeAddressDetails = false;
  20
  21     protected $aExcludePlaceIDs = array();
  22     protected $bReverseInPlan = true;
  23
  24     protected $iLimit = 20;
  25     protected $iFinalLimit = 10;
  26     protected $iOffset = 0;
  27     protected $bFallback = false;
  28
  29     protected $aCountryCodes = false;
  30
  31     protected $bBoundedSearch = false;
  32     protected $aViewBox = false;
  33     protected $aRoutePoints = false;
  34     protected $aRouteWidth = false;
  35
  36     protected $iMaxRank = 20;
  37     protected $iMinAddressRank = 0;
  38     protected $iMaxAddressRank = 30;
  39     protected $aAddressRankList = array();
  40
  41     protected $sAllowedTypesSQLList = false;
  42
  43     protected $sQuery = false;
  44     protected $aStructuredQuery = false;
  45
  46     protected $oNormalizer = null;
  47
  48
  49     public function __construct(&$oDB)
  50     {
  51         $this->oDB =& $oDB;
  52         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  53         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  54     }
  55
  56     private function normTerm($sTerm)
  57     {
  58         if ($this->oNormalizer === null) {
  59             return $sTerm;
  60         }
  61
  62         return $this->oNormalizer->transliterate($sTerm);
  63     }
  64
  65     public function setReverseInPlan($bReverse)
  66     {
  67         $this->bReverseInPlan = $bReverse;
  68     }
  69
  70     public function setLanguagePreference($aLangPref)
  71     {
  72         $this->aLangPrefOrder = $aLangPref;
  73     }
  74
  75     public function getMoreUrlParams()
  76     {
  77         if ($this->aStructuredQuery) {
  78             $aParams = $this->aStructuredQuery;
  79         } else {
  80             $aParams = array('q' => $this->sQuery);
  81         }
  82
  83         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  84
  85         if ($this->aExcludePlaceIDs) {
  86             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  87         }
  88
  89         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  90         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  91
  92         if ($this->aCountryCodes) {
  93             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  94         }
  95
  96         if ($this->aViewBox) {
  97             $aParams['viewbox'] = join(',', $this->aViewBox);
  98         }
  99
 100         return $aParams;
 101     }
 102
 103     public function setLimit($iLimit = 10)
 104     {
 105         if ($iLimit > 50) $iLimit = 50;
 106         if ($iLimit < 1) $iLimit = 1;
 107
 108         $this->iFinalLimit = $iLimit;
 109         $this->iLimit = $iLimit + min($iLimit, 10);
 110     }
 111
 112     public function setFeatureType($sFeatureType)
 113     {
 114         switch ($sFeatureType) {
 115             case 'country':
 116                 $this->setRankRange(4, 4);
 117                 break;
 118             case 'state':
 119                 $this->setRankRange(8, 8);
 120                 break;
 121             case 'city':
 122                 $this->setRankRange(14, 16);
 123                 break;
 124             case 'settlement':
 125                 $this->setRankRange(8, 20);
 126                 break;
 127         }
 128     }
 129
 130     public function setRankRange($iMin, $iMax)
 131     {
 132         $this->iMinAddressRank = $iMin;
 133         $this->iMaxAddressRank = $iMax;
 134     }
 135
 136     public function setViewbox($aViewbox)
 137     {
 138         $aBox = array_map('floatval', $aViewbox);
 139
 140         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 141         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 142         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 143         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 144
 145         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 146             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 147         ) {
 148             userError("Bad parameter 'viewbox'. Not a box.");
 149         }
 150     }
 151
 152     private function viewboxImportanceFactor($fX, $fY)
 153     {
 154         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 155         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 156
 157         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 158         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 159
 160         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 161             return 1;
 162         }
 163
 164         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 165             return 0.5;
 166         }
 167
 168         return 0.25;
 169     }
 170
 171     public function setQuery($sQueryString)
 172     {
 173         $this->sQuery = $sQueryString;
 174         $this->aStructuredQuery = false;
 175     }
 176
 177     public function getQueryString()
 178     {
 179         return $this->sQuery;
 180     }
 181
 182
 183     public function loadParamArray($oParams, $sForceGeometryType = null)
 184     {
 185         $this->bIncludeAddressDetails
 186          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 187
 188         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 189
 190         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 191         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 192
 193         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 194
 195         // List of excluded Place IDs - used for more acurate pageing
