Do not store the same malware several times

this decreases disk space consumption

+ also log sha sum

kippo/commands/wget.py

@@ -12,8 +12,10 @@ import urlparse
 import random
 import re
 import exceptions
-import os.path
+import os
 import getopt
+import hashlib
+import shutil
 
 commands = {}
 
@@ -93,13 +95,15 @@ class command_wget(HoneyPotCommand):
         if cfg.has_option('honeypot', 'download_limit_size'):
             self.limit_size = int(cfg.get('honeypot', 'download_limit_size'))
 
+        self.download_path = cfg.get('honeypot', 'download_path')
+
         self.safeoutfile = '%s/%s_%s' % \
-            (cfg.get('honeypot', 'download_path'),
+            (self.download_path,
             time.strftime('%Y%m%d%H%M%S'),
             re.sub('[^A-Za-z0-9]', '_', url))
         self.deferred = self.download(url, outfile, self.safeoutfile)
         if self.deferred:
-            self.deferred.addCallback(self.success)
+            self.deferred.addCallback(self.success, outfile)
             self.deferred.addErrback(self.error, url)
 
     def download(self, url, fakeoutfile, outputfile, *args, **kwargs):
@@ -137,7 +141,30 @@ class command_wget(HoneyPotCommand):
         self.writeln('^C')
         self.connection.transport.loseConnection()
 
-    def success(self, data):
+    def success(self, data, outfile):
+        if not os.path.isfile(self.safeoutfile):
+            print "there's no file " + self.safeoutfile
+            self.exit()
+
+        shasum = hashlib.sha256(open(self.safeoutfile, 'rb').read()).hexdigest()
+        hash_path = '%s/%s' % (self.download_path, shasum)
+
+        msg = 'SHA sum %s of URL %s in file %s' % \
+            (shasum, self.url, self.fileName)
+        print msg
+        self.honeypot.logDispatch(msg)
+
+        if not os.path.exists(hash_path):
+            print "moving " + self.safeoutfile + " -> " + hash_path
+            shutil.move(self.safeoutfile, hash_path)
+        else:
+            print "deleting " + self.safeoutfile + " SHA sum: " + shasum
+            os.remove(self.safeoutfile)
+        self.safeoutfile = hash_path
+
+        print "Updating realfile to " + hash_path
+        f = self.fs.getfile(outfile)
+        f[9] = hash_path
         self.exit()
 
     def error(self, error, url):
@@ -257,6 +284,8 @@ class HTTPProgressDownloader(client.HTTPDownloader):
         self.wget.fs.update_realfile(
             self.wget.fs.getfile(self.fakeoutfile),
             self.wget.safeoutfile)
+
+        self.wget.fileName = self.fileName
         return client.HTTPDownloader.pageEnd(self)
 
 # vim: set sw=4 et:

kippo/core/dblog.py

@@ -32,6 +32,10 @@ class DBLogger(object):
                 self.handleUnknownCommand),
             ('^:dispatch: Saving URL \((?P<url>.*)\) to (?P<outfile>.*)$',
                 self.handleFileDownload),
+            ('^:dispatch: SHA sum (?P<shasum>.*) of URL (?P<url>.*) in file (?P<outfile>.*)$',
+                self.handleShaSum),
+            ('^:dispatch: Updated outfile (?P<outfile>.*) to (?P<dl_file>.*) with SHA sum (?P<shasum>.*)$',
+                self.handleUpdatedFile),
             ('^INPUT \((?P<realm>[a-zA-Z0-9]+)\): (?P<input>.*)$',
                 self.handleInput),
             ('^Terminal size: (?P<height>[0-9]+) (?P<width>[0-9]+)$',
@@ -145,4 +149,12 @@ class DBLogger(object):
     def handleFileDownload(self, session, args):
         pass
 
+    # args has: shasum, url, outfile
+    def handleShaSum(self, session, args):
+        pass
+
+    # args has: outfile, dl_file, shasum
+    def handleUpdatedFile(self, session, args):
+        pass
+
 # vim: set sw=4 et:

kippo/dblog/mysql.py

@@ -146,4 +146,9 @@ class DBLogger(dblog.DBLogger):
             ' VALUES (%s, FROM_UNIXTIME(%s), %s, %s)',
             (session, self.nowUnix(), args['url'], args['outfile']))
 
+    def handleShaSum(self, session, args):
+         self.simpleQuery('UPDATE `downloads` SET `shasum` = %s' + \
+            '  WHERE `outfile` = %s',
+            (args['shasum'], args['outfile']))
+
 # vim: set sw=4 et:

kippo/dblog/textlog.py

@@ -56,4 +56,12 @@ class DBLogger(dblog.DBLogger):
         self.write(session, 'File download: [%s] -> %s' % \
             (args['url'], args['outfile']))
 
+    def handleShaSum(self, session, args):
+        self.write(session, 'File SHA sum: %s [%s] -> %s' % \
+            (args['shasum'], args['url'], args['outfile']))
+
+    def handleUpdatedFile(self, session, args):
+        self.write(session, 'Updated wget outfile %s to %s' % \
+            (args['outfile'], args['dl_file']))
+
 # vim: set sw=4 et: