lib-php/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_LibDir.'/PlaceLookup.php');
   6 require_once(CONST_LibDir.'/Phrase.php');
   7 require_once(CONST_LibDir.'/ReverseGeocode.php');
   8 require_once(CONST_LibDir.'/SearchDescription.php');
   9 require_once(CONST_LibDir.'/SearchContext.php');
  10 require_once(CONST_LibDir.'/TokenList.php');
  11 require_once(CONST_TokenizerDir.'/tokenizer.php');
  12
  13 class Geocode
  14 {
  15     protected $oDB;
  16
  17     protected $oPlaceLookup;
  18     protected $oTokenizer;
  19
  20     protected $aLangPrefOrder = array();
  21
  22     protected $aExcludePlaceIDs = array();
  23
  24     protected $iLimit = 20;
  25     protected $iFinalLimit = 10;
  26     protected $iOffset = 0;
  27     protected $bFallback = false;
  28
  29     protected $aCountryCodes = false;
  30
  31     protected $bBoundedSearch = false;
  32     protected $aViewBox = false;
  33     protected $aRoutePoints = false;
  34     protected $aRouteWidth = false;
  35
  36     protected $iMaxRank = 20;
  37     protected $iMinAddressRank = 0;
  38     protected $iMaxAddressRank = 30;
  39     protected $aAddressRankList = array();
  40
  41     protected $sAllowedTypesSQLList = false;
  42
  43     protected $sQuery = false;
  44     protected $aStructuredQuery = false;
  45
  46
  47     public function __construct(&$oDB)
  48     {
  49         $this->oDB =& $oDB;
  50         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  51         $this->oTokenizer = new \Nominatim\Tokenizer($this->oDB);
  52     }
  53
  54     public function setLanguagePreference($aLangPref)
  55     {
  56         $this->aLangPrefOrder = $aLangPref;
  57     }
  58
  59     public function getMoreUrlParams()
  60     {
  61         if ($this->aStructuredQuery) {
  62             $aParams = $this->aStructuredQuery;
  63         } else {
  64             $aParams = array('q' => $this->sQuery);
  65         }
  66
  67         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  68
  69         if ($this->aExcludePlaceIDs) {
  70             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  71         }
  72
  73         if ($this->bBoundedSearch) {
  74             $aParams['bounded'] = '1';
  75         }
  76
  77         if ($this->aCountryCodes) {
  78             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  79         }
  80
  81         if ($this->aViewBox) {
  82             $aParams['viewbox'] = join(',', $this->aViewBox);
  83         }
  84
  85         return $aParams;
  86     }
  87
  88     public function setLimit($iLimit = 10)
  89     {
  90         if ($iLimit > 50) {
  91             $iLimit = 50;
  92         } elseif ($iLimit < 1) {
  93             $iLimit = 1;
  94         }
  95
  96         $this->iFinalLimit = $iLimit;
  97         $this->iLimit = $iLimit + min($iLimit, 10);
  98     }
  99
 100     public function setFeatureType($sFeatureType)
 101     {
 102         switch ($sFeatureType) {
 103             case 'country':
 104                 $this->setRankRange(4, 4);
 105                 break;
 106             case 'state':
 107                 $this->setRankRange(8, 8);
 108                 break;
 109             case 'city':
 110                 $this->setRankRange(14, 16);
 111                 break;
 112             case 'settlement':
 113                 $this->setRankRange(8, 20);
 114                 break;
 115         }
 116     }
 117
 118     public function setRankRange($iMin, $iMax)
 119     {
 120         $this->iMinAddressRank = $iMin;
 121         $this->iMaxAddressRank = $iMax;
 122     }
 123
 124     public function setViewbox($aViewbox)
 125     {
 126         $aBox = array_map('floatval', $aViewbox);
 127
 128         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 129         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 130         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 131         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 132
 133         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 134             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 135         ) {
 136             userError("Bad parameter 'viewbox'. Not a box.");
 137         }
 138     }
 139
 140     private function viewboxImportanceFactor($fX, $fY)
 141     {
 142         if (!$this->aViewBox) {
 143             return 1;
 144         }
 145
 146         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 147         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 148
 149         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 150         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 151
 152         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 153             return 1;
 154         }
 155
 156         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 157             return 0.5;
 158         }
 159
 160         return 0.25;
 161     }
 162
 163     public function setQuery($sQueryString)
 164     {
 165         $this->sQuery = $sQueryString;
 166         $this->aStructuredQuery = false;
 167     }
 168
 169     public function getQueryString()
 170     {
 171         return $this->sQuery;
 172     }
 173
 174
 175     public function loadParamArray($oParams, $sForceGeometryType = null)
 176     {
 177         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 178
 179         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 180         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 181
 182         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 183
 184         // List of excluded Place IDs - used for more acurate pageing
