be more strict with us state replacement hack

[nominatim.git] / website / search.php

diff --git a/website/search.php b/website/search.php
index 37ddd23c4118c9885cfc54397745f4c7a270bc9a..53314d518ba4feeab8ed0aba669ddf5ff4149d59 100755 (executable)
--- a/website/search.php
+++ b/website/search.php
@@ -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


@@ -163,8 +162,7 @@
                        $sViewboxSmallSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxSmallSQL))
                        {
                        $sViewboxSmallSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxSmallSQL))
                        {

-                               var_dump($sViewboxSmallSQL);
-                               exit;
+                               failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL);

                        }
                        $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry";
 
                        }
                        $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry";
 


@@ -172,8 +170,7 @@
                        $sViewboxLargeSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxLargeSQL))
                        {
                        $sViewboxLargeSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxLargeSQL))
                        {

-                               var_dump($sViewboxLargeSQL);
-                               exit;
+                               failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL);

                        }
                        $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry";
                }
                        }
                        $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry";
                }


@@ -181,21 +178,36 @@
                // 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))
                {
                // 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))
                {
                }
                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)
                }
 
                if ($sQuery)


@@ -271,7 +283,8 @@
                                $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
                                if (PEAR::isError($aPhrase))
                                {
                                $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($aPhrase['string']))
                                        exit;
                                }
                                if (trim($aPhrase['string']))


@@ -296,7 +309,12 @@
                        // 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)).')';
                        // 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);
 //                     $sSQL .= ' group by word_token, word, class, type, location, country_code';
 
                        if (CONST_Debug) var_Dump($sSQL);


@@ -308,8 +326,7 @@
                                $aDatabaseWords = array();
                        if (PEAR::IsError($aDatabaseWords))
                        {
                                $aDatabaseWords = array();
                        if (PEAR::IsError($aDatabaseWords))
                        {

-                               var_dump($sSQL, $aDatabaseWords);
-                               exit;
+                               failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords);

                        }
                        $aPossibleMainWordIDs = array();
                        foreach($aDatabaseWords as $aToken)
                        }
                        $aPossibleMainWordIDs = array();
                        foreach($aDatabaseWords as $aToken)


@@ -378,7 +395,7 @@
                                // Unknown single word token with a number - assume it is a house number
                                if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken))
                                {
                                // Unknown single word token with a number - assume it is a house number
                                if (!isset($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'));

                                }
                        }
 
                                }
                        }
 


@@ -770,12 +787,16 @@
                                                                        $sSQL .= " limit ".$iLimit;
 
                                                                if (CONST_Debug) var_dump($sSQL);
                                                                        $sSQL .= " limit ".$iLimit;
 
                                                                if (CONST_Debug) var_dump($sSQL);

+                                                               $iStartTime = time();

                                                                $aViewBoxPlaceIDs = $oDB->getAll($sSQL);
                                                                if (PEAR::IsError($aViewBoxPlaceIDs))
                                                                {
                                                                $aViewBoxPlaceIDs = $oDB->getAll($sSQL);
                                                                if (PEAR::IsError($aViewBoxPlaceIDs))
                                                                {

-                                                                       var_dump($sSQL, $aViewBoxPlaceIDs);                                     
-                                                                       exit;
+                                                                       failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs);

                                                                }
                                                                }

+                                                               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?
                                                                $aPlaceIDs = array();
 //var_dump($aViewBoxPlaceIDs);
                                                                // Did we have an viewbox matches?
                                                                $aPlaceIDs = array();


@@ -798,7 +819,7 @@
                                                                $sPlaceIDs = join(',',$aPlaceIDs);
        
                                                                // Now they are indexed look for a house attached to a street we found
                                                                $sPlaceIDs = join(',',$aPlaceIDs);
        
                                                                // Now they are indexed look for a house attached to a street we found

-                                                               $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M';                                         
+                                                               $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M';                                                

                                                                $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'";
                                                                if (sizeof($aExcludePlaceIDs))
                                                                {
                                                                $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'";
                                                                if (sizeof($aExcludePlaceIDs))
                                                                {


@@ -955,8 +976,7 @@
 
                                                if (PEAR::IsError($aPlaceIDs))
                                                {
 
                                                if (PEAR::IsError($aPlaceIDs))
                                                {

-                                                       var_dump($sSQL, $aPlaceIDs);                                    
-                                                       exit;
+                                                       failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs);

                                                }
 
                                                if (CONST_Debug) var_Dump($aPlaceIDs);
                                                }
 
                                                if (CONST_Debug) var_Dump($aPlaceIDs);


@@ -1032,8 +1052,7 @@
 
                                        if (PEAR::IsError($aSearchResults))
                                        {
 
                                        if (PEAR::IsError($aSearchResults))
                                        {

-                                               var_dump($sSQL, $aSearchResults);                                       
-                                               exit;
+                                               failInternalError("Could not get details for place.", $sSQL, $aSearchResults);

                                        }
                                }
                        } // end if ($sQuery)
                                        }
                                }
                        } // end if ($sQuery)


@@ -1101,8 +1120,7 @@
 
                                        if (PEAR::IsError($aSearchResults))
                                        {
 
                                        if (PEAR::IsError($aSearchResults))
                                        {

-                                               var_dump($sSQL, $aSearchResults);                                       
-                                               exit;
+                        failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults);

                                        }
                                }
                        }
                                        }
                                }
                        }


@@ -1133,8 +1151,7 @@
                        $aPointPolygon = $oDB->getRow($sSQL);
                        if (PEAR::IsError($aPointPolygon))
                        {
                        $aPointPolygon = $oDB->getRow($sSQL);
                        if (PEAR::IsError($aPointPolygon))
                        {

-                               var_dump($sSQL, $aPointPolygon);
-                               exit;
+                               failInternalError("Could not get outline.", $sSQL, $aPointPolygon);

                        }
                        if ($aPointPolygon['place_id'])
                        {
                        }
                        if ($aPointPolygon['place_id'])
                        {