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
  11 class Geocode
  12 {
  13     protected $oDB;
  14
  15     protected $oPlaceLookup;
  16
  17     protected $aLangPrefOrder = array();
  18
  19     protected $bIncludeAddressDetails = false;
  20
  21     protected $aExcludePlaceIDs = array();
  22     protected $bReverseInPlan = false;
  23
  24     protected $iLimit = 20;
  25     protected $iFinalLimit = 10;
  26     protected $iOffset = 0;
  27     protected $bFallback = false;
  28
  29     protected $aCountryCodes = false;
  30
  31     protected $bBoundedSearch = false;
  32     protected $aViewBox = false;
  33     protected $aRoutePoints = false;
  34     protected $aRouteWidth = false;
  35
  36     protected $iMaxRank = 20;
  37     protected $iMinAddressRank = 0;
  38     protected $iMaxAddressRank = 30;
  39     protected $aAddressRankList = array();
  40
  41     protected $sAllowedTypesSQLList = false;
  42
  43     protected $sQuery = false;
  44     protected $aStructuredQuery = false;
  45
  46     protected $oNormalizer = null;
  47
  48
  49     public function __construct(&$oDB)
  50     {
  51         $this->oDB =& $oDB;
  52         $this->oPlaceLookup = new PlaceLookup($this->oDB);
  53         $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
  54     }
  55
  56     private function normTerm($sTerm)
  57     {
  58         if ($this->oNormalizer === null) {
  59             return $sTerm;
  60         }
  61
  62         return $this->oNormalizer->transliterate($sTerm);
  63     }
  64
  65     public function setReverseInPlan($bReverse)
  66     {
  67         $this->bReverseInPlan = $bReverse;
  68     }
  69
  70     public function setLanguagePreference($aLangPref)
  71     {
  72         $this->aLangPrefOrder = $aLangPref;
  73     }
  74
  75     public function getMoreUrlParams()
  76     {
  77         if ($this->aStructuredQuery) {
  78             $aParams = $this->aStructuredQuery;
  79         } else {
  80             $aParams = array('q' => $this->sQuery);
  81         }
  82
  83         $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams());
  84
  85         if ($this->aExcludePlaceIDs) {
  86             $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
  87         }
  88
  89         if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
  90         if ($this->bBoundedSearch) $aParams['bounded'] = '1';
  91
  92         if ($this->aCountryCodes) {
  93             $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
  94         }
  95
  96         if ($this->aViewBox) {
  97             $aParams['viewbox'] = join(',', $this->aViewBox);
  98         }
  99
 100         return $aParams;
 101     }
 102
 103     public function setLimit($iLimit = 10)
 104     {
 105         if ($iLimit > 50) $iLimit = 50;
 106         if ($iLimit < 1) $iLimit = 1;
 107
 108         $this->iFinalLimit = $iLimit;
 109         $this->iLimit = $iLimit + min($iLimit, 10);
 110     }
 111
 112     public function setFeatureType($sFeatureType)
 113     {
 114         switch ($sFeatureType) {
 115             case 'country':
 116                 $this->setRankRange(4, 4);
 117                 break;
 118             case 'state':
 119                 $this->setRankRange(8, 8);
 120                 break;
 121             case 'city':
 122                 $this->setRankRange(14, 16);
 123                 break;
 124             case 'settlement':
 125                 $this->setRankRange(8, 20);
 126                 break;
 127         }
 128     }
 129
 130     public function setRankRange($iMin, $iMax)
 131     {
 132         $this->iMinAddressRank = $iMin;
 133         $this->iMaxAddressRank = $iMax;
 134     }
 135
 136     public function setViewbox($aViewbox)
 137     {
 138         $aBox = array_map('floatval', $aViewbox);
 139
 140         $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2]));
 141         $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3]));
 142         $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2]));
 143         $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3]));
 144
 145         if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001
 146             || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001
 147         ) {
 148             userError("Bad parameter 'viewbox'. Not a box.");
 149         }
 150     }
 151
 152     private function viewboxImportanceFactor($fX, $fY)
 153     {
 154         $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
 155         $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
 156
 157         $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2);
 158         $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2);
 159
