lib-php/Geocode.php

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