lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10 require_once(CONST_BasePath.'/lib/TokenList.php');
  11
  12 class Geocode
  13 {
  14     protected $oDB;
  15
  16     protected $oPlaceLookup;
  17
  18     protected $aLangPrefOrder = array();
  19
  20     protected $bIncludeAddressDetails = false;
  21
  22     protected $aExcludePlaceIDs = array();
  23     protected $bReverseInPlan = false;
  24
  25     protected $iLimit = 20;
  26     protected $iFinalLimit = 10;
  27     protected $iOffset = 0;
  28     protected $bFallback = false;
  29
  30     protected $aCountryCodes = false;
  31
  32     protected $bBoundedSearch = false;
  33     protected $aViewBox = false;
  34     protected $aRoutePoints = false;
  35     protected $aRouteWidth = false;
  36
  37     protected $iMaxRank = 20;
  38     protected $iMinAddressRank = 0;
  39     protected $iMaxAddressRank = 30;
  40     protected $aAddressRankList = array();
  41
  42     protected $sAllowedTypesSQLList = false;
  43
  44     protected $sQuery = false;
  45     protected $aStructuredQuery = false;
  46
  47     protected $oNormalizer = null;
  48
  49
  50     public function __construct(&$oDB)
  51     {
  52         $this->oDB =& $oDB;
  53         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  54         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  55     }
  56
  57     private function normTerm($sTerm)
  58     {
  59         if ($this->oNormalizer === null) {
  60             return $sTerm;
  61         }
  62
  63         return $this->oNormalizer->transliterate($sTerm);
  64     }
  65
  66     public function setReverseInPlan($bReverse)
  67     {
  68         $this->bReverseInPlan = $bReverse;
  69     }
  70
  71     public function setLanguagePreference($aLangPref)
  72     {
  73         $this->aLangPrefOrder = $aLangPref;
  74     }
  75
  76     public function getMoreUrlParams()
  77     {
  78         if ($this->aStructuredQuery) {
  79             $aParams = $this->aStructuredQuery;
  80         } else {
  81             $aParams = array('q' => $this->sQuery);
  82         }
  83
  84         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  85
  86         if ($this->aExcludePlaceIDs) {
  87             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  88         }
  89
  90         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  91         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  92
  93         if ($this->aCountryCodes) {
  94             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  95         }
  96
  97         if ($this->aViewBox) {
  98             $aParams['viewbox'] = join(',', $this->aViewBox);
  99         }
 100
 101         return $aParams;
 102     }
 103
 104     public function setLimit($iLimit = 10)
 105     {
 106         if ($iLimit > 50) $iLimit = 50;
 107         if ($iLimit < 1) $iLimit = 1;
 108
 109         $this->iFinalLimit = $iLimit;
 110         $this->iLimit = $iLimit + min($iLimit, 10);
 111     }
 112
 113     public function setFeatureType($sFeatureType)
 114     {
 115         switch ($sFeatureType) {
 116             case 'country':
 117                 $this->setRankRange(4, 4);
 118                 break;
 119             case 'state':
 120                 $this->setRankRange(8, 8);
 121                 break;
 122             case 'city':
 123                 $this->setRankRange(14, 16);
 124                 break;
 125             case 'settlement':
 126                 $this->setRankRange(8, 20);
 127                 break;
 128         }
 129     }
 130
 131     public function setRankRange($iMin, $iMax)
 132     {
 133         $this->iMinAddressRank = $iMin;
 134         $this->iMaxAddressRank = $iMax;
 135     }
 136
 137     public function setViewbox($aViewbox)
 138     {
 139         $aBox = array_map('floatval', $aViewbox);
 140
 141         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 142         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 143         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 144         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 145
 146         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 147             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 148         ) {
 149             userError("Bad parameter 'viewbox'. Not a box.");
 150         }
 151     }
 152
 153     private function viewboxImportanceFactor($fX, $fY)
 154     {
 155         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 156         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 157
 158         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 159         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 160
 161         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 162             return 1;
 163         }
 164
 165         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 166             return 0.5;
 167         }
 168
 169         return 0.25;
 170     }
 171
 172     public function setQuery($sQueryString)
 173     {
 174         $this->sQuery = $sQueryString;
 175         $this->aStructuredQuery = false;
 176     }
 177
 178     public function getQueryString()
 179     {
 180         return $this->sQuery;
 181     }
 182
 183
 184     public function loadParamArray($oParams, $sForceGeometryType = null)
 185     {
 186         $this->bIncludeAddressDetails
 187          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 188
 189         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 190
 191         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 192         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 193
 194         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 195