 160         if ($fXDist <= $fWidth && $fYDist <= $fHeight) {
 161             return 1;
 162         }
 163
 164         if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) {
 165             return 0.5;
 166         }
 167
 168         return 0.25;
 169     }
 170
 171     public function setQuery($sQueryString)
 172     {
 173         $this->sQuery = $sQueryString;
 174         $this->aStructuredQuery = false;
 175     }
 176
 177     public function getQueryString()
 178     {
 179         return $this->sQuery;
 180     }
 181
 182
 183     public function loadParamArray($oParams, $sForceGeometryType = null)
 184     {
 185         $this->bIncludeAddressDetails
 186          = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
 187
 188         $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
 189
 190         $this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
 191         $this->iOffset = $oParams->getInt('offset', $this->iOffset);
 192
 193         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 194
 195         // List of excluded Place IDs - used for more acurate pageing
 196         $sExcluded = $oParams->getStringList('exclude_place_ids');
 197         if ($sExcluded) {
 198             foreach ($sExcluded as $iExcludedPlaceID) {
 199                 $iExcludedPlaceID = (int)$iExcludedPlaceID;
 200                 if ($iExcludedPlaceID)
 201                     $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
 202             }
 203
 204             if (isset($aExcludePlaceIDs))
 205                 $this->aExcludePlaceIDs = $aExcludePlaceIDs;
 206         }
 207
 208         // Only certain ranks of feature
 209         $sFeatureType = $oParams->getString('featureType');
 210         if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
 211         if ($sFeatureType) $this->setFeatureType($sFeatureType);
 212
 213         // Country code list
 214         $sCountries = $oParams->getStringList('countrycodes');
 215         if ($sCountries) {
 216             foreach ($sCountries as $sCountryCode) {
 217                 if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
 218                     $aCountries[] = strtolower($sCountryCode);
 219                 }
 220             }
 221             if (isset($aCountries))
 222                 $this->aCountryCodes = $aCountries;
 223         }
 224
 225         $aViewbox = $oParams->getStringList('viewboxlbrt');
 226         if ($aViewbox) {
 227             if (count($aViewbox) != 4) {
 228                 userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");
 229             }
 230             $this->setViewbox($aViewbox);
 231         } else {
 232             $aViewbox = $oParams->getStringList('viewbox');
 233             if ($aViewbox) {
 234                 if (count($aViewbox) != 4) {
 235                     userError("Bad parameter 'viewbox'. Expected 4 coordinates.");
 236                 }
 237                 $this->setViewBox($aViewbox);
 238             } else {
 239                 $aRoute = $oParams->getStringList('route');
 240                 $fRouteWidth = $oParams->getFloat('routewidth');
 241                 if ($aRoute && $fRouteWidth) {
 242                     $this->aRoutePoints = $aRoute;
 243                     $this->aRouteWidth = $fRouteWidth;
 244                 }
 245             }
 246         }
 247
 248         $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
 249         $this->oPlaceLookup->setIncludeAddressDetails(false);
 250         $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
 251     }
 252
 253     public function setQueryFromParams($oParams)
 254     {
 255         // Search query
 256         $sQuery = $oParams->getString('q');
 257         if (!$sQuery) {
 258             $this->setStructuredQuery(
 259                 $oParams->getString('amenity'),
 260                 $oParams->getString('street'),
 261                 $oParams->getString('city'),
 262                 $oParams->getString('county'),
 263                 $oParams->getString('state'),
 264                 $oParams->getString('country'),
 265                 $oParams->getString('postalcode')
 266             );
 267             $this->setReverseInPlan(false);
 268         } else {
 269             $this->setQuery($sQuery);
 270         }
 271     }
 272
 273     public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
 274     {
 275         $sValue = trim($sValue);
 276         if (!$sValue) return false;
 277         $this->aStructuredQuery[$sKey] = $sValue;
 278         if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
 279             $this->iMinAddressRank = $iNewMinAddressRank;
 280             $this->iMaxAddressRank = $iNewMaxAddressRank;
