From 0cf0541ad9c2f40aa987cb34be34bc33b7341232 Mon Sep 17 00:00:00 2001
From: jvoisin <julien.voisin@dustri.org>
Date: Sun, 1 Sep 2019 12:54:56 +0200
Subject: [PATCH] Remove nsid fields from MSOffice documents
nsids are random identifiers, usually used to ease merging
between documents, and can trivially be used for fingerprinting.
---
libmat2/office.py | 50 +++++++++++++++++++++++++++++++++++++++++------
1 file changed, 44 insertions(+), 6 deletions(-)
diff --git a/libmat2/office.py b/libmat2/office.py
index b769991..c9bed7a 100644
--- a/libmat2/office.py
+++ b/libmat2/office.py
@@ -62,9 +62,6 @@ class MSOfficeParser(ZipParser):
# Do we want to keep the following ones?
'application/vnd.openxmlformats-officedocument.wordprocessingml.settings+xml',
-
- # See https://0xacab.org/jvoisin/mat2/issues/71
- 'application/vnd.openxmlformats-officedocument.wordprocessingml.numbering+xml', # /word/numbering.xml
}
@@ -88,6 +85,7 @@ class MSOfficeParser(ZipParser):
r'^word/printerSettings/',
r'^word/theme',
r'^word/people\.xml$',
+ r'^word/numbering\.xml$',
# we have an allowlist in self.files_to_keep,
# so we can trash everything else
@@ -124,7 +122,7 @@ class MSOfficeParser(ZipParser):
@staticmethod
def __remove_rsid(full_path: str) -> bool:
- """ The method will remove "revision session ID". We're '}rsid'
+ """ The method will remove "revision session ID". We're using '}rsid'
instead of proper parsing, since rsid can have multiple forms, like
`rsidRDefault`, `rsidR`, `rsids`, …
@@ -137,7 +135,8 @@ class MSOfficeParser(ZipParser):
"""
try:
tree, namespace = _parse_xml(full_path)
- except ET.ParseError:
+ except ET.ParseError as e:
+ logging.error("Unable to parse %s: %s", full_path, e)
return False
# rsid, tags or attributes, are always under the `w` namespace
@@ -161,6 +160,41 @@ class MSOfficeParser(ZipParser):
tree.write(full_path, xml_declaration=True)
return True
+ @staticmethod
+ def __remove_nsid(full_path: str) -> bool:
+ """
+ NSID are random identifiers that can be used
+ to ease the merging of some components of a document.
+ They can also be used for fingerprinting.
+
+ See the spec for more details: https://docs.microsoft.com/en-us/dotnet/api/documentformat.openxml.wordprocessing.nsid?view=openxml-2.8.1
+
+ In this function, we're changing the XML document in several
+ different times, since we don't want to change the tree we're currently
+ iterating on.
+ """
+ try:
+ tree, namespace = _parse_xml(full_path)
+ except ET.ParseError as e: # pragma: no cover
+ logging.error("Unable to parse %s: %s", full_path, e)
+ return False
+
+ # The NSID tag is always under the `w` namespace
+ if 'w' not in namespace.keys():
+ return True
+
+ parent_map = {c:p for p in tree.iter() for c in p}
+
+ elements_to_remove = list()
+ for element in tree.iterfind('.//w:nsid', namespace):
+ elements_to_remove.append(element)
+ for element in elements_to_remove:
+ parent_map[element].remove(element)
+
+ tree.write(full_path, xml_declaration=True)
+ return True
+
+
@staticmethod
def __remove_revisions(full_path: str) -> bool:
""" In this function, we're changing the XML document in several
@@ -208,7 +242,8 @@ class MSOfficeParser(ZipParser):
"""
try:
tree, namespace = _parse_xml(full_path)
- except ET.ParseError: # pragma: no cover
+ except ET.ParseError as e: # pragma: no cover
+ logging.error("Unable to parse %s: %s", full_path, e)
return False
if len(namespace.items()) != 1:
@@ -269,6 +304,9 @@ class MSOfficeParser(ZipParser):
if self.__remove_rsid(full_path) is False:
return False
+ if self.__remove_nsid(full_path) is False:
+ return False # pragma: no cover
+
try:
_sort_xml_attributes(full_path)
except ET.ParseError as e: # pragma: no cover
--
GitLab