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 $aExcludePlaceIDs = array();
  21     protected $bReverseInPlan = false;
  22
  23     protected $iLimit = 20;
  24     protected $iFinalLimit = 10;
  25     protected $iOffset = 0;
  26     protected $bFallback = false;
  27
  28     protected $aCountryCodes = false;
  29
  30     protected $bBoundedSearch = false;
  31     protected $aViewBox = false;
  32     protected $aRoutePoints = false;
  33     protected $aRouteWidth = false;
  34
  35     protected $iMaxRank = 20;
  36     protected $iMinAddressRank = 0;
  37     protected $iMaxAddressRank = 30;
  38     protected $aAddressRankList = array();
  39
  40     protected $sAllowedTypesSQLList = false;
  41
  42     protected $sQuery = false;
  43     protected $aStructuredQuery = false;
  44
  45     protected $oNormalizer = null;
  46
  47
  48     public function __construct(&$oDB)
  49     {
  50         $this->oDB =& $oDB;
  51         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  52         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  53     }
  54
  55     private function normTerm($sTerm)
  56     {
  57         if ($this->oNormalizer === null) {
  58             return $sTerm;
  59         }
  60
  61         return $this->oNormalizer->transliterate($sTerm);
  62     }
  63
  64     public function setReverseInPlan($bReverse)
  65     {
  66         $this->bReverseInPlan = $bReverse;
  67     }
  68
  69     public function setLanguagePreference($aLangPref)
  70     {
  71         $this->aLangPrefOrder = $aLangPref;
  72     }
  73
  74     public function getMoreUrlParams()
  75     {
  76         if ($this->aStructuredQuery) {
  77             $aParams = $this->aStructuredQuery;
  78         } else {
  79             $aParams = array('q' => $this->sQuery);
  80         }
  81
  82         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  83
  84         if ($this->aExcludePlaceIDs) {
  85             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  86         }
  87
  88         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  89
  90         if ($this->aCountryCodes) {
  91             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  92         }
  93
  94         if ($this->aViewBox) {
  95             $aParams['viewbox'] = join(',', $this->aViewBox);
  96         }
  97
  98         return $aParams;
  99     }
 100
 101     public function setLimit($iLimit = 10)
 102     {
 103         if ($iLimit > 50) $iLimit = 50;
 104         if ($iLimit < 1) $iLimit = 1;
 105
 106         $this->iFinalLimit = $iLimit;
 107         $this->iLimit = $iLimit + min($iLimit, 10);
 108     }
 109
 110     public function setFeatureType($sFeatureType)
 111     {
 112         switch ($sFeatureType) {
 113             case 'country':
 114                 $this->setRankRange(4, 4);
 115                 break;
 116             case 'state':
 117                 $this->setRankRange(8, 8);
 118                 break;
 119             case 'city':
 120                 $this->setRankRange(14, 16);
 121                 break;
 122             case 'settlement':
 123                 $this->setRankRange(8, 20);
 124                 break;
 125         }
 126     }
 127
 128     public function setRankRange($iMin, $iMax)
 129     {
 130         $this->iMinAddressRank = $iMin;
 131         $this->iMaxAddressRank = $iMax;
 132     }
 133
 134     public function setViewbox($aViewbox)
 135     {
 136         $aBox = array_map('floatval', $aViewbox);
 137
 138         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 139         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 140         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 141         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 142
 143         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 144             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 145         ) {
 146             userError("Bad parameter 'viewbox'. Not a box.");
 147         }
 148     }
 149
 150     private function viewboxImportanceFactor($fX, $fY)
 151     {
 152         if (!$this->aViewBox) {
 153             return 1;
 154         }
 155
 156         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 157         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 158
 159         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 160         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 161
 162         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 163             return 1;
 164         }
 165
 166         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 167             return 0.5;
 168         }
 169
 170         return 0.25;
 171     }
 172
 173     public function setQuery($sQueryString)
 174     {
 175         $this->sQuery = $sQueryString;
 176         $this->aStructuredQuery = false;
 177     }
 178
 179     public function getQueryString()
 180     {
 181         return $this->sQuery;
 182     }
 183
 184
 185     public function loadParamArray($oParams, $sForceGeometryType = null)
 186     {
 187         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 188
 189         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 190         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 191
 192         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 193
 194         // List of excluded Place IDs - used for more acurate pageing