 281         }
 282         if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
 283         return true;
 284     }
 285
 286     public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
 287     {
 288         $this->sQuery = false;
 289
 290         // Reset
 291         $this->iMinAddressRank = 0;
 292         $this->iMaxAddressRank = 30;
 293         $this->aAddressRankList = array();
 294
 295         $this->aStructuredQuery = array();
 296         $this->sAllowedTypesSQLList = false;
 297
 298         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
 299         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
 300         $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
 301         $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
 302         $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
 303         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
 304         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 305
 306         if (!empty($this->aStructuredQuery)) {
 307             $this->sQuery = join(', ', $this->aStructuredQuery);
 308             if ($this->iMaxAddressRank < 30) {
 309                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
 310             }
 311         }
 312     }
 313
 314     public function fallbackStructuredQuery()
 315     {
 316         if (!$this->aStructuredQuery) return false;
 317
 318         $aParams = $this->aStructuredQuery;
 319
 320         if (count($aParams) == 1) return false;
 321
 322         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 323
 324         foreach ($aOrderToFallback as $sType) {
 325             if (isset($aParams[$sType])) {
 326                 unset($aParams[$sType]);
 327                 $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
 328                 return true;
 329             }
 330         }
 331
 332         return false;
 333     }
 334
 335     public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
 336     {
 337         /*
 338              Calculate all searches using aValidTokens i.e.
 339              'Wodsworth Road, Sheffield' =>
 340
 341              Phrase Wordset
 342              0      0       (wodsworth road)
 343              0      1       (wodsworth)(road)
 344              1      0       (sheffield)
 345
 346              Score how good the search is so they can be ordered
 347          */
 348         foreach ($aPhrases as $iPhrase => $oPhrase) {
 349             $aNewPhraseSearches = array();
 350             $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
 351
 352             foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
 353                 // Too many permutations - too expensive
 354                 if ($iWordSet > 120) break;
 355
 356                 $aWordsetSearches = $aSearches;
 357
 358                 // Add all words from this wordset
 359                 foreach ($aWordset as $iToken => $sToken) {
 360                     //echo "<br><b>$sToken</b>";
 361                     $aNewWordsetSearches = array();
 362
 363                     foreach ($aWordsetSearches as $oCurrentSearch) {
 364                         //echo "<i>";
 365                         //var_dump($oCurrentSearch);
 366                         //echo "</i>";
 367
 368                         // If the token is valid
 369                         if (isset($aValidTokens[' '.$sToken])) {
 370                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
 371                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
 372                                     $aSearchTerm,
 373                                     isset($aValidTokens[$sToken])
 374                                       && strpos($sToken, ' ') === false,
 375                                     $sPhraseType,
 376                                     $iToken == 0 && $iPhrase == 0,
 377                                     $iPhrase == 0,
 378                                     $iToken + 1 == count($aWordset)
 379                                       && $iPhrase + 1 == count($aPhrases)
 380                                 );
 381
 382                                 foreach ($aNewSearches as $oSearch) {
 383                                     if ($oSearch->getRank() < $this->iMaxRank) {
 384                                         $aNewWordsetSearches[] = $oSearch;
 385                                     }
 386                                 }
 387                             }
 388                         }
 389                         // Look for partial matches.
 390                         // Note that there is no point in adding country terms here
 391                         // because country is omitted in the address.