 185         $sExcluded = $oParams->getStringList('exclude_place_ids');
 186         if ($sExcluded) {
 187             foreach ($sExcluded as $iExcludedPlaceID) {
 188                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 189                 if ($iExcludedPlaceID) {
 190                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 191                 }
 192             }
 193
 194             if (isset($aExcludePlaceIDs)) {
 195                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 196             }
 197         }
 198
 199         // Only certain ranks of feature
 200         $sFeatureType = $oParams->getString('featureType');
 201         if (!$sFeatureType) {
 202             $sFeatureType = $oParams->getString('featuretype');
 203         }
 204         if ($sFeatureType) {
 205             $this->setFeatureType($sFeatureType);
 206         }
 207
 208         // Country code list
 209         $sCountries = $oParams->getStringList('countrycodes');
 210         if ($sCountries) {
 211             foreach ($sCountries as $sCountryCode) {
 212                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 213                     $aCountries[] = strtolower($sCountryCode);
 214                 }
 215             }
 216             if (isset($aCountries)) {
 217                 $this->aCountryCodes = $aCountries;
 218             }
 219         }
 220
 221         $aViewbox = $oParams->getStringList('viewboxlbrt');
 222         if ($aViewbox) {
 223             if (count($aViewbox) != 4) {
 224                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 225             }
 226             $this->setViewbox($aViewbox);
 227         } else {
 228             $aViewbox = $oParams->getStringList('viewbox');
 229             if ($aViewbox) {
 230                 if (count($aViewbox) != 4) {
 231                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 232                 }
 233                 $this->setViewBox($aViewbox);
 234             } else {
 235                 $aRoute = $oParams->getStringList('route');
 236                 $fRouteWidth = $oParams->getFloat('routewidth');
 237                 if ($aRoute && $fRouteWidth) {
 238                     $this->aRoutePoints = $aRoute;
 239                     $this->aRouteWidth = $fRouteWidth;
 240                 }
 241             }
 242         }
 243
 244         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 245         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 246     }
 247
 248     public function setQueryFromParams($oParams)
 249     {
 250         // Search query
 251         $sQuery = $oParams->getString('q');
 252         if (!$sQuery) {
 253             $this->setStructuredQuery(
 254                 $oParams->getString('amenity'),
 255                 $oParams->getString('street'),
 256                 $oParams->getString('city'),
 257                 $oParams->getString('county'),
 258                 $oParams->getString('state'),
 259                 $oParams->getString('country'),
 260                 $oParams->getString('postalcode')
 261             );
 262         } else {
 263             $this->setQuery($sQuery);
 264         }
 265     }
 266
 267     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 268     {
 269         $sValue = trim($sValue);
 270         if (!$sValue) {
 271             return false;
 272         }
 273         $this->aStructuredQuery[$sKey] = $sValue;
 274         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 275             $this->iMinAddressRank = $iNewMinAddressRank;
 276             $this->iMaxAddressRank = $iNewMaxAddressRank;
 277         }
 278         if ($aItemListValues) {
 279             $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 280         }
 281         return true;
 282     }
 283
 284     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 285     {
 286         $this->sQuery = false;
 287
 288         // Reset
 289         $this->iMinAddressRank = 0;
 290         $this->iMaxAddressRank = 30;
 291         $this->aAddressRankList = array();
 292
 293         $this->aStructuredQuery = array();
 294         $this->sAllowedTypesSQLList = false;
 295
 296         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 297         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 298         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 299         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 300         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 301         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 302         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 303
 304         if (!empty($this->aStructuredQuery)) {
 305             $this->sQuery = join(', ', $this->aStructuredQuery);
 306             if ($this->iMaxAddressRank < 30) {
 307                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 308             }
 309         }
 310     }
 311
 312     public function fallbackStructuredQuery()
 313     {
 314         $aParams = $this->aStructuredQuery;
 315
 316         if (!$aParams || count($aParams) == 1) {
 317             return false;
 318         }
 319
 320         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 321
 322         foreach ($aOrderToFallback as $sType) {
 323             if (isset($aParams[$sType])) {
 324                 unset($aParams[$sType]);
 325                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 326                 return true;
 327             }
 328         }
 329
 330         return false;
 331     }
 332
 333     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens)
