Bump the changelog

A zip file with a file pointing to /etc/passwd would, upon being cleaned by
mat2, produce a file with the filesystem's /etc/passwd file.

.gitlab-ci.yml

@@ -1,66 +1,88 @@
-image: debian
+include:
+  - template: Security/SAST.gitlab-ci.yml
+
+variables:
+  CONTAINER_REGISTRY: $CI_REGISTRY/georg/mat2-ci-images
 
 stages:
   - linting
   - test
 
-bandit:
+.prepare_env: &prepare_env
+  before_script:  # This is needed to not run the testsuite as root
+    - useradd --home-dir ${CI_PROJECT_DIR} mat2
+    - chown -R mat2 .
+
+linting:bandit:
+  image: $CONTAINER_REGISTRY:linting 
   stage: linting
   script:  # TODO: remove B405 and B314
-  - apt-get -qqy update
-  - apt-get -qqy install --no-install-recommends python3-bandit
-  - bandit ./mat2 --format txt --skip B101
-  - bandit -r ./nautilus/ --format txt --skip B101
-  - bandit -r ./libmat2 --format txt --skip B101,B404,B603,B405,B314
+    - bandit ./mat2 --format txt --skip B101
+    - bandit -r ./nautilus/ --format txt --skip B101
+    - bandit -r ./libmat2 --format txt --skip B101,B404,B603,B405,B314,B108,B311
 
-pylint:
+linting:codespell:
+  image: $CONTAINER_REGISTRY:linting
   stage: linting
   script:
-  - apt-get -qqy update
-  - apt-get -qqy install --no-install-recommends pylint3 python3-mutagen python3-gi-cairo gir1.2-poppler-0.18 gir1.2-gdkpixbuf-2.0
-  - pylint3 --extension-pkg-whitelist=cairo,gi ./libmat2 ./mat2
-  # Once nautilus-python is in Debian, decomment it form the line below
-  - pylint3 --extension-pkg-whitelist=Nautilus,GObject,Gtk,Gio,GLib,gi ./nautilus/mat2.py
-
-pyflakes:
+    # Run codespell to check for spelling errors; ignore errors about binary
+    # files, use a config with ignored words and exclude the git directory,
+    # which might contain false positives
+    - codespell -q 2 -I utils/ci/codespell/ignored_words.txt -S .git
+  
+linting:pylint:
+  image: $CONTAINER_REGISTRY:linting
   stage: linting
   script:
-  - apt-get -qqy update
-  - apt-get -qqy install --no-install-recommends pyflakes3
-  - pyflakes3 ./libmat2 ./mat2 ./tests/ ./nautilus
+    - pylint --disable=no-else-return,no-else-raise,no-else-continue,unnecessary-comprehension,raise-missing-from,unsubscriptable-object,use-dict-literal,unspecified-encoding,consider-using-f-string,use-list-literal,too-many-statements --extension-pkg-whitelist=cairo,gi ./libmat2 ./mat2
+    # Once nautilus-python is in Debian, decomment it form the line below
+    - pylint --disable=no-else-return,no-else-raise,no-else-continue,unnecessary-comprehension,raise-missing-from,unsubscriptable-object,use-list-literal --extension-pkg-whitelist=Nautilus,GObject,Gtk,Gio,GLib,gi ./nautilus/mat2.py
 
-mypy:
+linting:pyflakes:
+  image: $CONTAINER_REGISTRY:linting
   stage: linting
   script:
-  - apt-get -qqy update
-  - apt-get -qqy install --no-install-recommends python3-pip
-  - pip3 install mypy
-  - mypy mat2 libmat2/*.py --ignore-missing-imports
-  - mypy --ignore-missing-imports ./nautilus/mat2.py
+    - pyflakes3 ./libmat2 ./mat2 ./tests/ ./nautilus
 
