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
  12 class Geocode
  13 {
  14     protected $oDB;
  15
  16     protected $oPlaceLookup;
  17
  18     protected $aLangPrefOrder = array();
  19
  20     protected $aExcludePlaceIDs = array();
  21
  22     protected $iLimit = 20;
  23     protected $iFinalLimit = 10;
  24     protected $iOffset = 0;
  25     protected $bFallback = false;
  26
  27     protected $aCountryCodes = false;
  28
  29     protected $bBoundedSearch = false;
  30     protected $aViewBox = false;
  31     protected $aRoutePoints = false;
  32     protected $aRouteWidth = false;
  33
  34     protected $iMaxRank = 20;
  35     protected $iMinAddressRank = 0;
  36     protected $iMaxAddressRank = 30;
  37     protected $aAddressRankList = array();
  38
  39     protected $sAllowedTypesSQLList = false;
  40
  41     protected $sQuery = false;
  42     protected $aStructuredQuery = false;
  43
  44     protected $oNormalizer = null;
  45
  46
  47     public function __construct(&$oDB)
  48     {
  49         $this->oDB =& $oDB;
  50         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  51         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  52     }
  53
  54     private function normTerm($sTerm)
  55     {
  56         if ($this->oNormalizer === null) {
  57             return $sTerm;
  58         }
  59
  60         return $this->oNormalizer->transliterate($sTerm);
  61     }
  62
  63     public function setLanguagePreference($aLangPref)
  64     {
  65         $this->aLangPrefOrder = $aLangPref;
  66     }
  67
  68     public function getMoreUrlParams()
  69     {
  70         if ($this->aStructuredQuery) {
  71             $aParams = $this->aStructuredQuery;
  72         } else {
  73             $aParams = array('q' => $this->sQuery);
  74         }
  75
  76         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  77
  78         if ($this->aExcludePlaceIDs) {
  79             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  80         }
  81
  82         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  83
  84         if ($this->aCountryCodes) {
  85             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  86         }
  87
  88         if ($this->aViewBox) {
  89             $aParams['viewbox'] = join(',', $this->aViewBox);
  90         }
  91
  92         return $aParams;
  93     }
  94
  95     public function setLimit($iLimit = 10)
  96     {
  97         if ($iLimit > 50) $iLimit = 50;
  98         if ($iLimit < 1) $iLimit = 1;
  99
 100         $this->iFinalLimit = $iLimit;
 101         $this->iLimit = $iLimit + min($iLimit, 10);
 102     }
 103
 104     public function setFeatureType($sFeatureType)
 105     {
 106         switch ($sFeatureType) {
 107             case 'country':
 108                 $this->setRankRange(4, 4);
 109                 break;
 110             case 'state':
 111                 $this->setRankRange(8, 8);
 112                 break;
 113             case 'city':
 114                 $this->setRankRange(14, 16);
 115                 break;
 116             case 'settlement':
 117                 $this->setRankRange(8, 20);
 118                 break;
 119         }
 120     }
 121
 122     public function setRankRange($iMin, $iMax)
 123     {
 124         $this->iMinAddressRank = $iMin;
 125         $this->iMaxAddressRank = $iMax;
 126     }
 127
 128     public function setViewbox($aViewbox)
 129     {
 130         $aBox = array_map('floatval', $aViewbox);
 131
 132         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 133         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 134         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 135         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 136
 137         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 138             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 139         ) {
 140             userError("Bad parameter 'viewbox'. Not a box.");
 141         }
 142     }
 143
 144     private function viewboxImportanceFactor($fX, $fY)
 145     {
 146         if (!$this->aViewBox) {
 147             return 1;
 148         }
 149
 150         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 151         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 152
 153         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 154         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 155
 156         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 157             return 1;
 158         }
 159
 160         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 161             return 0.5;
 162         }
 163
 164         return 0.25;
 165     }
 166
 167     public function setQuery($sQueryString)
 168     {
 169         $this->sQuery = $sQueryString;
 170         $this->aStructuredQuery = false;
 171     }
 172
 173     public function getQueryString()
 174     {
 175         return $this->sQuery;
 176     }
 177
 178
 179     public function loadParamArray($oParams, $sForceGeometryType = null)
 180     {
 181         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 182
 183         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 184         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 185
 186         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 187