 392                         if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
 393                             // Allow searching for a word - but at extra cost
 394                             foreach ($aValidTokens[$sToken] as $aSearchTerm) {
 395                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
 396                                     $aSearchTerm,
 397                                     $bIsStructured,
 398                                     $iPhrase,
 399                                     isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
 400                                 );
 401
 402                                 foreach ($aNewSearches as $oSearch) {
 403                                     if ($oSearch->getRank() < $this->iMaxRank) {
 404                                         $aNewWordsetSearches[] = $oSearch;
 405                                     }
 406                                 }
 407                             }
 408                         }
 409                     }
 410                     // Sort and cut
 411                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 412                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
 413                 }
 414                 //var_Dump('<hr>',count($aWordsetSearches)); exit;
 415
 416                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 417                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 418
 419                 $aSearchHash = array();
 420                 foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
 421                     $sHash = serialize($aSearch);
 422                     if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
 423                     else $aSearchHash[$sHash] = 1;
 424                 }
 425
 426                 $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 427             }
 428
 429             // Re-group the searches by their score, junk anything over 20 as just not worth trying
 430             $aGroupedSearches = array();
 431             foreach ($aNewPhraseSearches as $aSearch) {
 432                 $iRank = $aSearch->getRank();
 433                 if ($iRank < $this->iMaxRank) {
 434                     if (!isset($aGroupedSearches[$iRank])) {
 435                         $aGroupedSearches[$iRank] = array();
 436                     }
 437                     $aGroupedSearches[$iRank][] = $aSearch;
 438                 }
 439             }
 440             ksort($aGroupedSearches);
 441
 442             $iSearchCount = 0;
 443             $aSearches = array();
 444             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
 445                 $iSearchCount += count($aNewSearches);
 446                 $aSearches = array_merge($aSearches, $aNewSearches);
 447                 if ($iSearchCount > 50) break;
 448             }
 449         }
 450
 451         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
 452         $aGroupedSearches = array();
 453         foreach ($aSearches as $oSearch) {
 454             if (!$oSearch->isValidSearch()) {
 455                 continue;
 456             }
 457
 458             $iRank = $oSearch->getRank();
 459             if (!isset($aGroupedSearches[$iRank])) {
 460                 $aGroupedSearches[$iRank] = array();
 461             }
 462             $aGroupedSearches[$iRank][] = $oSearch;
 463         }
 464         ksort($aGroupedSearches);
 465
 466         return $aGroupedSearches;
 467     }
 468
 469     /* Perform the actual query lookup.
 470
 471         Returns an ordered list of results, each with the following fields:
 472             osm_type: type of corresponding OSM object
 473                         N - node
 474                         W - way
 475                         R - relation
 476                         P - postcode (internally computed)
 477             osm_id: id of corresponding OSM object
 478             class: general object class (corresponds to tag key of primary OSM tag)
 479             type: subclass of object (corresponds to tag value of primary OSM tag)
 480             admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level
 481             rank_search: rank in search hierarchy
 482                         (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
 483             rank_address: rank in address hierarchy (determines orer in address)
 484             place_id: internal key (may differ between different instances)
 485             country_code: ISO country code
 486             langaddress: localized full address
 487             placename: localized name of object
 488             ref: content of ref tag (if available)
 489             lon: longitude
 490             lat: latitude
 491             importance: importance of place based on Wikipedia link count
 492             addressimportance: cumulated importance of address elements
 493             extra_place: type of place (for admin boundaries, if there is a place tag)
 494             aBoundingBox: bounding Box
 495             label: short description of the object class/type (English only)
 496             name: full name (currently the same as langaddress)
 497             foundorder: secondary ordering for places with same importance
 498     */
 499
 500
 501     public function lookup()
 502     {
 503         Debug::newFunction('Geocode::lookup');
 504         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 505
 506         Debug::printDebugArray('Geocode', $this);
 507
 508         $oCtx = new SearchContext();
 509
 510         if ($this->aRoutePoints) {
 511             $oCtx->setViewboxFromRoute(
 512                 $this->oDB,
 513                 $this->aRoutePoints,
 514                 $this->aRouteWidth,
 515                 $this->bBoundedSearch
 516             );
 517         } elseif ($this->aViewBox) {
 518             $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
 519         }
 520         if ($this->aExcludePlaceIDs) {
 521             $oCtx->setExcludeList($this->aExcludePlaceIDs);
 522         }
 523         if ($this->aCountryCodes) {
 524             $oCtx->setCountryList($this->aCountryCodes);
 525         }
 526
 527         Debug::newSection('Query Preprocessing');
 528
 529         $sNormQuery = $this->normTerm($this->sQuery);
 530         Debug::printVar('Normalized query', $sNormQuery);
 531
 532         $sLanguagePrefArraySQL = getArraySQL(
 533             array_map('getDBQuoted', $this->aLangPrefOrder)
 534         );
 535
 536         $sQuery = $this->sQuery;
 537         if (!preg_match('//u', $sQuery)) {
 538             userError('Query string is not UTF-8 encoded.');
 539         }
 540
 541         // Conflicts between US state abreviations and various words for 'the' in different languages
 542         if (isset($this->aLangPrefOrder['name:en'])) {
 543             $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
 544             $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
 545             $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
 546         }
 547
 548         // Do we have anything that looks like a lat/lon pair?