 196         // List of excluded Place IDs - used for more acurate pageing
 197         $sExcluded = $oParams->getStringList('exclude_place_ids');
 198         if ($sExcluded) {
 199             foreach ($sExcluded as $iExcludedPlaceID) {
 200                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 201                 if ($iExcludedPlaceID)
 202                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 203             }
 204
 205             if (isset($aExcludePlaceIDs))
 206                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 207         }
 208
 209         // Only certain ranks of feature
 210         $sFeatureType = $oParams->getString('featureType');
 211         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 212         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 213
 214         // Country code list
 215         $sCountries = $oParams->getStringList('countrycodes');
 216         if ($sCountries) {
 217             foreach ($sCountries as $sCountryCode) {
 218                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 219                     $aCountries[] = strtolower($sCountryCode);
 220                 }
 221             }
 222             if (isset($aCountries))
 223                 $this->aCountryCodes = $aCountries;
 224         }
 225
 226         $aViewbox = $oParams->getStringList('viewboxlbrt');
 227         if ($aViewbox) {
 228             if (count($aViewbox) != 4) {
 229                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 230             }
 231             $this->setViewbox($aViewbox);
 232         } else {
 233             $aViewbox = $oParams->getStringList('viewbox');
 234             if ($aViewbox) {
 235                 if (count($aViewbox) != 4) {
 236                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 237                 }
 238                 $this->setViewBox($aViewbox);
 239             } else {
 240                 $aRoute = $oParams->getStringList('route');
 241                 $fRouteWidth = $oParams->getFloat('routewidth');
 242                 if ($aRoute && $fRouteWidth) {
 243                     $this->aRoutePoints = $aRoute;
 244                     $this->aRouteWidth = $fRouteWidth;
 245                 }
 246             }
 247         }
 248
 249         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 250         $this->oPlaceLookup->setIncludeAddressDetails(false);
 251         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 252
 253         if ($this->bIncludeAddressDetails
 254             && $oParams->getString('format', '') == 'geocodejson'
 255            ) {
 256             $this->oPlaceLookup->setAddressAdminLevels(true);
 257         }
 258
 259     }
 260
 261     public function setQueryFromParams($oParams)
 262     {
 263         // Search query
 264         $sQuery = $oParams->getString('q');
 265         if (!$sQuery) {
 266             $this->setStructuredQuery(
 267                 $oParams->getString('amenity'),
 268                 $oParams->getString('street'),
 269                 $oParams->getString('city'),
 270                 $oParams->getString('county'),
 271                 $oParams->getString('state'),
 272                 $oParams->getString('country'),
 273                 $oParams->getString('postalcode')
 274             );
 275             $this->setReverseInPlan(false);
 276         } else {
 277             $this->setQuery($sQuery);
 278         }
 279     }
 280
 281     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 282     {
 283         $sValue = trim($sValue);
 284         if (!$sValue) return false;
 285         $this->aStructuredQuery[$sKey] = $sValue;
 286         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 287             $this->iMinAddressRank = $iNewMinAddressRank;
 288             $this->iMaxAddressRank = $iNewMaxAddressRank;
 289         }
 290         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 291         return true;
 292     }
 293
 294     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 295     {
 296         $this->sQuery = false;
 297
 298         // Reset
 299         $this->iMinAddressRank = 0;
 300         $this->iMaxAddressRank = 30;
 301         $this->aAddressRankList = array();
 302
 303         $this->aStructuredQuery = array();
 304         $this->sAllowedTypesSQLList = false;
 305
 306         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 307         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 308         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 309         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 310         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 311         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 312         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 313
 314         if (!empty($this->aStructuredQuery)) {
 315             $this->sQuery = join(', ', $this->aStructuredQuery);
 316             if ($this->iMaxAddressRank < 30) {
 317                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 318             }
 319         }
 320     }
 321
 322     public function fallbackStructuredQuery()
 323     {
 324         if (!$this->aStructuredQuery) return false;
 325
 326         $aParams = $this->aStructuredQuery;
 327
 328         if (count($aParams) == 1) return false;
 329
 330         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 331
 332         foreach ($aOrderToFallback as $sType) {
 333             if (isset($aParams[$sType])) {
 334                 unset($aParams[$sType]);
 335                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 336                 return true;
 337             }
 338         }
 339
 340         return false;
