Merge branch 'master' of http://github.com/twain47/Nominatim

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 4803599837843d5fb4f0598ff257b1dbf0c544fb..fd91f31daff2831f349c0b5f5ce8725c306b1436 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -101,7 +101,6 @@
                $aPhrases = array_reverse($aPhrases); 
                $sQuery = join(', ',$aPhrases);
        }
-
        if ($sQuery)
        {
                $hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
@@ -113,6 +112,7 @@
                        $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);
                }
 
                // If we have a view box create the SQL
@@ -181,21 +181,36 @@
                // Do we have anything that looks like a lat/lon pair?
                if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[ ]+([0-9.]+)?[, ]+([EW])[ ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?\\b/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
-                       $_GET['nearlon'] = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
+                       $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
                elseif (preg_match('/\\b([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([NS])[, ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([EW])\\b/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
-                       $_GET['nearlon'] = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
+                       $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
-               elseif (preg_match('/(\\[|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|\\b)/', $sQuery, $aData))
+               elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|$|\\b)/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = $aData[2];
-                       $_GET['nearlon'] = $aData[3];
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = $aData[2];
+                       $fQueryLon = $aData[3];
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
 
                if ($sQuery)
@@ -271,7 +286,8 @@
                                $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
                                if (PEAR::isError($aPhrase))
                                {
-                                       var_dump($aPhrase);
+                                       echo "Illegal query string (not an UTF-8 string): ".$sPhrase;
+                                       if (CONST_Debug) var_dump($aPhrase);
                                        exit;
                                }
                                if (trim($aPhrase['string']))
@@ -296,7 +312,12 @@
                        // 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\'))';
+                       // HACK WARNING
+                       // (mis)using search_name_count to exclude words that return too many
+                       // search results. saerch_name_count is currently set to 1 by hand
+                       // because there is no fast way to extract this count from a live database. 
+                       $sSQL .= ' and search_name_count = 0';
+//                     $sSQL .= ' and (class is null or class not in (\'highway\'))';
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';
 
                        if (CONST_Debug) var_Dump($sSQL);
@@ -354,7 +375,7 @@
                                }
                        }
                        if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion);
-/*
+
                        // Try and calculate GB postcodes we might be missing
                        foreach($aTokens as $sToken)
                        {
@@ -372,14 +393,13 @@
                                        }
                                }
                        }
-*/
 
                        foreach($aTokens as $sToken)
                        {
                                // Unknown single word token with a number - assume it is a house number
                                if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken))
                                {
-                                       $aValidTokens[' '.$sToken] = array('class'=>'place','type'=>'house');
+                                       $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house'));
                                }
                        }
 
@@ -771,12 +791,17 @@
                                                                        $sSQL .= " limit ".$iLimit;
 
                                                                if (CONST_Debug) var_dump($sSQL);
+                                                               $iStartTime = time();
                                                                $aViewBoxPlaceIDs = $oDB->getAll($sSQL);
                                                                if (PEAR::IsError($aViewBoxPlaceIDs))
                                                                {
                                                                        var_dump($sSQL, $aViewBoxPlaceIDs);                                     
                                                                        exit;
                                                                }
+                                                               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();
@@ -1161,7 +1186,7 @@
                                }
 
                                // Output data suitable for display (points and a bounding box)
-                               if ($bShowPolygons)
+                               if ($bShowPolygons && isset($aPolyPoints))
                                {
                                        $aResult['aPolyPoints'] = array();
                                        foreach($aPolyPoints as $aPoint)