Merge remote-tracking branch 'upstream/master'

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 24f21e12f31d1690aca25e4d3064d3e9b7c136a2..cfc1455f28c85b4bd5190efa415cbd2faa5c77dc 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -164,7 +164,10 @@
        }
        if (sizeof($aStructuredQuery) > 0) {
                $sQuery = join(', ', $aStructuredQuery);
        }
        if (sizeof($aStructuredQuery) > 0) {
                $sQuery = join(', ', $aStructuredQuery);

-               $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+               if ($iMaxAddressRank < 30)
+               {
+                       $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+               }

        }
 
        if ($sQuery)
        }
 
        if ($sQuery)


@@ -325,7 +328,7 @@
                                {
                                        foreach($aSearchWords as $aSearchTerm)
                                        {
                                {
                                        foreach($aSearchWords as $aSearchTerm)
                                        {

-                                               $aNewSearch = $aSearch;                 
+                                               $aNewSearch = $aSearch;

                                                if ($aSearchTerm['country_code'])
                                                {
                                                        $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
                                                if ($aSearchTerm['country_code'])
                                                {
                                                        $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);


@@ -393,7 +396,7 @@
                        {
 
                        // Check which tokens we have, get the ID numbers                       
                        {
 
                        // 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';
                        $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')';
                        $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';


@@ -420,7 +423,7 @@
                                {
                                        $aValidTokens[$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 ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1 + $aToken['search_name_count'];

                        }
                        if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
                        }
                        if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 


@@ -528,13 +531,18 @@
                                                                        {
                                                                                $aSearch = $aCurrentSearch;
                                                                                $aSearch['iSearchRank']++;
                                                                        {
                                                                                $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 ($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[] = $aSearch;
                                                                                        }
                                                                                }
                                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                }


@@ -827,12 +835,19 @@
                                                                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 .= " join placex using (place_id)";
                                                                $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)";
                                                                if ($sCountryCodesSQL) $sSQL .= " and country_code in ($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 ($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)";
                                                                {
                                                                        $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                                                        if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";


@@ -861,7 +876,7 @@
                                                        if (CONST_Debug) var_dump('<hr>',$aSearch);
                                                        if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);     
                                                        $aPlaceIDs = array();
                                                        if (CONST_Debug) var_dump('<hr>',$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();
                                                        // First we need a position, either aName or fLat or both
                                                        $aTerms = array();
                                                        $aOrder = array();


@@ -869,7 +884,20 @@
                                                        // 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'],",")."]";
                                                        // 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'] && $aSearch['fLat'])
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                                                        if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)";
                                                        if ($aSearch['fLon'] && $aSearch['fLat'])


@@ -1047,13 +1075,13 @@
                                                                                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)";
                                                                                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 ";
                                                                                $sSQL = "select distinct l.place_id".($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 ($sPlaceGeom)
                                                                                {
                                                                                }
                                                                                if ($sPlaceGeom)
                                                                                {


@@ -1080,7 +1108,7 @@
                                                                                else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
 
                                                                                $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where ";
                                                                                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))
                                                                                {
                                                                                $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' ";
                                                                                if (sizeof($aExcludePlaceIDs))
                                                                                {


@@ -1121,7 +1149,7 @@
                                        if ($iGroupLoop > 4) break;
                                        if ($iQueryLoop > 30) break;
                                }
                                        if ($iGroupLoop > 4) break;
                                        if ($iQueryLoop > 30) break;
                                }

-//exit;
+

                                // Did we find anything?        
                                if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs))
                                {
                                // Did we find anything?        
                                if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs))
                                {


@@ -1138,7 +1166,7 @@
                                        $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 .= "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 .= $sOrderSQL." as porder, ";
                                        $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance ";
                                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";


@@ -1208,7 +1236,7 @@
                                        $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 .= "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 .= $sOrderSQL." as porder, ";
                                        $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance ";
                                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";