 195         $sExcluded = $oParams->getStringList('exclude_place_ids');
 196         if ($sExcluded) {
 197             foreach ($sExcluded as $iExcludedPlaceID) {
 198                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 199                 if ($iExcludedPlaceID)
 200                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 201             }
 202
 203             if (isset($aExcludePlaceIDs))
 204                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 205         }
 206
 207         // Only certain ranks of feature
 208         $sFeatureType = $oParams->getString('featureType');
 209         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 210         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 211
 212         // Country code list
 213         $sCountries = $oParams->getStringList('countrycodes');
 214         if ($sCountries) {
 215             foreach ($sCountries as $sCountryCode) {
 216                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 217                     $aCountries[] = strtolower($sCountryCode);
 218                 }
 219             }
 220             if (isset($aCountries))
 221                 $this->aCountryCodes = $aCountries;
 222         }
 223
 224         $aViewbox = $oParams->getStringList('viewboxlbrt');
 225         if ($aViewbox) {
 226             if (count($aViewbox) != 4) {
 227                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 228             }
 229             $this->setViewbox($aViewbox);
 230         } else {
 231             $aViewbox = $oParams->getStringList('viewbox');
 232             if ($aViewbox) {
 233                 if (count($aViewbox) != 4) {
 234                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 235                 }
 236                 $this->setViewBox($aViewbox);
 237             } else {
 238                 $aRoute = $oParams->getStringList('route');
 239                 $fRouteWidth = $oParams->getFloat('routewidth');
 240                 if ($aRoute && $fRouteWidth) {
 241                     $this->aRoutePoints = $aRoute;
 242                     $this->aRouteWidth = $fRouteWidth;
 243                 }
 244             }
 245         }
 246
 247         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 248         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 249         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 250     }
 251
 252     public function setQueryFromParams($oParams)
 253     {
 254         // Search query
 255         $sQuery = $oParams->getString('q');
 256         if (!$sQuery) {
 257             $this->setStructuredQuery(
 258                 $oParams->getString('amenity'),
 259                 $oParams->getString('street'),
 260                 $oParams->getString('city'),
 261                 $oParams->getString('county'),
 262                 $oParams->getString('state'),
 263                 $oParams->getString('country'),
 264                 $oParams->getString('postalcode')
 265             );
 266             $this->setReverseInPlan(false);
 267         } else {
 268             $this->setQuery($sQuery);
 269         }
 270     }
 271
 272     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 273     {
 274         $sValue = trim($sValue);
 275         if (!$sValue) return false;
 276         $this->aStructuredQuery[$sKey] = $sValue;
 277         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 278             $this->iMinAddressRank = $iNewMinAddressRank;
 279             $this->iMaxAddressRank = $iNewMaxAddressRank;
 280         }
 281         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 282         return true;
 283     }
 284
 285     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 286     {
 287         $this->sQuery = false;
 288
 289         // Reset
 290         $this->iMinAddressRank = 0;
 291         $this->iMaxAddressRank = 30;
 292         $this->aAddressRankList = array();
 293
 294         $this->aStructuredQuery = array();
 295         $this->sAllowedTypesSQLList = false;
 296
 297         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 298         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 299         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 300         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 301         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 302         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 303         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 304
 305         if (!empty($this->aStructuredQuery)) {
 306             $this->sQuery = join(', ', $this->aStructuredQuery);
 307             if ($this->iMaxAddressRank < 30) {
 308                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 309             }
 310         }
 311     }
 312
 313     public function fallbackStructuredQuery()
 314     {
 315         if (!$this->aStructuredQuery) return false;
 316
 317         $aParams = $this->aStructuredQuery;
 318
 319         if (count($aParams) == 1) return false;
 320
 321         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 322
 323         foreach ($aOrderToFallback as $sType) {
 324             if (isset($aParams[$sType])) {
 325                 unset($aParams[$sType]);
 326                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 327                 return true;
 328             }
 329         }
 330
 331         return false;