 188         // List of excluded Place IDs - used for more acurate pageing
 189         $sExcluded = $oParams->getStringList('exclude_place_ids');
 190         if ($sExcluded) {
 191             foreach ($sExcluded as $iExcludedPlaceID) {
 192                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 193                 if ($iExcludedPlaceID)
 194                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 195             }
 196
 197             if (isset($aExcludePlaceIDs))
 198                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 199         }
 200
 201         // Only certain ranks of feature
 202         $sFeatureType = $oParams->getString('featureType');
 203         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 204         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 205
 206         // Country code list
 207         $sCountries = $oParams->getStringList('countrycodes');
 208         if ($sCountries) {
 209             foreach ($sCountries as $sCountryCode) {
 210                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 211                     $aCountries[] = strtolower($sCountryCode);
 212                 }
 213             }
 214             if (isset($aCountries))
 215                 $this->aCountryCodes = $aCountries;
 216         }
 217
 218         $aViewbox = $oParams->getStringList('viewboxlbrt');
 219         if ($aViewbox) {
 220             if (count($aViewbox) != 4) {
 221                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 222             }
 223             $this->setViewbox($aViewbox);
 224         } else {
 225             $aViewbox = $oParams->getStringList('viewbox');
 226             if ($aViewbox) {
 227                 if (count($aViewbox) != 4) {
 228                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 229                 }
 230                 $this->setViewBox($aViewbox);
 231             } else {
 232                 $aRoute = $oParams->getStringList('route');
 233                 $fRouteWidth = $oParams->getFloat('routewidth');
 234                 if ($aRoute && $fRouteWidth) {
 235                     $this->aRoutePoints = $aRoute;
 236                     $this->aRouteWidth = $fRouteWidth;
 237                 }
 238             }
 239         }
 240
 241         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 242         $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
 243     }
 244
 245     public function setQueryFromParams($oParams)
 246     {
 247         // Search query
 248         $sQuery = $oParams->getString('q');
 249         if (!$sQuery) {
 250             $this->setStructuredQuery(
 251                 $oParams->getString('amenity'),
 252                 $oParams->getString('street'),
 253                 $oParams->getString('city'),
 254                 $oParams->getString('county'),
 255                 $oParams->getString('state'),
 256                 $oParams->getString('country'),
 257                 $oParams->getString('postalcode')
 258             );
 259         } else {
 260             $this->setQuery($sQuery);
 261         }
 262     }
 263
 264     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 265     {
 266         $sValue = trim($sValue);
 267         if (!$sValue) return false;
 268         $this->aStructuredQuery[$sKey] = $sValue;
 269         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 270             $this->iMinAddressRank = $iNewMinAddressRank;
 271             $this->iMaxAddressRank = $iNewMaxAddressRank;
 272         }
 273         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 274         return true;
 275     }
 276
 277     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 278     {
 279         $this->sQuery = false;
 280
 281         // Reset
 282         $this->iMinAddressRank = 0;
 283         $this->iMaxAddressRank = 30;
 284         $this->aAddressRankList = array();
 285
 286         $this->aStructuredQuery = array();
 287         $this->sAllowedTypesSQLList = false;
 288
 289         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 290         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 291         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 292         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 293         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 294         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 295         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 296
 297         if (!empty($this->aStructuredQuery)) {
 298             $this->sQuery = join(', ', $this->aStructuredQuery);
 299             if ($this->iMaxAddressRank < 30) {
 300                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 301             }
 302         }
 303     }
 304
 305     public function fallbackStructuredQuery()
 306     {
 307         if (!$this->aStructuredQuery) return false;
 308
 309         $aParams = $this->aStructuredQuery;
 310
 311         if (count($aParams) == 1) return false;
 312
 313         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 314
 315         foreach ($aOrderToFallback as $sType) {
 316             if (isset($aParams[$sType])) {
 317                 unset($aParams[$sType]);
 318                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 319                 return true;
 320             }
 321         }
 322
 323         return false;
