lib/Geocode.php

   1 <?php
   2
   3 namespace Nominatim;
   4
   5 require_once(CONST_BasePath.'/lib/PlaceLookup.php');
   6 require_once(CONST_BasePath.'/lib/Phrase.php');
   7 require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
   8 require_once(CONST_BasePath.'/lib/SearchDescription.php');
   9 require_once(CONST_BasePath.'/lib/SearchContext.php');
  10 require_once(CONST_BasePath.'/lib/TokenList.php');
  11
  12 class Geocode
  13 {
  14     protected $oDB;
  15
  16     protected $oPlaceLookup;
  17
  18     protected $aLangPrefOrder = array();
  19
  20     protected $aExcludePlaceIDs = array();
  21     protected $bReverseInPlan = false;
  22
  23     protected $iLimit = 20;
  24     protected $iFinalLimit = 10;
  25     protected $iOffset = 0;
  26     protected $bFallback = false;
  27
  28     protected $aCountryCodes = false;
  29
  30     protected $bBoundedSearch = false;
  31     protected $aViewBox = false;
  32     protected $aRoutePoints = false;
  33     protected $aRouteWidth = false;
  34
  35     protected $iMaxRank = 20;
  36     protected $iMinAddressRank = 0;
  37     protected $iMaxAddressRank = 30;
  38     protected $aAddressRankList = array();
  39
  40     protected $sAllowedTypesSQLList = false;
  41
  42     protected $sQuery = false;
  43     protected $aStructuredQuery = false;
  44
  45     protected $oNormalizer = null;
  46
  47
  48     public function __construct(&$oDB)
  49     {
  50         $this->oDB =& $oDB;
  51         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  52         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  53     }
  54
  55     private function normTerm($sTerm)
  56     {
  57         if ($this->oNormalizer === null) {
  58             return $sTerm;
  59         }
  60
  61         return $this->oNormalizer->transliterate($sTerm);
  62     }
  63
  64     public function setReverseInPlan($bReverse)
  65     {
  66         $this->bReverseInPlan = $bReverse;
  67     }
  68
  69     public function setLanguagePreference($aLangPref)
  70     {
  71         $this->aLangPrefOrder = $aLangPref;
  72     }
  73
  74     public function getMoreUrlParams()
  75     {
  76         if ($this->aStructuredQuery) {
  77             $aParams = $this->aStructuredQuery;
  78         } else {
  79             $aParams = array('q' => $this->sQuery);
  80         }
  81
  82         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  83
  84         if ($this->aExcludePlaceIDs) {
  85             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  86         }
  87
  88         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  89
  90         if ($this->aCountryCodes) {
  91             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  92         }
  93
  94         if ($this->aViewBox) {
  95             $aParams['viewbox'] = join(',', $this->aViewBox);
  96         }
  97
  98         return $aParams;
  99     }
 100
 101     public function setLimit($iLimit = 10)
 102     {
 103         if ($iLimit > 50) $iLimit = 50;
 104         if ($iLimit < 1) $iLimit = 1;
 105
 106         $this->iFinalLimit = $iLimit;
 107         $this->iLimit = $iLimit + min($iLimit, 10);
 108     }
 109
 110     public function setFeatureType($sFeatureType)
 111     {
 112         switch ($sFeatureType) {
 113             case 'country':
 114                 $this->setRankRange(4, 4);
 115                 break;
 116             case 'state':
 117                 $this->setRankRange(8, 8);
 118                 break;
 119             case 'city':
 120                 $this->setRankRange(14, 16);
 121                 break;
 122             case 'settlement':
 123                 $this->setRankRange(8, 20);
 124                 break;
 125         }
 126     }
 127
 128     public function setRankRange($iMin, $iMax)
 129     {
 130         $this->iMinAddressRank = $iMin;
 131         $this->iMaxAddressRank = $iMax;
 132     }
 133
 134     public function setViewbox($aViewbox)
 135     {
 136         $aBox = array_map('floatval', $aViewbox);
 137
 138         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 139         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 140         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 141         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 142
 143         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 144             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 145         ) {
 146             userError("Bad parameter 'viewbox'. Not a box.");
 147         }
 148     }
 149
 150     private function viewboxImportanceFactor($fX, $fY)
 151     {
 152         if (!$this->aViewBox) {
 153             return 1;
 154         }
 155
 156         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 157         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 158
 159         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 160         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 161
 162         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 163             return 1;
 164         }
 165
 166         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 167             return 0.5;
 168         }
 169
 170         return 0.25;
 171     }
 172
 173     public function setQuery($sQueryString)
 174     {
 175         $this->sQuery = $sQueryString;
 176         $this->aStructuredQuery = false;
 177     }
 178
 179     public function getQueryString()
 180     {
 181         return $this->sQuery;
 182     }
 183
 184
 185     public function loadParamArray($oParams, $sForceGeometryType = null)
 186     {
 187         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 188
