security-and-privacy/mat2
security-and-privacy/mat2
1
0
Fork 0
mirror of synced 2024-11-24 10:14:24 +01:00
Code Releases Activity
543 Commits 1 Branch 26 Tags 12 MiB
660f0dea73
Commit Graph

214 Commits

Author SHA1 Message Date
jvoisin
efa525c102 Improve the robustness of the HTML parser 2019-12-15 06:50:54 -08:00
jvoisin
f67cd9d7dc Improve the robustness of the CSS parser 2019-12-15 06:44:21 -08:00
jvoisin
e4114af3b5 Improve a bit ppt support 2019-11-30 11:38:22 +01:00
jvoisin
d56f83bed1 Improve a bit odt handling 2019-11-30 10:25:24 +01:00
georg
697cb36b81 This is mat2, not MAT2 
Closes #131
 2019-11-30 01:14:41 -08:00
jvoisin
df1eb98a40 Please the new version of pylint 2019-11-26 22:12:56 +01:00
jvoisin
655c19d17d Improve a bit the support for ppt files 2019-10-17 23:02:17 +02:00
jvoisin
5f0b3beb46 Add a way to disable the sandbox 
Due to bubblewrap's pickiness, mat2 can now be run
without a sandbox, even if bubblewrap is installed.
 2019-10-12 16:13:49 -07:00
jvoisin
3cef7fe7fc Refactor tests 2019-10-12 13:32:04 -07:00
jvoisin
12489bb682 Remove a useless \ 2019-10-12 21:36:28 +02:00
jvoisin
bb903ec309 Remove useless parenthesis 2019-10-12 21:36:19 +02:00
jvoisin
4483c06f19 Replace abstractstaticmethod with abstractmethod 
Apparently, abstractstaticmethod is deprecated
since python3.3.
 2019-10-12 21:28:27 +02:00
madaidan
58773088ac Mount a new tmpfs on /tmp and drop all capabilities 
This mounts a new tmpfs on /tmp so any files residing there would be hidden
from the sandbox. Many programs store some files in there that might be useful
to an attacker.  It also drops all capabilities incase it is ever run with
extra capabilities for whatever reason.
 2019-10-05 15:21:40 +02:00
jvoisin
3714553185 Fix bubblewrap 
On some machines (like mine), `/proc` has to be mounted.  Also, since
sandboxing with bubblewrap is best effort and assumes that an attacker doesn't
have control outside of the file to clean, it's safe to __try__ to enable some
bubblewrap features, and to silently fail otherwise.
 2019-09-21 14:14:39 +02:00
jvoisin
1678d37856 Mark a comment as FP 2019-09-01 19:01:33 +02:00
jvoisin
397a18b0cc Add support for ppm 2019-09-01 09:28:46 -07:00
jvoisin
0170f0e37e Improve a bit the comments in the code 
This is related to the previous commit
 2019-09-01 13:52:02 +02:00
jvoisin
0cf0541ad9 Remove nsid fields from MSOffice documents 
nsids are random identifiers, usually used to ease merging
between documents, and can trivially be used for fingerprinting.
 2019-09-01 13:52:02 +02:00
jvoisin
0c75cd15dc Remove a mypy workaround to bump coverage back to 100% 2019-07-22 23:28:51 +02:00
jvoisin
5280b6c2b3 Add a test for svg namespace 2019-07-22 23:21:06 +02:00
georg
8bb2826f7a CI: Add job to run codespell, a spell checking software 2019-07-22 13:31:40 -07:00
jvoisin
5c33b290ae Fix mypy 2019-07-20 16:05:55 +02:00
jvoisin
dc5603eb1d Please mypy 2019-07-13 23:25:44 +02:00
jvoisin
4999209f9c Add support for svg 2019-07-13 21:26:05 +02:00
jvoisin
bdd5581033 Compress cleaned zip archives by default 2019-07-13 15:04:43 +02:00
jvoisin
47f9cb33bf Please mypy 2019-07-13 15:03:40 +02:00
jvoisin
35d550d229 Use memoization get _*_path() functions 
This shouldn't make a big difference in the CLI/extension
usage, but might improve the performances of long-running
instances, or people misusing the API.
 2019-05-16 00:31:40 +02:00
jvoisin
aa52a5c91c Please mypy wrt. the last two commits 2019-05-14 00:50:17 +02:00
Antoine Tenart
f19f6ed8b6 Rework the dependency checks to distinguish required/optional ones 
Rework the dependencies definition to include a 'required' flags, which
is passed by the check_dependencies helper to the callers, so that they
can distinguish between required and optional dependencies.

