From c186fc42929b2660e5c507adeb8a8fb406593b11 Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Sun, 1 Apr 2018 00:17:06 +0200
Subject: Clean deep metadata for zip files

---
 src/libreoffice.py | 12 +++++++++++-
 src/office.py      | 14 +++++++++++++-
 2 files changed, 24 insertions(+), 2 deletions(-)

(limited to 'src')

diff --git a/src/libreoffice.py b/src/libreoffice.py
index a3481a1..809ae3c 100644
--- a/src/libreoffice.py
+++ b/src/libreoffice.py
@@ -34,6 +34,13 @@ class LibreOfficeParser(abstract.AbstractParser):
         zipin.close()
         return metadata
 
+    def __clean_zipinfo(self, zipinfo:zipfile.ZipInfo) -> zipfile.ZipInfo:
+        zipinfo.compress_type = zipfile.ZIP_DEFLATED
+        zipinfo.create_system = 3  # Linux
+        zipinfo.comment = b''
+        zipinfo.date_time = (1980, 1, 1, 0, 0, 0)
+        return zipinfo
+
     def remove_all(self):
         zin = zipfile.ZipFile(self.filename, 'r')
         zout = zipfile.ZipFile(self.output_filename, 'w')
@@ -51,7 +58,10 @@ class LibreOfficeParser(abstract.AbstractParser):
                 print("%s isn't supported" % item.filename)
                 continue
             tmp_parser.remove_all()
-            zout.write(tmp_parser.output_filename, item.filename)
+            zinfo = zipfile.ZipInfo(item.filename)
+            item = self.__clean_zipinfo(item)
+            with open(tmp_parser.output_filename, 'rb') as f:
+                zout.writestr(zinfo, f.read())
         shutil.rmtree(temp_folder)
         zout.close()
         zin.close()
diff --git a/src/office.py b/src/office.py
index 5de0597..a729f2f 100644
--- a/src/office.py
+++ b/src/office.py
@@ -33,6 +33,13 @@ class OfficeParser(abstract.AbstractParser):
         zipin.close()
         return metadata
 
+    def __clean_zipinfo(self, zipinfo:zipfile.ZipInfo) -> zipfile.ZipInfo:
+        zipinfo.compress_type = zipfile.ZIP_DEFLATED
+        zipinfo.create_system = 3  # Linux
+        zipinfo.comment = b''
+        zipinfo.date_time = (1980, 1, 1, 0, 0, 0)
+        return zipinfo
+
     def remove_all(self):
         zin = zipfile.ZipFile(self.filename, 'r')
         zout = zipfile.ZipFile(self.output_filename, 'w')
@@ -45,6 +52,7 @@ class OfficeParser(abstract.AbstractParser):
                 if not item.filename.endswith('.rels'):
                     continue  # don't keep metadata files
             if item.filename in self.files_to_keep:
+                item = self.__clean_zipinfo(item)
                 zout.writestr(item, zin.read(item))
                 continue
 
@@ -54,7 +62,11 @@ class OfficeParser(abstract.AbstractParser):
                 print("%s isn't supported" % item.filename)
                 continue
             tmp_parser.remove_all()
-            zout.write(tmp_parser.output_filename, item.filename)
+            zinfo = zipfile.ZipInfo(item.filename)
+            item = self.__clean_zipinfo(item)
+            with open(tmp_parser.output_filename, 'rb') as f:
+                zout.writestr(zinfo, f.read())
+
         shutil.rmtree(temp_folder)
         zout.close()
         zin.close()
-- 
cgit v1.3