X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/55ca68671d495ad05559a010ae9884a7e2307be6..b19e1907624f6ad06252b4a385598f73d743ed36:/website/search.php
diff --git a/website/search.php b/website/search.php
index f0bcf077..c165b92b 100755
--- a/website/search.php
+++ b/website/search.php
@@ -36,11 +36,12 @@
// 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'])) $bReverseInPlan = true;
+ if (isset($aLangPrefOrder['name:pl'])) $bReverseInPlan = true;
$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]";
@@ -53,7 +54,7 @@
}
}
- // 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 +93,11 @@
// 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 +110,9 @@
$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
@@ -134,6 +134,8 @@
$aCoOrdinates[1] += $fWidth;
$aCoOrdinates[3] -= $fWidth;
$sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aCoOrdinates[0].",".(float)$aCoOrdinates[1]."),ST_Point(".(float)$aCoOrdinates[2].",".(float)$aCoOrdinates[3].")),4326)";
+ } else {
+ $bBoundingBoxSearch = false;
}
if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth'])
{
@@ -244,7 +246,8 @@
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
$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';
+ $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\')) or country_code is not null';
+ if (CONST_Debug) var_Dump($sSQL);
$aSearchWords = $oDB->getAll($sSQL);
$aNewSearches = array();
foreach($aSearches as $aSearch)
@@ -310,12 +313,7 @@
// 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)).')';
- // 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 .= ' and search_name_count < '.CONST_Max_Word_Frequency;
// $sSQL .= ' group by word_token, word, class, type, location, country_code';
if (CONST_Debug) var_Dump($sSQL);
@@ -376,7 +374,8 @@
// 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) != ' ')
{
@@ -418,17 +417,16 @@
Score how good the search is so they can be ordered
*/
-
foreach($aPhrases as $iPhrase => $sPhrase)
{
$aNewPhraseSearches = array();
- foreach($aPhrases[$iPhrase]['wordsets'] as $iWordset => $aWordset)
+ foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset)
{
$aWordsetSearches = $aSearches;
// Add all words from this wordset
- foreach($aWordset as $sToken)
+ foreach($aWordset as $iToken => $sToken)
{
//echo "
$sToken";
$aNewWordsetSearches = array();
@@ -451,12 +449,12 @@
if ($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)
- if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
+ // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
+ if ($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
- elseif ($aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
+ elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
{
if ($aSearch['fLat'] === '')
{
@@ -505,7 +503,7 @@
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
- else
+ elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
{
if (sizeof($aSearch['aName']))
{
@@ -532,6 +530,8 @@
// 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)
@@ -551,6 +551,7 @@
$aSearch['iNamePhrase'] = $iPhrase;
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
+ }
}
}
else
@@ -705,22 +706,22 @@
$sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
if ($oDB->getOne($sSQL))
{
- $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType'];
+ $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
- $sSQL .= " where st_contains($sViewboxSmallSQL, centroid)";
+ $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)";
if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
- if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
+ 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))
{
- $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType'];
+ $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
- $sSQL .= " where st_contains($sViewboxLargeSQL, centroid)";
+ $sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)";
if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
- if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
+ if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $oDB->getCol($sSQL);
@@ -729,7 +730,7 @@
else
{
$sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
- $sSQL .= " and st_contains($sViewboxSmallSQL, centroid)";
+ $sSQL .= " and st_contains($sViewboxSmallSQL, geometry)";
if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
$sSQL .= " limit $iLimit";
@@ -747,6 +748,9 @@
// 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'])."'";
@@ -1015,6 +1019,7 @@
$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, $sLanguagePrefArraySQL) ";
@@ -1135,20 +1140,26 @@
//var_Dump($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($aPointPolygon))
{
@@ -1156,6 +1167,10 @@
}
if ($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))
{
@@ -1251,6 +1266,16 @@
//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*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
+
//if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
/*
if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance'])
@@ -1272,7 +1297,6 @@
$aResult['foundorder'] = $iResNum;
$aSearchResults[$iResNum] = $aResult;
}
-
uasort($aSearchResults, 'byImportance');
//var_dump($aSearchResults);exit;
@@ -1295,10 +1319,10 @@
$bFirst = false;
}
if (!$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;
}