 196         $sExcluded = $oParams->getStringList('exclude_place_ids');
 197         if ($sExcluded) {
 198             foreach ($sExcluded as $iExcludedPlaceID) {
 199                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 200                 if ($iExcludedPlaceID)
 201                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 202             }
 203
 204             if (isset($aExcludePlaceIDs))
 205                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 206         }
 207
 208         // Only certain ranks of feature
 209         $sFeatureType = $oParams->getString('featureType');
 210         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 211         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 212
 213         // Country code list
 214         $sCountries = $oParams->getStringList('countrycodes');
 215         if ($sCountries) {
 216             foreach ($sCountries as $sCountryCode) {
 217                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 218                     $aCountries[] = strtolower($sCountryCode);
 219                 }
 220             }
 221             if (isset($aCountries))
 222                 $this->aCountryCodes = $aCountries;
 223         }
 224
 225         $aViewbox = $oParams->getStringList('viewboxlbrt');
 226         if ($aViewbox) {
 227             if (count($aViewbox) != 4) {
 228                 userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
 229             }
 230             $this->setViewbox($aViewbox);
 231         } else {
 232             $aViewbox = $oParams->getStringList('viewbox');
 233             if ($aViewbox) {
 234                 if (count($aViewbox) != 4) {
 235                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
 236                 }
 237                 $this->setViewBox($aViewbox);
 238             } else {
 239                 $aRoute = $oParams->getStringList('route');
 240                 $fRouteWidth = $oParams->getFloat('routewidth');
 241                 if ($aRoute && $fRouteWidth) {
 242                     $this->aRoutePoints = $aRoute;
 243                     $this->aRouteWidth = $fRouteWidth;
 244                 }
 245             }
 246         }
 247
 248         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 249         $this->oPlaceLookup->setIncludeAddressDetails(false);
 250         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 251     }
 252
 253     public function setQueryFromParams($oParams)
 254     {
 255         // Search query
 256         $sQuery = $oParams->getString('q');
 257         if (!$sQuery) {
 258             $this->setStructuredQuery(
 259                 $oParams->getString('amenity'),
 260                 $oParams->getString('street'),
 261                 $oParams->getString('city'),
 262                 $oParams->getString('county'),
 263                 $oParams->getString('state'),
 264                 $oParams->getString('country'),
 265                 $oParams->getString('postalcode')
 266             );
 267             $this->setReverseInPlan(false);
 268         } else {
 269             $this->setQuery($sQuery);
 270         }
 271     }
 272
 273     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 274     {
 275         $sValue = trim($sValue);
 276         if (!$sValue) return false;
 277         $this->aStructuredQuery[$sKey] = $sValue;
 278         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 279             $this->iMinAddressRank = $iNewMinAddressRank;
 280             $this->iMaxAddressRank = $iNewMaxAddressRank;
 281         }
 282         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 283         return true;
 284     }
 285
 286     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 287     {
 288         $this->sQuery = false;
 289
 290         // Reset
 291         $this->iMinAddressRank = 0;
 292         $this->iMaxAddressRank = 30;
 293         $this->aAddressRankList = array();
 294
 295         $this->aStructuredQuery = array();
 296         $this->sAllowedTypesSQLList = false;
 297
 298         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 299         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 300         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 301         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 302         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 303         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 304         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 305
 306         if (sizeof($this->aStructuredQuery) > 0) {
 307             $this->sQuery = join(', ', $this->aStructuredQuery);
 308             if ($this->iMaxAddressRank < 30) {
 309                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 310             }
 311         }
 312     }
 313
 314     public function fallbackStructuredQuery()
 315     {
 316         if (!$this->aStructuredQuery) return false;
 317
 318         $aParams = $this->aStructuredQuery;
 319
 320         if (sizeof($aParams) == 1) return false;
 321
 322         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 323
 324         foreach ($aOrderToFallback as $sType) {
 325             if (isset($aParams[$sType])) {
 326                 unset($aParams[$sType]);
 327                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 328                 return true;
 329             }
 330         }
 331
 332         return false;