 549         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 550
 551         $aResults = array();
 552         if ($sQuery || $this->aStructuredQuery) {
 553             // Start with a single blank search
 554             $aSearches = array(new SearchDescription($oCtx));
 555
 556             if ($sQuery) {
 557                 $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
 558             }
 559
 560             $sSpecialTerm = '';
 561             if ($sQuery) {
 562                 preg_match_all(
 563                     '/\\[([\\w ]*)\\]/u',
 564                     $sQuery,
 565                     $aSpecialTermsRaw,
 566                     PREG_SET_ORDER
 567                 );
 568                 if (!empty($aSpecialTermsRaw)) {
 569                     Debug::printVar('Special terms', $aSpecialTermsRaw);
 570                 }
 571
 572                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
 573                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
 574                     if (!$sSpecialTerm) {
 575                         $sSpecialTerm = $aSpecialTerm[1];
 576                     }
 577                 }
 578             }
 579             if (!$sSpecialTerm && $this->aStructuredQuery
 580                 && isset($this->aStructuredQuery['amenity'])) {
 581                 $sSpecialTerm = $this->aStructuredQuery['amenity'];
 582                 unset($this->aStructuredQuery['amenity']);
 583             }
 584
 585             if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
 586                 $sSpecialTerm = pg_escape_string($sSpecialTerm);
 587                 $sToken = chksql(
 588                     $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
 589                     'Cannot decode query. Wrong encoding?'
 590                 );
 591                 $sSQL = 'SELECT class, type FROM word ';
 592                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
 593                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
 594
 595                 Debug::printSQL($sSQL);
 596                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
 597                 $aNewSearches = array();
 598                 foreach ($aSearches as $oSearch) {
 599                     foreach ($aSearchWords as $aSearchTerm) {
 600                         $oNewSearch = clone $oSearch;
 601                         $oNewSearch->setPoiSearch(
 602                             Operator::TYPE,
 603                             $aSearchTerm['class'],
 604                             $aSearchTerm['type']
 605                         );
 606                         $aNewSearches[] = $oNewSearch;
 607                     }
 608                 }
 609                 $aSearches = $aNewSearches;
 610             }
 611
 612             // Split query into phrases
 613             // Commas are used to reduce the search space by indicating where phrases split
 614             if ($this->aStructuredQuery) {
 615                 $aInPhrases = $this->aStructuredQuery;
 616                 $bStructuredPhrases = true;
 617             } else {
 618                 $aInPhrases = explode(',', $sQuery);
 619                 $bStructuredPhrases = false;
 620             }
 621
 622             Debug::printDebugArray('Search context', $oCtx);
 623             Debug::printDebugArray('Base search', $aSearches[0]);
 624             Debug::printVar('Final query phrases', $aInPhrases);
 625
 626             // Convert each phrase to standard form
 627             // Create a list of standard words
 628             // Get all 'sets' of words
 629             // Generate a complete list of all
 630             Debug::newSection('Tokenization');
 631             $aTokens = array();
 632             $aPhrases = array();
 633             foreach ($aInPhrases as $iPhrase => $sPhrase) {
 634                 $sPhrase = chksql(
 635                     $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
 636                     'Cannot normalize query string (is it a UTF-8 string?)'
 637                 );
 638                 if (trim($sPhrase)) {
 639                     $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : '');
 640                     $oPhrase->addTokens($aTokens);
 641                     $aPhrases[] = $oPhrase;
 642                 }
 643             }
 644
 645             Debug::printDebugTable('Phrases', $aPhrases);
 646             Debug::printVar('Tokens', $aTokens);
 647
 648             if (!empty($aTokens)) {
 649                 // Check which tokens we have, get the ID numbers
 650                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
 651                 $sSQL .= ' FROM word ';
 652                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 653
 654                 Debug::printSQL($sSQL);
 655
 656                 $aValidTokens = array();
 657                 $aDatabaseWords = chksql(
 658                     $this->oDB->getAll($sSQL),
 659                     'Could not get word tokens.'