 324     }
 325
 326     public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
 327     {
 328         /*
 329              Calculate all searches using oValidTokens i.e.
 330              'Wodsworth Road, Sheffield' =>
 331
 332              Phrase Wordset
 333              0      0       (wodsworth road)
 334              0      1       (wodsworth)(road)
 335              1      0       (sheffield)
 336
 337              Score how good the search is so they can be ordered
 338          */
 339         foreach ($aPhrases as $iPhrase => $oPhrase) {
 340             $aNewPhraseSearches = array();
 341             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 342
 343             foreach ($oPhrase->getWordSets() as $aWordset) {
 344                 $aWordsetSearches = $aSearches;
 345
 346                 // Add all words from this wordset
 347                 foreach ($aWordset as $iToken => $sToken) {
 348                     //echo "<br><b>$sToken</b>";
 349                     $aNewWordsetSearches = array();
 350
 351                     foreach ($aWordsetSearches as $oCurrentSearch) {
 352                         //echo "<i>";
 353                         //var_dump($oCurrentSearch);
 354                         //echo "</i>";
 355
 356                         // Tokens with full name matches.
 357                         foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
 358                             $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 359                                 $oSearchTerm,
 360                                 $oValidTokens->contains($sToken)
 361                                   && strpos($sToken, ' ') === false,
 362                                 $sPhraseType,
 363                                 $iToken == 0 && $iPhrase == 0,
 364                                 $iPhrase == 0,
 365                                 $iToken + 1 == count($aWordset)
 366                                   && $iPhrase + 1 == count($aPhrases)
 367                             );
 368
 369                             foreach ($aNewSearches as $oSearch) {
 370                                 if ($oSearch->getRank() < $this->iMaxRank) {
 371                                     $aNewWordsetSearches[] = $oSearch;
 372                                 }
 373                             }
 374                         }
 375                         // Look for partial matches.
 376                         // Note that there is no point in adding country terms here
 377                         // because country is omitted in the address.
 378                         if ($sPhraseType != 'country') {
 379                             // Allow searching for a word - but at extra cost
 380                             foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
 381                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 382                                     $sToken,
 383                                     $oSearchTerm,
 384                                     $bIsStructured,
 385                                     $iPhrase,
 386                                     $oValidTokens->get(' '.$sToken)
 387                                 );
 388
 389                                 foreach ($aNewSearches as $oSearch) {
 390                                     if ($oSearch->getRank() < $this->iMaxRank) {
 391                                         $aNewWordsetSearches[] = $oSearch;
 392                                     }
 393                                 }
 394                             }
 395                         }
 396                     }
 397                     // Sort and cut
 398                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 399                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 400                 }
 401                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 402
 403                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 404                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 405
 406                 $aSearchHash = array();
 407                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 408                     $sHash = serialize($aSearch);
 409                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 410                     else $aSearchHash[$sHash] = 1;
 411                 }
 412
 413                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 414             }
 415
 416             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 417             $aGroupedSearches = array();
 418             foreach ($aNewPhraseSearches as $aSearch) {
 419                 $iRank = $aSearch->getRank();
 420                 if ($iRank < $this->iMaxRank) {
 421                     if (!isset($aGroupedSearches[$iRank])) {
 422                         $aGroupedSearches[$iRank] = array();
 423                     }
 424                     $aGroupedSearches[$iRank][] = $aSearch;
 425                 }
 426             }
 427             ksort($aGroupedSearches);
 428
 429             $iSearchCount = 0;
 430             $aSearches = array();
 431             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 432                 $iSearchCount += count($aNewSearches);
 433                 $aSearches = array_merge($aSearches, $aNewSearches);
 434                 if ($iSearchCount > 50) break;
 435             }
 436         }
 437
 438         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 439         $aGroupedSearches = array();
 440         foreach ($aSearches as $oSearch) {
 441             if (!$oSearch->isValidSearch()) {
 442                 continue;
 443             }
 444
 445             $iRank = $oSearch->getRank();
 446             if (!isset($aGroupedSearches[$iRank])) {
 447                 $aGroupedSearches[$iRank] = array();
 448             }
 449             $aGroupedSearches[$iRank][] = $oSearch;
 450         }
 451         ksort($aGroupedSearches);
 452
 453         return $aGroupedSearches;
 454     }
 455
 456     /* Perform the actual query lookup.