 333     }
 334
 335     public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
 336     {
 337         /*
 338              Calculate all searches using aValidTokens i.e.
 339              'Wodsworth Road, Sheffield' =>
 340
 341              Phrase Wordset
 342              0      0       (wodsworth road)
 343              0      1       (wodsworth)(road)
 344              1      0       (sheffield)
 345
 346              Score how good the search is so they can be ordered
 347          */
 348         foreach ($aPhrases as $iPhrase => $oPhrase) {
 349             $aNewPhraseSearches = array();
 350             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 351
 352             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 353                 // Too many permutations - too expensive
 354                 if ($iWordSet > 120) break;
 355
 356                 $aWordsetSearches = $aSearches;
 357
 358                 // Add all words from this wordset
 359                 foreach ($aWordset as $iToken => $sToken) {
 360                     //echo "<br><b>$sToken</b>";
 361                     $aNewWordsetSearches = array();
 362
 363                     foreach ($aWordsetSearches as $oCurrentSearch) {
 364                         //echo "<i>";
 365                         //var_dump($oCurrentSearch);
 366                         //echo "</i>";
 367
 368                         // If the token is valid
 369                         if (isset($aValidTokens[' '.$sToken])) {
 370                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 371                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 372                                     $aSearchTerm,
 373                                     isset($aValidTokens[$sToken])
 374                                       && strpos($sToken, ' ') === false,
 375                                     $sPhraseType,
 376                                     $iToken == 0 && $iPhrase == 0,
 377                                     $iPhrase == 0,
 378                                     $iToken + 1 == sizeof($aWordset)
 379                                       && $iPhrase + 1 == sizeof($aPhrases)
 380                                 );
 381
 382                                 foreach ($aNewSearches as $oSearch) {
 383                                     if ($oSearch->getRank() < $this->iMaxRank) {
 384                                         $aNewWordsetSearches[] = $oSearch;
 385                                     }
 386                                 }
 387                             }
 388                         }
 389                         // Look for partial matches.
 390                         // Note that there is no point in adding country terms here
 391                         // because country is omitted in the address.
 392                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 393                             // Allow searching for a word - but at extra cost
 394                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 395                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 396                                     $aSearchTerm,
 397                                     $bIsStructured,
 398                                     $iPhrase,
 399                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 400                                 );
 401
 402                                 foreach ($aNewSearches as $oSearch) {
 403                                     if ($oSearch->getRank() < $this->iMaxRank) {
 404                                         $aNewWordsetSearches[] = $oSearch;
 405                                     }
 406                                 }
 407                             }
 408                         }
 409                     }
 410                     // Sort and cut
 411                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 412                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 413                 }
 414                 //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
 415
 416                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 417                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 418
 419                 $aSearchHash = array();
 420                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 421                     $sHash = serialize($aSearch);
 422                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 423                     else $aSearchHash[$sHash] = 1;
 424                 }
 425
 426                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 427             }
 428
 429             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 430             $aGroupedSearches = array();
 431             foreach ($aNewPhraseSearches as $aSearch) {
 432                 $iRank = $aSearch->getRank();
 433                 if ($iRank < $this->iMaxRank) {
 434                     if (!isset($aGroupedSearches[$iRank])) {
 435                         $aGroupedSearches[$iRank] = array();
 436                     }
 437                     $aGroupedSearches[$iRank][] = $aSearch;
 438                 }
 439             }
 440             ksort($aGroupedSearches);
 441
 442             $iSearchCount = 0;
 443             $aSearches = array();
 444             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 445                 $iSearchCount += sizeof($aNewSearches);
 446                 $aSearches = array_merge($aSearches, $aNewSearches);
 447                 if ($iSearchCount > 50) break;
 448             }
 449
 450             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 451         }
 452
 453         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 454         $aGroupedSearches = array();
 455         foreach ($aSearches as $oSearch) {
 456             if (!$oSearch->isValidSearch()) {
 457                 continue;
 458             }
 459
 460             $iRank = $oSearch->getRank();
 461             if (!isset($aGroupedSearches[$iRank])) {
 462                 $aGroupedSearches[$iRank] = array();
 463             }
 464             $aGroupedSearches[$iRank][] = $oSearch;
 465         }
 466         ksort($aGroupedSearches);
 467
 468         return $aGroupedSearches;
 469     }
 470
 471     /* Perform the actual query lookup.
