X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/fa1f23edfbd0672ae430722152314ecc69615b56..5502c223756ed755edae9670497f1456e528d3f0:/website/search.php
diff --git a/website/search.php b/website/search.php
index 5ba2571b..77a3d135 100755
--- a/website/search.php
+++ b/website/search.php
@@ -71,6 +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))."]";
@@ -163,7 +164,10 @@
}
if (sizeof($aStructuredQuery) > 0) {
$sQuery = join(', ', $aStructuredQuery);
- $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ if ($iMaxAddressRank < 30)
+ {
+ $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ }
}
if ($sQuery)
@@ -324,7 +328,7 @@
{
foreach($aSearchWords as $aSearchTerm)
{
- $aNewSearch = $aSearch;
+ $aNewSearch = $aSearch;
if ($aSearchTerm['country_code'])
{
$aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
@@ -392,7 +396,7 @@
{
// 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';
@@ -419,7 +423,7 @@
{
$aValidTokens[$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);
@@ -527,13 +531,18 @@
{
$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)
{
$aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
// Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
- if ($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
+ // If reverse order is enabled, it may appear at the beginning as well.
+ if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) &&
+ (!$bReverseInPlan || $iToken > 0 || $iPhrase > 0))
+ {
+ $aSearch['iSearchRank'] += 5;
+ }
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
@@ -800,6 +809,10 @@
{
$iQueryLoop++;
+ if (CONST_Debug) { echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop"; }
+ if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
+
+
// Must have a location term
if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon'])
{
@@ -807,9 +820,10 @@
{
if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank)
{
- $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " order by st_area(geometry) desc limit 1";
+ if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $oDB->getCol($sSQL);
}
}
@@ -817,26 +831,31 @@
{
if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue;
if (!$aSearch['sClass']) continue;
- if (CONST_Debug) var_dump('
',$aSearch);
- if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
$sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
if ($oDB->getOne($sSQL))
{
$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
$sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
+ if (sizeof($aExcludePlaceIDs))
+ {
+ $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")";
+ }
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $oDB->getCol($sSQL);
- if (!sizeof($aPlaceIDs))
+ // If excluded place IDs are given, it is fair to assume that
+ // there have been results in the small box, so no further
+ // expansion in that case.
+ if (!sizeof($aPlaceIDs) && !sizeof($aExcludePlaceIDs))
{
$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
$sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
@@ -847,7 +866,7 @@
{
$sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
$sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
@@ -857,10 +876,8 @@
}
else
{
- 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();
$aOrder = array();
@@ -868,7 +885,20 @@
// TODO: filter out the pointless search terms (2 letter name tokens and less)
// they might be right - but they are just too darned expensive to run
if (sizeof($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'])
@@ -907,12 +937,17 @@
else
$sSQL .= " limit ".$iLimit;
- if (CONST_Debug) { var_dump($sSQL); }
+ if (CONST_Debug) var_dump($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();
@@ -989,7 +1024,7 @@
// If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match
$sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
$sSQL .= " and linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " order by rank_search asc limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
$aClassPlaceIDs = $oDB->getCol($sSQL);
@@ -1041,13 +1076,13 @@
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)
{
@@ -1058,7 +1093,7 @@
{
$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
if ($iOffset) $sSQL .= " offset $iOffset";
$sSQL .= " limit $iLimit";
@@ -1074,13 +1109,13 @@
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))
{
$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
if ($iOffset) $sSQL .= " offset $iOffset";
$sSQL .= " limit $iLimit";
@@ -1101,7 +1136,7 @@
failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs);
}
- if (CONST_Debug) var_Dump($aPlaceIDs);
+ if (CONST_Debug) { echo "
Place IDs: "; var_Dump($aPlaceIDs); }
foreach($aPlaceIDs as $iPlaceID)
{
@@ -1115,7 +1150,7 @@
if ($iGroupLoop > 4) break;
if ($iQueryLoop > 30) break;
}
-//exit;
+
// Did we find anything?
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs))
{
@@ -1128,18 +1163,18 @@
$sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' ';
}
$sOrderSQL .= ' ELSE 10000000 END';
- $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,country_code,";
+ $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,";
$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,0.75-(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";
+ $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
if (!$bDeDupe) $sSQL .= ",place_id";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
@@ -1171,7 +1206,7 @@
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= "order by importance desc";
// $sSQL .= "order by rank_search,rank_address,porder asc";
- if (CONST_Debug) var_dump('
',$sSQL);
+ if (CONST_Debug) { echo "
"; var_dump($sSQL); }
$aSearchResults = $oDB->getAll($sSQL);
//var_dump($sSQL,$aSearchResults);exit;
@@ -1198,16 +1233,16 @@
$sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' ';
}
$sOrderSQL .= ' ELSE 10000000 END';
- $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,country_code,";
+ $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,";
$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,0.75-(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";
+ $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
if (!$bDeDupe) $sSQL .= ",place_id";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
@@ -1239,7 +1274,7 @@
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= "order by importance desc";
// $sSQL .= "order by rank_search,rank_address,porder asc";
- if (CONST_Debug) var_dump('
',$sSQL);
+ if (CONST_Debug) { echo "
", var_dump($sSQL); }
$aSearchResults = $oDB->getAll($sSQL);
//var_dump($sSQL,$aSearchResults);exit;