 334     {
 335         /*
 336              Calculate all searches using oValidTokens i.e.
 337              'Wodsworth Road, Sheffield' =>
 338
 339              Phrase Wordset
 340              0      0       (wodsworth road)
 341              0      1       (wodsworth)(road)
 342              1      0       (sheffield)
 343
 344              Score how good the search is so they can be ordered
 345          */
 346         foreach ($aPhrases as $iPhrase => $oPhrase) {
 347             $aNewPhraseSearches = array();
 348             $sPhraseType = $oPhrase->getPhraseType();
 349
 350             foreach ($oPhrase->getWordSets() as $aWordset) {
 351                 $aWordsetSearches = $aSearches;
 352
 353                 // Add all words from this wordset
 354                 foreach ($aWordset as $iToken => $sToken) {
 355                     //echo "<br><b>$sToken</b>";
 356                     $aNewWordsetSearches = array();
 357
 358                     foreach ($aWordsetSearches as $oCurrentSearch) {
 359                         //echo "<i>";
 360                         //var_dump($oCurrentSearch);
 361                         //echo "</i>";
 362
 363                         // Tokens with full name matches.
 364                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 365                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 366                                 $oSearchTerm,
 367                                 $sPhraseType,
 368                                 $iToken == 0 && $iPhrase == 0,
 369                                 $iPhrase == 0,
 370                                 $iToken + 1 == count($aWordset)
 371                                   && $iPhrase + 1 == count($aPhrases)
 372                             );
 373
 374                             foreach ($aNewSearches as $oSearch) {
 375                                 if ($oSearch->getRank() < $this->iMaxRank) {
 376                                     $aNewWordsetSearches[] = $oSearch;
 377                                 }
 378                             }
 379                         }
 380                         // Look for partial matches.
 381                         // Note that there is no point in adding country terms here
 382                         // because country is omitted in the address.
 383                         if ($sPhraseType != 'country') {
 384                             // Allow searching for a word - but at extra cost
 385                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 386                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 387                                     $sToken,
 388                                     $oSearchTerm,
 389                                     (bool) $sPhraseType,
 390                                     $iPhrase,
 391                                     $oValidTokens->get(' '.$sToken)
 392                                 );
 393
 394                                 foreach ($aNewSearches as $oSearch) {
 395                                     if ($oSearch->getRank() < $this->iMaxRank) {
 396                                         $aNewWordsetSearches[] = $oSearch;
 397                                     }
 398                                 }
 399                             }
 400                         }
 401                     }
 402                     // Sort and cut
 403                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 404                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 405                 }
 406                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 407
 408                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 409                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 410
 411                 $aSearchHash = array();
 412                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 413                     $sHash = serialize($aSearch);
 414                     if (isset($aSearchHash[$sHash])) {
 415                         unset($aNewPhraseSearches[$iSearch]);
 416                     } else {
 417                         $aSearchHash[$sHash] = 1;
 418                     }
 419                 }
 420
 421                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 422             }
 423
 424             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 425             $aGroupedSearches = array();
 426             foreach ($aNewPhraseSearches as $aSearch) {
 427                 $iRank = $aSearch->getRank();
 428                 if ($iRank < $this->iMaxRank) {
 429                     if (!isset($aGroupedSearches[$iRank])) {
 430                         $aGroupedSearches[$iRank] = array();
 431                     }
 432                     $aGroupedSearches[$iRank][] = $aSearch;
 433                 }
 434             }
 435             ksort($aGroupedSearches);
 436
 437             $iSearchCount = 0;
 438             $aSearches = array();
 439             foreach ($aGroupedSearches as $aNewSearches) {
 440                 $iSearchCount += count($aNewSearches);
 441                 $aSearches = array_merge($aSearches, $aNewSearches);
 442                 if ($iSearchCount > 50) {
 443                     break;
 444                 }
 445             }
 446         }
 447
 448         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 449         $aGroupedSearches = array();
 450         foreach ($aSearches as $oSearch) {
 451             if (!$oSearch->isValidSearch()) {
 452                 continue;
 453             }
 454
 455             $iRank = $oSearch->getRank();
 456             if (!isset($aGroupedSearches[$iRank])) {
 457                 $aGroupedSearches[$iRank] = array();
 458             }
 459             $aGroupedSearches[$iRank][] = $oSearch;
 460         }
 461         ksort($aGroupedSearches);
 462
 463         return $aGroupedSearches;
 464     }
 465
 466     /* Perform the actual query lookup.