 472
 473         Returns an ordered list of results, each with the following fields:
 474             osm_type: type of corresponding OSM object
 475                         N - node
 476                         W - way
 477                         R - relation
 478                         P - postcode (internally computed)
 479             osm_id: id of corresponding OSM object
 480             class: general object class (corresponds to tag key of primary OSM tag)
 481             type: subclass of object (corresponds to tag value of primary OSM tag)
 482             admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
 483             rank_search: rank in search hierarchy
 484                         (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 485             rank_address: rank in address hierarchy (determines orer in address)
 486             place_id: internal key (may differ between different instances)
 487             country_code: ISO country code
 488             langaddress: localized full address
 489             placename: localized name of object
 490             ref: content of ref tag (if available)
 491             lon: longitude
 492             lat: latitude
 493             importance: importance of place based on Wikipedia link count
 494             addressimportance: cumulated importance of address elements
 495             extra_place: type of place (for admin boundaries, if there is a place tag)
 496             aBoundingBox: bounding Box
 497             label: short description of the object class/type (English only)
 498             name: full name (currently the same as langaddress)
 499             foundorder: secondary ordering for places with same importance
 500     */
 501
 502
 503     public function lookup()
 504     {
 505         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 506
 507         $oCtx = new SearchContext();
 508
 509         if ($this->aRoutePoints) {
 510             $oCtx->setViewboxFromRoute(
 511                 $this->oDB,
 512                 $this->aRoutePoints,
 513                 $this->aRouteWidth,
 514                 $this->bBoundedSearch
 515             );
 516         } elseif ($this->aViewBox) {
 517             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 518         }
 519         if ($this->aExcludePlaceIDs) {
 520             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 521         }
 522         if ($this->aCountryCodes) {
 523             $oCtx->setCountryList($this->aCountryCodes);
 524         }
 525
 526         $sNormQuery = $this->normTerm($this->sQuery);
 527         $sLanguagePrefArraySQL = getArraySQL(
 528             array_map('getDBQuoted', $this->aLangPrefOrder)
 529         );
 530
 531         $sQuery = $this->sQuery;
 532         if (!preg_match('//u', $sQuery)) {
 533             userError('Query string is not UTF-8 encoded.');
 534         }
 535
 536         // Conflicts between US state abreviations and various words for 'the' in different languages
 537         if (isset($this->aLangPrefOrder['name:en'])) {
 538             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
 539             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
 540             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
 541         }
 542
 543         // Do we have anything that looks like a lat/lon pair?
 544         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 545
 546         $aResults = array();
 547         if ($sQuery || $this->aStructuredQuery) {
 548             // Start with a single blank search
 549             $aSearches = array(new SearchDescription($oCtx));
 550
 551             if ($sQuery) {
 552                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 553             }
 554
 555             $sSpecialTerm = '';
 556             if ($sQuery) {
 557                 preg_match_all(
 558                     '/\\[([\\w ]*)\\]/u',
 559                     $sQuery,
 560                     $aSpecialTermsRaw,
 561                     PREG_SET_ORDER
 562                 );
 563                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 564                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 565                     if (!$sSpecialTerm) {
 566                         $sSpecialTerm = $aSpecialTerm[1];
 567                     }
 568                 }
 569             }
 570             if (!$sSpecialTerm && $this->aStructuredQuery
 571                 && isset($this->aStructuredQuery['amenity'])) {
 572                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 573                 unset($this->aStructuredQuery['amenity']);
 574             }
 575
 576             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 577                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 578                 $sToken = chksql(
 579                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 580                     'Cannot decode query. Wrong encoding?'
 581                 );
 582                 $sSQL = 'SELECT class, type FROM word ';
 583                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 584                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 585                 if (CONST_Debug) var_Dump($sSQL);
 586                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 587                 $aNewSearches = array();
 588                 foreach ($aSearches as $oSearch) {
 589                     foreach ($aSearchWords as $aSearchTerm) {
 590                         $oNewSearch = clone $oSearch;
 591                         $oNewSearch->setPoiSearch(
 592                             Operator::TYPE,
 593                             $aSearchTerm['class'],
 594                             $aSearchTerm['type']
 595                         );
 596                         $aNewSearches[] = $oNewSearch;
 597                     }
 598                 }
 599                 $aSearches = $aNewSearches;
 600             }
 601
 602             // Split query into phrases
 603             // Commas are used to reduce the search space by indicating where phrases split
 604             if ($this->aStructuredQuery) {
 605                 $aInPhrases = $this->aStructuredQuery;
 606                 $bStructuredPhrases = true;
 607             } else {
 608                 $aInPhrases = explode(',', $sQuery);
 609                 $bStructuredPhrases = false;
 610             }
 611
 612             // Convert each phrase to standard form
 613             // Create a list of standard words
 614             // Get all 'sets' of words
 615             // Generate a complete list of all
 616             $aTokens = array();
 617             $aPhrases = array();
 618             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 619                 $sPhrase = chksql(
 620                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 621                     'Cannot normalize query string (is it a UTF-8 string?)'
