]> git.openstreetmap.org Git - nominatim.git/blobdiff - website/search.php
calculate search position based to 'importance' rather than address rank
[nominatim.git] / website / search.php
index 8c92919c85086c782a4062ba35c4171fcbf5da51..80a77dfc65ea941a13d791b7ade7195a23315087 100755 (executable)
@@ -20,6 +20,8 @@
        $iOffset = isset($_GET['offset'])?(int)$_GET['offset']:0;
        $iMaxRank = 20;
        if ($iLimit > 100) $iLimit = 100;
+       $iMinAddressRank = 0;
+       $iMaxAddressRank = 30;
 
        // Format for output
        if (isset($_GET['format']) && ($_GET['format'] == 'html' || $_GET['format'] == 'xml' || $_GET['format'] == 'json' ||  $_GET['format'] == 'jsonv2'))
 
        // Prefered language    
        $aLangPrefOrder = getPrefferedLangauges();
-//     if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
+       if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
+       if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true;
+       if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true;
+
+$bReverseInPlan = true;
+
        $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]";
 
        if (isset($_GET['exclude_place_ids']) && $_GET['exclude_place_ids'])
                        if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
                }
        }
+
+  // Only certain ranks of feature
+       if (isset($_GET['featuretype']))
+       {
+               switch($_GET['featuretype'])
+               {
+               case 'country':
+                       $iMinAddressRank = $iMaxAddressRank = 4;
+                       break;
+               case 'state':
+                       $iMinAddressRank = $iMaxAddressRank = 8;
+                       break;
+               case 'city':
+                       $iMinAddressRank = 14;
+                       $iMaxAddressRank = 16;
+                       break;
+               case 'settlement':
+                       $iMinAddressRank = 8;
+                       $iMaxAddressRank = 20;
+                       break;
+               }
+       }
                
        // Search query
        $sQuery = (isset($_GET['q'])?trim($_GET['q']):'');
@@ -64,7 +93,7 @@
                $hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
 
                // Hack to make it handle "new york, ny" (and variants) correctly
-                $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, new york', $sQuery);
+                $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $sQuery);
 
                // If we have a view box create the SQL
                // Small is the actual view box, Large is double (on each axis) that 
                                var_dump($sSQL, $aDatabaseWords);
                                exit;
                        }
+                       $aPossibleMainWordIDs = array();
                        foreach($aDatabaseWords as $aToken)
                        {
                                if (isset($aValidTokens[$aToken['word_token']]))
                                {
                                        $aValidTokens[$aToken['word_token']] = array($aToken);
                                }
+                               if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
                        }
                        if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
                                                                                {
                                                                                        if (sizeof($aSearch['aName']))
                                                                                        {
-                                                                                               if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4)
+                                                                                               if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)
                                                                                                {
                                                                                                        $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                }
                                                                                }
                                                                        }
                                                                }
-                                                               if (isset($aValidTokens[$sToken]) && strlen($sToken) >= 4)
+                                                               if (isset($aValidTokens[$sToken]))
                                                                {
                                                                        // Allow searching for a word - but at extra cost
                                                                        foreach($aValidTokens[$sToken] as $aSearchTerm)
                                                                        {
 //var_Dump('<hr>',$aSearch['aName']);
 
-                                                                               if (sizeof($aCurrentSearch['aName']))
+                                                                               if (sizeof($aCurrentSearch['aName'])  && strlen($sToken) >= 4)
                                                                                {
                                                                                $aSearch = $aCurrentSearch;
                                                                                        $aSearch['iSearchRank'] += 1;
                                                                                if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
                                                                                {
                                                                                $aSearch = $aCurrentSearch;
-                                                                                       $aSearch['iSearchRank'] += 4;
+                                                                                       $aSearch['iSearchRank'] += 2;
+                                                                                       if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
                                                                                        $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                        $aSearch['iNamePhrase'] = $iPhrase;
                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
 
                                                $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
                                                usort($aNewPhraseSearches, 'bySearchRank');
+
+          $aSearchHash = array();
+          foreach($aNewPhraseSearches as $iSearch => $aSearch)
+          {
+            $sHash = serialize($aSearch);
+            if (isset($aSearchHash[$sHash]))
+            {
+              unset($aNewPhraseSearches[$iSearch]);
+            }
+            else
+            {
+              $aSearchHash[$sHash] = 1;
+            }
+          }
+
                                                $aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
                                        }
 
                                
                                if (CONST_Debug) var_Dump($aGroupedSearches);
 