 660                 );
 661                 $aWordFrequencyScores = array();
 662                 foreach ($aDatabaseWords as $aToken) {
 663                     // Filter country tokens that do not match restricted countries.
 664                     if ($this->aCountryCodes
 665                         && $aToken['country_code']
 666                         && !in_array($aToken['country_code'], $this->aCountryCodes)
 667                     ) {
 668                         continue;
 669                     }
 670
 671                     // Special terms need to appear in their normalized form.
 672                     if ($aToken['word'] && $aToken['class']) {
 673                         $sNormWord = $this->normTerm($aToken['word']);
 674                         if (strpos($sNormQuery, $sNormWord) === false) {
 675                             continue;
 676                         }
 677                     }
 678
 679                     if (isset($aValidTokens[$aToken['word_token']])) {
 680                         $aValidTokens[$aToken['word_token']][] = $aToken;
 681                     } else {
 682                         $aValidTokens[$aToken['word_token']] = array($aToken);
 683                     }
 684                     $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
 685                 }
 686
 687                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
 688                 foreach ($aTokens as $sToken) {
 689                     if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
 690                         if (isset($aValidTokens[$aData[1]])) {
 691                             foreach ($aValidTokens[$aData[1]] as $aToken) {
 692                                 if (!$aToken['class']) {
 693                                     if (isset($aValidTokens[$sToken])) {
 694                                         $aValidTokens[$sToken][] = $aToken;
 695                                     } else {
 696                                         $aValidTokens[$sToken] = array($aToken);
 697                                     }
 698                                 }
 699                             }
 700                         }
 701                     }
 702                 }
 703
 704                 foreach ($aTokens as $sToken) {
 705                     // Unknown single word token with a number - assume it is a house number
 706                     if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) {
 707                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
 708                     }
 709                 }
 710                 Debug::printGroupTable('Valid Tokens', $aValidTokens);
 711
 712                 // Any words that have failed completely?
 713                 // TODO: suggestions
 714                 Debug::newSection('Search candidates');
 715
 716                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 717
 718                 if ($this->bReverseInPlan) {
 719                     // Reverse phrase array and also reverse the order of the wordsets in
 720                     // the first and final phrase. Don't bother about phrases in the middle
 721                     // because order in the address doesn't matter.
 722                     $aPhrases = array_reverse($aPhrases);
 723                     $aPhrases[0]->invertWordSets();
 724                     if (count($aPhrases) > 1) {
 725                         $aPhrases[count($aPhrases)-1]->invertWordSets();
 726                     }
 727                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 728
 729                     foreach ($aGroupedSearches as $aSearches) {
 730                         foreach ($aSearches as $aSearch) {
 731                             if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
 732                                 $aReverseGroupedSearches[$aSearch->getRank()] = array();
 733                             }
 734                             $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
 735                         }
 736                     }
 737
 738                     $aGroupedSearches = $aReverseGroupedSearches;
 739                     ksort($aGroupedSearches);
 740                 }
 741             } else {
 742                 // Re-group the searches by their score, junk anything over 20 as just not worth trying
 743                 $aGroupedSearches = array();
 744                 foreach ($aSearches as $aSearch) {
 745                     if ($aSearch->getRank() < $this->iMaxRank) {
 746                         if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
 747                         $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
 748                     }
 749                 }
 750                 ksort($aGroupedSearches);
 751             }
 752
 753             // Filter out duplicate searches
 754             $aSearchHash = array();
 755             foreach ($aGroupedSearches as $iGroup => $aSearches) {
 756                 foreach ($aSearches as $iSearch => $aSearch) {
 757                     $sHash = serialize($aSearch);
 758                     if (isset($aSearchHash[$sHash])) {
 759                         unset($aGroupedSearches[$iGroup][$iSearch]);
 760                         if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);
 761                     } else {
 762                         $aSearchHash[$sHash] = 1;
 763                     }
 764                 }
 765             }
 766
 767             if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
 768
 769             // Start the search process
