From e163520a130145496ceeeeeb94d950f17e7fd76e Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Thu, 23 Jun 2011 19:00:44 +0200
Subject: Implementation of the --backup option, to keep a backup of the
 treated file (disabled by default)

---
 cli.py        | 21 +++++++++++++++++++--
 lib/mat.py    | 10 +++++-----
 lib/misc.py   |  5 +++++
 lib/parser.py | 28 +++++++++++++++++++++++++++-
 4 files changed, 56 insertions(+), 8 deletions(-)

diff --git a/cli.py b/cli.py
index f2bec52..1a8713d 100755
--- a/cli.py
+++ b/cli.py
@@ -18,7 +18,10 @@ def parse():
         help='Check if a file is free of harmfull metadatas')
     common.add_option('--version', action='callback', callback=displayVersion,
         help='Display version and exit')
-
+    common.add_option('--backup', '-b', action='store_true', default=False,
+        help='Keep a backup copy')
+    common.add_option('--ugly', '-u', action='store_true', default=False,
+        help='Remove harmful meta, but with loss')
     parser.add_option_group(common)
 
     values, arguments = parser.parse_args()
@@ -60,6 +63,18 @@ def clean_meta(class_file, filename):
         class_file.remove_all()
         print('%s cleaned !' % filename)
 
+def clean_meta_ugly(class_file, filename):
+    '''
+	Clean the file 'filename', ugly way
+    '''
+    print('[+] Cleaning %s' % filename)
+    if class_file.is_clean():
+        print('%s is already clean' % filename)
+    else:
+        class_file.remove_all_ugly()
+        print('%s cleaned !' % filename)
+
+
 def main():
     args, filenames = parse()
 
@@ -68,11 +83,13 @@ def main():
         func = list_meta
     elif args.check is True: #only check if the file is clean
         func = is_clean
+    elif args.ugly is True:
+        func = clean_meta_ugly
     else: #clean the file
         func = clean_meta
 
     for filename in filenames:
-        class_file = mat.create_class_file(filename)
+        class_file = mat.create_class_file(filename, args.backup)
         func(class_file, filename)
         print('\n')
 
diff --git a/lib/mat.py b/lib/mat.py
index a9b8e17..6abcd64 100644
--- a/lib/mat.py
+++ b/lib/mat.py
@@ -14,7 +14,7 @@ import hachoir_editor
 import images
 import audio
 import misc
-import archive
+#import archive
 
 __version__ = "0.1"
 __author__ = "jvoisin"
@@ -24,10 +24,10 @@ strippers = {
     hachoir_parser.image.PngFile: images.PngStripper,
     hachoir_parser.audio.MpegAudioFile: audio.MpegAudioStripper,
     hachoir_parser.misc.PDFDocument: misc.PdfStripper,
-    hachoir_parser.archive.TarFile: archive.TarStripper,
+    #hachoir_parser.archive.TarFile: archive.TarStripper,
 }
 
-def create_class_file(name):
+def create_class_file(name, backup):
     '''
         return a $FILETYPEStripper() class,
         corresponding to the filetype of the given file
@@ -57,5 +57,5 @@ def create_class_file(name):
         print("Don't have stripper for file type: %s" % editor.description)
         sys.exit(1)
     if editor.input.__class__ == hachoir_parser.misc.PDFDocument:
-        return stripper_class(filename)
-    return stripper_class(realname, filename, parser, editor)
+        return stripper_class(filename, backup)
+    return stripper_class(realname, filename, parser, editor, backup)
diff --git a/lib/misc.py b/lib/misc.py
index 56c2274..a8070f1 100644
--- a/lib/misc.py
+++ b/lib/misc.py
@@ -1,5 +1,6 @@
 import parser
 import pdfrw
+import shutil
 
 class PdfStripper(parser.Generic_parser):
     '''
@@ -23,6 +24,10 @@ class PdfStripper(parser.Generic_parser):
 
         self.writer.trailer = self.trailer
         self.writer.write(self.filename + parser.POSTFIX)
+        if self.backup is False:
+            self.secure_remove() #remove the old file
+            shutil.rename(self.filename+ POSTFIX, self.filename)#rename the new
+
 
     def is_clean(self):
         '''
diff --git a/lib/parser.py b/lib/parser.py
index 1084de5..d629619 100644
--- a/lib/parser.py
+++ b/lib/parser.py
@@ -6,15 +6,27 @@ import hachoir_core.error
 import hachoir_parser
 import hachoir_editor
 import sys
+import shutil
 
 POSTFIX = ".cleaned"
 
 class Generic_parser():
-    def __init__(self, realname, filename, parser, editor):
+    def __init__(self, realname, filename, parser, editor, backup):
         self.filename = filename
         self.realname = realname
         self.parser = parser
         self.editor = editor
+        self.backup = backup
+
+    def secure_remove(self):
+        '''
+            securely remove the file
+        '''
+        #FIXME : not secure at all !
+        try:
+            shutil.rmtree(self.filename)
+        except:
+            print('Unable to remove %s' % self.filename)
 
     def is_clean(self):
         '''
@@ -33,6 +45,20 @@ class Generic_parser():
             if self._should_remove(field):
                 self._remove(field)
         hachoir_core.field.writeIntoFile(self.editor, self.filename + POSTFIX)
+        if self.backup is False:
+            self.secure_remove() #remove the old file
+            shutil.rename(self.filename+ POSTFIX, self.filename)#rename the new
+
+    def remove_all_ugly(self):
+        '''
+            If the remove_all() is not efficient enough,
+            this method is implemented :
+            It is efficient, but destructive.
+            In a perfect world, with nice fileformat,
+            this method does not exist.
+        '''
+        raise NotImplementedError()
+
 
     def _remove(self, field):
         '''
-- 
cgit v1.3