From e96e5b8ac44095668198e3e6797a3f77d9fb4634 Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Sat, 21 Jan 2012 23:56:52 +0100
Subject: Preliminary handling of ctime/mtime reset

---
 mat/archive.py  |  9 +++++++++
 mat/audio.py    |  2 ++
 mat/exiftool.py |  1 +
 mat/misc.py     |  1 +
 mat/office.py   |  4 ++++
 mat/parser.py   | 31 +++++++++++++++++++++++++++++--
 6 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/mat/archive.py b/mat/archive.py
index 9993102..84ef10a 100644
--- a/mat/archive.py
+++ b/mat/archive.py
@@ -5,6 +5,7 @@
 import zipfile
 import shutil
 import os
+import time
 import logging
 import tempfile
 
@@ -12,6 +13,10 @@ import parser
 import mat
 from tarfile import tarfile
 
+#Zip fileformat does not handle dates
+#prior to 1980
+ZIP_TIME = 315529200 #1jan1980
+
 
 class GenericArchiveStripper(parser.GenericParser):
     '''
@@ -161,6 +166,9 @@ harmless format' % item.filename)
         zipout.close()
         logging.info('%s treated' % self.filename)
         self.do_backup()
+        #time = 1980
+        # TODO
+        #self.set_time(time)
         return True
 
 
@@ -204,6 +212,7 @@ class TarStripper(GenericArchiveStripper):
         tarin.close()
         tarout.close()
         self.do_backup()
+        self.set_time(parser.EPOCH)
         return True
 
     def is_file_clean(self, current_file):
diff --git a/mat/audio.py b/mat/audio.py
index ed849ee..4682db1 100644
--- a/mat/audio.py
+++ b/mat/audio.py
@@ -35,6 +35,7 @@ class OggStripper(parser.GenericParser):
         mfile = OggVorbis(self.filename)
         mfile.delete()
         mfile.save()
+        self.set_time(parser.EPOCH)
         return True
 
     def is_clean(self):
@@ -74,6 +75,7 @@ class FlacStripper(parser.GenericParser):
         mfile.delete()
         mfile.clear_pictures()
         mfile.save()
+        self.set_time(parser.EPOCH)
         return True
 
     def is_clean(self):
diff --git a/mat/exiftool.py b/mat/exiftool.py
index 758a094..47e7dfc 100644
--- a/mat/exiftool.py
+++ b/mat/exiftool.py
@@ -43,6 +43,7 @@ class ExiftoolStripper(parser.GenericParser):
                     [ 'exiftool', '-m', '-All=', '-overwrite_original', self.filename ],
                     stdout=open('/dev/null'))
                 process.wait()
+            self.set_time(parser.EPOCH)
             return True
         except:
             return False
diff --git a/mat/misc.py b/mat/misc.py
index d084861..80ccc79 100644
--- a/mat/misc.py
+++ b/mat/misc.py
@@ -60,4 +60,5 @@ class TorrentStripper(parser.GenericParser):
         with open(self.output, 'w') as f:  # encode the decoded torrent
             f.write(bencode.bencode(decoded))  # and write it in self.output
         self.do_backup()
+        self.set_time(parser.EPOCH)
         return True
diff --git a/mat/office.py b/mat/office.py
index e1d738e..5ff9a83 100644
--- a/mat/office.py
+++ b/mat/office.py
@@ -100,6 +100,7 @@ class OpenDocumentStripper(archive.GenericArchiveStripper):
         zipin.close()
         zipout.close()
         self.do_backup()
+        self.set_time(archive.ZIP_TIME)
         return True
 
     def is_clean(self):
@@ -213,6 +214,8 @@ class PdfStripper(parser.GenericParser):
         if processed is False:
             logging.error('Please install either pdfrw, or exiftool to\
                     fully handle PDF files')
+
+        self.set_time(parser.EPOCH)
         return processed
 
     def get_meta(self):
@@ -274,6 +277,7 @@ class OpenXmlStripper(archive.GenericArchiveStripper):
         zipin.close()
         zipout.close()
         self.do_backup()
+        self.set_time(archive.ZIP_TIME)
         return True
 
     def is_clean(self):
diff --git a/mat/parser.py b/mat/parser.py
index 6dc5d0b..0c6f0ee 100644
--- a/mat/parser.py
+++ b/mat/parser.py
@@ -6,15 +6,18 @@ import hachoir_core
 import hachoir_editor
 
 import os
+import time
+import sys
 
 import mat
 
 NOMETA = ('.bmp', '.rdf', '.txt', '.xml', '.rels')
+EPOCH = 0
 #bmp : image
 #rdf : text
 #txt : plain text
 #xml : formated text
-#rels : openxml foramted text
+#rels : openxml formated text
 
 
 FIELD = object()
@@ -55,7 +58,14 @@ class GenericParser(object):
             if remove is FIELD:
                 if not self._is_clean(field):
                     return False
-        return True
+
+    def is_time_clean(self):
+        '''
+            Check if the atime and the mtime
+            of self.filename is Epoch
+        '''
+        stat = os.stat(self.filename)
+        return stat.st_atime == 0 and stat.st_mtime == 0
 
     def remove_all(self):
         '''
@@ -64,6 +74,7 @@ class GenericParser(object):
         state = self._remove_all(self.editor)
         hachoir_core.field.writeIntoFile(self.editor, self.output)
         self.do_backup()
+        self.set_time(EPOCH)
         return state
 
     def _remove_all(self, fieldset):
@@ -128,3 +139,19 @@ class GenericParser(object):
         if self.backup is False:
             mat.secure_remove(self.filename)
             os.rename(self.output, self.filename)
+
+    def set_time(self, time):
+        '''
+            Set the ctime of the file to $time
+        '''
+        filename = ''
+        if self.backup is True:
+            filename = self.output
+        else:
+            filename = self.filename
+
+        try:
+            os.utime(filename, (time, time))
+        except:
+            print "Unable to set %s's date to %s" % (filename, time)
+            sys.exit(1)
-- 
cgit v1.3