 189         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 190         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 191
 192         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 193
 194         // List of excluded Place IDs - used for more acurate pageing
 195         $sExcluded = $oParams->getStringList('exclude_place_ids');
 196         if ($sExcluded) {
 197             foreach ($sExcluded as $iExcludedPlaceID) {
 198                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 199                 if ($iExcludedPlaceID)
 200                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 201             }
 202
 203             if (isset($aExcludePlaceIDs))
 204                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 205         }
 206
 207         // Only certain ranks of feature
 208         $sFeatureType = $oParams->getString('featureType');
 209         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 210         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 211
 212         // Country code list
 213         $sCountries = $oParams->getStringList('countrycodes');
 214         if ($sCountries) {
 215             foreach ($sCountries as $sCountryCode) {
 216                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 217                     $aCountries[] = strtolower($sCountryCode);
 218                 }
 219             }
 220             if (isset($aCountries))
 221                 $this->aCountryCodes = $aCountries;
 222         }
 223
 224         $aViewbox = $oParams->getStringList('viewboxlbrt');
 225         if ($aViewbox) {
 226             if (count($aViewbox) != 4) {
 227                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 228             }
 229             $this->setViewbox($aViewbox);
 230         } else {
 231             $aViewbox = $oParams->getStringList('viewbox');
 232             if ($aViewbox) {
 233                 if (count($aViewbox) != 4) {
 234                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 235                 }
 236                 $this->setViewBox($aViewbox);
 237             } else {
 238                 $aRoute = $oParams->getStringList('route');
 239                 $fRouteWidth = $oParams->getFloat('routewidth');
 240                 if ($aRoute && $fRouteWidth) {
 241                     $this->aRoutePoints = $aRoute;
 242                     $this->aRouteWidth = $fRouteWidth;
 243                 }
 244             }
 245         }
 246
 247         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 248         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 249         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 250     }
 251
 252     public function setQueryFromParams($oParams)
 253     {
 254         // Search query
 255         $sQuery = $oParams->getString('q');
 256         if (!$sQuery) {
 257             $this->setStructuredQuery(
 258                 $oParams->getString('amenity'),
 259                 $oParams->getString('street'),
 260                 $oParams->getString('city'),
 261                 $oParams->getString('county'),
 262                 $oParams->getString('state'),
 263                 $oParams->getString('country'),
 264                 $oParams->getString('postalcode')
 265             );
 266             $this->setReverseInPlan(false);
 267         } else {
 268             $this->setQuery($sQuery);
 269         }
 270     }
 271
 272     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 273     {
 274         $sValue = trim($sValue);
 275         if (!$sValue) return false;
 276         $this->aStructuredQuery[$sKey] = $sValue;
 277         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 278             $this->iMinAddressRank = $iNewMinAddressRank;
 279             $this->iMaxAddressRank = $iNewMaxAddressRank;
 280         }
 281         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 282         return true;
 283     }
 284
 285     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 286     {
 287         $this->sQuery = false;
 288
 289         // Reset
 290         $this->iMinAddressRank = 0;
 291         $this->iMaxAddressRank = 30;
 292         $this->aAddressRankList = array();
 293
 294         $this->aStructuredQuery = array();
 295         $this->sAllowedTypesSQLList = false;
 296
 297         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 298         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 299         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 300         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 301         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 302         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 303         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 304
 305         if (!empty($this->aStructuredQuery)) {
 306             $this->sQuery = join(', ', $this->aStructuredQuery);
 307             if ($this->iMaxAddressRank < 30) {
 308                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 309             }
 310         }
 311     }
 312
 313     public function fallbackStructuredQuery()
 314     {
 315         if (!$this->aStructuredQuery) return false;
 316
 317         $aParams = $this->aStructuredQuery;
 318
 319         if (count($aParams) == 1) return false;
 320
 321         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 322
 323         foreach ($aOrderToFallback as $sType) {
 324             if (isset($aParams[$sType])) {
 325                 unset($aParams[$sType]);
 326                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 327                 return true;
 328             }
 329         }
 330
 331         return false;