 332     }
 333
 334     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
 335     {
 336         /*
 337              Calculate all searches using oValidTokens i.e.
 338              'Wodsworth Road, Sheffield' =>
 339
 340              Phrase Wordset
 341              0      0       (wodsworth road)
 342              0      1       (wodsworth)(road)
 343              1      0       (sheffield)
 344
 345              Score how good the search is so they can be ordered
 346          */
 347         foreach ($aPhrases as $iPhrase => $oPhrase) {
 348             $aNewPhraseSearches = array();
 349             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 350
 351             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 352                 // Too many permutations - too expensive
 353                 if ($iWordSet > 120) break;
 354
 355                 $aWordsetSearches = $aSearches;
 356
 357                 // Add all words from this wordset
 358                 foreach ($aWordset as $iToken => $sToken) {
 359                     //echo "<br><b>$sToken</b>";
 360                     $aNewWordsetSearches = array();
 361
 362                     foreach ($aWordsetSearches as $oCurrentSearch) {
 363                         //echo "<i>";
 364                         //var_dump($oCurrentSearch);
 365                         //echo "</i>";
 366
 367                         // Tokens with full name matches.
 368                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 369                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 370                                 $oSearchTerm,
 371                                 $oValidTokens->contains($sToken)
 372                                   && strpos($sToken, ' ') === false,
 373                                 $sPhraseType,
 374                                 $iToken == 0 && $iPhrase == 0,
 375                                 $iPhrase == 0,
 376                                 $iToken + 1 == count($aWordset)
 377                                   && $iPhrase + 1 == count($aPhrases)
 378                             );
 379
 380                             foreach ($aNewSearches as $oSearch) {
 381                                 if ($oSearch->getRank() < $this->iMaxRank) {
 382                                     $aNewWordsetSearches[] = $oSearch;
 383                                 }
 384                             }
 385                         }
 386                         // Look for partial matches.
 387                         // Note that there is no point in adding country terms here
 388                         // because country is omitted in the address.
 389                         if ($sPhraseType != 'country') {
 390                             // Allow searching for a word - but at extra cost
 391                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 392                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 393                                     $sToken,
 394                                     $oSearchTerm,
 395                                     $bIsStructured,
 396                                     $iPhrase,
 397                                     $oValidTokens->get(' '.$sToken)
 398                                 );
 399
 400                                 foreach ($aNewSearches as $oSearch) {
 401                                     if ($oSearch->getRank() < $this->iMaxRank) {
 402                                         $aNewWordsetSearches[] = $oSearch;
 403                                     }
 404                                 }
 405                             }
 406                         }
 407                     }
 408                     // Sort and cut
 409                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 410                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 411                 }
 412                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 413
 414                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 415                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 416
 417                 $aSearchHash = array();
 418                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 419                     $sHash = serialize($aSearch);
 420                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 421                     else $aSearchHash[$sHash] = 1;
 422                 }
 423
 424                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 425             }
 426
 427             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 428             $aGroupedSearches = array();
 429             foreach ($aNewPhraseSearches as $aSearch) {
 430                 $iRank = $aSearch->getRank();
 431                 if ($iRank < $this->iMaxRank) {
 432                     if (!isset($aGroupedSearches[$iRank])) {
 433                         $aGroupedSearches[$iRank] = array();
 434                     }
 435                     $aGroupedSearches[$iRank][] = $aSearch;
 436                 }
 437             }
 438             ksort($aGroupedSearches);
 439
 440             $iSearchCount = 0;
 441             $aSearches = array();
 442             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 443                 $iSearchCount += count($aNewSearches);
 444                 $aSearches = array_merge($aSearches, $aNewSearches);
 445                 if ($iSearchCount > 50) break;
 446             }
 447         }
 448
 449         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 450         $aGroupedSearches = array();
 451         foreach ($aSearches as $oSearch) {
 452             if (!$oSearch->isValidSearch()) {
 453                 continue;
 454             }
 455
 456             $iRank = $oSearch->getRank();
 457             if (!isset($aGroupedSearches[$iRank])) {
 458                 $aGroupedSearches[$iRank] = array();
 459             }
 460             $aGroupedSearches[$iRank][] = $oSearch;
 461         }
 462         ksort($aGroupedSearches);
 463
 464         return $aGroupedSearches;
 465     }
 466
 467     /* Perform the actual query lookup.