 467
 468         Returns an ordered list of results, each with the following fields:
 469             osm_type: type of corresponding OSM object
 470                         N - node
 471                         W - way
 472                         R - relation
 473                         P - postcode (internally computed)
 474             osm_id: id of corresponding OSM object
 475             class: general object class (corresponds to tag key of primary OSM tag)
 476             type: subclass of object (corresponds to tag value of primary OSM tag)
 477             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 478             rank_search: rank in search hierarchy
 479                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 480             rank_address: rank in address hierarchy (determines orer in address)
 481             place_id: internal key (may differ between different instances)
 482             country_code: ISO country code
 483             langaddress: localized full address
 484             placename: localized name of object
 485             ref: content of ref tag (if available)
 486             lon: longitude
 487             lat: latitude
 488             importance: importance of place based on Wikipedia link count
 489             addressimportance: cumulated importance of address elements
 490             extra_place: type of place (for admin boundaries, if there is a place tag)
 491             aBoundingBox: bounding Box
 492             label: short description of the object class/type (English only)
 493             name: full name (currently the same as langaddress)
 494             foundorder: secondary ordering for places with same importance
 495     */
 496
 497
 498     public function lookup()
 499     {
 500         Debug::newFunction('Geocode::lookup');
 501         if (!$this->sQuery && !$this->aStructuredQuery) {
 502             return array();
 503         }
 504
 505         Debug::printDebugArray('Geocode', $this);
 506
 507         $oCtx = new SearchContext();
 508
 509         if ($this->aRoutePoints) {
 510             $oCtx->setViewboxFromRoute(
 511                 $this->oDB,
 512                 $this->aRoutePoints,
 513                 $this->aRouteWidth,
 514                 $this->bBoundedSearch
 515             );
 516         } elseif ($this->aViewBox) {
 517             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 518         }
 519         if ($this->aExcludePlaceIDs) {
 520             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 521         }
 522         if ($this->aCountryCodes) {
 523             $oCtx->setCountryList($this->aCountryCodes);
 524         }
 525         $this->oTokenizer->setCountryRestriction($this->aCountryCodes);
 526
 527         Debug::newSection('Query Preprocessing');
 528
 529         $sQuery = $this->sQuery;
 530         if (!preg_match('//u', $sQuery)) {
 531             userError('Query string is not UTF-8 encoded.');
 532         }
 533
 534         // Conflicts between US state abreviations and various words for 'the' in different languages
 535         if (isset($this->aLangPrefOrder['name:en'])) {
 536             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 537             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 538             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 539         }
 540
 541         // Do we have anything that looks like a lat/lon pair?