 622                 );
 623                 if (trim($sPhrase)) {
 624                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 625                     $oPhrase->addTokens($aTokens);
 626                     $aPhrases[] = $oPhrase;
 627                 }
 628             }
 629
 630             if (sizeof($aTokens)) {
 631                 // Check which tokens we have, get the ID numbers
 632                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 633                 $sSQL .= ' FROM word ';
 634                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 635
 636                 if (CONST_Debug) var_Dump($sSQL);
 637
 638                 $aValidTokens = array();
 639                 $aDatabaseWords = chksql(
 640                     $this->oDB->getAll($sSQL),
 641                     'Could not get word tokens.'
 642                 );
 643                 $aWordFrequencyScores = array();
 644                 foreach ($aDatabaseWords as $aToken) {
 645                     // Filter country tokens that do not match restricted countries.
 646                     if ($this->aCountryCodes
 647                         && $aToken['country_code']
 648                         && !in_array($aToken['country_code'], $this->aCountryCodes)
 649                     ) {
 650                         continue;
 651                     }
 652
 653                     // Special terms need to appear in their normalized form.
 654                     if ($aToken['word'] && $aToken['class']) {
 655                         $sNormWord = $this->normTerm($aToken['word']);
 656                         if (strpos($sNormQuery, $sNormWord) === false) {
 657                             continue;
 658                         }
 659                     }
 660
 661                     if (isset($aValidTokens[$aToken['word_token']])) {
 662                         $aValidTokens[$aToken['word_token']][] = $aToken;
 663                     } else {
 664                         $aValidTokens[$aToken['word_token']] = array($aToken);
 665                     }
 666                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
 667                 }
 668                 if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 669
 670                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
 671                 foreach ($aTokens as $sToken) {
 672                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 673                         if (isset($aValidTokens[$aData[1]])) {
 674                             foreach ($aValidTokens[$aData[1]] as $aToken) {
 675                                 if (!$aToken['class']) {
 676                                     if (isset($aValidTokens[$sToken])) {
 677                                         $aValidTokens[$sToken][] = $aToken;
 678                                     } else {
 679                                         $aValidTokens[$sToken] = array($aToken);
 680                                     }
 681                                 }
 682                             }
 683                         }
 684                     }
 685                 }
 686
 687                 foreach ($aTokens as $sToken) {
 688                     // Unknown single word token with a number - assume it is a house number
 689                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
 690                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
 691                     }
 692                 }
 693
 694                 // Any words that have failed completely?
 695                 // TODO: suggestions
 696
 697                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 698
 699                 if ($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                     $aPhrases[0]->invertWordSets();
 705                     if (sizeof($aPhrases) > 1) {
 706                         $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
 707                     }
 708                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 709
 710                     foreach ($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][$iSearch]);
 741                         if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
 742                     } else {
 743                         $aSearchHash[$sHash] = 1;
 744                     }
 745                 }
 746             }
 747
 748             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 749
 750             // Start the search process
 751             $iGroupLoop = 0;
 752             $iQueryLoop = 0;
 753             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 754                 $iGroupLoop++;
 755                 foreach ($aSearches as $oSearch) {
 756                     $iQueryLoop++;
 757
 758                     if (CONST_Debug) {
 759                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
 760                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
 761                     }
 762
 763                     $aResults += $oSearch->query(
 764                         $this->oDB,
 765                         $aWordFrequencyScores,
 766                         $this->iMinAddressRank,
 767                         $this->iMaxAddressRank,
 768                         $this->iLimit
 769                     );
 770
 771                     if ($iQueryLoop > 20) break;
 772                 }
 773
 774                 if (sizeof($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 775                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 776                     // reduces the number of place ids, like a filter
 777                     // rank_address is 30 for interpolated housenumbers
 778                     $aFilterSql = array();
 779                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 780                     if ($sPlaceIds) {
 781                         $sSQL = 'SELECT place_id FROM placex ';
 782                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 783                         $sSQL .= '  AND (';