 468
 469         Returns an ordered list of results, each with the following fields:
 470             osm_type: type of corresponding OSM object
 471                         N - node
 472                         W - way
 473                         R - relation
 474                         P - postcode (internally computed)
 475             osm_id: id of corresponding OSM object
 476             class: general object class (corresponds to tag key of primary OSM tag)
 477             type: subclass of object (corresponds to tag value of primary OSM tag)
 478             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 479             rank_search: rank in search hierarchy
 480                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 481             rank_address: rank in address hierarchy (determines orer in address)
 482             place_id: internal key (may differ between different instances)
 483             country_code: ISO country code
 484             langaddress: localized full address
 485             placename: localized name of object
 486             ref: content of ref tag (if available)
 487             lon: longitude
 488             lat: latitude
 489             importance: importance of place based on Wikipedia link count
 490             addressimportance: cumulated importance of address elements
 491             extra_place: type of place (for admin boundaries, if there is a place tag)
 492             aBoundingBox: bounding Box
 493             label: short description of the object class/type (English only)
 494             name: full name (currently the same as langaddress)
 495             foundorder: secondary ordering for places with same importance
 496     */
 497
 498
 499     public function lookup()
 500     {
 501         Debug::newFunction('Geocode::lookup');
 502         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 503
 504         Debug::printDebugArray('Geocode', $this);
 505
 506         $oCtx = new SearchContext();
 507
 508         if ($this->aRoutePoints) {
 509             $oCtx->setViewboxFromRoute(
 510                 $this->oDB,
 511                 $this->aRoutePoints,
 512                 $this->aRouteWidth,
 513                 $this->bBoundedSearch
 514             );
 515         } elseif ($this->aViewBox) {
 516             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 517         }
 518         if ($this->aExcludePlaceIDs) {
 519             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 520         }
 521         if ($this->aCountryCodes) {
 522             $oCtx->setCountryList($this->aCountryCodes);
 523         }
 524
 525         Debug::newSection('Query Preprocessing');
 526
 527         $sNormQuery = $this->normTerm($this->sQuery);
 528         Debug::printVar('Normalized query', $sNormQuery);
 529
 530         $sLanguagePrefArraySQL = getArraySQL(
 531             array_map('getDBQuoted', $this->aLangPrefOrder)
 532         );
 533
 534         $sQuery = $this->sQuery;
 535         if (!preg_match('//u', $sQuery)) {
 536             userError('Query string is not UTF-8 encoded.');
 537         }
 538
 539         // Conflicts between US state abreviations and various words for 'the' in different languages
 540         if (isset($this->aLangPrefOrder['name:en'])) {
 541             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 542             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 543             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 544         }
 545
 546         // Do we have anything that looks like a lat/lon pair?
 547         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 548
 549         if ($sQuery || $this->aStructuredQuery) {
 550             // Start with a single blank search
 551             $aSearches = array(new SearchDescription($oCtx));
 552
 553             if ($sQuery) {
 554                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 555             }
 556
 557             $sSpecialTerm = '';
 558             if ($sQuery) {
 559                 preg_match_all(
 560                     '/\\[([\\w ]*)\\]/u',
 561                     $sQuery,
 562                     $aSpecialTermsRaw,
 563                     PREG_SET_ORDER
 564                 );
 565                 if (!empty($aSpecialTermsRaw)) {
 566                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 567                 }
 568
 569                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 570                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 571                     if (!$sSpecialTerm) {
 572                         $sSpecialTerm = $aSpecialTerm[1];
 573                     }
 574                 }
 575             }
 576             if (!$sSpecialTerm && $this->aStructuredQuery
 577                 && isset($this->aStructuredQuery['amenity'])) {
 578                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 579                 unset($this->aStructuredQuery['amenity']);
 580             }
 581
 582             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 583                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 584                 $sToken = chksql(
 585                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 586                     'Cannot decode query. Wrong encoding?'
 587                 );
 588                 $sSQL = 'SELECT class, type FROM word ';
 589                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 590                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 591
 592                 Debug::printSQL($sSQL);
 593                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 594                 $aNewSearches = array();
 595                 foreach ($aSearches as $oSearch) {
 596                     foreach ($aSearchWords as $aSearchTerm) {
 597                         $oNewSearch = clone $oSearch;
 598                         $oNewSearch->setPoiSearch(
 599                             Operator::TYPE,
 600                             $aSearchTerm['class'],
 601                             $aSearchTerm['type']
 602                         );
 603                         $aNewSearches[] = $oNewSearch;
 604                     }
 605                 }
 606                 $aSearches = $aNewSearches;
 607             }
 608
 609             // Split query into phrases
 610             // Commas are used to reduce the search space by indicating where phrases split
 611             if ($this->aStructuredQuery) {
 612                 $aInPhrases = $this->aStructuredQuery;
 613                 $bStructuredPhrases = true;
 614             } else {
 615                 $aInPhrases = explode(',', $sQuery);
 616                 $bStructuredPhrases = false;
 617             }
 618
 619             Debug::printDebugArray('Search context', $oCtx);
 620             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 621             Debug::printVar('Final query phrases', $aInPhrases);
 622
 623             // Convert each phrase to standard form
 624             // Create a list of standard words
 625             // Get all 'sets' of words
 626             // Generate a complete list of all
 627             Debug::newSection('Tokenization');
 628             $aTokens = array();
 629             $aPhrases = array();
 630             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 631                 $sPhrase = chksql(
 632                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 633                     'Cannot normalize query string (is it a UTF-8 string?)'
