X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/4d7e0d97b2a644cf5479c3301eb20f5f46d6509e..932d945b7f897fd18be7d4c2fac3eda3915aa296:/website/search.php diff --git a/website/search.php b/website/search.php index 50d009e1..cc73513e 100755 --- a/website/search.php +++ b/website/search.php @@ -36,8 +36,8 @@ // Show address breakdown $bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails']; - // Prefered language - $aLangPrefOrder = getPrefferedLangauges(); + // Preferred language + $aLangPrefOrder = getPreferredLanguages(); if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true; @@ -53,7 +53,7 @@ } } - // Only certain ranks of feature + // Only certain ranks of feature if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType']; if (isset($_GET['featuretype'])) @@ -92,11 +92,11 @@ // Search query $sQuery = (isset($_GET['q'])?trim($_GET['q']):''); - if (!$sQuery && $_SERVER['PATH_INFO'] && $_SERVER['PATH_INFO'][0] == '/') + if (!$sQuery && isset($_SERVER['PATH_INFO']) && $_SERVER['PATH_INFO'][0] == '/') { $sQuery = substr($_SERVER['PATH_INFO'], 1); - // reverse order of '/' seperated string + // reverse order of '/' separated string $aPhrases = explode('/', $sQuery); $aPhrases = array_reverse($aPhrases); $sQuery = join(', ',$aPhrases); @@ -109,10 +109,9 @@ $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $sQuery); if (isset($aLangPrefOrder['name:en'])) { - $sQuery = preg_replace('/\\bil\\b/','illinois', $sQuery); - $sQuery = preg_replace('/\\bal\\b/','alabama', $sQuery); - $sQuery = preg_replace('/\\bla\\b/','louisiana', $sQuery); - $sQuery = preg_replace('/\\bde louisiana\\b/','de la', $sQuery); + $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery); + $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery); + $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery); } // If we have a view box create the SQL @@ -245,6 +244,7 @@ $sToken = $oDB->getOne("select make_standard_name('".$aSpecialTerm[1]."') as string"); $sSQL = 'select * from (select word_id,word_token, word, class, type, location, country_code, operator'; $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\',\'highway\')) or country_code is not null'; + if (CONST_Debug) var_Dump($sSQL); $aSearchWords = $oDB->getAll($sSQL); $aNewSearches = array(); foreach($aSearches as $aSearch) @@ -310,7 +310,12 @@ // Check which tokens we have, get the ID numbers $sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator'; $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')'; - $sSQL .= ' and (class is null or class not in (\'highway\'))'; + // HACK WARNING + // (mis)using search_name_count to exclude words that return too many + // search results. saerch_name_count is currently set to 1 by hand + // because there is no fast way to extract this count from a live database. + $sSQL .= ' and search_name_count = 0'; +// $sSQL .= ' and (class is null or class not in (\'highway\'))'; // $sSQL .= ' group by word_token, word, class, type, location, country_code'; if (CONST_Debug) var_Dump($sSQL); @@ -371,7 +376,8 @@ // Try and calculate GB postcodes we might be missing foreach($aTokens as $sToken) { - if (!isset($aValidTokens[$sToken]) && !isset($aValidTokens[' '.$sToken]) && preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) + // Source of gb postcodes is now definitive - always use + if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) { if (substr($aData[1],-2,1) != ' ') { @@ -413,7 +419,6 @@ Score how good the search is so they can be ordered */ - foreach($aPhrases as $iPhrase => $sPhrase) { $aNewPhraseSearches = array(); @@ -446,12 +451,12 @@ if ($aSearch['sCountryCode'] === false) { $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); - // Country is almost always at the end of the string - increase score for finding it anywhere else (opimisation) + // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation) if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - elseif ($aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) + elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) { if ($aSearch['fLat'] === '') { @@ -500,7 +505,7 @@ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } } - else + elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { if (sizeof($aSearch['aName'])) { @@ -527,6 +532,8 @@ // Allow searching for a word - but at extra cost foreach($aValidTokens[$sToken] as $aSearchTerm) { + if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) + { //var_Dump('