-tests:debian:
-  stage: test
+linting:mypy:
+  image: $CONTAINER_REGISTRY:linting
+  stage: linting
   script:
-  - apt-get -qqy update
-  - apt-get -qqy install --no-install-recommends python3-mutagen python3-gi-cairo gir1.2-poppler-0.18 gir1.2-gdkpixbuf-2.0 libimage-exiftool-perl python3-coverage ffmpeg
-  - python3-coverage run --branch -m unittest discover -s tests/
-  - python3-coverage report --fail-under=100 -m --include 'libmat2/*'
-
-tests:fedora:
-  image: fedora
-  stage: test
-  tags:
-    - whitewhale
-  script:
-  - dnf install -y python3 python3-mutagen python3-gobject gdk-pixbuf2 poppler-glib gdk-pixbuf2 gdk-pixbuf2-modules cairo-gobject cairo python3-cairo perl-Image-ExifTool mailcap
-  - gdk-pixbuf-query-loaders-64 > /usr/lib64/gdk-pixbuf-2.0/2.10.0/loaders.cache
-  - python3 setup.py test
+    - mypy --ignore-missing-imports mat2 libmat2/*.py ./nautilus/mat2.py
 
 tests:archlinux:
-  image: archlinux/base
+  image: $CONTAINER_REGISTRY:archlinux
   stage: test
-  tags:
-    - whitewhale
   script:
-  - pacman -Sy --noconfirm python-mutagen python-gobject gdk-pixbuf2 poppler-glib gdk-pixbuf2 python-cairo perl-image-exiftool python-setuptools mailcap ffmpeg
-  - python3 setup.py test
+    - python3 -m unittest discover -v
+  
+tests:debian:
+  image: $CONTAINER_REGISTRY:debian
+  stage: test
+  <<: *prepare_env
+  script:
+    - apt-get -qqy purge bubblewrap
+    - su - mat2 -c "python3-coverage run --branch -m unittest discover -s tests/"
+    - su - mat2 -c "python3-coverage report --fail-under=95 -m --include 'libmat2/*'"
+
+tests:debian_with_bubblewrap:
+  image: $CONTAINER_REGISTRY:debian
+  stage: test
+  allow_failure: true
+  <<: *prepare_env
+  script:
+    - apt-get -qqy install bubblewrap
+    - python3 -m unittest discover -v
+
+tests:fedora:
+  image: $CONTAINER_REGISTRY:fedora
+  stage: test
+  script:
+    - python3 -m unittest discover -v
+
+tests:gentoo:
+  image: $CONTAINER_REGISTRY:gentoo
+  stage: test
+  <<: *prepare_env
+  script:
+    - su - mat2 -c "python3 -m unittest discover -v"

.pylintrc

@@ -14,4 +14,5 @@ disable=
     catching-non-exception,
     cell-var-from-loop,
     locally-disabled,
+		raise-missing-from,
     invalid-sequence-index,  # pylint doesn't like things like `Tuple[int, bytes]` in type annotation