 341     }
 342
 343     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
 344     {
 345         /*
 346              Calculate all searches using oValidTokens i.e.
 347              'Wodsworth Road, Sheffield' =>
 348
 349              Phrase Wordset
 350              0      0       (wodsworth road)
 351              0      1       (wodsworth)(road)
 352              1      0       (sheffield)
 353
 354              Score how good the search is so they can be ordered
 355          */
 356         foreach ($aPhrases as $iPhrase => $oPhrase) {
 357             $aNewPhraseSearches = array();
 358             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 359
 360             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 361                 // Too many permutations - too expensive
 362                 if ($iWordSet > 120) break;
 363
 364                 $aWordsetSearches = $aSearches;
 365
 366                 // Add all words from this wordset
 367                 foreach ($aWordset as $iToken => $sToken) {
 368                     //echo "<br><b>$sToken</b>";
 369                     $aNewWordsetSearches = array();
 370
 371                     foreach ($aWordsetSearches as $oCurrentSearch) {
 372                         //echo "<i>";
 373                         //var_dump($oCurrentSearch);
 374                         //echo "</i>";
 375
 376                         // Tokens with full name matches.
 377                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 378                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 379                                 $oSearchTerm,
 380                                 $oValidTokens->contains($sToken)
 381                                   && strpos($sToken, ' ') === false,
 382                                 $sPhraseType,
 383                                 $iToken == 0 && $iPhrase == 0,
 384                                 $iPhrase == 0,
 385                                 $iToken + 1 == count($aWordset)
 386                                   && $iPhrase + 1 == count($aPhrases)
 387                             );
 388
 389                             foreach ($aNewSearches as $oSearch) {
 390                                 if ($oSearch->getRank() < $this->iMaxRank) {
 391                                     $aNewWordsetSearches[] = $oSearch;
 392                                 }
 393                             }
 394                         }
 395                         // Look for partial matches.
 396                         // Note that there is no point in adding country terms here
 397                         // because country is omitted in the address.
 398                         if ($sPhraseType != 'country') {
 399                             // Allow searching for a word - but at extra cost
 400                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 401                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 402                                     $sToken,
 403                                     $oSearchTerm,
 404                                     $bIsStructured,
 405                                     $iPhrase,
 406                                     $oValidTokens->get(' '.$sToken)
 407                                 );
 408
 409                                 foreach ($aNewSearches as $oSearch) {
 410                                     if ($oSearch->getRank() < $this->iMaxRank) {
 411                                         $aNewWordsetSearches[] = $oSearch;
 412                                     }
 413                                 }
 414                             }
 415                         }
 416                     }
 417                     // Sort and cut
 418                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 419                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 420                 }
 421                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 422
 423                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 424                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 425
 426                 $aSearchHash = array();
 427                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 428                     $sHash = serialize($aSearch);
 429                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 430                     else $aSearchHash[$sHash] = 1;
 431                 }
 432
 433                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 434             }
 435
 436             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 437             $aGroupedSearches = array();
 438             foreach ($aNewPhraseSearches as $aSearch) {
 439                 $iRank = $aSearch->getRank();
 440                 if ($iRank < $this->iMaxRank) {
 441                     if (!isset($aGroupedSearches[$iRank])) {
 442                         $aGroupedSearches[$iRank] = array();
 443                     }
 444                     $aGroupedSearches[$iRank][] = $aSearch;
 445                 }
 446             }
 447             ksort($aGroupedSearches);
 448
 449             $iSearchCount = 0;
 450             $aSearches = array();
 451             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 452                 $iSearchCount += count($aNewSearches);
 453                 $aSearches = array_merge($aSearches, $aNewSearches);
 454                 if ($iSearchCount > 50) break;
 455             }
 456         }
 457
 458         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 459         $aGroupedSearches = array();
 460         foreach ($aSearches as $oSearch) {
 461             if (!$oSearch->isValidSearch()) {
 462                 continue;
 463             }
 464
 465             $iRank = $oSearch->getRank();
 466             if (!isset($aGroupedSearches[$iRank])) {
 467                 $aGroupedSearches[$iRank] = array();
 468             }
 469             $aGroupedSearches[$iRank][] = $oSearch;
 470         }
 471         ksort($aGroupedSearches);
 472
 473         return $aGroupedSearches;
 474     }
 475
 476     /* Perform the actual query lookup.
