From b1dab4c844e2fae4909eca47c2b42698d12c55fc Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Sat, 18 Jun 2011 02:13:37 +0200
Subject: Hachoir is so ugly : I am able to display meta, but not to remove
 them (because I didn't manage to find an elegant/efficient way to find the
 "path" to a field). So I remove a whole "Rootfield" (exif, author, comment,
 ...). It's efficient, but not very elegant.

---
 cli.py | 12 ------------
 mat.py | 44 +++++++++++++++++---------------------------
 2 files changed, 17 insertions(+), 39 deletions(-)

diff --git a/cli.py b/cli.py
index b5bea20..bd87409 100755
--- a/cli.py
+++ b/cli.py
@@ -16,8 +16,6 @@ def parse():
         help='List all the meta of a file without removing them')
     common.add_option('--check', '-c',  action='store_true', default=False,
         help='Check if a file is free of harmfull metadatas')
-    common.add_option('--harmful',  action='store_true', default=False,
-        help='List all the harmful meta of a file without removing them')
     common.add_option('--version', action='callback', callback=displayVersion,
         help='Display version and exit')
 
@@ -42,14 +40,6 @@ def list_meta(class_file, filename):
     for key, item in class_file.get_meta().iteritems():
         print('\t%s : %s' % (key, item) )
 
-def list_harmful_meta(class_file, filename):
-    '''
-	Print all the harmful meta of 'filename' on stdout
-    '''
-    print('[+] File %s :' % filename)
-    for key, item in class_file.get_harmful().iteritems():
-        print('\t%s : %s' % (key, item) )
-
 def is_clean(class_file, filename):
     '''
 	Say if 'filename' is clean or not
@@ -78,8 +68,6 @@ def main():
         func = list_meta
     elif args.check is True: #only check if the file is clean
         func = is_clean
-    elif args.harmful is True: #only print harmful metadatas
-        func = list_harmful_meta
     else: #clean the file
         func = clean_meta
 
diff --git a/mat.py b/mat.py
index f59bdce..200fc04 100755
--- a/mat.py
+++ b/mat.py
@@ -18,6 +18,7 @@ import hachoir_parser.image
 __version__ = "0.1"
 __author__ = "jvoisin"
 
+POSTFIX = ".cleaned"
 
 class file():
     def __init__(self, realname, filename, parser, editor):
@@ -52,8 +53,8 @@ class file():
         '''
             Check if the file is clean from harmful metadatas
         '''
-        for key, field in self.meta.iteritems():
-            if self._should_remove(key):
+        for field in self.editor:
+            if self._should_remove(field):
                 return False
         return True
 
@@ -61,37 +62,25 @@ class file():
         '''
             Remove all the files that are compromizing
         '''
-        for key, field in self.meta.iteritems():
-            if self._should_remove(key):
-                self._remove(key)
-        hachoir_core.field.writeIntoFile(self.editor, self.filename + "cleaned")
+        for field in self.editor:
+            if self._should_remove(field):
+                self._remove(field)
+        hachoir_core.field.writeIntoFile(self.editor, self.filename + POSTFIX)
 
-
-    def _remove(self, key):
+    def _remove(self, field):
         '''
             Remove the given field
         '''
-        del self.editor[key]
+        del self.editor[field.name]
 
 
     def get_meta(self):
         '''
             return a dict with all the meta of the file
         '''
+        #am I useless ?
         return self.meta
 
-    def get_harmful(self):
-        '''
-            return a dict with all the harmful meta of the file
-        '''
-        harmful = {}
-        for key, value in self.meta.iteritems():
-            if self._should_remove(key):
-                harmful[key] = value
-        return harmful
-
-
-
     def _should_remove(self, key):
         '''
             return True if the field is compromizing
@@ -100,19 +89,20 @@ class file():
         raise NotImplementedError()
 
 class JpegStripper(file):
-    def _should_remove(self, key):
-        if key in ('comment', 'author'):
+    def _should_remove(self, field):
+        if field.name.startswith('comment'):
+            return True
+        elif field.name in ("photoshop", "exif", "adobe"):
             return True
         else:
             return False
 
 class PngStripper(file):
-    def _should_remove(self, key):
-        if key in ('comment', 'author'):
+    def _should_remove(self, field):
+        if field.name in ('comment'):
             return True
         else:
             return False
-        return False
 
 strippers = {
     hachoir_parser.image.JpegFile: JpegStripper,
@@ -131,7 +121,7 @@ def create_class_file(name):
     filename = ""
     realname = name
     filename = hachoir_core.cmd_line.unicodeFilename(name)
-    parser = hachoir_parser.createParser(filename, realname)
+    parser = hachoir_parser.createParser(filename)
     if not parser:
         print("Unable to parse the file %s : sorry" % filename)
         sys.exit(1)
-- 
cgit v1.3