 542         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 543
 544         if ($sQuery || $this->aStructuredQuery) {
 545             // Start with a single blank search
 546             $aSearches = array(new SearchDescription($oCtx));
 547
 548             if ($sQuery) {
 549                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 550             }
 551
 552             $sSpecialTerm = '';
 553             if ($sQuery) {
 554                 preg_match_all(
 555                     '/\\[([\\w ]*)\\]/u',
 556                     $sQuery,
 557                     $aSpecialTermsRaw,
 558                     PREG_SET_ORDER
 559                 );
 560                 if (!empty($aSpecialTermsRaw)) {
 561                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 562                 }
 563
 564                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 565                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 566                     if (!$sSpecialTerm) {
 567                         $sSpecialTerm = $aSpecialTerm[1];
 568                     }
 569                 }
 570             }
 571             if (!$sSpecialTerm && $this->aStructuredQuery
 572                 && isset($this->aStructuredQuery['amenity'])) {
 573                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 574                 unset($this->aStructuredQuery['amenity']);
 575             }
 576
 577             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 578                 $aTokens = $this->oTokenizer->tokensForSpecialTerm($sSpecialTerm);
 579
 580                 if (!empty($aTokens)) {
 581                     $aNewSearches = array();
 582                     foreach ($aSearches as $oSearch) {
 583                         foreach ($aTokens as $oToken) {
 584                             $oNewSearch = clone $oSearch;
 585                             $oNewSearch->setPoiSearch(
 586                                 $oToken->iOperator,
 587                                 $oToken->sClass,
 588                                 $oToken->sType
 589                             );
 590                             $aNewSearches[] = $oNewSearch;
 591                         }
 592                     }
 593                     $aSearches = $aNewSearches;
 594                 }
 595             }
 596
 597             // Split query into phrases
 598             // Commas are used to reduce the search space by indicating where phrases split
 599             $aPhrases = array();
 600             if ($this->aStructuredQuery) {
 601                 foreach ($this->aStructuredQuery as $iPhrase => $sPhrase) {
 602                     $aPhrases[] = new Phrase($sPhrase, $iPhrase);
 603                 }
 604             } else {
 605                 foreach (explode(',', $sQuery) as $sPhrase) {
 606                     $aPhrases[] = new Phrase($sPhrase, '');
 607                 }
 608             }
 609
 610             Debug::printDebugArray('Search context', $oCtx);
 611             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 612
 613             Debug::newSection('Tokenization');
 614             $oValidTokens = $this->oTokenizer->extractTokensFromPhrases($aPhrases);
 615
 616             if ($oValidTokens->count() > 0) {
 617                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 618
 619                 $aPhrases = array_filter($aPhrases, function ($oPhrase) {
 620                     return $oPhrase->getWordSets() !== null;
 621                 });
 622
 623                 // Any words that have failed completely?
 624                 // TODO: suggestions
 625
 626                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 627                 Debug::printDebugTable('Phrases', $aPhrases);
 628
 629                 Debug::newSection('Search candidates');
 630
 631                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 632
 633                 if (!$this->aStructuredQuery) {
 634                     // Reverse phrase array and also reverse the order of the wordsets in
 635                     // the first and final phrase. Don't bother about phrases in the middle
 636                     // because order in the address doesn't matter.
 637                     $aPhrases = array_reverse($aPhrases);
 638                     $aPhrases[0]->invertWordSets();
 639                     if (count($aPhrases) > 1) {
 640                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 641                     }
 642                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 643
 644                     foreach ($aGroupedSearches as $aSearches) {
 645                         foreach ($aSearches as $aSearch) {
 646                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 647                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 648                             }
 649                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 650                         }
 651                     }
 652
 653                     $aGroupedSearches = $aReverseGroupedSearches;
 654                     ksort($aGroupedSearches);
 655                 }
 656             } else {
 657                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 658                 $aGroupedSearches = array();
 659                 foreach ($aSearches as $aSearch) {
 660                     if ($aSearch->getRank() < $this->iMaxRank) {
 661                         if (!isset($aGroupedSearches[$aSearch->getRank()])) {
 662                             $aGroupedSearches[$aSearch->getRank()] = array();
 663                         }
 664                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 665                     }
 666                 }
 667                 ksort($aGroupedSearches);
 668             }
 669
 670             // Filter out duplicate searches
 671             $aSearchHash = array();
 672             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 673                 foreach ($aSearches as $iSearch => $aSearch) {
 674                     $sHash = serialize($aSearch);
 675                     if (isset($aSearchHash[$sHash])) {
 676                         unset($aGroupedSearches[$iGroup][$iSearch]);
 677                         if (empty($aGroupedSearches[$iGroup])) {
 678                             unset($aGroupedSearches[$iGroup]);
 679                         }
 680                     } else {
 681                         $aSearchHash[$sHash] = 1;
 682                     }
 683                 }
 684             }
 685
 686             Debug::printGroupedSearch(
 687                 $aGroupedSearches,
 688                 $oValidTokens->debugTokenByWordIdList()
 689             );
 690
 691             // Start the search process
 692             $iGroupLoop = 0;
 693             $iQueryLoop = 0;
 694             $aNextResults = array();
 695             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 696                 $iGroupLoop++;
 697                 $aResults = $aNextResults;
 698                 foreach ($aSearches as $oSearch) {
 699                     $iQueryLoop++;
 700
 701                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 702                     Debug::printGroupedSearch(
 703                         array($iGroupedRank => array($oSearch)),
 704                         $oValidTokens->debugTokenByWordIdList()
 705                     );
 706
 707                     $aNewResults = $oSearch->query(
 708                         $this->oDB,
 709                         $this->iMinAddressRank,
 710                         $this->iMaxAddressRank,
 711                         $this->iLimit
 712                     );
 713
 714                     // The same result may appear in different rounds, only
 715                     // use the one with minimal rank.