 477
 478         Returns an ordered list of results, each with the following fields:
 479             osm_type: type of corresponding OSM object
 480                         N - node
 481                         W - way
 482                         R - relation
 483                         P - postcode (internally computed)
 484             osm_id: id of corresponding OSM object
 485             class: general object class (corresponds to tag key of primary OSM tag)
 486             type: subclass of object (corresponds to tag value of primary OSM tag)
 487             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 488             rank_search: rank in search hierarchy
 489                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 490             rank_address: rank in address hierarchy (determines orer in address)
 491             place_id: internal key (may differ between different instances)
 492             country_code: ISO country code
 493             langaddress: localized full address
 494             placename: localized name of object
 495             ref: content of ref tag (if available)
 496             lon: longitude
 497             lat: latitude
 498             importance: importance of place based on Wikipedia link count
 499             addressimportance: cumulated importance of address elements
 500             extra_place: type of place (for admin boundaries, if there is a place tag)
 501             aBoundingBox: bounding Box
 502             label: short description of the object class/type (English only)
 503             name: full name (currently the same as langaddress)
 504             foundorder: secondary ordering for places with same importance
 505     */
 506
 507
 508     public function lookup()
 509     {
 510         Debug::newFunction('Geocode::lookup');
 511         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 512
 513         Debug::printDebugArray('Geocode', $this);
 514
 515         $oCtx = new SearchContext();
 516
 517         if ($this->aRoutePoints) {
 518             $oCtx->setViewboxFromRoute(
 519                 $this->oDB,
 520                 $this->aRoutePoints,
 521                 $this->aRouteWidth,
 522                 $this->bBoundedSearch
 523             );
 524         } elseif ($this->aViewBox) {
 525             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 526         }
 527         if ($this->aExcludePlaceIDs) {
 528             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 529         }
 530         if ($this->aCountryCodes) {
 531             $oCtx->setCountryList($this->aCountryCodes);
 532         }
 533
 534         Debug::newSection('Query Preprocessing');
 535
 536         $sNormQuery = $this->normTerm($this->sQuery);
 537         Debug::printVar('Normalized query', $sNormQuery);
 538
 539         $sLanguagePrefArraySQL = getArraySQL(
 540             array_map('getDBQuoted', $this->aLangPrefOrder)
 541         );
 542
 543         $sQuery = $this->sQuery;
 544         if (!preg_match('//u', $sQuery)) {
 545             userError('Query string is not UTF-8 encoded.');
 546         }
 547
 548         // Conflicts between US state abreviations and various words for 'the' in different languages
 549         if (isset($this->aLangPrefOrder['name:en'])) {
 550             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 551             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 552             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 553         }
 554
 555         // Do we have anything that looks like a lat/lon pair?
 556         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 557
 558         $aResults = array();
 559         if ($sQuery || $this->aStructuredQuery) {
 560             // Start with a single blank search
 561             $aSearches = array(new SearchDescription($oCtx));
 562
 563             if ($sQuery) {
 564                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 565             }
 566
 567             $sSpecialTerm = '';
 568             if ($sQuery) {
 569                 preg_match_all(
 570                     '/\\[([\\w ]*)\\]/u',
 571                     $sQuery,
 572                     $aSpecialTermsRaw,
 573                     PREG_SET_ORDER
 574                 );
 575                 if (!empty($aSpecialTermsRaw)) {
 576                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 577                 }
 578
 579                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 580                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 581                     if (!$sSpecialTerm) {
 582                         $sSpecialTerm = $aSpecialTerm[1];
 583                     }
 584                 }
 585             }
 586             if (!$sSpecialTerm && $this->aStructuredQuery
 587                 && isset($this->aStructuredQuery['amenity'])) {
 588                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 589                 unset($this->aStructuredQuery['amenity']);
 590             }
 591
 592             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 593                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 594                 $sToken = chksql(
 595                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 596                     'Cannot decode query. Wrong encoding?'
