Add is_clean() method to archives, and move the 'secure_remove' function from parser.py to mat.py

3 files changed, 52 insertions, 33 deletions

diff --git a/lib/archive.py b/lib/archive.py
index 89425cf..446e493 100644
--- a/lib/archive.py
+++ b/lib/archive.py
@@ -5,47 +5,63 @@ import mat
 import shutil
 
 class TarStripper(parser.Generic_parser):
-    def compression_type(self):
+    def __init__(self, realname, filename, parser, editor, backup):
+        super(TarStripper, self).__init__(realname,
+            filename, parser, editor, backup)
         self.compression = ''
+        self.tarin = tarfile.open(self.filename, 'r' + self.compression)
+        self.folder_list = []
 
     def remove_all(self):
-        self.compression_type()
-        if not tarfile.is_tarfile(self.filename):
-            print('%s is not a valid tar file' % self.filename)
-            sys.exit(1)
-
-        tarin = tarfile.open(self.filename, 'r' + self.compression)
-        tarout = tarfile.open(self.filename + parser.POSTFIX,
+        self.tarout = tarfile.open(self.filename + parser.POSTFIX,
             'w' + self.compression)
-        folder_list = []
-
-        for current_file in tarin.getmembers():
-            tarin.extract(current_file)
+        for current_file in self.tarin.getmembers():
+            self.tarin.extract(current_file)
             if current_file.type is '0': #is current_file a regular file ?
                 #no backup file
                 class_file = mat.create_class_file(current_file.name, False)
                 class_file.remove_all()
-                tarout.add(current_file.name)
-                class_file.secure_remove()
+                self.tarout.add(current_file.name)
+                mat.secure_remove(current_file.name)
             else:
-                folder_list.insert(0, current_file.name)
+                self.folder_list.insert(0, current_file.name)
 
-        for folder in folder_list: #delete remainings folders
+        for folder in self.folder_list: #delete remainings folders
             shutil.rmtree(folder)
 
         #meta from the tar itself
-        tarout.mtime = None
+        self.tarout.mtime = None
 
-        tarout.close()
-        tarin.close()
+        self.tarout.close()
+        self.tarin.close()
 
     def is_clean(self):
+        for current_file in self.tarin.getmembers():
+            self.tarin.extract(current_file)
+            if current_file.type is '0': #is current_file a regular file ?
+                #no backup file
+                class_file = mat.create_class_file(current_file.name, False)
+                if not class_file.is_clean():
+                    self.folder_list = []
+                    return False
+                mat.secure_remove(current_file.name)
+            else:
+                self.folder_list.insert(0, current_file.name)
+        self.tarin.close()
+
+        for folder in self.folder_list: #delete remainings folders
+            shutil.rmtree(folder)
+        self.folder_list = []
         return False
 
 class GzipStripper(TarStripper):
-    def compression_type(self):
+    def __init__(self, realname, filename, parser, editor, backup):
+        super(GzipStripper, self).__init__(realname,
+            filename, parser, editor, backup)
         self.compression = ':gz'
 
 class Bzip2Stripper(TarStripper):
-    def compression_type(self):
+    def __init__(self, realname, filename, parser, editor, backup):
+        super(Bzip2Stripper, self).__init__(realname,
+            filename, parser, editor, backup)
         self.compression = ':bz2'
diff --git a/lib/mat.py b/lib/mat.py
index 9624511..c9547c1 100644
--- a/lib/mat.py
+++ b/lib/mat.py
@@ -6,6 +6,7 @@
 
 import sys
 import os
+import subprocess
 
 import hachoir_core.cmd_line
 import hachoir_parser
@@ -29,6 +30,17 @@ strippers = {
     hachoir_parser.archive.bzip2_parser.Bzip2Parser: archive.Bzip2Stripper,
 }
 
+def secure_remove(filename):
+    '''
+        securely remove the file
+    '''
+    #FIXME : Vulnerable to shell injection ?
+    try:
+        subprocess.call('shred --remove %s' % filename, shell=True)
+    except:
+        print('Unable to remove %s' % filename)
+
+
 def is_secure(filename):
     '''
         Prevent shell injection
diff --git a/lib/parser.py b/lib/parser.py
index c7e189e..6b899f3 100644
--- a/lib/parser.py
+++ b/lib/parser.py
@@ -8,10 +8,11 @@ import hachoir_editor
 import sys
 import os
 import subprocess
+import mat
 
 POSTFIX = ".cleaned"
 
-class Generic_parser():
+class Generic_parser(object):
     def __init__(self, realname, filename, parser, editor, backup):
         self.filename = filename
         self.realname = realname
@@ -19,16 +20,6 @@ class Generic_parser():
         self.editor = editor
         self.backup = backup
 
-    def secure_remove(self):
-        '''
-            securely remove the file
-        '''
-        #FIXME : Vulnerable to shell injection ?
-        try:
-            subprocess.call('shred --remove %s' % self.filename, shell=True)
-        except:
-            print('Unable to remove %s' % self.filename)
-
     def is_clean(self):
         '''
             Check if the file is clean from harmful metadatas
@@ -47,7 +38,7 @@ class Generic_parser():
                 self._remove(field)
         hachoir_core.field.writeIntoFile(self.editor, self.filename + POSTFIX)
         if self.backup is False:
-            self.secure_remove() #remove the old file
+            mat.secure_remove(self.filename) #remove the old file
             os.rename(self.filename+ POSTFIX, self.filename)#rename the new
 
     def remove_all_ugly(self):