 332     }
 333
 334     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
 335     {
 336         /*
 337              Calculate all searches using oValidTokens i.e.
 338              'Wodsworth Road, Sheffield' =>
 339
 340              Phrase Wordset
 341              0      0       (wodsworth road)
 342              0      1       (wodsworth)(road)
 343              1      0       (sheffield)
 344
 345              Score how good the search is so they can be ordered
 346          */
 347         foreach ($aPhrases as $iPhrase => $oPhrase) {
 348             $aNewPhraseSearches = array();
 349             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 350
 351             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 352                 // Too many permutations - too expensive
 353                 if ($iWordSet > 120) break;
 354
 355                 $aWordsetSearches = $aSearches;
 356
 357                 // Add all words from this wordset
 358                 foreach ($aWordset as $iToken => $sToken) {
 359                     //echo "<br><b>$sToken</b>";
 360                     $aNewWordsetSearches = array();
 361
 362                     foreach ($aWordsetSearches as $oCurrentSearch) {
 363                         //echo "<i>";
 364                         //var_dump($oCurrentSearch);
 365                         //echo "</i>";
 366
 367                         // Tokens with full name matches.
 368                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 369                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 370                                 $oSearchTerm,
 371                                 $oValidTokens->contains($sToken)
 372                                   && strpos($sToken, ' ') === false,
 373                                 $sPhraseType,
 374                                 $iToken == 0 && $iPhrase == 0,
 375                                 $iPhrase == 0,
 376                                 $iToken + 1 == count($aWordset)
 377                                   && $iPhrase + 1 == count($aPhrases)
 378                             );
 379
 380                             foreach ($aNewSearches as $oSearch) {
 381                                 if ($oSearch->getRank() < $this->iMaxRank) {
 382                                     $aNewWordsetSearches[] = $oSearch;
 383                                 }
 384                             }
 385                         }
 386                         // Look for partial matches.
 387                         // Note that there is no point in adding country terms here
 388                         // because country is omitted in the address.
 389                         if ($sPhraseType != 'country') {
 390                             // Allow searching for a word - but at extra cost
 391                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 392                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 393                                     $sToken,
 394                                     $oSearchTerm,
 395                                     $bIsStructured,
 396                                     $iPhrase,
 397                                     $oValidTokens->get(' '.$sToken)
 398                                 );
 399
 400                                 foreach ($aNewSearches as $oSearch) {
 401                                     if ($oSearch->getRank() < $this->iMaxRank) {
 402                                         $aNewWordsetSearches[] = $oSearch;
 403                                     }
 404                                 }
 405                             }
 406                         }
 407                     }
 408                     // Sort and cut
 409                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 410                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 411                 }
 412                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 413
 414                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 415                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 416
 417                 $aSearchHash = array();
 418                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 419                     $sHash = serialize($aSearch);
 420                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 421                     else $aSearchHash[$sHash] = 1;
 422                 }
 423
 424                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 425             }
 426
 427             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 428             $aGroupedSearches = array();
 429             foreach ($aNewPhraseSearches as $aSearch) {
 430                 $iRank = $aSearch->getRank();
 431                 if ($iRank < $this->iMaxRank) {
 432                     if (!isset($aGroupedSearches[$iRank])) {
 433                         $aGroupedSearches[$iRank] = array();
 434                     }
 435                     $aGroupedSearches[$iRank][] = $aSearch;
 436                 }
 437             }
 438             ksort($aGroupedSearches);
 439
 440             $iSearchCount = 0;
 441             $aSearches = array();
 442             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 443                 $iSearchCount += count($aNewSearches);
 444                 $aSearches = array_merge($aSearches, $aNewSearches);
 445                 if ($iSearchCount > 50) break;
 446             }
 447         }
 448
 449         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 450         $aGroupedSearches = array();
 451         foreach ($aSearches as $oSearch) {
 452             if (!$oSearch->isValidSearch()) {
 453                 continue;
 454             }
 455
 456             $iRank = $oSearch->getRank();
 457             if (!isset($aGroupedSearches[$iRank])) {
 458                 $aGroupedSearches[$iRank] = array();
 459             }
 460             $aGroupedSearches[$iRank][] = $oSearch;
 461         }
 462         ksort($aGroupedSearches);
 463
 464         return $aGroupedSearches;
 465     }
 466
 467     /* Perform the actual query lookup.
