Add the possibility to add (or not) unsupported fileformat to outputed archives fileformat (invasive patch, sorry about that).

7 files changed, 56 insertions, 29 deletions

diff --git a/cli.py b/cli.py
index a19b8ce..b9c8a5c 100755
--- a/cli.py
+++ b/cli.py
@@ -12,6 +12,8 @@ __version__ = '0.1'
 
 def parse():
     parser = optparse.OptionParser(usage='%prog [options] filename')
+    parser.add_option('--add2archive', '-a', action='store_true', default=False,
+        help='Add to outputed archive non-supported filetypes')
     parser.add_option('--backup', '-b', action='store_true', default=False,
         help='Keep a backup copy')
     parser.add_option('--check', '-c',  action='store_true', default=False,
@@ -91,7 +93,8 @@ def main():
         func = clean_meta
 
     for filename in filenames:
-        class_file = mat.create_class_file(filename, args.backup)
+        class_file = mat.create_class_file(filename, args.backup,
+            args.add2archive)
         if class_file is not None:
             func(class_file, filename)
 
diff --git a/gui.py b/gui.py
index 9294fcc..978bd4b 100644
--- a/gui.py
+++ b/gui.py
@@ -22,10 +22,10 @@ class cfile(GObject.GObject):
         This class exist just to be "around" my parser.Generic_parser class,
         since Gtk.ListStore does not accept it.
     '''
-    def __init__(self, path, backup):
+    def __init__(self, path, backup, add2archive):
         GObject.GObject.__init__(self)
         try:
-            self.file = mat.create_class_file(path, backup)
+            self.file = mat.create_class_file(path, backup, add2archive)
         except:
             self.file = None
 
@@ -37,6 +37,7 @@ class ListStoreApp:
         #preferences
         self.backup = True
         self.force = False
+        self.add2archive = True
 
         self.window = Gtk.Window()
         self.window.set_title('Metadata Anonymisation Toolkit %s' % __version__)
@@ -223,7 +224,7 @@ class ListStoreApp:
         '''
             Append selected files by add_file to the self.liststore
         '''
-        cf = cfile(item, self.backup)
+        cf = cfile(item, self.backup, self.add2archive)
         if cf.file is not None:
             self.liststore.append([cf, cf.file.filename, cf.file.mime,'unknow'])
 
@@ -253,7 +254,7 @@ class ListStoreApp:
 
         hbox.pack_start(icon, False, False, 0)
 
-        table = Gtk.Table(2, 2, False)#nb rows, nb lines
+        table = Gtk.Table(3, 2, False)#nb rows, nb lines
         table.set_row_spacings(4)
         table.set_col_spacings(4)
         hbox.pack_start(table, True, True, 0)
@@ -268,8 +269,15 @@ class ListStoreApp:
         backup.set_tooltip_text('Keep a backup copy.')
         backup.set_active(self.backup)
 
+        add2archive = Gtk.CheckButton('Add unsupported file to archives', False)
+        add2archive.connect('toggled', self.invert, 'add2archive')
+        add2archive.set_tooltip_text('Add non-supported (and so non-anonymised)\
+ file to outputed archive.')
+        add2archive.set_active(self.add2archive)
+
         table.attach_defaults(force, 0, 1, 0, 1)
         table.attach_defaults(backup, 0, 1, 1, 2)
+        table.attach_defaults(add2archive, 0, 1, 2, 3)
 
         hbox.show_all()
         response = dialog.run()
@@ -313,7 +321,7 @@ class ListStoreApp:
                 if self.force:
                     self.liststore[i][0].file.remove_all()
                 else:
-                    if not self.liststore[i][0].is_clean():
+                    if not self.liststore[i][0].file.is_clean():
                         self.liststore[i][0].file.remove_all()
             self.liststore[i][3] = 'clean'
 
@@ -326,7 +334,7 @@ class ListStoreApp:
                 if self.force:
                     self.liststore[i][0].file.remove_all_ugly()
                 else:
-                    if not self.liststore[i][0].is_clean():
+                    if not self.liststore[i][0].file.is_clean():
                         self.liststore[i][0].file.remove_all_ugly()
             self.liststore[i][3] = 'clean'
 
diff --git a/lib/archive.py b/lib/archive.py
index aa2a47f..c31fac6 100644
--- a/lib/archive.py
+++ b/lib/archive.py
@@ -13,11 +13,12 @@ class GenericArchiveStripper(parser.Generic_parser):
     '''
         Represent a generic archive
     '''
-    def __init__(self, realname, filename, parser, editor, backup):
+    def __init__(self, realname, filename, parser, editor, backup, add2archive):
         super(GenericArchiveStripper, self).__init__(realname,
-            filename, parser, editor, backup)
+            filename, parser, editor, backup, add2archive)
         self.compression = ''
         self.folder_list = []
+        self.add2archive = add2archive
 