 770             $iGroupLoop = 0;
 771             $iQueryLoop = 0;
 772             foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
 773                 $iGroupLoop++;
 774                 foreach ($aSearches as $oSearch) {
 775                     $iQueryLoop++;
 776
 777                     if (CONST_Debug) {
 778                         echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
 779                         _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
 780                     }
 781
 782                     $aResults += $oSearch->query(
 783                         $this->oDB,
 784                         $aWordFrequencyScores,
 785                         $this->iMinAddressRank,
 786                         $this->iMaxAddressRank,
 787                         $this->iLimit
 788                     );
 789
 790                     if ($iQueryLoop > 20) break;
 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                         if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
 805                             $sSQL .= "     OR (extratags->'place') = 'city'";
 806                         }
 807                         if ($this->aAddressRankList) {
 808                             $sSQL .= '     OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
 809                         }
 810                         $sSQL .= ')';
 811                         $aFilterSql[] = $sSQL;
 812                     }
 813                     $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE);
 814                     if ($sPlaceIds) {
 815                         $sSQL = ' SELECT place_id FROM location_postcode lp ';
 816                         $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') ';
 817                         $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
 818                         if ($this->aAddressRankList) {
 819                             $sSQL .= '     OR lp.rank_address in ('.join(',', $this->aAddressRankList).')';
 820                         }
 821                         $sSQL .= ') ';
 822                         $aFilterSql[] = $sSQL;
 823                     }
 824
 825                     $aFilteredIDs = array();
 826                     if ($aFilterSql) {
 827                         $sSQL = join(' UNION ', $aFilterSql);
 828                         Debug::printSQL($sSQL);
 829                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
 830                     }
 831
 832                     $tempIDs = array();
 833                     foreach ($aResults as $oResult) {
 834                         if (($this->iMaxAddressRank == 30 &&
 835                              ($oResult->iTable == Result::TABLE_OSMLINE
 836                               || $oResult->iTable == Result::TABLE_AUX
 837                               || $oResult->iTable == Result::TABLE_TIGER))
 838                             || in_array($oResult->iId, $aFilteredIDs)
 839                         ) {
 840                             $tempIDs[$oResult->iId] = $oResult;
 841                         }
 842                     }
 843                     $aResults = $tempIDs;
 844                 }
 845
 846                 if (!empty($aResults)) break;
 847                 if ($iGroupLoop > 4) break;
 848                 if ($iQueryLoop > 30) break;
 849             }
 850         } else {
 851             // Just interpret as a reverse geocode
 852             $oReverse = new ReverseGeocode($this->oDB);
 853             $oReverse->setZoom(18);
 854
 855             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 856
 857             Debug::printVar('Reverse search', $oLookup);
 858
 859             if ($oLookup) {
 860                 $aResults = array($oLookup->iId => $oLookup);
 861             }
 862         }
 863
 864         // No results? Done
 865         if (empty($aResults)) {
 866             if ($this->bFallback) {
 867                 if ($this->fallbackStructuredQuery()) {
 868                     return $this->lookup();
 869                 }
 870             }
 871
 872             return array();
 873         }
 874
 875         if ($this->aAddressRankList) {
 876             $this->oPlaceLookup->setAddressRankList($this->aAddressRankList);
 877         }
 878         $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList);
 879         $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder);
 880         if ($oCtx->hasNearPoint()) {
 881             $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear);
 882         }
 883
 884         $aSearchResults = $this->oPlaceLookup->lookup($aResults);
 885
 886         $aClassType = getClassTypesWithImportance();
 887         $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
 888         foreach ($aRecheckWords as $i => $sWord) {
 889             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
 890         }
 891
 892         Debug::printVar('Recheck words', $aRecheckWords);
 893
 894         foreach ($aSearchResults as $iIdx => $aResult) {
 895             // Default
 896             $fDiameter = getResultDiameter($aResult);
 897
 898             $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
 899             if ($aOutlineResult) {
 900                 $aResult = array_merge($aResult, $aOutlineResult);
 901             }
 902
 903             if ($aResult['extra_place'] == 'city') {
 904                 $aResult['class'] = 'place';
 905                 $aResult['type'] = 'city';
 906                 $aResult['rank_search'] = 16;
 907             }
 908
 909             // Is there an icon set for this type of result?