 468
 469         Returns an ordered list of results, each with the following fields:
 470             osm_type: type of corresponding OSM object
 471                         N - node
 472                         W - way
 473                         R - relation
 474                         P - postcode (internally computed)
 475             osm_id: id of corresponding OSM object
 476             class: general object class (corresponds to tag key of primary OSM tag)
 477             type: subclass of object (corresponds to tag value of primary OSM tag)
 478             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 479             rank_search: rank in search hierarchy
 480                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 481             rank_address: rank in address hierarchy (determines orer in address)
 482             place_id: internal key (may differ between different instances)
 483             country_code: ISO country code
 484             langaddress: localized full address
 485             placename: localized name of object
 486             ref: content of ref tag (if available)
 487             lon: longitude
 488             lat: latitude
 489             importance: importance of place based on Wikipedia link count
 490             addressimportance: cumulated importance of address elements
 491             extra_place: type of place (for admin boundaries, if there is a place tag)
 492             aBoundingBox: bounding Box
 493             label: short description of the object class/type (English only)
 494             name: full name (currently the same as langaddress)
 495             foundorder: secondary ordering for places with same importance
 496     */
 497
 498
 499     public function lookup()
 500     {
 501         Debug::newFunction('Geocode::lookup');
 502         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 503
 504         Debug::printDebugArray('Geocode', $this);
 505
 506         $oCtx = new SearchContext();
 507
 508         if ($this->aRoutePoints) {
 509             $oCtx->setViewboxFromRoute(
 510                 $this->oDB,
 511                 $this->aRoutePoints,
 512                 $this->aRouteWidth,
 513                 $this->bBoundedSearch
 514             );
 515         } elseif ($this->aViewBox) {
 516             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 517         }
 518         if ($this->aExcludePlaceIDs) {
 519             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 520         }
 521         if ($this->aCountryCodes) {
 522             $oCtx->setCountryList($this->aCountryCodes);
 523         }
 524
 525         Debug::newSection('Query Preprocessing');
 526
 527         $sNormQuery = $this->normTerm($this->sQuery);
 528         Debug::printVar('Normalized query', $sNormQuery);
 529
 530         $sLanguagePrefArraySQL = getArraySQL(
 531             array_map('getDBQuoted', $this->aLangPrefOrder)
 532         );
 533
 534         $sQuery = $this->sQuery;
 535         if (!preg_match('//u', $sQuery)) {
 536             userError('Query string is not UTF-8 encoded.');
 537         }
 538
 539         // Conflicts between US state abreviations and various words for 'the' in different languages
 540         if (isset($this->aLangPrefOrder['name:en'])) {
 541             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 542             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 543             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 544         }
 545
 546         // Do we have anything that looks like a lat/lon pair?
 547         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 548
 549         $aResults = array();
 550         if ($sQuery || $this->aStructuredQuery) {
 551             // Start with a single blank search
 552             $aSearches = array(new SearchDescription($oCtx));
 553
 554             if ($sQuery) {
 555                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 556             }
 557
 558             $sSpecialTerm = '';
 559             if ($sQuery) {
 560                 preg_match_all(
 561                     '/\\[([\\w ]*)\\]/u',
 562                     $sQuery,
 563                     $aSpecialTermsRaw,
 564                     PREG_SET_ORDER
 565                 );
 566                 if (!empty($aSpecialTermsRaw)) {
 567                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 568                 }
 569
 570                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 571                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 572                     if (!$sSpecialTerm) {
 573                         $sSpecialTerm = $aSpecialTerm[1];
 574                     }
 575                 }
 576             }
 577             if (!$sSpecialTerm && $this->aStructuredQuery
 578                 && isset($this->aStructuredQuery['amenity'])) {
 579                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 580                 unset($this->aStructuredQuery['amenity']);
 581             }
 582
 583             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 584                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 585                 $sToken = chksql(
 586                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 587                     'Cannot decode query. Wrong encoding?'