 716                     foreach ($aNewResults as $iPlace => $oRes) {
 717                         if (!isset($aResults[$iPlace])
 718                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 719                             $aResults[$iPlace] = $oRes;
 720                         }
 721                     }
 722
 723                     if ($iQueryLoop > 20) {
 724                         break;
 725                     }
 726                 }
 727
 728                 if (!empty($aResults)) {
 729                     $aSplitResults = Result::splitResults($aResults);
 730                     Debug::printVar('Split results', $aSplitResults);
 731                     if ($iGroupLoop <= 4
 732                         && reset($aSplitResults['head'])->iResultRank > 0
 733                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 734                         // Haven't found an exact match for the query yet.
 735                         // Therefore add result from the next group level.
 736                         $aNextResults = $aSplitResults['head'];
 737                         foreach ($aNextResults as $oRes) {
 738                             $oRes->iResultRank--;
 739                         }
 740                         foreach ($aSplitResults['tail'] as $oRes) {
 741                             $oRes->iResultRank--;
 742                             $aNextResults[$oRes->iId] = $oRes;
 743                         }
 744                         $aResults = array();
 745                     } else {
 746                         $aResults = $aSplitResults['head'];
 747                     }
 748                 }
 749
 750                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 751                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 752                     // reduces the number of place ids, like a filter
 753                     // rank_address is 30 for interpolated housenumbers
 754                     $aFilterSql = array();
 755                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 756                     if ($sPlaceIds) {
 757                         $sSQL = 'SELECT place_id FROM placex ';
 758                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 759                         $sSQL .= '  AND (';
 760                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 761                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 762                         if ($this->aAddressRankList) {
 763                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 764                         }
 765                         $sSQL .= ')';
 766                         $aFilterSql[] = $sSQL;
 767                     }
 768                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 769                     if ($sPlaceIds) {
 770                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 771                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 772                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 773                         if ($this->aAddressRankList) {
 774                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 775                         }
 776                         $sSQL .= ') ';
 777                         $aFilterSql[] = $sSQL;
 778                     }
 779
 780                     $aFilteredIDs = array();
 781                     if ($aFilterSql) {
 782                         $sSQL = join(' UNION ', $aFilterSql);
 783                         Debug::printSQL($sSQL);
 784                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 785                     }
 786
 787                     $tempIDs = array();
 788                     foreach ($aResults as $oResult) {
 789                         if (($this->iMaxAddressRank == 30 &&
 790                              ($oResult->iTable == Result::TABLE_OSMLINE
 791                               || $oResult->iTable == Result::TABLE_TIGER))
 792                             || in_array($oResult->iId, $aFilteredIDs)
 793                         ) {
 794                             $tempIDs[$oResult->iId] = $oResult;
 795                         }
 796                     }
 797                     $aResults = $tempIDs;
 798                 }
 799
 800                 if (!empty($aResults) || $iGroupLoop > 4 || $iQueryLoop > 30) {
 801                     break;
 802                 }
 803             }
 804         } else {
 805             // Just interpret as a reverse geocode
 806             $oReverse = new ReverseGeocode($this->oDB);
 807             $oReverse->setZoom(18);
 808
 809             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 810
 811             Debug::printVar('Reverse search', $oLookup);
 812
 813             if ($oLookup) {
 814                 $aResults = array($oLookup->iId => $oLookup);
 815             }
 816         }
 817
 818         // No results? Done
 819         if (empty($aResults)) {
 820             if ($this->bFallback) {
 821                 if ($this->fallbackStructuredQuery()) {
 822                     return $this->lookup();
 823                 }
 824             }
 825
 826             return array();
 827         }
 828
 829         if ($this->aAddressRankList) {
 830             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 831         }
 832         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 833         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 834         if ($oCtx->hasNearPoint()) {
 835             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 836         }
 837
 838         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 839
 840         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 841         foreach ($aRecheckWords as $i => $sWord) {
 842             if (!preg_match('/[\pL\pN]/', $sWord)) {
 843                 unset($aRecheckWords[$i]);
 844             }
 845         }
 846
 847         Debug::printVar('Recheck words', $aRecheckWords);
 848
 849         foreach ($aSearchResults as $iIdx => $aResult) {
 850             $fRadius = ClassTypes\getDefRadius($aResult);
 851
 852             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 853             if ($aOutlineResult) {
 854                 $aResult = array_merge($aResult, $aOutlineResult);
 855             }
 856
 857             // Is there an icon set for this type of result?
