X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/9eb5b62382a93d07d4c9643e49a8c764a3e50b93..3b960a5bf343f5617326f162c6df3399dce34930:/website/search.php?ds=sidebyside diff --git a/website/search.php b/website/search.php index c4d37f9b..a92ac30a 100755 --- a/website/search.php +++ b/website/search.php @@ -1,8 +1,11 @@ 100) $iLimit = 100; + if ($iFinalLimit > 50) $iFinalLimit = 50; + $iLimit = $iFinalLimit + min($iFinalLimit, 10); $iMinAddressRank = 0; $iMaxAddressRank = 30; + $sAllowedTypesSQLList = false; // Format for output if (isset($_GET['format']) && ($_GET['format'] == 'html' || $_GET['format'] == 'xml' || $_GET['format'] == 'json' || $_GET['format'] == 'jsonv2')) @@ -66,6 +71,7 @@ if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true; if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true; + if (isset($aLangPrefOrder['name:pl'])) $bReverseInPlan = true; $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]"; @@ -151,12 +157,14 @@ array('postalcode', 16, 25), ); $aStructuredQuery = array(); + $sAllowedTypesSQLList = ''; foreach($aStructuredOptions as $aStructuredOption) { loadStructuredAddressElement($aStructuredQuery, $iMinAddressRank, $iMaxAddressRank, $_GET, $aStructuredOption[0], $aStructuredOption[1], $aStructuredOption[2]); } if (sizeof($aStructuredQuery) > 0) { $sQuery = join(', ', $aStructuredQuery); + $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } if ($sQuery) @@ -317,7 +325,7 @@ { foreach($aSearchWords as $aSearchTerm) { - $aNewSearch = $aSearch; + $aNewSearch = $aSearch; if ($aSearchTerm['country_code']) { $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); @@ -385,7 +393,7 @@ { // Check which tokens we have, get the ID numbers - $sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator'; + $sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator, search_name_count'; $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')'; $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency; // $sSQL .= ' group by word_token, word, class, type, location, country_code'; @@ -412,7 +420,7 @@ { $aValidTokens[$aToken['word_token']] = array($aToken); } - if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1; + if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1 + $aToken['search_name_count']; } if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); @@ -520,7 +528,7 @@ { $aSearch = $aCurrentSearch; $aSearch['iSearchRank']++; - if (($sPhraseType == '' || $sPhraseType == 'country') && $aSearchTerm['country_code'] !== null && $aSearchTerm['country_code'] != '0') + if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0') { if ($aSearch['sCountryCode'] === false) { @@ -727,6 +735,42 @@ } } + if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0) + { + $aCopyGroupedSearches = $aGroupedSearches; + foreach($aCopyGroupedSearches as $iGroup => $aSearches) + { + foreach($aSearches as $iSearch => $aSearch) + { + $aReductionsList = array($aSearch['aAddress']); + $iSearchRank = $aSearch['iSearchRank']; + while(sizeof($aReductionsList) > 0) + { + $iSearchRank += 5; + if ($iSearchRank > iMaxRank) break 3; + $aNewReductionsList = array(); + foreach($aReductionsList as $aReductionsWordList) + { + for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) + { + $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1)); + $aReverseSearch = $aSearch; + $aSearch['aAddress'] = $aReductionsWordListResult; + $aSearch['iSearchRank'] = $iSearchRank; + $aGroupedSearches[$iSearchRank][] = $aReverseSearch; + if (sizeof($aReductionsWordListResult) > 0) + { + $aNewReductionsList[] = $aReductionsWordListResult; + } + } + } + $aReductionsList = $aNewReductionsList; + } + } + } + ksort($aGroupedSearches); + } + // Filter out duplicate searches $aSearchHash = array(); foreach($aGroupedSearches as $iGroup => $aSearches) @@ -760,7 +804,7 @@ // Must have a location term if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) { - if ($aSearch['sCountryCode'] && !$aSearch['sClass']) + if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) { if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank) { @@ -817,7 +861,7 @@ if (CONST_Debug) var_dump('
',$aSearch); if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); $aPlaceIDs = array(); - + // First we need a position, either aName or fLat or both $aTerms = array(); $aOrder = array(); @@ -825,7 +869,20 @@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($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 (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) + { + // For infrequent name terms disable index usage for address + if (CONST_Search_NameOnlySearchFrequencyThreshold && + sizeof($aSearch['aName']) == 1 && + $aPossibleMainWordIDs[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold) + { + $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]"; + } + else + { + $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; + } + } if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)"; if ($aSearch['fLon'] && $aSearch['fLat']) @@ -845,7 +902,7 @@ if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end'; + $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end'; if ($sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END"; if ($sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END"; @@ -865,11 +922,16 @@ $sSQL .= " limit ".$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? $aPlaceIDs = array(); @@ -998,13 +1060,13 @@ if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.centroid)"; else if ($sPlaceIDs) $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)"; else if ($sPlaceGeom) $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)"; - + $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l"; if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)"; if ($sPlaceIDs) { $sSQL .= ",placex as f where "; - $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) "; + $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, f.centroid, $fRange) "; } if ($sPlaceGeom) { @@ -1031,7 +1093,7 @@ else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)"; $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where "; - $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) "; + $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, f.centroid, $fRange) "; $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' "; if (sizeof($aExcludePlaceIDs)) { @@ -1089,11 +1151,12 @@ $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; - $sSQL .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, "; + $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; // $sSQL .= $sOrderSQL." as porder, "; - $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance "; + $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; + if ($sAllowedTypesSQLList) $sSQL .= "and placex.class in $sAllowedTypesSQLList "; $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"; @@ -1158,9 +1221,9 @@ $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; - $sSQL .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, "; + $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; // $sSQL .= $sOrderSQL." as porder, "; - $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance "; + $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank "; $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance"; @@ -1390,8 +1453,6 @@ } uasort($aSearchResults, 'byImportance'); -//var_dump($aSearchResults);exit; - $aOSMIDDone = array(); $aClassTypeNameDone = array(); $aToFilter = $aSearchResults; @@ -1418,7 +1479,7 @@ } // Absolute limit on number of results - if (sizeof($aSearchResults) >= $iLimit) break; + if (sizeof($aSearchResults) >= $iFinalLimit) break; } $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1"); @@ -1433,7 +1494,7 @@ logEnd($oDB, $hLog, sizeof($aToFilter)); } $sMoreURL = CONST_Website_BaseURL.'search?format='.urlencode($sOutputFormat).'&exclude_place_ids='.join(',',$aExcludePlaceIDs); - $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"]; + if (isset($_SERVER["HTTP_ACCEPT_LANGUAGE"])) $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"]; if ($bShowPolygons) $sMoreURL .= '&polygon=1'; if ($bShowAddressDetails) $sMoreURL .= '&addressdetails=1'; if (isset($_GET['viewbox']) && $_GET['viewbox']) $sMoreURL .= '&viewbox='.urlencode($_GET['viewbox']);