 588                 );
 589                 $sSQL = 'SELECT class, type FROM word ';
 590                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 591                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 592
 593                 Debug::printSQL($sSQL);
 594                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 595                 $aNewSearches = array();
 596                 foreach ($aSearches as $oSearch) {
 597                     foreach ($aSearchWords as $aSearchTerm) {
 598                         $oNewSearch = clone $oSearch;
 599                         $oNewSearch->setPoiSearch(
 600                             Operator::TYPE,
 601                             $aSearchTerm['class'],
 602                             $aSearchTerm['type']
 603                         );
 604                         $aNewSearches[] = $oNewSearch;
 605                     }
 606                 }
 607                 $aSearches = $aNewSearches;
 608             }
 609
 610             // Split query into phrases
 611             // Commas are used to reduce the search space by indicating where phrases split
 612             if ($this->aStructuredQuery) {
 613                 $aInPhrases = $this->aStructuredQuery;
 614                 $bStructuredPhrases = true;
 615             } else {
 616                 $aInPhrases = explode(',', $sQuery);
 617                 $bStructuredPhrases = false;
 618             }
 619
 620             Debug::printDebugArray('Search context', $oCtx);
 621             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 622             Debug::printVar('Final query phrases', $aInPhrases);
 623
 624             // Convert each phrase to standard form
 625             // Create a list of standard words
 626             // Get all 'sets' of words
 627             // Generate a complete list of all
 628             Debug::newSection('Tokenization');
 629             $aTokens = array();
 630             $aPhrases = array();
 631             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 632                 $sPhrase = chksql(
 633                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 634                     'Cannot normalize query string (is it a UTF-8 string?)'
 635                 );
 636                 if (trim($sPhrase)) {
 637                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 638                     $oPhrase->addTokens($aTokens);
 639                     $aPhrases[] = $oPhrase;
 640                 }
 641             }
 642
 643             Debug::printDebugTable('Phrases', $aPhrases);
 644             Debug::printVar('Tokens', $aTokens);
 645
 646             $oValidTokens = new TokenList();
 647
 648             if (!empty($aTokens)) {
 649                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 650                 $sSQL .= ' FROM word ';
 651                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 652
 653                 Debug::printSQL($sSQL);
 654
 655                 $oValidTokens->addTokensFromDB(
 656                     $this->oDB,
 657                     $aTokens,
 658                     $this->aCountryCodes,
 659                     $sNormQuery,
 660                     $this->oNormalizer
 661                 );
 662
 663                 // Try more interpretations for Tokens that could not be matched.
 664                 foreach ($aTokens as $sToken) {
 665                     if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
 666                         if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 667                             // US ZIP+4 codes - merge in the 5-digit ZIP code
 668                             $oValidTokens->addToken(
 669                                 $sToken,
 670                                 new Token\Postcode(null, $aData[1], 'us')
 671                             );
 672                         } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
 673                             // Unknown single word token with a number.
 674                             // Assume it is a house number.
 675                             $oValidTokens->addToken(
 676                                 $sToken,
 677                                 new Token\HouseNumber(null, trim($sToken))
 678                             );
 679                         }
 680                     }
 681                 }
 682
 683                 // Any words that have failed completely?
 684                 // TODO: suggestions
 685
 686                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 687
 688                 Debug::newSection('Search candidates');
 689
 690                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
 691
 692                 if ($this->bReverseInPlan) {
 693                     // Reverse phrase array and also reverse the order of the wordsets in
 694                     // the first and final phrase. Don't bother about phrases in the middle
 695                     // because order in the address doesn't matter.