CHANGELOG.md

@@ -1,3 +1,116 @@
+# 0.13.0 - 2022-07-06
+
+- Fix an arbitrary file read
+- Add support for heic files 
+
+# 0.12.4 - 2022-04-30
+
+- Fix possible errors/crashes when processing multiple files
+  via the command line interface
+- Use a fixed PDF version for the output
+- Improve compatibility with modern versions of rsvg
+- Improve the robustness of the command line interface with
+  regard to control characters
+
+# 0.12.3 - 2022-01-06
+
+- Implement code for internationalization
+- Keep individual files compression type in zip files
+- Increase the robustness of mat2 against weird/corrupted files
+- Fix the dolphin integration
+- Add a fuzzer
+
+# 0.12.2 - 2021-08-29
+
+- Add support for aiff files
+- Improve MS Office support
+- Improve compatibility with newer/older version of mat2's dependencies
+- Fix possible issues with the resolution of processed pdf
+
+# 0.12.1 - 2021-03-19
+
+- Improve epub support
+- Improve MS Office support
+
+# 0.12.0 - 2020-12-18
+
+- Improve significantly MS Office formats support
+- Fix some typos in the Nautilus extension
+- Improve reliability of the mp3, pdf and svg parsers
+- Improve compatibility with ffmpeg when sandboxing is used
+- Improve the dolphin extension usability
+- libmat2 now raises a ValueError on malformed files while trying to 
+  find the right parser, instead of returning None
+
+# 0.11.0 - 2020-03-29
+
+- Improve significantly MS Office formats support
+- Refactor how mat2 looks for executables
+
+# 0.10.1 - 2020-02-09
+
+- Improve the documentation and the manpage
+- Improve the robustness of css, html, png, gdk-based, exiftool-based parsers
+- Future-proof a bit the testsuite
+- Handle tiff files with a .tif extension
+- Improve the sandbox' usability
+- Add support for wav files
+
+# 0.10.0 - 2019-11-30
+
+- Make mat2 work on Python3.8
+- Minor improvement of ppt handling
+- Minor improvement of odt handling
+- Add an integration KDE's file manager: Dolphin
+- mat2 now copies file permissions on the cleaned files
+- Add a flag to disable sandboxing
+- Tighten a bit the sandboxing
+- Improve handling of MSOffice documents
+- Add support for inplace cleaning
+- Better handling of mutually-exclusive arguments in the command line
+- Add support for svg
+- Add support for ppm
+- Cleaned zip files are compressed by default
+- Minor performances improvement when dealing with images/video files
+- Better handling of optional dependencies
+
+# 0.9.0 - 2019-05-10
+
+- Add tar/tar.gz/tar.bz2/tar.zx archives support
+- Add support for xhtml files
+- Improve handling of read-only files
+- Improve a bit the command line's documentation
+- Fix a confusing error message
+- Add even more tests
+- Usuals internal cleanups/refactorings
+
+# 0.8.0 - 2019-02-28
+
+- Add support for epub files
+- Fix the setup.py file crashing on non-utf8 platforms
+- Improve css support
+- Improve html support
+
+# 0.7.0 - 2019-02-17
+
+- Add support for wmv files
+- Add support for gif files
+- Add support for html files
+- Sandbox external processes via bubblewrap
+- Simplify archive-based formats processing
+- The Nautilus extension now plays nicer with other extensions
+
+# 0.6.0 - 2018-11-10
+
+- Add lightweight cleaning for jpeg
+- Add support for zip files
+- Add support for mp4 files
+- Improve metadata extraction for archives
+- Improve robustness against corrupted embedded files
+- Fix a possible security issue on some terminals (control character
+	injection via --show)
+- Various internal cleanup/improvements
+
 # 0.5.0 - 2018-10-23
 
 - Video (.avi files for now) support, via FFmpeg, optionally
@@ -5,7 +118,7 @@
 - Processing files starting with a dash is now quicker
 - Metadata are now displayed sorted
 - Recursive metadata support for FLAC files
-- Unsupported extensions aren't displayed in `/.mat -l` anymore
+- Unsupported extensions aren't displayed in `./mat2 -l` anymore
 - Improve the display when no metadata are found
 - Update the logo according to the GNOME guidelines
 - The testsuite is now runnable on the installed version of mat2
@@ -30,12 +143,12 @@
 
 # 0.3.1 - 2018-09-01
 
-- Document how to install MAT2 for various distributions
+- Document how to install mat2 for various distributions
 - Fix various typos in the documentation/comments
-- Add ArchLinux to the CI to ensure that MAT2 is running on it
+- Add ArchLinux to the CI to ensure that mat2 is running on it
 - Fix the handling of files with a name ending in `.JPG`
 - Improve the detection of unsupported extensions in upper-case
-- Streamline MAT2's logging
+- Streamline mat2's logging
 
 
 # 0.3.0 - 2018-08-03
@@ -55,14 +168,14 @@
 - Simplify various code-paths
 - Remove superfluous debug message
 - Remove the `--check` option that never was implemented anyway
-- Add a `-c` option to check for MAT2's dependencies
+- Add a `-c` option to check for mat2's dependencies
 
 
 # 0.1.3 - 2018-07-06
 
-- Improve MAT2 resilience against corrupted images
+- Improve mat2 resilience against corrupted images
 - Check that the minimal version of Poppler is available
-- Simplify how MAT2 deals with office files
+- Simplify how mat2 deals with office files
 - Improve cleaning of office files
 	- Thumbnails are removed
 	- Revisions are removed
@@ -74,8 +187,8 @@
 - Rename some files to ease the packaging
 - Add linters to the CI (mypy, bandit and pyflakes)
 - Prevent exitftool-related parameters injections
-- Improve MAT2's resilience against corrupted files
-- Make MAT2 work on fedora, thanks to @atenart
+- Improve mat2's resilience against corrupted files
+- Make mat2 work on fedora, thanks to @atenart
 - Tighten the threat model
 - Simplify and improve how office files are handled
 

CONTRIBUTING.md

@@ -1,11 +1,17 @@
-# Contributing to MAT2
+# Contributing to mat2
 
-The main repository for MAT2 is on [0xacab]( https://0xacab.org/jvoisin/mat2 ),
-with a mirror on [gitlab.com]( https://gitlab.com/jvoisin/mat2 ).
+The main repository for mat2 is on [0xacab]( https://0xacab.org/jvoisin/mat2 ),
+but you can send patches to jvoisin by [email](https://dustri.org/) if you prefer.
 