 457
 458         Returns an ordered list of results, each with the following fields:
 459             osm_type: type of corresponding OSM object
 460                         N - node
 461                         W - way
 462                         R - relation
 463                         P - postcode (internally computed)
 464             osm_id: id of corresponding OSM object
 465             class: general object class (corresponds to tag key of primary OSM tag)
 466             type: subclass of object (corresponds to tag value of primary OSM tag)
 467             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 468             rank_search: rank in search hierarchy
 469                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 470             rank_address: rank in address hierarchy (determines orer in address)
 471             place_id: internal key (may differ between different instances)
 472             country_code: ISO country code
 473             langaddress: localized full address
 474             placename: localized name of object
 475             ref: content of ref tag (if available)
 476             lon: longitude
 477             lat: latitude
 478             importance: importance of place based on Wikipedia link count
 479             addressimportance: cumulated importance of address elements
 480             extra_place: type of place (for admin boundaries, if there is a place tag)
 481             aBoundingBox: bounding Box
 482             label: short description of the object class/type (English only)
 483             name: full name (currently the same as langaddress)
 484             foundorder: secondary ordering for places with same importance
 485     */
 486
 487
 488     public function lookup()
 489     {
 490         Debug::newFunction('Geocode::lookup');
 491         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 492
 493         Debug::printDebugArray('Geocode', $this);
 494
 495         $oCtx = new SearchContext();
 496
 497         if ($this->aRoutePoints) {
 498             $oCtx->setViewboxFromRoute(
 499                 $this->oDB,
 500                 $this->aRoutePoints,
 501                 $this->aRouteWidth,
 502                 $this->bBoundedSearch
 503             );
 504         } elseif ($this->aViewBox) {
 505             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 506         }
 507         if ($this->aExcludePlaceIDs) {
 508             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 509         }
 510         if ($this->aCountryCodes) {
 511             $oCtx->setCountryList($this->aCountryCodes);
 512         }
 513
 514         Debug::newSection('Query Preprocessing');
 515
 516         $sNormQuery = $this->normTerm($this->sQuery);
 517         Debug::printVar('Normalized query', $sNormQuery);
 518
 519         $sLanguagePrefArraySQL = $this->oDB->getArraySQL(
 520             $this->oDB->getDBQuotedList($this->aLangPrefOrder)
 521         );
 522
 523         $sQuery = $this->sQuery;
 524         if (!preg_match('//u', $sQuery)) {
 525             userError('Query string is not UTF-8 encoded.');
 526         }
 527
 528         // Conflicts between US state abreviations and various words for 'the' in different languages
 529         if (isset($this->aLangPrefOrder['name:en'])) {
 530             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 531             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 532             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 533         }
 534
 535         // Do we have anything that looks like a lat/lon pair?