This help in two ways:
- The unit test for the dependencies was now failing when an optional
  one was missing, due to a previous rework.
- Mat2's --check-dependencies was referring to "required dependencies"
  and was misleading for the user as some of them could be optional.

Signed-off-by: Antoine Tenart <antoine.tenart@ack.tf>
 2019-05-13 23:35:26 +02:00
jvoisin
97abafdc58 Minor code cleanup 2019-05-09 09:41:05 +02:00
fuzzy
7e031c9757 typo 2019-05-03 02:39:15 -07:00
jvoisin
9516990693 Add some verification for "dangerous" tarfiles 2019-05-01 17:55:35 +02:00
jvoisin
a7ebb587e1 Handle weird permissions in tar archives 2019-04-27 22:48:40 +02:00
jvoisin
14a4cddb8b Improve the display of tarfile's members mtime 2019-04-27 21:15:06 +02:00
jvoisin
8e41b098d6 Add support for compressed tar files 2019-04-27 06:03:09 -07:00
jvoisin
82cc822a1d Add tar archive support 2019-04-27 04:05:36 -07:00
jvoisin
05f429b197 Add support for xhtml files 2019-04-14 20:36:33 +02:00
jvoisin
1e325c5b5b Please mypy 
Apparently, mypy isn't able (yet?) to deal
with variables that are changing their types
at runtime.

Python is wonderful.
 2019-03-30 10:33:16 +01:00
Antoine Tenart
d454ef5b8e libmat2: fix dependency checks for cmd line utilities 
The command line checks for command line utilities are done by trying to
access the executables and by throwing an exception when not found. This
lead to:
- The mat2 cmd line --check-dependencies option failing.
- The ffmpeg unit tests failing when ffmpeg isn't installed (even though
  it's an optional dependency).

This patch fixes it.

Signed-off-by: Antoine Tenart <antoine.tenart@ack.tf>
 2019-03-29 19:29:28 +01:00
Antoine Tenart
c824a68dd8 libmat2: reshape the dependencies list 
Invert the keys and values in DEPENDENCIES. It seems more natural to use
the key as a key in check_dependencies(), and the value as the value.
This also help in preparing for reworking the check_dependencies()
helper.

Signed-off-by: Antoine Tenart <antoine.tenart@ack.tf>
 2019-03-29 19:29:28 +01:00
jvoisin
b8c92fec09 Fix the testsuite 2019-03-23 00:41:23 +01:00
Antoine Tenart
0e3c2c9b1b libmat2: audio: not all id3 types have a text attribute 
Not all id3 types have a text attribute (such as mutagen.id3.APIC or
mutagen.id3.UFID). This leads to the get_meta helper to crash when
trying to access the text attribute of an object which does not have it.
Fixes it by checking the text attribute is available before accessing
it.