     def remove_folder(self):
         [shutil.rmtree(folder) for folder in self.folder_list]
@@ -49,13 +50,18 @@ class ZipStripper(GenericArchiveStripper):
             zipin.extract(item)
             if os.path.isfile(item.filename):
                 try:
-                    cfile = mat.create_class_file(item.filename, False)
+                    cfile = mat.create_class_file(item.filename, False,
+                        self.add2archive)
                     cfile.remove_all()
                     logging.debug('Processing %s from %s' % (item.filename,
                         self.filename))
+                    zipout.write(item.filename)
                 except:
-                    print('%s\' filefomart is not supported'%item.filename)
-                zipout.write(item.filename)
+                    logging.info('%s\' filefomart is not supported' %
+                        item.filename)
+                    if self.add2archive:
+                        zipout.write(item.filename)
+                mat.secure_remove(item.filename)
             else:
                 self.folder_list.insert(0, item.filename)
         logging.info('%s treated' % self.filename)
@@ -84,11 +90,15 @@ class TarStripper(GenericArchiveStripper):
             if current_file.type is '0': #is current_file a regular file ?
                 #no backup file
                 try:
-                    cfile = mat.create_class_file(current_file.name, False)
+                    cfile = mat.create_class_file(current_file.name, False,
+                    self.add2archive)
                     cfile.remove_all()
+                    tarout.add(current_file.name, filter=self._remove)
                 except:
-                    print('%s\' format is not supported'%current_file.name)
-                tarout.add(current_file.name, filter=self._remove)
+                    logging.info('%s\' format is not supported' %
+                        current_file.name)
+                    if self.add2archive:
+                        tarout.add(current_file.name, filter=self._remove)
                 mat.secure_remove(current_file.name)
             else:
                 self.folder_list.insert(0, current_file.name)
@@ -123,7 +133,8 @@ class TarStripper(GenericArchiveStripper):
             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 = mat.create_class_file(current_file.name, False,
+                    self.add2archive)
                 if not class_file.is_clean():#if the extracted file is not clean
                     mat.secure_remove(current_file.name) #remove it
                     self.remove_folder() #remove all the remaining folders
diff --git a/lib/mat.py b/lib/mat.py
index 1d7f53b..49a84f1 100644
--- a/lib/mat.py
+++ b/lib/mat.py
@@ -52,14 +52,16 @@ def is_secure(filename):
 
     if not(os.path.isfile(filename)): #check if the file exist
         logging.error("Error: %s is not a valid file" % filename)
-        return
+        return False
 
-def create_class_file(name, backup):
+def create_class_file(name, backup, add2archive):
     '''
         return a $FILETYPEStripper() class,
         corresponding to the filetype of the given file
     '''
-    is_secure(name)
+    if is_secure(name):
+        print 'a'
+        return
 
     filename = ""
     realname = name
@@ -83,4 +85,5 @@ def create_class_file(name, backup):
         return
     if editor.input.__class__ == hachoir_parser.misc.PDFDocument:
         return stripper_class(filename, realname, backup)
-    return stripper_class(realname, filename, parser, editor, backup)
+    return stripper_class(realname, filename, parser, editor, backup,
+        add2archive)
diff --git a/lib/parser.py b/lib/parser.py
index 2b7b52e..6e6ee34 100644
--- a/lib/parser.py
+++ b/lib/parser.py
@@ -5,16 +5,18 @@
 import hachoir_core.error
 import hachoir_parser
 import hachoir_editor
+
 import sys
 import os
 import subprocess
 import mimetypes
+
 import mat
 
 POSTFIX = ".cleaned"
 
 class Generic_parser(object):
-    def __init__(self, realname, filename, parser, editor, backup):
+    def __init__(self, realname, filename, parser, editor, backup, add2archive):
         self.filename = filename
         self.realname = realname
         self.shortname = os.path.basename(filename)
diff --git a/test/clitest.py b/test/clitest.py
index 91b71ca..f603f97 100644
--- a/test/clitest.py
+++ b/test/clitest.py
@@ -17,14 +17,14 @@ class Test_Remove_cli(test.MATTest):
         '''make sure that the cli remove all compromizing meta'''
         for clean, dirty in self.file_list:
             subprocess.call(['../cli.py', dirty])
-            current_file = mat.create_class_file(dirty, False)
+            current_file = mat.create_class_file(dirty, False, True)
             self.assertTrue(current_file.is_clean())
 
     def test_remove_empty(self):
         '''Test removal with clean files'''
         for clean, dirty in self.file_list:
             subprocess.call(['../cli.py', clean])
-            current_file = mat.create_class_file(clean, False)
+            current_file = mat.create_class_file(clean, False, True)
             self.assertTrue(current_file.is_clean())
 
 
diff --git a/test/libtest.py b/test/libtest.py
index b35b4f3..eea8117 100644
--- a/test/libtest.py
+++ b/test/libtest.py
@@ -14,14 +14,14 @@ class Test_Remove_lib(test.MATTest):
     def test_remove(self):
         '''make sure that the lib remove all compromizing meta'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(dirty, False)
+            current_file = mat.create_class_file(dirty, False, True)
             current_file.remove_all()
             self.assertTrue(current_file.is_clean())
 
     def test_remove_empty(self):
         '''Test removal with clean files'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(clean, False)
+            current_file = mat.create_class_file(clean, False, True)
             current_file.remove_all()
             self.assertTrue(current_file.is_clean())
 
@@ -30,14 +30,14 @@ class Test_List_lib(test.MATTest):
     def test_list(self):
         '''check if get_meta returns all the expected meta'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(dirty, False)
+            current_file = mat.create_class_file(dirty, False, True)
             #FIXME assertisNotNone() : python 2.7
             self.assertTrue(current_file.get_meta())
 
     def testlist_list_empty(self):
         '''check that a listing of a clean file return an empty dict'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(clean, False)
+            current_file = mat.create_class_file(clean, False, True)
             self.assertEqual(current_file.get_meta(), dict())
 
 
@@ -45,13 +45,13 @@ class Test_isClean_lib(test.MATTest):
     def test_dirty(self):
         '''test is_clean on clean files'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(dirty, False)
+            current_file = mat.create_class_file(dirty, False, True)
             self.assertFalse(current_file.is_clean())
 
     def test_clean(self):
         '''test is_clean on dirty files'''
         for clean, dirty in self.file_list:
-            current_file = mat.create_class_file(clean, False)
+            current_file = mat.create_class_file(clean, False, True)
             self.assertTrue(current_file.is_clean())