 597                 );
 598                 $sSQL = 'SELECT class, type FROM word ';
 599                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 600                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 601
 602                 Debug::printSQL($sSQL);
 603                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 604                 $aNewSearches = array();
 605                 foreach ($aSearches as $oSearch) {
 606                     foreach ($aSearchWords as $aSearchTerm) {
 607                         $oNewSearch = clone $oSearch;
 608                         $oNewSearch->setPoiSearch(
 609                             Operator::TYPE,
 610                             $aSearchTerm['class'],
 611                             $aSearchTerm['type']
 612                         );
 613                         $aNewSearches[] = $oNewSearch;
 614                     }
 615                 }
 616                 $aSearches = $aNewSearches;
 617             }
 618
 619             // Split query into phrases
 620             // Commas are used to reduce the search space by indicating where phrases split
 621             if ($this->aStructuredQuery) {
 622                 $aInPhrases = $this->aStructuredQuery;
 623                 $bStructuredPhrases = true;
 624             } else {
 625                 $aInPhrases = explode(',', $sQuery);
 626                 $bStructuredPhrases = false;
 627             }
 628
 629             Debug::printDebugArray('Search context', $oCtx);
 630             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 631             Debug::printVar('Final query phrases', $aInPhrases);
 632
 633             // Convert each phrase to standard form
 634             // Create a list of standard words
 635             // Get all 'sets' of words
 636             // Generate a complete list of all
 637             Debug::newSection('Tokenization');
 638             $aTokens = array();
 639             $aPhrases = array();
 640             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 641                 $sPhrase = chksql(
 642                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 643                     'Cannot normalize query string (is it a UTF-8 string?)'
 644                 );
 645                 if (trim($sPhrase)) {
 646                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 647                     $oPhrase->addTokens($aTokens);
 648                     $aPhrases[] = $oPhrase;
 649                 }
 650             }
 651
 652             Debug::printDebugTable('Phrases', $aPhrases);
 653             Debug::printVar('Tokens', $aTokens);
 654
 655             $oValidTokens = new TokenList();
 656
 657             if (!empty($aTokens)) {
 658                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 659                 $sSQL .= ' FROM word ';
 660                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 661
 662                 Debug::printSQL($sSQL);
 663
 664                 $oValidTokens->addTokensFromDB(
 665                     $this->oDB,
 666                     $aTokens,
 667                     $this->aCountryCodes,
 668                     $sNormQuery,
 669                     $this->oNormalizer
 670                 );
 671
 672                 // Try more interpretations for Tokens that could not be matched.
 673                 foreach ($aTokens as $sToken) {
 674                     if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
 675                         if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 676                             // US ZIP+4 codes - merge in the 5-digit ZIP code
 677                             $oValidTokens->addToken(
 678                                 $sToken,
 679                                 new Token\Postcode(null, $aData[1], 'us')
 680                             );
 681                         } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
 682                             // Unknown single word token with a number.
 683                             // Assume it is a house number.
 684                             $oValidTokens->addToken(
 685                                 $sToken,
 686                                 new Token\HouseNumber(null, trim($sToken))
 687                             );
 688                         }
 689                     }
 690                 }
 691
 692                 // Any words that have failed completely?
 693                 // TODO: suggestions
 694
 695                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 696
 697                 Debug::newSection('Search candidates');
 698
 699                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
 700
 701                 if ($this->bReverseInPlan) {
 702                     // Reverse phrase array and also reverse the order of the wordsets in
 703                     // the first and final phrase. Don't bother about phrases in the middle
 704                     // because order in the address doesn't matter.