Signed-off-by: Antoine Tenart <antoine.tenart@ack.tf>
 2019-03-23 00:32:44 +01:00
Brolf
5ac91cd4f9
Refactor {black,white}list into {block,allow}list 
Closes #96
 2019-03-05 23:13:42 +00:00
georg
c3f097a82b
fix typo 2019-03-01 22:00:23 +00:00
jvoisin
55214206b5 Improve the previous commit 
- More tests
- More documentation
- Minor code cleanup
 2019-02-27 23:53:07 +01:00
jvoisin
73d2966e8c Improve epub support 2019-02-27 23:04:38 +01:00
jvoisin
eb2e702f37 Document the previous commit 2019-02-25 15:37:44 +01:00
jvoisin
545dccc352 In archive-based formats, the mimetype file comes first 
This should improve epub compatibility,
along with other formats as a side-effect
 2019-02-24 23:32:32 +01:00
jvoisin
524bae5972 <title> is also an html metadata 2019-02-23 20:47:26 +01:00
jvoisin
c757a9b7ef Fix a bug in css cleaning 
It's not mandatory to actually have a comment inside
comment delimiter, like `/**/`.
 2019-02-23 20:21:11 +01:00
jvoisin
02ff21b158 Implement epub support 2019-02-20 16:28:11 -08:00
jvoisin
a81b7658a8 Make the mandatory metadata warning generic 
This should close #95.
 2019-02-10 21:46:13 +01:00
jvoisin
6e63e03b86 Streamline a bit the previous commit 2019-02-09 15:23:16 +01:00
Poncho
a71488d459 bind mount /etc/ld.so.cache to the sandbox 
without /etc/ld.so.cache available in the sandbox, tests fail on gentoo with:
/usr/bin/ffmpeg: error while loading shared libraries: libstdc++.so.6:
    cannot open shared object file: No such file or directory
 2019-02-09 09:49:51 +01:00
jvoisin
6ef6aaa222 Improve a bit get_meta for libreoffice files 2019-02-08 23:23:56 +01:00
jvoisin
6cc034e81b Add support for html files 2019-02-08 23:05:18 +01:00
jvoisin
e1dd439fc8 Use of the archive refactoring for the office documents too 2019-02-07 22:19:37 +01:00
jvoisin
b9a62d798a Refactor a bit office get_meta handling 
This should make easier to get more metadata from
archive-based file formats.
 2019-02-04 00:31:26 +01:00
jvoisin
433609f8ea Implement .gif support 2019-02-03 21:01:58 +01:00
intrigeri
e8c1bb0e3c Whenever possible, use bwrap for subprocesses 
This should closes  #90
 2019-02-03 19:18:41 +01:00
jvoisin
8e84ba547a Add support for wmv 2019-02-02 19:19:36 +01:00
jvoisin
04bb8c8ccf Add mp4 support 2018-10-28 07:41:04 -07:00
jvoisin
3a070b0ab7 Add support for zip files 2018-10-25 11:56:46 +02:00
jvoisin
283e5e5787 Improve archive-based parser's robustness against corrupted embedded files 2018-10-25 11:56:12 +02:00
jvoisin
513d897ea0 Implement get_meta() for archives 2018-10-25 11:29:50 +02:00
jvoisin
5a9dc388ad Minor refactorisation of how we're checking for exiftool's presence 2018-10-25 11:05:06 +02:00
jvoisin
fe885babee Implement lightweight cleaning for jpg 2018-10-24 19:35:07 +02:00
jvoisin
9a81b3adfd Improve type annotation coverage 2018-10-23 16:32:28 +02:00
jvoisin
f1a071d460 Implement lightweight cleaning for png and tiff 2018-10-23 16:22:11 +02:00
jvoisin
38df679a88 Optimize the handling of problematic files 2018-10-23 13:49:58 +02:00
jvoisin
44f267a596 Improve problematic filenames support 2018-10-22 16:56:05 +02:00
jvoisin
83389a63e9 Test mat2's reliability wrt. corrupted video files 2018-10-22 13:42:04 +02:00
jvoisin
e70ea811c9 Implement support for .avi files, via ffmpeg 
- This commit introduces optional dependencies (namely ffmpeg):
  mat2 will spit a warning when trying to process an .avi file
  if ffmpeg isn't installed.
- Since metadata are obtained via exiftool, this commit
  also refactors a bit our exfitool wrapper.
 2018-10-22 12:58:01 +02:00
jvoisin
2ba38dd2a1 Bump mypy typing coverage 2018-10-12 14:32:09 +02:00
jvoisin
b832a59414 Refactor lightweight mode implementation 2018-10-12 11:49:24 +02:00
jvoisin
b9dbd12ef9 Implement recursive metadata for FLAC files 
Since FLAC files can contain covers, it makes sense
to parse their metadata
 2018-10-11 19:52:47 +02:00
jvoisin
b2e153b69c Delete pictures of FLAC files 2018-10-11 18:15:11 +02:00
jvoisin
0d25b18d26 Improve both the typing and the comments 2018-10-05 17:07:58 +02:00
jvoisin
d0f3534eff Hide unsupported extensions in mat2 -l 2018-10-05 12:43:21 +02:00
jvoisin
8e98593b02 Trash word/people.xml in office files 2018-10-04 16:28:20 +02:00
georg
34fbd633fd
libmat2: fix shebang 
Relates 0a2a398c9c
 2018-10-03 18:38:28 +00:00
jvoisin
5a5c642a46 Don't break office files for MS Office 
We didn't take the whitelist into account while
removing dangling files from [Content_types].xml
 2018-10-03 16:38:05 +02:00
jvoisin
1b356b8c6f Improve mat2's cli reliability 
- Replace some class members by instance members
- Don't thread the cleaning process anymore for now
 2018-10-03 15:22:36 +02:00
jvoisin
c67bbafb2c Use [Content_Types].xml to improve MS Office coverage 2018-10-02 11:55:42 -07:00
georg
5b606f939d
fix typo 2018-10-02 16:01:24 +00:00
jvoisin
652b8e519f Files processed via MAT2 are now accepted without warnings by MS Office 2018-10-01 12:25:37 -07:00
jvoisin
81a3881aa4 Please mypy 2018-09-30 19:55:17 +02:00
jvoisin
e342671ead Remove dangling references in MS Office's [Content_types].xml 2018-09-30 19:53:18 +02:00
jvoisin
719cdf20fa Second pass of minor formatting 2018-09-24 20:15:07 +02:00
jvoisin
2e243355f5 Fix some minor formatting issues 2018-09-24 19:50:24 +02:00
jvoisin
174d4a0ac0 Implement rsid stripping for office files 
MS Office XML rsid is a "unique identifier used to track the editing session
when the physical character representing this section mark was last formatted."

