Preliminary implementation of PIL for images.

PIL (Python Image Library) is used to open, then save
images, in order to remove unknown metadata.

4 files changed, 47 insertions, 4 deletions

diff --git a/README b/README
index 218655f..9e49136 100644
--- a/README
+++ b/README
@@ -21,6 +21,7 @@ DEPENDENCIES
 ============
  * python2.7 (at least)
  * python-hachoir-core and python-hachoir-parser
+ * python-pil for more secure images handling
  * python-pdfrw, gir-poppler and python-gi-cairo for full PDF support
  * python-gi for the GUI
  * shred (should be already installed)
diff --git a/libmat/exiftool.py b/libmat/exiftool.py
index 9e38f04..2c2a2b8 100644
--- a/libmat/exiftool.py
+++ b/libmat/exiftool.py
@@ -1,11 +1,13 @@
 ''' Care about images with help of the amazing (perl) library Exiftool.
 '''
 
-import parser
 import subprocess
 
+import parser
+import pillow
+
 
-class ExiftoolStripper(parser.GenericParser):
+class ExiftoolStripper(parser.GenericParser, pillow.PillowStripper):
     ''' A generic stripper class using exiftool as backend
     '''
 
@@ -25,6 +27,7 @@ class ExiftoolStripper(parser.GenericParser):
     def remove_all(self):
         ''' Remove all metadata with help of exiftool
         '''
+        self.open_and_save()
         try:
             if self.backup:
                 self.create_backup_copy()
diff --git a/libmat/images.py b/libmat/images.py
index 67c710f..2daea88 100644
--- a/libmat/images.py
+++ b/libmat/images.py
@@ -7,9 +7,21 @@ References:
 '''
 
 import parser
+import pillow
 
 
-class JpegStripper(parser.GenericParser):
+class ImageStripper(parser.GenericParser, pillow.PillowStripper):
+    ''' Common stripper for images.
+    Its purpose is to open then save
+    images with PIL, the goal being to remove
+    unknown metadata.
+    '''
+    def remove_all(self):
+        self.open_and_save()
+        super(ImageStripper, self).remove_all()
+
+
+class JpegStripper(ImageStripper):
     ''' Represents a jpeg file.
         Custom Huffman and Quantization tables
         are stripped: they may leak
@@ -34,7 +46,7 @@ class JpegStripper(parser.GenericParser):
         return True
 
 
-class PngStripper(parser.GenericParser):
+class PngStripper(ImageStripper):
     ''' Represents a png file
     '''
     def _should_remove(self, field):
diff --git a/libmat/pillow.py b/libmat/pillow.py
new file mode 100644
index 0000000..556d58a
--- /dev/null
+++ b/libmat/pillow.py
@@ -0,0 +1,27 @@
+''' Care about images with help of PIL.
+This class doesn't remove metadata in the "conventional way";
+it opens, then saves the image. This should remove unknown metadata.
+'''
+
+# FIXME Implement this with a decorator instead
+
+import parser
+
+
+class PillowStripper(object):
+    ''' This class implements a single method, "open_and_save".
+        It's a class instead of a function so it can be inherited.
+    '''
+    def open_and_save(self):
+        ''' Open and save the image with PIL.
+        This should remove a lot of unknown metadata.
+        '''
+        try:
+            from PIL import Image
+        except ImportError:
+            logging.error('Unable to import PIL, image support degraded. Be careful.')
+
+        try:
+            Image.open(self.filename).save(self.filename)
+        except IOError:
+            logging.error('Can not save %s.' % self.filename)