 705                     $aPhrases = array_reverse($aPhrases);
 706                     $aPhrases[0]->invertWordSets();
 707                     if (count($aPhrases) > 1) {
 708                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 709                     }
 710                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
 711
 712                     foreach ($aGroupedSearches as $aSearches) {
 713                         foreach ($aSearches as $aSearch) {
 714                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 715                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 716                             }
 717                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 718                         }
 719                     }
 720
 721                     $aGroupedSearches = $aReverseGroupedSearches;
 722                     ksort($aGroupedSearches);
 723                 }
 724             } else {
 725                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 726                 $aGroupedSearches = array();
 727                 foreach ($aSearches as $aSearch) {
 728                     if ($aSearch->getRank() < $this->iMaxRank) {
 729                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 730                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 731                     }
 732                 }
 733                 ksort($aGroupedSearches);
 734             }
 735
 736             // Filter out duplicate searches
 737             $aSearchHash = array();
 738             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 739                 foreach ($aSearches as $iSearch => $aSearch) {
 740                     $sHash = serialize($aSearch);
 741                     if (isset($aSearchHash[$sHash])) {
 742                         unset($aGroupedSearches[$iGroup][$iSearch]);
 743                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 744                     } else {
 745                         $aSearchHash[$sHash] = 1;
 746                     }
 747                 }
 748             }
 749
 750             Debug::printGroupedSearch(
 751                 $aGroupedSearches,
 752                 $oValidTokens->debugTokenByWordIdList()
 753             );
 754
 755             // Start the search process
 756             $iGroupLoop = 0;
 757             $iQueryLoop = 0;
 758             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 759                 $iGroupLoop++;
 760                 foreach ($aSearches as $oSearch) {
 761                     $iQueryLoop++;
 762
 763                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 764                     Debug::printGroupedSearch(
 765                         array($iGroupedRank => array($oSearch)),
 766                         $oValidTokens->debugTokenByWordIdList()
 767                     );
 768
 769                     $aResults += $oSearch->query(
 770                         $this->oDB,
 771                         $this->iMinAddressRank,
 772                         $this->iMaxAddressRank,
 773                         $this->iLimit
 774                     );
 775
 776                     if ($iQueryLoop > 20) break;
 777                 }
 778
 779                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 780                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 781                     // reduces the number of place ids, like a filter
 782                     // rank_address is 30 for interpolated housenumbers
 783                     $aFilterSql = array();
 784                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 785                     if ($sPlaceIds) {
 786                         $sSQL = 'SELECT place_id FROM placex ';
 787                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 788                         $sSQL .= '  AND (';
 789                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 790                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 791                             $sSQL .= "     OR (extratags->'place') = 'city'";
 792                         }
 793                         if ($this->aAddressRankList) {
 794                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 795                         }
 796                         $sSQL .= ')';
 797                         $aFilterSql[] = $sSQL;
 798                     }
 799                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 800                     if ($sPlaceIds) {
 801                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 802                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 803                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 804                         if ($this->aAddressRankList) {
 805                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 806                         }
 807                         $sSQL .= ') ';
 808                         $aFilterSql[] = $sSQL;
 809                     }
 810
 811                     $aFilteredIDs = array();
 812                     if ($aFilterSql) {
 813                         $sSQL = join(' UNION ', $aFilterSql);
 814                         Debug::printSQL($sSQL);
 815                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 816                     }
 817
 818                     $tempIDs = array();
 819                     foreach ($aResults as $oResult) {
 820                         if (($this->iMaxAddressRank == 30 &&
 821                              ($oResult->iTable == Result::TABLE_OSMLINE
 822                               || $oResult->iTable == Result::TABLE_AUX
 823                               || $oResult->iTable == Result::TABLE_TIGER))
 824                             || in_array($oResult->iId, $aFilteredIDs)
 825                         ) {
 826                             $tempIDs[$oResult->iId] = $oResult;
 827                         }
 828                     }
 829                     $aResults = $tempIDs;
 830                 }
 831
 832                 if (!empty($aResults)) break;
 833                 if ($iGroupLoop > 4) break;
 834                 if ($iQueryLoop > 30) break;
 835             }
 836         } else {
 837             // Just interpret as a reverse geocode
 838             $oReverse = new ReverseGeocode($this->oDB);
 839             $oReverse->setZoom(18);
 840
 841             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 842
 843             Debug::printVar('Reverse search', $oLookup);
 844
 845             if ($oLookup) {
 846                 $aResults = array($oLookup->iId => $oLookup);
 847             }
 848         }
 849
 850         // No results? Done
 851         if (empty($aResults)) {
 852             if ($this->bFallback) {
 853                 if ($this->fallbackStructuredQuery()) {
 854                     return $this->lookup();
 855                 }
 856             }
 857
 858             return array();
 859         }
 860
 861         if ($this->aAddressRankList) {
 862             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 863         }
 864         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 865         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 866         if ($oCtx->hasNearPoint()) {
 867             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 868         }
 869
 870         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 871
 872         $aClassType = getClassTypesWithImportance();
 873         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 874         foreach ($aRecheckWords as $i => $sWord) {
 875             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 876         }
 877
 878         Debug::printVar('Recheck words', $aRecheckWords);
 879
 880         foreach ($aSearchResults as $iIdx => $aResult) {
 881             // Default
 882             $fDiameter = getResultDiameter($aResult);
 883
 884             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 885             if ($aOutlineResult) {
 886                 $aResult = array_merge($aResult, $aOutlineResult);
 887             }
 888
 889             if ($aResult['extra_place'] == 'city') {
 890                 $aResult['class'] = 'place';
 891                 $aResult['type'] = 'city';
 892                 $aResult['rank_search'] = 16;
 893             }
 894
 895             // Is there an icon set for this type of result?
