script for scraper blocking using apache log files

[nominatim.git] / lib / log.php

diff --git a/lib/log.php b/lib/log.php
index 12e2953ce496d896f410cc8d49eec0b078ef5489..d38bb892b78a0d876e5c93a79e5b85032413b500 100644 (file)
--- a/lib/log.php
+++ b/lib/log.php
@@ -3,45 +3,83 @@
        function logStart(&$oDB, $sType = '', $sQuery = '', $aLanguageList = array())
        {
                $aStartTime = explode('.',microtime(true));
-               if (!$aStartTime[1]) $aStartTime[1] = '0';
+               if (!isset($aStartTime[1])) $aStartTime[1] = '0';
+
+               if (isset($_GET['email']))
+                               $sUserAgent = $_GET['email'];
+                       elseif (isset($_SERVER['HTTP_REFERER']))
+                               $sUserAgent = $_SERVER['HTTP_REFERER'];
+                       elseif (isset($_SERVER['HTTP_USER_AGENT']))
+                               $sUserAgent = $_SERVER['HTTP_USER_AGENT'];
+                       else
+                               $sUserAgent = '';
+
+               $sOutputFormat = '';
+               if (isset($_GET['format'])) $sOutputFormat = $_GET['format'];
+
+               if ($sType == 'reverse')
+               {
+                       $sOutQuery = (isset($_GET['lat'])?$_GET['lat']:'').'/';
+                       if (isset($_GET['lon'])) $sOutQuery .= $_GET['lon'];
+                       if (isset($_GET['zoom'])) $sOutQuery .= '/'.$_GET['zoom'];
+               }
+               else
+                       $sOutQuery = $sQuery;
 
                $hLog = array(
                                date('Y-m-d H:i:s',$aStartTime[0]).'.'.$aStartTime[1],
                                $_SERVER["REMOTE_ADDR"],
                                $_SERVER['QUERY_STRING'],
-                               $sQuery
-                       );
+                               $sOutQuery,
+                               $sType,
+                               $sUserAgent,
+                               $sOutputFormat
+                               );
 
-                // Log
-               if ($sType == 'search')
+               if (CONST_Log_DB)
                {
-                       $oDB->query('insert into query_log values ('.getDBQuoted($hLog[0]).','.getDBQuoted($hLog[3]).','.getDBQuoted($hLog[1]).')');
+                       if (isset($_GET['email']))
+                               $sUserAgent = $_GET['email'];
+                       elseif (isset($_SERVER['HTTP_REFERER']))
+                               $sUserAgent = $_SERVER['HTTP_REFERER'];
+                       elseif (isset($_SERVER['HTTP_USER_AGENT']))
+                               $sUserAgent = $_SERVER['HTTP_USER_AGENT'];
+                       else
+                               $sUserAgent = '';
+                       $sSQL = 'insert into new_query_log (type,starttime,query,ipaddress,useragent,language,format,searchterm)';
+                       $sSQL .= ' values ('.getDBQuoted($sType).','.getDBQuoted($hLog[0]).','.getDBQuoted($hLog[2]);
+                       $sSQL .= ','.getDBQuoted($hLog[1]).','.getDBQuoted($sUserAgent).','.getDBQuoted(join(',',$aLanguageList)).','.getDBQuoted($sOutputFormat).','.getDBQuoted($hLog[3]).')';
+                       $oDB->query($sSQL);
                }
 
-               $sSQL = 'insert into new_query_log (type,starttime,query,ipaddress,useragent,language,format)';
-               $sSQL .= ' values ('.getDBQuoted($sType).','.getDBQuoted($hLog[0]).','.getDBQuoted($hLog[2]);
-               $sSQL .= ','.getDBQuoted($hLog[1]).','.getDBQuoted($_SERVER['HTTP_USER_AGENT']).','.getDBQuoted(join(',',$aLanguageList)).','.getDBQuoted($_GET['format']).')';
-               $oDB->query($sSQL);
-
-
                return $hLog;
        }
 
        function logEnd(&$oDB, $hLog, $iNumResults)
        {
-                $aEndTime = explode('.',microtime(true));
-                if (!$aEndTime[1]) $aEndTime[1] = '0';
-                $sEndTime = date('Y-m-d H:i:s',$aEndTime[0]).'.'.$aEndTime[1];
-
-               $sSQL = 'update query_log set endtime = '.getDBQuoted($sEndTime).', results = '.$iNumResults;
-               $sSQL .= ' where starttime = '.getDBQuoted($hLog[0]);
-               $sSQL .= ' and ipaddress = '.getDBQuoted($hLog[1]);
-               $sSQL .= ' and query = '.getDBQuoted($hLog[3]);
-                $oDB->query($sSQL);
-
-               $sSQL = 'update new_query_log set endtime = '.getDBQuoted($sEndTime).', results = '.$iNumResults;
-               $sSQL .= ' where starttime = '.getDBQuoted($hLog[0]);
-               $sSQL .= ' and ipaddress = '.getDBQuoted($hLog[1]);
-               $sSQL .= ' and query = '.getDBQuoted($hLog[2]);
-               $oDB->query($sSQL);
+               $aEndTime = explode('.',microtime(true));
+               if (!$aEndTime[1]) $aEndTime[1] = '0';
+               $sEndTime = date('Y-m-d H:i:s',$aEndTime[0]).'.'.$aEndTime[1];
+
+               if (CONST_Log_DB)
+               {
+                       $sSQL = 'update new_query_log set endtime = '.getDBQuoted($sEndTime).', results = '.$iNumResults;
+                       $sSQL .= ' where starttime = '.getDBQuoted($hLog[0]);
+                       $sSQL .= ' and ipaddress = '.getDBQuoted($hLog[1]);
+                       $sSQL .= ' and query = '.getDBQuoted($hLog[2]);
+                       $oDB->query($sSQL);
+               }
+
+               if (CONST_Log_File)
+               {
+                       array_unshift($hLog, $sEndTime);
+                       $hLog[] = $iNumResults;
+                       $fLog = fopen(CONST_Log_File, 'a');
+                       flock($fLog, LOCK_EX);
+                       fputcsv($fLog, $hLog);
+                       fflush($fLog);
+                       flock($fLog, LOCK_UN);
+                       fclose($fLog);
+               }
+
        }