 634                 );
 635                 if (trim($sPhrase)) {
 636                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 637                     $oPhrase->addTokens($aTokens);
 638                     $aPhrases[] = $oPhrase;
 639                 }
 640             }
 641
 642             Debug::printDebugTable('Phrases', $aPhrases);
 643             Debug::printVar('Tokens', $aTokens);
 644
 645             $oValidTokens = new TokenList();
 646
 647             if (!empty($aTokens)) {
 648                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 649                 $sSQL .= ' FROM word ';
 650                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 651
 652                 Debug::printSQL($sSQL);
 653
 654                 $oValidTokens->addTokensFromDB(
 655                     $this->oDB,
 656                     $aTokens,
 657                     $this->aCountryCodes,
 658                     $sNormQuery,
 659                     $this->oNormalizer
 660                 );
 661
 662                 // Try more interpretations for Tokens that could not be matched.
 663                 foreach ($aTokens as $sToken) {
 664                     if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
 665                         if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 666                             // US ZIP+4 codes - merge in the 5-digit ZIP code
 667                             $oValidTokens->addToken(
 668                                 $sToken,
 669                                 new Token\Postcode(null, $aData[1], 'us')
 670                             );
 671                         } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
 672                             // Unknown single word token with a number.
 673                             // Assume it is a house number.
 674                             $oValidTokens->addToken(
 675                                 $sToken,
 676                                 new Token\HouseNumber(null, trim($sToken))
 677                             );
 678                         }
 679                     }
 680                 }
 681
 682                 // Any words that have failed completely?
 683                 // TODO: suggestions
 684
 685                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 686
 687                 Debug::newSection('Search candidates');
 688
 689                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
 690
 691                 if ($this->bReverseInPlan) {
 692                     // Reverse phrase array and also reverse the order of the wordsets in
 693                     // the first and final phrase. Don't bother about phrases in the middle
 694                     // because order in the address doesn't matter.
 695                     $aPhrases = array_reverse($aPhrases);
 696                     $aPhrases[0]->invertWordSets();
 697                     if (count($aPhrases) > 1) {
 698                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 699                     }
 700                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
 701
 702                     foreach ($aGroupedSearches as $aSearches) {
 703                         foreach ($aSearches as $aSearch) {
 704                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 705                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 706                             }
 707                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 708                         }
 709                     }
 710
 711                     $aGroupedSearches = $aReverseGroupedSearches;
 712                     ksort($aGroupedSearches);
 713                 }
 714             } else {
 715                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 716                 $aGroupedSearches = array();
 717                 foreach ($aSearches as $aSearch) {
 718                     if ($aSearch->getRank() < $this->iMaxRank) {
 719                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 720                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 721                     }
 722                 }
 723                 ksort($aGroupedSearches);
 724             }
 725
 726             // Filter out duplicate searches
 727             $aSearchHash = array();
 728             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 729                 foreach ($aSearches as $iSearch => $aSearch) {
 730                     $sHash = serialize($aSearch);
 731                     if (isset($aSearchHash[$sHash])) {
 732                         unset($aGroupedSearches[$iGroup][$iSearch]);
 733                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 734                     } else {
 735                         $aSearchHash[$sHash] = 1;
 736                     }
 737                 }
 738             }
 739
 740             Debug::printGroupedSearch(
 741                 $aGroupedSearches,
 742                 $oValidTokens->debugTokenByWordIdList()
 743             );
 744
 745             // Start the search process
 746             $iGroupLoop = 0;
 747             $iQueryLoop = 0;
 748             $aNextResults = array();
 749             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 750                 $iGroupLoop++;
 751                 $aResults = $aNextResults;
 752                 foreach ($aSearches as $oSearch) {
 753                     $iQueryLoop++;
 754
 755                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 756                     Debug::printGroupedSearch(
 757                         array($iGroupedRank => array($oSearch)),
 758                         $oValidTokens->debugTokenByWordIdList()
 759                     );
 760
 761                     $aNewResults = $oSearch->query(
 762                         $this->oDB,
 763                         $this->iMinAddressRank,
 764                         $this->iMaxAddressRank,
 765                         $this->iLimit
 766                     );
 767
 768                     // The same result may appear in different rounds, only
 769                     // use the one with minimal rank.