 536         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 537
 538         if ($sQuery || $this->aStructuredQuery) {
 539             // Start with a single blank search
 540             $aSearches = array(new SearchDescription($oCtx));
 541
 542             if ($sQuery) {
 543                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 544             }
 545
 546             $sSpecialTerm = '';
 547             if ($sQuery) {
 548                 preg_match_all(
 549                     '/\\[([\\w ]*)\\]/u',
 550                     $sQuery,
 551                     $aSpecialTermsRaw,
 552                     PREG_SET_ORDER
 553                 );
 554                 if (!empty($aSpecialTermsRaw)) {
 555                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 556                 }
 557
 558                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 559                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 560                     if (!$sSpecialTerm) {
 561                         $sSpecialTerm = $aSpecialTerm[1];
 562                     }
 563                 }
 564             }
 565             if (!$sSpecialTerm && $this->aStructuredQuery
 566                 && isset($this->aStructuredQuery['amenity'])) {
 567                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 568                 unset($this->aStructuredQuery['amenity']);
 569             }
 570
 571             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 572                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 573                 $sToken = $this->oDB->getOne(
 574                     'SELECT make_standard_name(:term)',
 575                     array(':term' => $sSpecialTerm),
 576                     'Cannot decode query. Wrong encoding?'
 577                 );
 578                 $sSQL = 'SELECT class, type FROM word ';
 579                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 580                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 581
 582                 Debug::printSQL($sSQL);
 583                 $aSearchWords = $this->oDB->getAll($sSQL);
 584                 $aNewSearches = array();
 585                 foreach ($aSearches as $oSearch) {
 586                     foreach ($aSearchWords as $aSearchTerm) {
 587                         $oNewSearch = clone $oSearch;
 588                         $oNewSearch->setPoiSearch(
 589                             Operator::TYPE,
 590                             $aSearchTerm['class'],
 591                             $aSearchTerm['type']
 592                         );
 593                         $aNewSearches[] = $oNewSearch;
 594                     }
 595                 }
 596                 $aSearches = $aNewSearches;
 597             }
 598
 599             // Split query into phrases
 600             // Commas are used to reduce the search space by indicating where phrases split
 601             if ($this->aStructuredQuery) {
 602                 $aInPhrases = $this->aStructuredQuery;
 603                 $bStructuredPhrases = true;
 604             } else {
 605                 $aInPhrases = explode(',', $sQuery);
 606                 $bStructuredPhrases = false;
 607             }
 608
 609             Debug::printDebugArray('Search context', $oCtx);
 610             Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
 611             Debug::printVar('Final query phrases', $aInPhrases);
 612
 613             // Convert each phrase to standard form
 614             // Create a list of standard words
 615             // Get all 'sets' of words
 616             // Generate a complete list of all
 617             Debug::newSection('Tokenization');
 618             $aTokens = array();
 619             $aPhrases = array();
 620             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 621                 $sPhrase = $this->oDB->getOne(
 622                     'SELECT make_standard_name(:phrase)',
 623                     array(':phrase' => $sPhrase),
 624                     'Cannot normalize query string (is it a UTF-8 string?)'
 625                 );
 626                 if (trim($sPhrase)) {
 627                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 628                     $oPhrase->addTokens($aTokens);
 629                     $aPhrases[] = $oPhrase;
 630                 }
 631             }
 632
 633             Debug::printVar('Tokens', $aTokens);
 634
 635             $oValidTokens = new TokenList();
 636
 637             if (!empty($aTokens)) {
 638                 $oValidTokens->addTokensFromDB(
 639                     $this->oDB,
 640                     $aTokens,
 641                     $this->aCountryCodes,
 642                     $sNormQuery,
 643                     $this->oNormalizer
 644                 );
 645
 646                 $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
 647
 648                 // Try more interpretations for Tokens that could not be matched.
 649                 foreach ($aTokens as $sToken) {
 650                     if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
 651                         if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 652                             // US ZIP+4 codes - merge in the 5-digit ZIP code
 653                             $oValidTokens->addToken(
 654                                 $sToken,
 655                                 new Token\Postcode(null, $aData[1], 'us')
 656                             );
 657                         } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
 658                             // Unknown single word token with a number.
