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