 784                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 785                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 786                             $sSQL .= "     OR (extratags->'place') = 'city'";
 787                         }
 788                         if ($this->aAddressRankList) {
 789                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 790                         }
 791                         $sSQL .= ')';
 792                         $aFilterSql[] = $sSQL;
 793                     }
 794                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 795                     if ($sPlaceIds) {
 796                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 797                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 798                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 799                         if ($this->aAddressRankList) {
 800                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 801                         }
 802                         $sSQL .= ') ';
 803                         $aFilterSql[] = $sSQL;
 804                     }
 805
 806                     $aFilteredIDs = array();
 807                     if ($aFilterSql) {
 808                         $sSQL = join(' UNION ', $aFilterSql);
 809                         if (CONST_Debug) var_dump($sSQL);
 810                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 811                     }
 812
 813                     $tempIDs = array();
 814                     foreach ($aResults as $oResult) {
 815                         if (($this->iMaxAddressRank == 30 &&
 816                              ($oResult->iTable == Result::TABLE_OSMLINE
 817                               || $oResult->iTable == Result::TABLE_AUX
 818                               || $oResult->iTable == Result::TABLE_TIGER))
 819                             || in_array($oResult->iId, $aFilteredIDs)
 820                         ) {
 821                             $tempIDs[$oResult->iId] = $oResult;
 822                         }
 823                     }
 824                     $aResults = $tempIDs;
 825                 }
 826
 827                 if (sizeof($aResults)) break;
 828                 if ($iGroupLoop > 4) break;
 829                 if ($iQueryLoop > 30) break;
 830             }
 831         } else {
 832             // Just interpret as a reverse geocode
 833             $oReverse = new ReverseGeocode($this->oDB);
 834             $oReverse->setZoom(18);
 835
 836             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 837
 838             if (CONST_Debug) var_dump('Reverse search', $aLookup);
 839
 840             if ($oLookup) {
 841                 $aResults = array($oLookup->iId => $oLookup);
 842             }
 843         }
 844
 845         // No results? Done
 846         if (!sizeof($aResults)) {
 847             if ($this->bFallback) {
 848                 if ($this->fallbackStructuredQuery()) {
 849                     return $this->lookup();
 850                 }
 851             }
 852
 853             return array();
 854         }
 855
 856         if ($this->aAddressRankList) {
 857             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 858         }
 859         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 860         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 861         if ($oCtx->hasNearPoint()) {
 862             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 863         }
 864
 865         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 866
 867         $aClassType = getClassTypesWithImportance();
 868         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 869         foreach ($aRecheckWords as $i => $sWord) {
 870             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 871         }
 872
 873         if (CONST_Debug) {
 874             echo '<i>Recheck words:<\i>';
 875             var_dump($aRecheckWords);
 876         }
 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                 // Adjust importance for the number of exact string matches in the result
 925                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 926                     $aResult['lon'],
 927                     $aResult['lat']
 928                 );
 929                 $aResult['importance'] = max(0.001, $aResult['importance']);
 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                     $aResult['foundorder'] += 0.01;
 953                 }
 954             }
 955             if (CONST_Debug) var_dump($aResult);
 956             $aSearchResults[$iIdx] = $aResult;
 957         }
 958         uasort($aSearchResults, 'byImportance');
 959
 960         $aOSMIDDone = array();
 961         $aClassTypeNameDone = array();
 962         $aToFilter = $aSearchResults;
 963         $aSearchResults = array();
 964
 965         if (CONST_Debug) var_dump($aToFilter);
 966
 967         $bFirst = true;
 968         foreach ($aToFilter as $aResult) {
 969             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 970             if ($bFirst) {
 971                 $fLat = $aResult['lat'];
 972                 $fLon = $aResult['lon'];
 973                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 974                 $bFirst = false;
 975             }
 976             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 977                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 978             ) {
 979                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 980                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 981                 $aSearchResults[] = $aResult;
 982             }
 983
 984             // Absolute limit on number of results
 985             if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
 986         }
 987
 988         if (CONST_Debug) var_dump($aSearchResults);
 989         return $aSearchResults;
 990     } // end lookup()
 991 } // end class