From c308cf7daaa4fa46377e2df0f2e9a397981e19b2 Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Fri, 10 Jun 2011 01:29:29 +0200
Subject: The current version is (mostly) working

---
 mat.py | 117 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 117 insertions(+)
 create mode 100644 mat.py

(limited to 'mat.py')

diff --git a/mat.py b/mat.py
new file mode 100644
index 0000000..8a3afcd
--- /dev/null
+++ b/mat.py
@@ -0,0 +1,117 @@
+import hachoir_core.error
+import hachoir_core.cmd_line
+import hachoir_parser
+import hachoir_metadata
+import hachoir_editor
+
+import sys
+import os
+import hachoir_parser.image
+
+__version__ = "0.1"
+__author__ = "jvoisin"
+
+
+class file():
+    def __init__(self, realname, filename, parser, editor):
+        self.meta = {}
+        self.clean = False
+        self.filename = filename
+        self.realname = realname
+        self.parser = parser
+        self.editor = editor
+        self.meta = self.__fill_meta()
+
+    def __fill_meta(self):
+        metadata = {}
+        try:
+            meta = hachoir_metadata.extractMetadata(self.parser)
+        except hachoir_core.error.HachoirError, err:
+            print("Metadata extraction error: %s" % err)
+
+        if not meta:
+            print("Unable to extract metadata from the file %s" % self.filename)
+            sys.exit(1)
+
+        for title in meta:
+            #fixme i'm so dirty
+            if title.values != []: #if the field is not empty
+                value = ""
+                for item in title.values:
+                    value = item.text
+                metadata[title.key] = value
+        return metadata
+
+    def is_clean(self):
+        '''
+            Return true if the file is clean from any compromizing meta
+        '''
+        return self.clean
+
+    def remove_all(self):
+        '''
+            Remove all the files that are compromizing
+        '''
+        for key, field in self.meta.iteritems():
+            if self._should_remove(key):
+                print "BLEH" #DEBUG
+                #__remove(self, key)
+        #self.clean = True
+
+    def __remove(self, field):
+        '''
+            Remove the given file
+        '''
+        del self.editor[field]
+
+
+    def get_meta(self):
+        '''
+            return a dict with all the meta of the file
+        '''
+        return self.meta
+
+    def _should_remove(self, field):
+        '''
+            return True if the field is compromizing
+            abstract method
+        '''
+        raise NotImplementedError()
+
+class JpegStripper(file):
+    def _should_remove(self, field):
+        return False
+
+strippers = {
+    hachoir_parser.image.JpegFile: JpegStripper,
+}
+
+def create_class_file(name):
+    '''
+        return a $FILETYPEStripper() class,
+        corresponding to the filetype of the given file
+    '''
+    if not(os.path.isfile(name)): #check if the file exist
+        print("Error: %s is not a valid file" % name)
+        sys.exit(1)
+
+    filename = ""
+    realname = name
+    filename = hachoir_core.cmd_line.unicodeFilename(name)
+    parser = hachoir_parser.createParser(filename, realname)
+    if not parser:
+        print("Unable to parse the file %s : sorry" % filename)
+        sys.exit(1)
+
+    editor = hachoir_editor.createEditor(parser)
+    try:
+        '''this part is a little tricky :
+        stripper_class will receice the name of the class $FILETYPEStripper,
+        (which herits from the "file" class), based on the editor
+        of given file (name)
+        '''
+        stripper_class = strippers[editor.input.__class__]
+    except KeyError:
+        print("Don't have stripper for file type: %s" % editor.description)
+        sys.exit(1)
+    return stripper_class(realname, filename, parser, editor)
-- 
cgit v1.3