 659                             // Assume it is a house number.
 660                             $oValidTokens->addToken(
 661                                 $sToken,
 662                                 new Token\HouseNumber(null, trim($sToken))
 663                             );
 664                         }
 665                     }
 666                 }
 667
 668                 // Any words that have failed completely?
 669                 // TODO: suggestions
 670
 671                 Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
 672
 673                 foreach ($aPhrases as $oPhrase) {
 674                     $oPhrase->computeWordSets($oValidTokens);
 675                 }
 676                 Debug::printDebugTable('Phrases', $aPhrases);
 677
 678                 Debug::newSection('Search candidates');
 679
 680                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
 681
 682                 if (!$this->aStructuredQuery) {
 683                     // Reverse phrase array and also reverse the order of the wordsets in
 684                     // the first and final phrase. Don't bother about phrases in the middle
 685                     // because order in the address doesn't matter.
 686                     $aPhrases = array_reverse($aPhrases);
 687                     $aPhrases[0]->invertWordSets();
 688                     if (count($aPhrases) > 1) {
 689                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 690                     }
 691                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
 692
 693                     foreach ($aGroupedSearches as $aSearches) {
 694                         foreach ($aSearches as $aSearch) {
 695                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 696                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 697                             }
 698                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 699                         }
 700                     }
 701
 702                     $aGroupedSearches = $aReverseGroupedSearches;
 703                     ksort($aGroupedSearches);
 704                 }
 705             } else {
 706                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 707                 $aGroupedSearches = array();
 708                 foreach ($aSearches as $aSearch) {
 709                     if ($aSearch->getRank() < $this->iMaxRank) {
 710                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 711                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 712                     }
 713                 }
 714                 ksort($aGroupedSearches);
 715             }
 716
 717             // Filter out duplicate searches
 718             $aSearchHash = array();
 719             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 720                 foreach ($aSearches as $iSearch => $aSearch) {
 721                     $sHash = serialize($aSearch);
 722                     if (isset($aSearchHash[$sHash])) {
 723                         unset($aGroupedSearches[$iGroup][$iSearch]);
 724                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 725                     } else {
 726                         $aSearchHash[$sHash] = 1;
 727                     }
 728                 }
 729             }
 730
 731             Debug::printGroupedSearch(
 732                 $aGroupedSearches,
 733                 $oValidTokens->debugTokenByWordIdList()
 734             );
 735
 736             // Start the search process
 737             $iGroupLoop = 0;
 738             $iQueryLoop = 0;
 739             $aNextResults = array();
 740             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 741                 $iGroupLoop++;
 742                 $aResults = $aNextResults;
 743                 foreach ($aSearches as $oSearch) {
 744                     $iQueryLoop++;
 745
 746                     Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
 747                     Debug::printGroupedSearch(
 748                         array($iGroupedRank => array($oSearch)),
 749                         $oValidTokens->debugTokenByWordIdList()
 750                     );
 751
 752                     $aNewResults = $oSearch->query(
 753                         $this->oDB,
 754                         $this->iMinAddressRank,
 755                         $this->iMaxAddressRank,
 756                         $this->iLimit
 757                     );
 758
 759                     // The same result may appear in different rounds, only
 760                     // use the one with minimal rank.
 761                     foreach ($aNewResults as $iPlace => $oRes) {
 762                         if (!isset($aResults[$iPlace])
 763                             || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) {
 764                             $aResults[$iPlace] = $oRes;
 765                         }
 766                     }
 767
 768                     if ($iQueryLoop > 20) break;
 769                 }
 770
 771                 if (!empty($aResults)) {
 772                     $aSplitResults = Result::splitResults($aResults);
 773                     Debug::printVar('Split results', $aSplitResults);
 774                     if ($iGroupLoop <= 4
 775                         && reset($aSplitResults['head'])->iResultRank > 0
 776                         && $iGroupedRank !== array_key_last($aGroupedSearches)) {
 777                         // Haven't found an exact match for the query yet.