 910             if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon'])
 911                 && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']
 912             ) {
 913                 $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
 914             }
 915
 916             if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
 917                 && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
 918             ) {
 919                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'];
 920             } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label'])
 921                 && $aClassType[$aResult['class'].':'.$aResult['type']]['label']
 922             ) {
 923                 $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
 924             }
 925             // if tag '&addressdetails=1' is set in query
 926             if ($this->bIncludeAddressDetails) {
 927                 // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
 928                 $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber);
 929                 if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
 930                     $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
 931                 }
 932             }
 933
 934             $aResult['name'] = $aResult['langaddress'];
 935
 936             if ($oCtx->hasNearPoint()) {
 937                 $aResult['importance'] = 0.001;
 938                 $aResult['foundorder'] = $aResult['addressimportance'];
 939             } else {
 940                 $aResult['importance'] = max(0.001, $aResult['importance']);
 941                 $aResult['importance'] *= $this->viewboxImportanceFactor(
 942                     $aResult['lon'],
 943                     $aResult['lat']
 944                 );
 945                 // Adjust importance for the number of exact string matches in the result
 946                 $iCountWords = 0;
 947                 $sAddress = $aResult['langaddress'];
 948                 foreach ($aRecheckWords as $i => $sWord) {
 949                     if (stripos($sAddress, $sWord)!==false) {
 950                         $iCountWords++;
 951                         if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1;
 952                     }
 953                 }
 954
 955                 $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
 956
 957                 // secondary ordering (for results with same importance (the smaller the better):
 958                 // - approximate importance of address parts
 959                 $aResult['foundorder'] = -$aResult['addressimportance']/10;
 960                 // - number of exact matches from the query
 961                 $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
 962                 // - importance of the class/type
 963                 if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
 964                     && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
 965                 ) {
 966                     $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
 967                 } else {
 968                     $aResult['foundorder'] += 0.01;
 969                 }
 970             }
 971             $aSearchResults[$iIdx] = $aResult;
 972         }
 973         uasort($aSearchResults, 'byImportance');
 974         Debug::printVar('Pre-filter results', $aSearchResults);
 975
 976         $aOSMIDDone = array();
 977         $aClassTypeNameDone = array();
 978         $aToFilter = $aSearchResults;
 979         $aSearchResults = array();
 980
 981         $bFirst = true;
 982         foreach ($aToFilter as $aResult) {
 983             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
 984             if ($bFirst) {
 985                 $fLat = $aResult['lat'];
 986                 $fLon = $aResult['lon'];
 987                 if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
 988                 $bFirst = false;
 989             }
 990             if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
 991                 && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
 992             ) {
 993                 $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
 994                 $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
 995                 $aSearchResults[] = $aResult;
 996             }
 997
 998             // Absolute limit on number of results
 999             if (count($aSearchResults) >= $this->iFinalLimit) break;
1000         }
1001
1002         Debug::printVar('Post-filter results', $aSearchResults);
1003         return $aSearchResults;
1004     } // end lookup()
1005
1006     public function debugInfo()
1007     {
1008         return array(
1009                 'Query' => $this->sQuery,
1010                 'Structured query' => $this->aStructuredQuery,
1011                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
1012                 'Include address' => $this->bIncludeAddressDetails,
1013                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
1014                 'Try reversed query'=> $this->bReverseInPlan,
1015                 'Limit (for searches)' => $this->iLimit,
1016                 'Limit (for results)'=> $this->iFinalLimit,
1017                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
1018                 'Bounded search' => $this->bBoundedSearch,
1019                 'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
1020                 'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
1021                 'Route width' => $this->aRouteWidth,
1022                 'Max rank' => $this->iMaxRank,
1023                 'Min address rank' => $this->iMinAddressRank,
1024                 'Max address rank' => $this->iMaxAddressRank,
1025                 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
1026                );
1027     }
1028 } // end class