 770                     foreach ($aNewResults as $iPlace => $oRes) {
 771                         if (!isset($aResults[$iPlace])
 772                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 773                             $aResults[$iPlace] = $oRes;
 774                         }
 775                     }
 776
 777                     if ($iQueryLoop > 20) break;
 778                 }
 779
 780                 if (!empty($aResults)) {
 781                     $aSplitResults = Result::splitResults($aResults);
 782                     Debug::printVar('Split results', $aSplitResults);
 783                     if ($iGroupLoop <= 4 && empty($aSplitResults['tail'])
 784                         && reset($aSplitResults['head'])->iResultRank > 0) {
 785                         // Haven't found an exact match for the query yet.
 786                         // Therefore add result from the next group level.
 787                         $aNextResults = $aSplitResults['head'];
 788                         foreach ($aNextResults as $oRes) {
 789                             $oRes->iResultRank--;
 790                         }
 791                         $aResults = array();
 792                     } else {
 793                         $aResults = $aSplitResults['head'];
 794                     }
 795                 }
 796
 797                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 798                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 799                     // reduces the number of place ids, like a filter
 800                     // rank_address is 30 for interpolated housenumbers
 801                     $aFilterSql = array();
 802                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 803                     if ($sPlaceIds) {
 804                         $sSQL = 'SELECT place_id FROM placex ';
 805                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 806                         $sSQL .= '  AND (';
 807                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 808                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 809                             $sSQL .= "     OR (extratags->'place') = 'city'";
 810                         }
 811                         if ($this->aAddressRankList) {
 812                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 813                         }
 814                         $sSQL .= ')';
 815                         $aFilterSql[] = $sSQL;
 816                     }
 817                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 818                     if ($sPlaceIds) {
 819                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 820                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 821                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 822                         if ($this->aAddressRankList) {
 823                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 824                         }
 825                         $sSQL .= ') ';
 826                         $aFilterSql[] = $sSQL;
 827                     }
 828
 829                     $aFilteredIDs = array();
 830                     if ($aFilterSql) {
 831                         $sSQL = join(' UNION ', $aFilterSql);
 832                         Debug::printSQL($sSQL);
 833                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 834                     }
 835
 836                     $tempIDs = array();
 837                     foreach ($aResults as $oResult) {
 838                         if (($this->iMaxAddressRank == 30 &&
 839                              ($oResult->iTable == Result::TABLE_OSMLINE
 840                               || $oResult->iTable == Result::TABLE_AUX
 841                               || $oResult->iTable == Result::TABLE_TIGER))
 842                             || in_array($oResult->iId, $aFilteredIDs)
 843                         ) {
 844                             $tempIDs[$oResult->iId] = $oResult;
 845                         }
 846                     }
 847                     $aResults = $tempIDs;
 848                 }
 849
 850                 if (!empty($aResults)) break;
 851                 if ($iGroupLoop > 4) break;
 852                 if ($iQueryLoop > 30) break;
 853             }
 854         } else {
 855             // Just interpret as a reverse geocode
 856             $oReverse = new ReverseGeocode($this->oDB);
 857             $oReverse->setZoom(18);
 858
 859             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 860
 861             Debug::printVar('Reverse search', $oLookup);
 862
 863             if ($oLookup) {
 864                 $aResults = array($oLookup->iId => $oLookup);
 865             }
 866         }
 867
 868         // No results? Done
 869         if (empty($aResults)) {
 870             if ($this->bFallback) {
 871                 if ($this->fallbackStructuredQuery()) {
 872                     return $this->lookup();
 873                 }
 874             }
 875
 876             return array();
 877         }
 878
 879         if ($this->aAddressRankList) {
 880             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 881         }
 882         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 883         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 884         if ($oCtx->hasNearPoint()) {
 885             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 886         }
 887
 888         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 889
 890         $aClassType = ClassTypes\getListWithImportance();
 891         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 892         foreach ($aRecheckWords as $i => $sWord) {
 893             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 894         }
 895
 896         Debug::printVar('Recheck words', $aRecheckWords);
 897
 898         foreach ($aSearchResults as $iIdx => $aResult) {
 899             // Default
 900             $fDiameter = ClassTypes\getProperty($aResult, 'defdiameter', 0.0001);
 901
 902             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 903             if ($aOutlineResult) {
 904                 $aResult = array_merge($aResult, $aOutlineResult);
 905             }
 906
 907             if ($aResult['extra_place'] == 'city') {
 908                 $aResult['class'] = 'place';
 909                 $aResult['type'] = 'city';
 910                 $aResult['rank_search'] = 16;
 911             }
 912
 913             // Is there an icon set for this type of result?