 778                         // Therefore add result from the next group level.
 779                         $aNextResults = $aSplitResults['head'];
 780                         foreach ($aNextResults as $oRes) {
 781                             $oRes->iResultRank--;
 782                         }
 783                         foreach ($aSplitResults['tail'] as $oRes) {
 784                             $oRes->iResultRank--;
 785                             $aNextResults[$oRes->iId] = $oRes;
 786                         }
 787                         $aResults = array();
 788                     } else {
 789                         $aResults = $aSplitResults['head'];
 790                     }
 791                 }
 792
 793                 if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
 794                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
 795                     // reduces the number of place ids, like a filter
 796                     // rank_address is 30 for interpolated housenumbers
 797                     $aFilterSql = array();
 798                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
 799                     if ($sPlaceIds) {
 800                         $sSQL = 'SELECT place_id FROM placex ';
 801                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 802                         $sSQL .= '  AND (';
 803                         $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 804                         $sSQL .= "         OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
 805                         if ($this->aAddressRankList) {
 806                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 807                         }
 808                         $sSQL .= ')';
 809                         $aFilterSql[] = $sSQL;
 810                     }
 811                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 812                     if ($sPlaceIds) {
 813                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 814                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 815                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 816                         if ($this->aAddressRankList) {
 817                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 818                         }
 819                         $sSQL .= ') ';
 820                         $aFilterSql[] = $sSQL;
 821                     }
 822
 823                     $aFilteredIDs = array();
 824                     if ($aFilterSql) {
 825                         $sSQL = join(' UNION ', $aFilterSql);
 826                         Debug::printSQL($sSQL);
 827                         $aFilteredIDs = $this->oDB->getCol($sSQL);
 828                     }
 829
 830                     $tempIDs = array();
 831                     foreach ($aResults as $oResult) {
 832                         if (($this->iMaxAddressRank == 30 &&
 833                              ($oResult->iTable == Result::TABLE_OSMLINE
 834                               || $oResult->iTable == Result::TABLE_AUX
 835                               || $oResult->iTable == Result::TABLE_TIGER))
 836                             || in_array($oResult->iId, $aFilteredIDs)
 837                         ) {
 838                             $tempIDs[$oResult->iId] = $oResult;
 839                         }
 840                     }
 841                     $aResults = $tempIDs;
 842                 }
 843
 844                 if (!empty($aResults)) break;
 845                 if ($iGroupLoop > 4) break;
 846                 if ($iQueryLoop > 30) break;
 847             }
 848         } else {
 849             // Just interpret as a reverse geocode
 850             $oReverse = new ReverseGeocode($this->oDB);
 851             $oReverse->setZoom(18);
 852
 853             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 854
 855             Debug::printVar('Reverse search', $oLookup);
 856
 857             if ($oLookup) {
 858                 $aResults = array($oLookup->iId => $oLookup);
 859             }
 860         }
 861
 862         // No results? Done
 863         if (empty($aResults)) {
 864             if ($this->bFallback) {
 865                 if ($this->fallbackStructuredQuery()) {
 866                     return $this->lookup();
 867                 }
 868             }
 869
 870             return array();
 871         }
 872
 873         if ($this->aAddressRankList) {
 874             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 875         }
 876         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 877         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 878         if ($oCtx->hasNearPoint()) {
 879             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 880         }
 881
 882         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 883
 884         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 885         foreach ($aRecheckWords as $i => $sWord) {
 886             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 887         }
 888
 889         Debug::printVar('Recheck words', $aRecheckWords);
 890
 891         foreach ($aSearchResults as $iIdx => $aResult) {
 892             $fRadius = ClassTypes\getDefRadius($aResult);
 893
 894             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
 895             if ($aOutlineResult) {
 896                 $aResult = array_merge($aResult, $aOutlineResult);
 897             }
 898
 899             // Is there an icon set for this type of result?
