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