Merge remote-tracking branch 'upstream/master'

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index be74cc9b2d7061b9d695f8557d938edbbb17aad5..a6e94f5e2c807842ed981063577565649ae8811d 100755 (executable)
--- 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: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))."]";
 
 
        $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]";
 


@@ -285,7 +286,7 @@
                {
                        // Start with a blank search
                        $aSearches = array(
                {
                        // Start with a blank search
                        $aSearches = array(

-                               array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 
+                               array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 'aFullNameAddress'=>array(),

                                        'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(),
                                        'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'')
                        );
                                        'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(),
                                        'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'')
                        );


@@ -431,7 +432,7 @@
                                {
                                        $aValidTokens[$aToken['word_token']] = array($aToken);
                                }
                                {
                                        $aValidTokens[$aToken['word_token']] = array($aToken);
                                }

-                               if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
+                               if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;

                                $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
                        }
                        if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
                                $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
                        }
                        if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);


@@ -614,6 +615,7 @@
                                                                                                }
                                                                                                else
                                                                                                {
                                                                                                }
                                                                                                else
                                                                                                {

+                                                                                                       $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];

                                                                                                        $aSearch['iSearchRank'] += 1000; // skip;
                                                                                                }
                                                                                        }
                                                                                                        $aSearch['iSearchRank'] += 1000; // skip;
                                                                                                }
                                                                                        }


@@ -955,11 +957,13 @@
                                                        if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
                                                        if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
 
                                                        if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
                                                        if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
 

-                                                       $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) 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";
                                                        $aOrder[] = "$sImportanceSQL DESC";
                                                        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($aSearch['aFullNameAddress']))
+                                                               $aOrder[] = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) DESC';
+

                                                
                                                        if (sizeof($aTerms))
                                                        {
                                                
                                                        if (sizeof($aTerms))
                                                        {


@@ -974,12 +978,17 @@
                                                                else
                                                                        $sSQL .= " limit ".$iLimit;
 
                                                                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);
                                                                }
                                                                $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?
                                                                $aPlaceIDs = array();
 //var_dump($aViewBoxPlaceIDs);
                                                                // Did we have an viewbox matches?
                                                                $aPlaceIDs = array();


@@ -1201,16 +1210,17 @@
                                        $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
 //                                     $sSQL .= $sOrderSQL." as porder, ";
                                        $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
                                        $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 .= "coalesce(importance,0.75-(rank_search::float/40)) as importance, ";
+                                       $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(placex.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";

                                        $sSQL .= "from placex where place_id in ($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 ";
                                        $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= "from placex where place_id in ($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 ";
                                        $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
                                        if (!$bDeDupe) $sSQL .= ",place_id";

-                                       $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
-                                       $sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
-                                       $sSQL .= ",get_name_by_language(name, ARRAY['ref']) ";
+                                       $sSQL .= ",langaddress ";
+                                       $sSQL .= ",placename ";
+                                       $sSQL .= ",ref ";

                                        $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 .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
                                        $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 .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";


@@ -1218,7 +1228,8 @@
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
 //                                     $sSQL .= $sOrderSQL." as porder, ";
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
 //                                     $sSQL .= $sOrderSQL." as porder, ";

-                                       $sSQL .= "-0.15 as importance ";
+                                       $sSQL .= "-0.15 as importance, ";
+                                       $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";

                                        $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
                                        $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";
                                        $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
                                        $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";


@@ -1230,7 +1241,8 @@
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
 //                                     $sSQL .= $sOrderSQL." as porder, ";
                                        $sSQL .= "null as ref,";
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
 //                                     $sSQL .= $sOrderSQL." as porder, ";

-                                       $sSQL .= "-0.10 as importance ";
+                                       $sSQL .= "-0.10 as importance, ";
+                                       $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";

                                        $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
                                        $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";
                                        $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
                                        $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";


@@ -1476,6 +1488,7 @@
                {
                        if (stripos($sAddress, $sWord)!==false) $iCountWords++;
                }
                {
                        if (stripos($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']);
                $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']);


@@ -1496,7 +1509,7 @@
                }
 */
                $aResult['name'] = $aResult['langaddress'];
                }
 */
                $aResult['name'] = $aResult['langaddress'];

-               $aResult['foundorder'] = $iResNum;
+               $aResult['foundorder'] = -$aResult['addressimportance'];

                $aSearchResults[$iResNum] = $aResult;
        }
        uasort($aSearchResults, 'byImportance');
                $aSearchResults[$iResNum] = $aResult;
        }
        uasort($aSearchResults, 'byImportance');


@@ -1519,10 +1532,10 @@
                        $bFirst = false;
                }
                if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
                        $bFirst = false;
                }
                if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])

-                       && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']])))
+                       && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name']])))

                {
                        $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
                {
                        $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;

-                       $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true;
+                       $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name']] = true;

                        $aSearchResults[] = $aResult;
                }
 
                        $aSearchResults[] = $aResult;
                }