 900             $sIcon = ClassTypes\getIconFile($aResult);
 901             if (isset($sIcon)) {
 902                 $aResult['icon'] = $sIcon;
 903             }
 904
 905             $sLabel = ClassTypes\getLabel($aResult);
 906             if (isset($sLabel)) {
 907                 $aResult['label'] = $sLabel;
 908             }
 909             $aResult['name'] = $aResult['langaddress'];
 910
 911             if ($oCtx->hasNearPoint()) {
 912                 $aResult['importance'] = 0.001;
 913                 $aResult['foundorder'] = $aResult['addressimportance'];
 914             } else {
 915                 $aResult['importance'] = max(0.001, $aResult['importance']);
 916                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 917                     $aResult['lon'],
 918                     $aResult['lat']
 919                 );
 920
 921                 // secondary ordering (for results with same importance (the smaller the better):
 922                 // - approximate importance of address parts
 923                 if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
 924                     $aResult['foundorder'] = -$aResult['addressimportance']/10;
 925                 } else {
 926                     $aResult['foundorder'] = -$aResult['importance'];
 927                 }
 928                 // - number of exact matches from the query
 929                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 930                 // - importance of the class/type
 931                 $iClassImportance = ClassTypes\getImportance($aResult);
 932                 if (isset($iClassImportance)) {
 933                     $aResult['foundorder'] += 0.0001 * $iClassImportance;
 934                 } else {
 935                     $aResult['foundorder'] += 0.01;
 936                 }
 937                 // - rank
 938                 $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
 939
 940                 // Adjust importance for the number of exact string matches in the result
 941                 $iCountWords = 0;
 942                 $sAddress = $aResult['langaddress'];
 943                 foreach ($aRecheckWords as $i => $sWord) {
 944                     if (stripos($sAddress, $sWord)!==false) {
 945                         $iCountWords++;
 946                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 947                     }
 948                 }
 949
 950                 // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 951                 $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1);
 952             }
 953             $aSearchResults[$iIdx] = $aResult;
 954         }
 955         uasort($aSearchResults, 'byImportance');
 956         Debug::printVar('Pre-filter results', $aSearchResults);
 957
 958         $aOSMIDDone = array();
 959         $aClassTypeNameDone = array();
 960         $aToFilter = $aSearchResults;
 961         $aSearchResults = array();
 962
 963         $bFirst = true;
 964         foreach ($aToFilter as $aResult) {
 965             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 966             if ($bFirst) {
 967                 $fLat = $aResult['lat'];
 968                 $fLon = $aResult['lon'];
 969                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 970                 $bFirst = false;
 971             }
 972             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 973                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 974             ) {
 975                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 976                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 977                 $aSearchResults[] = $aResult;
 978             }
 979
 980             // Absolute limit on number of results
 981             if (count($aSearchResults) >= $this->iFinalLimit) break;
 982         }
 983
 984         Debug::printVar('Post-filter results', $aSearchResults);
 985         return $aSearchResults;
 986     } // end lookup()
 987
 988     public function debugInfo()
 989     {
 990         return array(
 991                 'Query' => $this->sQuery,
 992                 'Structured query' => $this->aStructuredQuery,
 993                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
 994                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
 995                 'Limit (for searches)' => $this->iLimit,
 996                 'Limit (for results)'=> $this->iFinalLimit,
 997                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
 998                 'Bounded search' => $this->bBoundedSearch,
 999                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
1000                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
1001                 'Route width' => $this->aRouteWidth,
1002                 'Max rank' => $this->iMaxRank,
1003                 'Min address rank' => $this->iMinAddressRank,
1004                 'Max address rank' => $this->iMaxAddressRank,
1005                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
1006                );
1007     }
1008 } // end class