deduplicate search term graph slightly earlier

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 8c92919c85086c782a4062ba35c4171fcbf5da51..d6007ef591730fd918f0801bddc70a0b09191fbc 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -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' ||  $_GET['format'] == 'jsonv2'))
@@ -46,6 +48,24 @@
                        if ($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 = 18;
+                       break;
+               }
+       }
                
        // Search query
        $sQuery = (isset($_GET['q'])?trim($_GET['q']):'');
@@ -64,7 +84,7 @@
                $hLog = logStart($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 
@@ -432,7 +452,7 @@
                                                                                {
                                                                                        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']] = $aSearchTerm['word_id'];
                                                                                                }
@@ -450,14 +470,14 @@
                                                                                }
                                                                        }
                                                                }
-                                                               if (isset($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;
                                                                                        $aSearch['iSearchRank'] += 1;
@@ -468,7 +488,8 @@
                                                                                if (!sizeof($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;
@@ -491,6 +512,21 @@
 
                                                $aNewPhraseSearches = array_merge($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);
                                        }
 
@@ -588,8 +624,11 @@
                                                {
                                                        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
                                                        {
@@ -847,6 +886,7 @@
                                        $sSQL .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, ";
                                        $sSQL .= $sOrderSQL." as porder ";
                                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
@@ -860,6 +900,7 @@
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
                                        $sSQL .= $sOrderSQL." as porder ";
                                        $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
+                                       $sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
                                        $sSQL .= "group by place_id";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= " union ";
@@ -870,6 +911,7 @@
                                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
                                        $sSQL .= $sOrderSQL." as porder ";
                                        $sSQL .= "from location_property_aux where place_id in ($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) ";