 914             $aClassInfo = ClassTypes\getInfo($aResult);
 915
 916             if ($aClassInfo) {
 917                 if (isset($aClassInfo['icon'])) {
 918                     $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassInfo['icon'].'.p.20.png';
 919                 }
 920
 921                 if (isset($aClassInfo['label'])) {
 922                     $aResult['label'] = $aClassInfo['label'];
 923                 }
 924             }
 925
 926             $aResult['name'] = $aResult['langaddress'];
 927
 928             if ($oCtx->hasNearPoint()) {
 929                 $aResult['importance'] = 0.001;
 930                 $aResult['foundorder'] = $aResult['addressimportance'];
 931             } else {
 932                 $aResult['importance'] = max(0.001, $aResult['importance']);
 933                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 934                     $aResult['lon'],
 935                     $aResult['lat']
 936                 );
 937                 // Adjust importance for the number of exact string matches in the result
 938                 $iCountWords = 0;
 939                 $sAddress = $aResult['langaddress'];
 940                 foreach ($aRecheckWords as $i => $sWord) {
 941                     if (stripos($sAddress, $sWord)!==false) {
 942                         $iCountWords++;
 943                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 944                     }
 945                 }
 946
 947                 $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
 948
 949                 // secondary ordering (for results with same importance (the smaller the better):
 950                 // - approximate importance of address parts
 951                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
 952                 // - number of exact matches from the query
 953                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 954                 // - importance of the class/type
 955                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
 956                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
 957                 ) {
 958                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
 959                 } else {
 960                     $aResult['foundorder'] += 0.01;
 961                 }
 962             }
 963             $aSearchResults[$iIdx] = $aResult;
 964         }
 965         uasort($aSearchResults, 'byImportance');
 966         Debug::printVar('Pre-filter results', $aSearchResults);
 967
 968         $aOSMIDDone = array();
 969         $aClassTypeNameDone = array();
 970         $aToFilter = $aSearchResults;
 971         $aSearchResults = array();
 972
 973         $bFirst = true;
 974         foreach ($aToFilter as $aResult) {
 975             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 976             if ($bFirst) {
 977                 $fLat = $aResult['lat'];
 978                 $fLon = $aResult['lon'];
 979                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 980                 $bFirst = false;
 981             }
 982             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 983                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 984             ) {
 985                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 986                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 987                 $aSearchResults[] = $aResult;
 988             }
 989
 990             // Absolute limit on number of results
 991             if (count($aSearchResults) >= $this->iFinalLimit) break;
 992         }
 993
 994         Debug::printVar('Post-filter results', $aSearchResults);
 995         return $aSearchResults;
 996     } // end lookup()
 997
 998     public function debugInfo()
 999     {
1000         return array(
1001                 'Query' => $this->sQuery,
1002                 'Structured query' => $this->aStructuredQuery,
1003                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
1004                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
1005                 'Try reversed query'=> $this->bReverseInPlan,
1006                 'Limit (for searches)' => $this->iLimit,
1007                 'Limit (for results)'=> $this->iFinalLimit,
1008                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
1009                 'Bounded search' => $this->bBoundedSearch,
1010                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
1011                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
1012                 'Route width' => $this->aRouteWidth,
1013                 'Max rank' => $this->iMaxRank,
1014                 'Min address rank' => $this->iMinAddressRank,
1015                 'Max address rank' => $this->iMaxAddressRank,
1016                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
1017                );
1018     }
1019 } // end class