 858             $sIcon = ClassTypes\getIconFile($aResult);
 859             if (isset($sIcon)) {
 860                 $aResult['icon'] = $sIcon;
 861             }
 862
 863             $sLabel = ClassTypes\getLabel($aResult);
 864             if (isset($sLabel)) {
 865                 $aResult['label'] = $sLabel;
 866             }
 867             $aResult['name'] = $aResult['langaddress'];
 868
 869             if ($oCtx->hasNearPoint()) {
 870                 $aResult['importance'] = 0.001;
 871                 $aResult['foundorder'] = $aResult['addressimportance'];
 872             } else {
 873                 $aResult['importance'] = max(0.001, $aResult['importance']);
 874                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 875                     $aResult['lon'],
 876                     $aResult['lat']
 877                 );
 878
 879                 // secondary ordering (for results with same importance (the smaller the better):
 880                 // - approximate importance of address parts
 881                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 882                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 883                 } else {
 884                     $aResult['foundorder'] = -$aResult['importance'];
 885                 }
 886                 // - number of exact matches from the query
 887                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 888                 // - importance of the class/type
 889                 $iClassImportance = ClassTypes\getImportance($aResult);
 890                 if (isset($iClassImportance)) {
 891                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 892                 } else {
 893                     $aResult['foundorder'] += 0.01;
 894                 }
 895                 // - rank
 896                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 897
 898                 // Adjust importance for the number of exact string matches in the result
 899                 $iCountWords = 0;
 900                 $sAddress = $aResult['langaddress'];
 901                 foreach ($aRecheckWords as $i => $sWord) {
 902                     if (stripos($sAddress, $sWord)!==false) {
 903                         $iCountWords++;
 904                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) {
 905                             $iCountWords += 0.1;
 906                         }
 907                     }
 908                 }
 909
 910                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 911                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 912             }
 913             $aSearchResults[$iIdx] = $aResult;
 914         }
 915         uasort($aSearchResults, 'byImportance');
 916         Debug::printVar('Pre-filter results', $aSearchResults);
 917
 918         $aOSMIDDone = array();
 919         $aClassTypeNameDone = array();
 920         $aToFilter = $aSearchResults;
 921         $aSearchResults = array();
 922
 923         foreach ($aToFilter as $aResult) {
 924             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 925             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 926                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 927             ) {
 928                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 929                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 930                 $aSearchResults[] = $aResult;
 931             }
 932
 933             // Absolute limit on number of results
 934             if (count($aSearchResults) >= $this->iFinalLimit) {
 935                 break;
 936             }
 937         }
 938
 939         Debug::printVar('Post-filter results', $aSearchResults);
 940         return $aSearchResults;
 941     } // end lookup()
 942
 943     public function debugInfo()
 944     {
 945         return array(
 946                 'Query' => $this->sQuery,
 947                 'Structured query' => $this->aStructuredQuery,
 948                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 949                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 950                 'Limit (for searches)' => $this->iLimit,
 951                 'Limit (for results)'=> $this->iFinalLimit,
 952                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 953                 'Bounded search' => $this->bBoundedSearch,
 954                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 955                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 956                 'Route width' => $this->aRouteWidth,
 957                 'Max rank' => $this->iMaxRank,
 958                 'Min address rank' => $this->iMinAddressRank,
 959                 'Max address rank' => $this->iMaxAddressRank,
 960                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 961                );
 962     }
 963 } // end class