 896             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
 897                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
 898             ) {
 899                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
 900             }
 901
 902             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
 903                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
 904             ) {
 905                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
 906             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
 907                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
 908             ) {
 909                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
 910             }
 911             // if tag '&addressdetails=1' is set in query
 912             if ($this->bIncludeAddressDetails) {
 913                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
 914                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber);
 915                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
 916                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
 917                 }
 918             }
 919
 920             $aResult['name'] = $aResult['langaddress'];
 921
 922             if ($oCtx->hasNearPoint()) {
 923                 $aResult['importance'] = 0.001;
 924                 $aResult['foundorder'] = $aResult['addressimportance'];
 925             } else {
 926                 $aResult['importance'] = max(0.001, $aResult['importance']);
 927                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 928                     $aResult['lon'],
 929                     $aResult['lat']
 930                 );
 931                 // Adjust importance for the number of exact string matches in the result
 932                 $iCountWords = 0;
 933                 $sAddress = $aResult['langaddress'];
 934                 foreach ($aRecheckWords as $i => $sWord) {
 935                     if (stripos($sAddress, $sWord)!==false) {
 936                         $iCountWords++;
 937                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 938                     }
 939                 }
 940
 941                 $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
 942
 943                 // secondary ordering (for results with same importance (the smaller the better):
 944                 // - approximate importance of address parts
 945                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
 946                 // - number of exact matches from the query
 947                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 948                 // - importance of the class/type
 949                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
 950                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
 951                 ) {
 952                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
 953                 } else {
 954                     $aResult['foundorder'] += 0.01;
 955                 }
 956             }
 957             $aSearchResults[$iIdx] = $aResult;
 958         }
 959         uasort($aSearchResults, 'byImportance');
 960         Debug::printVar('Pre-filter results', $aSearchResults);
 961
 962         $aOSMIDDone = array();
 963         $aClassTypeNameDone = array();
 964         $aToFilter = $aSearchResults;
 965         $aSearchResults = array();
 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[] = $aResult;
 982             }
 983
 984             // Absolute limit on number of results
 985             if (count($aSearchResults) >= $this->iFinalLimit) break;
 986         }
 987
 988         Debug::printVar('Post-filter results', $aSearchResults);
 989         return $aSearchResults;
 990     } // end lookup()
 991
 992     public function debugInfo()
 993     {
 994         return array(
 995                 'Query' => $this->sQuery,
 996                 'Structured query' => $this->aStructuredQuery,
 997                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 998                 'Include address' => $this->bIncludeAddressDetails,
 999                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
1000                 'Try reversed query'=> $this->bReverseInPlan,
1001                 'Limit (for searches)' => $this->iLimit,
1002                 'Limit (for results)'=> $this->iFinalLimit,
1003                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
1004                 'Bounded search' => $this->bBoundedSearch,
1005                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
1006                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
1007                 'Route width' => $this->aRouteWidth,
1008                 'Max rank' => $this->iMaxRank,
1009                 'Min address rank' => $this->iMinAddressRank,
1010                 'Max address rank' => $this->iMaxAddressRank,
1011                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
1012                );
1013     }
1014 } // end class