',$aSearch['aName']); if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) @@ -546,6 +553,7 @@ $aSearch['iNamePhrase'] = $iPhrase; if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch; } + } } } else @@ -742,6 +750,9 @@ // First we need a position, either aName or fLat or both $aTerms = array(); $aOrder = array(); + + // TODO: filter out the pointless search terms (2 letter name tokens and less) + // they might be right - but they are just too darned expensive to run if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]"; if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; @@ -783,11 +794,16 @@ $sSQL .= " limit ".$iLimit; if (CONST_Debug) var_dump($sSQL); + $iStartTime = time(); $aViewBoxPlaceIDs = $oDB->getAll($sSQL); if (PEAR::IsError($aViewBoxPlaceIDs)) { failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs); } + if (time() - $iStartTime > 60) { + file_put_contents(CONST_BasePath.'/log/long_queries.log', date('Y-m-d H:i:s', $iStartTime).' '.$sSQL."\n", FILE_APPEND); + } + //var_dump($aViewBoxPlaceIDs); // Did we have an viewbox matches? $aPlaceIDs = array(); @@ -810,7 +826,7 @@ $sPlaceIDs = join(',',$aPlaceIDs); // Now they are indexed look for a house attached to a street we found - $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M'; + $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M'; $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'"; if (sizeof($aExcludePlaceIDs)) { @@ -1005,6 +1021,7 @@ $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; + $sSQL .= "and linked_place_id is null "; $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance"; if (!$bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) "; @@ -1125,20 +1142,26 @@ //var_Dump($aSearchResults); //exit; $aClassType = getClassTypesWithImportance(); + $aRecheckWords = preg_split('/\b/',$sQuery); + foreach($aRecheckWords as $i => $sWord) + { + if (!$sWord) unset($aRecheckWords[$i]); + } foreach($aSearchResults as $iResNum => $aResult) { - if (CONST_Search_AreaPolygons || true) + if (CONST_Search_AreaPolygons) { // Get the bounding box and outline polygon $sSQL = "select place_id,numfeatures,area,outline,"; - $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),2)) as maxlat,"; - $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),3)) as maxlon,"; + $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),2)) as maxlat,"; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),3)) as maxlon,"; $sSQL .= "ST_AsText(outline) as outlinestring from get_place_boundingbox_quick(".$aResult['place_id'].")"; $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,"; - $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),2)) as maxlat,"; - $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),3)) as maxlon,"; - $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(ST_Box2D(geometry)) = \'ST_Polygon\''; + $sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,"; + $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,"; + $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon,"; + $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\''; $aPointPolygon = $oDB->getRow($sSQL); if (PEAR::IsError($aPointPolygon)) { @@ -1146,6 +1169,10 @@ } if ($aPointPolygon['place_id']) { + if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) { + $aResult['lat'] = $aPointPolygon['centrelat']; + $aResult['lon'] = $aPointPolygon['centrelon']; + } // Translate geometary string to point array if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch)) { @@ -1241,6 +1268,16 @@ //exit; } + // Adjust importance for the number of exact string matches in the result + $aResult['importance'] = max(0.001,$aResult['importance']); + $iCountWords = 0; + $sAddress = $aResult['langaddress']; + foreach($aRecheckWords as $i => $sWord) + { + if (stripos($sAddress, $sWord)!==false) $iCountWords++; + } + $aResult['importance'] = $aResult['importance'] + $iCountWords; + //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']); /* if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) @@ -1262,7 +1299,6 @@ $aResult['foundorder'] = $iResNum; $aSearchResults[$iResNum] = $aResult; } - uasort($aSearchResults, 'byImportance'); //var_dump($aSearchResults);exit; @@ -1285,10 +1321,10 @@ $bFirst = false; } if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']]) - && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']]))) + && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']]))) { $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true; - $aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']] = true; + $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true; $aSearchResults[] = $aResult; }