See the following links for details:
- https://msdn.microsoft.com/en-us/library/office/documentformat.openxml.wordprocessing.previoussectionproperties.rsidrpr.aspx
- https://blogs.msdn.microsoft.com/brian_jones/2006/12/11/whats-up-with-all-those-rsids/.
 2018-09-24 18:03:59 +02:00
jvoisin
fbcf68c280 Lexicographical sort on xml attributes for office files 
In XML, the order of the attributes shouldn't be meaningful,
however, MS Office sorts attributes for a given XML tag
differently than LibreOffice.
 2018-09-24 17:45:09 +02:00
jvoisin
a1a06d023e Insert archive members in lexicographic order 2018-09-18 22:44:21 +02:00
jvoisin
5cf94bd256 Bump coverage back to 100% 2018-09-12 14:54:54 +02:00
jvoisin
de65f4f4d4 Improve the resilience of MAT2 wrt. corrupted PNG 2018-09-09 19:09:05 +02:00
jvoisin
9fe6f1023b Make pylint happy 2018-09-06 11:36:04 +02:00
jvoisin
e3d817f57e Split office and archives 2018-09-06 11:34:14 +02:00
jvoisin
120b204988 Change a bit the previous commit 2018-09-06 11:13:11 +02:00
Daniel Kahn Gillmor
f3cef319b9 Unknown Members: make policy use an Enum 
Closes #60

Note: this changeset also ensures that clean.cleaned.docx is removed
up after the pytest is over.
 2018-09-05 18:59:33 -04:00
jvoisin
072ee1814d Remove defusedxml support and document why 2018-09-05 18:41:08 +02:00