-                               if ($bReverseInPlan && false)
+                               if ($bReverseInPlan)
                                {
-                                       foreach($aGroupedSearches as $iGroup => $aSearches)
+                                       $aCopyGroupedSearches = $aGroupedSearches;
+                                       foreach($aCopyGroupedSearches as $iGroup => $aSearches)
                                        {
                                                foreach($aSearches as $iSearch => $aSearch)
                                                {
                                                        if (sizeof($aSearch['aAddress']))
                                                        {
-                                                               $aReverseSearch = $aSearch;
                                                                $iReverseItem = array_pop($aSearch['aAddress']);
-                                                               $aReverseSearch['aName'][$iReverseItem] = $iReverseItem;
-                                                               $aGroupedSearches[$iGroup][] = $aReverseSearch;
+                                                               if (isset($aPossibleMainWordIDs[$iReverseItem]))
+                                                               {
+                                                                       $aSearch['aAddress'] = array_merge($aSearch['aAddress'], $aSearch['aName']);
+                                                                       $aSearch['aName'] = array($iReverseItem);
+                                                                       $aGroupedSearches[$iGroup][] = $aSearch;
+                                                               }
+//                                                             $aReverseSearch['aName'][$iReverseItem] = $iReverseItem;
+       //                                                      $aGroupedSearches[$iGroup][] = $aReverseSearch;
                                                        }
                                                }
                                        }
                                                {
                                                        if ($aSearch['sCountryCode'] && !$aSearch['sClass'])
                                                        {
-                                                               $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4 order by st_area(geometry) desc limit 1";
-                                                               $aPlaceIDs = $oDB->getCol($sSQL);
+                                                               if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank)
+                                                               {
+                                                                       $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4 order by st_area(geometry) desc limit 1";
+                                                                       $aPlaceIDs = $oDB->getCol($sSQL);
+                                                               }
                                                        }
                                                        else
                                                        {
                                                        }
                                                        if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
                                                        if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
-                                                       if ($sViewboxSmallSQL) $aOrder[] = "ST_Contains($sViewboxSmallSQL, centroid) desc";
-                                                       if ($sViewboxLargeSQL) $aOrder[] = "ST_Contains($sViewboxLargeSQL, centroid) desc";
-                                                       $aOrder[] = "search_rank ASC";
+
+                                                       $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end';
+
+                                                       if ($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";
+                                                       $aOrder[] = "$sImportanceSQL DESC";
                                                
                                                        if (sizeof($aTerms))
                                                        {
                                                                $sSQL = "select place_id";
-                                                               if ($sViewboxSmallSQL) $sSQL .= ",ST_Contains($sViewboxSmallSQL, centroid) as in_small";
-                                                               else $sSQL .= ",false as in_small";
-                                                               if ($sViewboxLargeSQL) $sSQL .= ",ST_Contains($sViewboxLargeSQL, centroid) as in_large";
-                                                               else $sSQL .= ",false as in_large";
                                                                $sSQL .= " from search_name";
                                                                $sSQL .= " where ".join(' and ',$aTerms);
                                                                $sSQL .= " order by ".join(', ',$aOrder);
                                                                        var_dump($sSQL, $aViewBoxPlaceIDs);                                     
                                                                        exit;
                                                                }
-
+//var_dump($aViewBoxPlaceIDs);
                                                                // Did we have an viewbox matches?
                                                                $aPlaceIDs = array();
                                                                $bViewBoxMatch = false;
                                                                foreach($aViewBoxPlaceIDs as $aViewBoxRow)
                                                                {
-                                                                       if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
-                                                                       if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
-                                                                       if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
-                                                                       else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
+//                                                                     if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
+//                                                                     if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
+//                                                                     if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
+//                                                                     else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
                                                                        $aPlaceIDs[] = $aViewBoxRow['place_id'];
                                                                }
                                                        }
+//var_Dump($aPlaceIDs);
+//exit;
 
                                                        if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs))
                                                        {
                                                                        $aPlaceIDs = $oDB->getCol($sSQL);
                                                                        $sPlaceIDs = join(',',$aPlaceIDs);
 
+                                                                       if ($sPlaceIDs)
+                                                                       {
+
                                                                        $fRange = 0.01;
                                                                        $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
                                                                        if ($oDB->getOne($sSQL))
                                                                        {
                                                                                if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius'];
                                                                                $sSQL = "select l.place_id 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, st_centroid(f.geometry), $fRange) ";
                                                                                $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' ";
                                                                                if (sizeof($aExcludePlaceIDs))
                                                                                {
                                                                                if (CONST_Debug) var_dump($sSQL);
                                                                                $aPlaceIDs = $oDB->getCol($sSQL);
                                                                        }
+                                                                       }
                                                                }
                                                        }
                                                
                                        $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 .= $sOrderSQL." as porder ";
+                                       $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance ";
                                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
-                                       $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code";
+                                       $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
+                                       $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
                                        $sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
                                        $sSQL .= "null as placename,";
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-                                       $sSQL .= $sOrderSQL." as porder ";
+                                       $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "-0.15 as importance ";
                                        $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= " union ";
-                                       $sSQL .= "select 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,";
+                                       $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,";
                                        $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
                                        $sSQL .= "null as placename,";
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-                                       $sSQL .= $sOrderSQL." as porder ";
+                                       $sSQL .= $sOrderSQL." as porder, ";
+                                       $sSQL .= "-0.15 as importance ";
                                        $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
-                                       $sSQL .= "order by rank_search,rank_address,porder asc";
+                                       $sSQL .= "order by porder asc";
+//                                     $sSQL .= "order by rank_search,rank_address,porder asc";
                                        if (CONST_Debug) var_dump('<hr>',$sSQL);
                                        $aSearchResults = $oDB->getAll($sSQL);
 //var_dump($sSQL,$aSearchResults);exit;
        {
                $sSearchResult = 'No Results Found';
        }
-       
+//var_Dump($aSearchResults);
+//exit;
        $aClassType = getClassTypesWithImportance();
-
        foreach($aSearchResults as $iResNum => $aResult)
        {
                if (CONST_Search_AreaPolygons || true)
                }
 
 //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
-
+/*
                if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) 
                        && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance'])
                {
                {
                        $aResult['importance'] = 1000000000000000;
                }
-
+*/
                $aResult['name'] = $aResult['langaddress'];
                $aResult['foundorder'] = $iResNum;
                $aSearchResults[$iResNum] = $aResult;
        }
-
-//var_dump($aSearchResults);exit;
        
        uasort($aSearchResults, 'byImportance');
+
+//var_dump($aSearchResults);exit;
        
        $aOSMIDDone = array();
        $aClassTypeNameDone = array();