 696                     $aPhrases = array_reverse($aPhrases);
 697                     $aPhrases[0]->invertWordSets();
 698                     if (count($aPhrases) > 1) {
 699                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 700                     }
 701                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
 702
 703                     foreach ($aGroupedSearches as $aSearches) {
 704                         foreach ($aSearches as $aSearch) {
 705                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 706                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 707                             }
 708                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 709                         }
 710                     }
 711
 712                     $aGroupedSearches = $aReverseGroupedSearches;
 713                     ksort($aGroupedSearches);
 714                 }
 715             } else {
 716                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 717                 $aGroupedSearches = array();
 718                 foreach ($aSearches as $aSearch) {
 719                     if ($aSearch->getRank() < $this->iMaxRank) {
 720                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 721                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 722                     }
 723                 }
 724                 ksort($aGroupedSearches);
 725             }
 726
 727             // Filter out duplicate searches
 728             $aSearchHash = array();
 729             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 730                 foreach ($aSearches as $iSearch => $aSearch) {
 731                     $sHash = serialize($aSearch);
 732                     if (isset($aSearchHash[$sHash])) {
 733                         unset($aGroupedSearches[$iGroup][$iSearch]);
 734                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 735                     } else {
 736                         $aSearchHash[$sHash] = 1;
 737                     }
 738                 }
 739             }
 740
 741             Debug::printGroupedSearch(
 742                 $aGroupedSearches,
 743                 $oValidTokens->debugTokenByWordIdList()
 744             );
 745
 746             // Start the search process
 747             $iGroupLoop = 0;
 748             $iQueryLoop = 0;
 749             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 750                 $iGroupLoop++;
 751                 foreach ($aSearches as $oSearch) {
 752                     $iQueryLoop++;
 753
 754                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 755                     Debug::printGroupedSearch(
 756                         array($iGroupedRank => array($oSearch)),
 757                         $oValidTokens->debugTokenByWordIdList()
 758                     );
 759
 760                     $aResults += $oSearch->query(
 761                         $this->oDB,
 762                         $this->iMinAddressRank,
 763                         $this->iMaxAddressRank,
 764                         $this->iLimit
 765                     );
 766
 767                     if ($iQueryLoop > 20) break;
 768                 }
 769
 770                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 771                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 772                     // reduces the number of place ids, like a filter
 773                     // rank_address is 30 for interpolated housenumbers
 774                     $aFilterSql = array();
 775                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 776                     if ($sPlaceIds) {
 777                         $sSQL = 'SELECT place_id FROM placex ';
 778                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 779                         $sSQL .= '  AND (';
 780                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 781                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 782                             $sSQL .= "     OR (extratags->'place') = 'city'";
 783                         }
 784                         if ($this->aAddressRankList) {
 785                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 786                         }
 787                         $sSQL .= ')';
 788                         $aFilterSql[] = $sSQL;
 789                     }
 790                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 791                     if ($sPlaceIds) {
 792                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 793                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 794                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 795                         if ($this->aAddressRankList) {
 796                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 797                         }
 798                         $sSQL .= ') ';
 799                         $aFilterSql[] = $sSQL;
 800                     }
 801
 802                     $aFilteredIDs = array();
 803                     if ($aFilterSql) {
 804                         $sSQL = join(' UNION ', $aFilterSql);
 805                         Debug::printSQL($sSQL);
 806                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 807                     }
 808
 809                     $tempIDs = array();
 810                     foreach ($aResults as $oResult) {
 811                         if (($this->iMaxAddressRank == 30 &&
 812                              ($oResult->iTable == Result::TABLE_OSMLINE
 813                               || $oResult->iTable == Result::TABLE_AUX
 814                               || $oResult->iTable == Result::TABLE_TIGER))
 815                             || in_array($oResult->iId, $aFilteredIDs)
 816                         ) {
 817                             $tempIDs[$oResult->iId] = $oResult;
 818                         }
 819                     }
 820                     $aResults = $tempIDs;
 821                 }
 822
 823                 if (!empty($aResults)) break;
 824                 if ($iGroupLoop > 4) break;
 825                 if ($iQueryLoop > 30) break;
 826             }
 827         } else {
 828             // Just interpret as a reverse geocode
 829             $oReverse = new ReverseGeocode($this->oDB);
 830             $oReverse->setZoom(18);
 831
 832             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 833
 834             Debug::printVar('Reverse search', $oLookup);
 835
 836             if ($oLookup) {
 837                 $aResults = array($oLookup->iId => $oLookup);
 838             }
 839         }
 840
 841         // No results? Done
 842         if (empty($aResults)) {
 843             if ($this->bFallback) {
 844                 if ($this->fallbackStructuredQuery()) {
 845                     return $this->lookup();
 846                 }
 847             }
 848
 849             return array();
 850         }
 851
 852         if ($this->aAddressRankList) {
 853             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 854         }
 855         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 856         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 857         if ($oCtx->hasNearPoint()) {
 858             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 859         }
 860
 861         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 862
 863         $aClassType = ClassTypes\getListWithImportance();
 864         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 865         foreach ($aRecheckWords as $i => $sWord) {
 866             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 867         }
 868
 869         Debug::printVar('Recheck words', $aRecheckWords);
 870
 871         foreach ($aSearchResults as $iIdx => $aResult) {
 872             // Default
 873             $fDiameter = ClassTypes\getProperty($aResult, 'defdiameter', 0.0001);
 874
 875             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 876             if ($aOutlineResult) {
 877                 $aResult = array_merge($aResult, $aOutlineResult);
 878             }
 879
 880             if ($aResult['extra_place'] == 'city') {
 881                 $aResult['class'] = 'place';
 882                 $aResult['type'] = 'city';
 883                 $aResult['rank_search'] = 16;
 884             }
 885
 886             // Is there an icon set for this type of result?