 Do feel free to pick up [an issue]( https://0xacab.org/jvoisin/mat2/issues )
-and to send a pull-request. Please do check that everything is fine by running the
-testsuite with `python3 -m unittest discover -v` before submitting one :)
+and to send a pull-request.
+
+Before sending the pull-request, please do check that everything is fine by
+running the full test suite in GitLab. To do that, after forking mat2 in GitLab,
+you need to go in Settings -> CI/CD -> Runner and there enable shared runners.
+
+Mat2 also has unit tests (that are also run in the full test suite). You can run
+them with `python3 -m unittest discover -v`.
 
 If you're fixing a bug or adding a new feature, please add tests accordingly,
 this will greatly improve the odds of your merge-request getting merged.
@@ -16,7 +22,7 @@ If you're adding a new fileformat, please add tests for:
 2. Cleaning metadata
 3. Raising `ValueError` upon a corrupted file
 
-Since MAT2 is written in Python3, please conform as much as possible to the
+Since mat2 is written in Python3, please conform as much as possible to the
 [pep8]( https://pep8.org/ ) style; except where it makes no sense of course.
 
 # Doing a release
@@ -29,9 +35,10 @@ Since MAT2 is written in Python3, please conform as much as possible to the
 6. Create a tag with `git tag -s $VERSION`
 7. Push the commit with `git push origin master`
 8. Push the tag with `git push --tags`
-9. Create the signed tarball with `git archive --format=tar.xz --prefix=mat-$VERSION/ $VERSION > mat-$VERSION.tar.xz`
-10. Sign the tarball with `gpg --armor --detach-sign mat-$VERSION.tar.xz`
-11. Upload the result on Gitlab's [tag page](https://0xacab.org/jvoisin/mat2/tags) and add the changelog there
-12. Announce the release on the [mailing list](https://mailman.boum.org/listinfo/mat-dev)
-13. Tell the [downstreams](https://0xacab.org/jvoisin/mat2/blob/master/INSTALL.md) about it
-14. Do the secret release dance
+9. Download the gitlab archive of the release
+10. Diff it against the local copy
+11. If there is no difference, sign the archive with `gpg --armor --detach-sign mat2-$VERSION.tar.xz`
+12. Upload the signature on Gitlab's [tag page](https://0xacab.org/jvoisin/mat2/tags) and add the changelog there
+13. Announce the release on the [mailing list](https://mailman.boum.org/listinfo/mat-dev)
+14. Sign'n'upload the new version on pypi with `python3 setup.py sdist bdist_wheel` then `twine upload -s dist/*`
+15. Do the secret release dance

INSTALL.md

@@ -1,5 +1,30 @@
+# Python ecosystem
+
+If you feel like running arbitrary code downloaded over the
+internet (pypi doesn't support gpg signatures [anymore](https://github.com/pypa/python-packaging-user-guide/pull/466)),
+mat2 is [available on pypi](https://pypi.org/project/mat2/), and can be
+installed like this:
+
+```
+pip3 install mat2
+```
+
 # GNU/Linux
 
+## Optional dependencies
+
+When [bubblewrap](https://github.com/projectatomic/bubblewrap) is
+installed, mat2 uses it to sandbox any external processes it invokes.
+
+## Arch Linux
+
+Thanks to [Francois_B](https://www.sciunto.org/), there is an package available on
+[Arch linux's AUR](https://aur.archlinux.org/packages/mat2/).
+
+## Debian
+
+There is a package available in [Debian](https://packages.debian.org/search?keywords=mat2&searchon=names&section=all).
+
 ## Fedora
 
