correct penalty for multi-word countries

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 71041080a1fb440adb7b1462c3cc5e3a8535d082..91869c3efaf22dc26e3f081893ecf948e405035c 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']))


@@ -96,7 +96,7 @@
        {
                $sQuery = substr($_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'])
                {


@@ -244,7 +245,7 @@
                                $sQuery = str_replace($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';
                                $sQuery = str_replace($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);
                                $aNewSearches = array();
                                if (CONST_Debug) var_Dump($sSQL);
                                $aSearchWords = $oDB->getAll($sSQL);
                                $aNewSearches = array();


@@ -311,7 +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 (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);
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';
 
                        if (CONST_Debug) var_Dump($sSQL);


@@ -419,12 +420,12 @@
                                {
                                        $aNewPhraseSearches = array();
 
                                {
                                        $aNewPhraseSearches = array();
 

-                                       foreach($aPhrases[$iPhrase]['wordsets'] as $iWordset => $aWordset)
+                                       foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset)

                                        {
                                                $aWordsetSearches = $aSearches;
 
                                                // Add all words from this wordset
                                        {
                                                $aWordsetSearches = $aSearches;
 
                                                // Add all words from this wordset

-                                               foreach($aWordset as $sToken)
+                                               foreach($aWordset as $iToken => $sToken)

                                                {
 //echo "<br><b>$sToken</b>";
                                                        $aNewWordsetSearches = array();
                                                {
 //echo "<br><b>$sToken</b>";
                                                        $aNewWordsetSearches = array();


@@ -447,8 +448,8 @@
                                                                                        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)
-                                                                                               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[] = $aSearch;
                                                                                        }
                                                                                }
                                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                }


@@ -704,22 +705,22 @@
                                                                $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$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);


@@ -728,7 +729,7 @@
                                                        else
                                                        {
                                                                $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
                                                        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 $iLimit";
                                                                if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";                                                             
                                                                if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
                                                                $sSQL .= " limit $iLimit";


@@ -1267,7 +1268,7 @@
                {
                        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']);
 /*