 887             $aClassInfo = ClassTypes\getInfo($aResult);
 888
 889             if ($aClassInfo) {
 890                 if (isset($aClassInfo['icon'])) {
 891                     $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassInfo['icon'].'.p.20.png';
 892                 }
 893
 894                 if (isset($aClassInfo['label'])) {
 895                     $aResult['label'] = $aClassInfo['label'];
 896                 }
 897             }
 898
 899             $aResult['name'] = $aResult['langaddress'];
 900
 901             if ($oCtx->hasNearPoint()) {
 902                 $aResult['importance'] = 0.001;
 903                 $aResult['foundorder'] = $aResult['addressimportance'];
 904             } else {
 905                 $aResult['importance'] = max(0.001, $aResult['importance']);
 906                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 907                     $aResult['lon'],
 908                     $aResult['lat']
 909                 );
 910                 // Adjust importance for the number of exact string matches in the result
 911                 $iCountWords = 0;
 912                 $sAddress = $aResult['langaddress'];
 913                 foreach ($aRecheckWords as $i => $sWord) {
 914                     if (stripos($sAddress, $sWord)!==false) {
 915                         $iCountWords++;
 916                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 917                     }
 918                 }
 919
 920                 $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
 921
 922                 // secondary ordering (for results with same importance (the smaller the better):
 923                 // - approximate importance of address parts
 924                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
 925                 // - number of exact matches from the query
 926                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 927                 // - importance of the class/type
 928                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
 929                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
 930                 ) {
 931                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
 932                 } else {
 933                     $aResult['foundorder'] += 0.01;
 934                 }
 935             }
 936             $aSearchResults[$iIdx] = $aResult;
 937         }
 938         uasort($aSearchResults, 'byImportance');
 939         Debug::printVar('Pre-filter results', $aSearchResults);
 940
 941         $aOSMIDDone = array();
 942         $aClassTypeNameDone = array();
 943         $aToFilter = $aSearchResults;
 944         $aSearchResults = array();
 945
 946         $bFirst = true;
 947         foreach ($aToFilter as $aResult) {
 948             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 949             if ($bFirst) {
 950                 $fLat = $aResult['lat'];
 951                 $fLon = $aResult['lon'];
 952                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 953                 $bFirst = false;
 954             }
 955             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 956                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 957             ) {
 958                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 959                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 960                 $aSearchResults[] = $aResult;
 961             }
 962
 963             // Absolute limit on number of results
 964             if (count($aSearchResults) >= $this->iFinalLimit) break;
 965         }
 966
 967         Debug::printVar('Post-filter results', $aSearchResults);
 968         return $aSearchResults;
 969     } // end lookup()
 970
 971     public function debugInfo()
 972     {
 973         return array(
 974                 'Query' => $this->sQuery,
 975                 'Structured query' => $this->aStructuredQuery,
 976                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 977                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 978                 'Try reversed query'=> $this->bReverseInPlan,
 979                 'Limit (for searches)' => $this->iLimit,
 980                 'Limit (for results)'=> $this->iFinalLimit,
 981                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 982                 'Bounded search' => $this->bBoundedSearch,
 983                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
 984                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
 985                 'Route width' => $this->aRouteWidth,
 986                 'Max rank' => $this->iMaxRank,
 987                 'Min address rank' => $this->iMinAddressRank,
 988                 'Max address rank' => $this->iMaxAddressRank,
 989                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
 990                );
 991     }
 992 } // end class