ignore frequent partial search terms

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 7c0838c9f57436467b3f10b26b7c3be833ea9491..1d877e77ebfdf76564e1d8b0cab03bc47cb06fa1 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -36,8 +36,8 @@
        // Show address breakdown
        $bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails'];
 
        // 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:de'])) $bReverseInPlan = true;
        if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true;
        if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true;


@@ -53,7 +53,7 @@
                }
        }
 
                }
        }
 

-  // Only certain ranks of feature
+       // Only certain ranks of feature

        if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType'];
 
        if (isset($_GET['featuretype']))
        if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType'];
 
        if (isset($_GET['featuretype']))


@@ -92,11 +92,11 @@
                
        // Search query
        $sQuery = (isset($_GET['q'])?trim($_GET['q']):'');
                
        // 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);
 
        {
                $sQuery = substr($_SERVER['PATH_INFO'], 1);
 

-               // reverse order of '/' seperated string
+               // reverse order of '/' separated string

                $aPhrases = explode('/', $sQuery);              
                $aPhrases = array_reverse($aPhrases); 
                $sQuery = join(', ',$aPhrases);
                $aPhrases = explode('/', $sQuery);              
                $aPhrases = array_reverse($aPhrases); 
                $sQuery = join(', ',$aPhrases);


@@ -109,10 +109,9 @@
                 $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 = 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
                }
 
                // If we have a view box create the SQL


@@ -134,6 +133,8 @@
                        $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)";
                        $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']) && $_GET['routewidth'])
                {
                }
                if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth'])
                {


@@ -311,6 +312,7 @@
                        // 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)).')';
                        // 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)).')';

+                       $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;

                        $sSQL .= ' and (class is null or class not in (\'highway\'))';
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';
 
                        $sSQL .= ' and (class is null or class not in (\'highway\'))';
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';
 


@@ -447,12 +449,12 @@
                                                                                        if ($aSearch['sCountryCode'] === false)
                                                                                        {
                                                                                                $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
                                                                                        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)
+                                                                                               // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)

                                                                                                if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
                                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                }
                                                                                                if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($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 ($aSearch['fLat'] === '')
                                                                                        {
                                                                                {
                                                                                        if ($aSearch['fLat'] === '')
                                                                                        {


@@ -704,22 +706,22 @@
                                                                $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
                                                                if ($oDB->getOne($sSQL))
                                                                {
                                                                $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$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) $sSQL .= " join placex using (place_id)";
                                                                if ($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 ($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 .= " 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) $sSQL .= " join placex using (place_id)";
                                                                        if ($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 ($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);
                                                                        $sSQL .= " limit $iLimit";
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $oDB->getCol($sSQL);


@@ -1144,15 +1146,15 @@
                {
                        // Get the bounding box and outline polygon
                        $sSQL = "select place_id,numfeatures,area,outline,";
                {
                        // 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(centroid) as centrelat,ST_X(centroid) as centrelon,";
                        $sSQL .= "ST_AsText(outline) as outlinestring from get_place_boundingbox_quick(".$aResult['place_id'].")";
 
                        $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,";
                        $sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,";

-                       $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(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))
                        {
                        $aPointPolygon = $oDB->getRow($sSQL);
                        if (PEAR::IsError($aPointPolygon))
                        {


@@ -1267,7 +1269,7 @@
                {
                        if (stripos($sAddress, $sWord)!==false) $iCountWords++;
                }
                {
                        if (stripos($sAddress, $sWord)!==false) $iCountWords++;
                }

-               $aResult['importance'] = $aResult['importance'] + $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 (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
 /*


@@ -1312,10 +1314,10 @@
                        $bFirst = false;
                }
                if (!$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'].$aResult['osm_class'].$aResult['name']])))
+                       && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']])))

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