 Thanks to [atenart](https://ack.tf/), there is a package available on
@@ -21,39 +46,6 @@ Then you can install both the Mat2 command and Nautilus extension:
 dnf -y install mat2 mat2-nautilus
 ```
 
-## Debian
-
-There is currently no package for Debian. If you want to help to make this
-happen, there is an [issue](https://0xacab.org/jvoisin/mat2/issues/16) open.
-
-But fear not, there is a way to install it *manually*:
-
-```
-# apt install python3-mutagen python3-gi-cairo gir1.2-gdkpixbuf-2.0 libimage-exiftool-perl gir1.2-glib-2.0 gir1.2-poppler-0.18
-$ git clone https://0xacab.org/jvoisin/mat2.git
-$ cd mat2
-$ ./mat2
-```
-
-and if you want to install the über-fancy Nautilus extension:
-
-```
-# apt install gnome-common gtk-doc-tools libnautilus-extension-dev python-gi-dev
-$ git clone https://github.com/GNOME/nautilus-python
-$ cd nautilus-python
-$ PYTHON=/usr/bin/python3 ./autogen.sh
-$ make
-# make install
-$ mkdir -p ~/.local/share/nautilus-python/extensions/
-$ cp ../nautilus/mat2.py ~/.local/share/nautilus-python/extensions/
-$ PYTHONPATH=/home/$USER/mat2 PYTHON=/usr/bin/python3 nautilus
-```
-
-## Arch Linux
-
-Thanks to [Francois_B](https://www.sciunto.org/), there is an package available on
-[Arch linux's AUR](https://aur.archlinux.org/packages/mat2/).
-
 ## Gentoo
 
-MAT2 is available in the [torbrowser overlay](https://github.com/MeisterP/torbrowser-overlay).
+mat2 is available in the [torbrowser overlay](https://github.com/MeisterP/torbrowser-overlay).

README.md

@@ -6,9 +6,6 @@
 
 ```
 
-This software is currently in **beta**, please don't use it for anything
-critical.
-
 # Metadata and privacy
 
 Metadata consist of information that characterizes data.
@@ -20,21 +17,32 @@ Metadata within a file can tell a lot about you.
 Cameras record data about when a picture was taken and what
 camera was used. Office documents like PDF or Office automatically adds
 author and company information to documents and spreadsheets.
-Maybe you don't want to disclose those information on the web.
+Maybe you don't want to disclose those information.
 
-This is precisely the job of MAT2: getting rid, as much as possible, of
+This is precisely the job of mat2: getting rid, as much as possible, of
 metadata.
 
+mat2 provides a command line tool, and graphical user interfaces via a service
+menu for Dolphin, the default file manager of KDE, and an extension for
+Nautilus, the default file manager of GNOME.
+
 # Requirements
 
 - `python3-mutagen` for audio support
 - `python3-gi-cairo` and `gir1.2-poppler-0.18` for PDF support
 - `gir1.2-gdkpixbuf-2.0` for images support
-- `FFmpeg`, optionally, for video support 
+- `gir1.2-rsvg-2.0` for svg support
+- `FFmpeg`, optionally, for video support
 - `libimage-exiftool-perl` for everything else
+- `bubblewrap`, optionally, for sandboxing
 
-Please note that MAT2 requires at least Python3.5, meaning that it
-doesn't run on [Debian Jessie](https://packages.debian.org/jessie/python3).
+Please note that mat2 requires at least Python3.5.
+
+# Requirements setup on macOS (OS X) using [Homebrew](https://brew.sh/)
+
+```bash
+brew install exiftool cairo pygobject3 poppler gdk-pixbuf librsvg ffmpeg
+```
 
 # Running the test suite
 
@@ -42,11 +50,18 @@ doesn't run on [Debian Jessie](https://packages.debian.org/jessie/python3).
 $ python3 -m unittest discover -v
 ```
 
-# How to use MAT2
+And if you want to see the coverage:
 
 ```bash
-usage: mat2 [-h] [-v] [-l] [--check-dependencies] [-V]
-            [--unknown-members policy] [-s | -L]
+$ python3-coverage run --branch -m unittest discover -s tests/
+$ python3-coverage report --include -m --include /libmat2/*'
+```
+
+# How to use mat2
+
+```
+usage: mat2 [-h] [-V] [--unknown-members policy] [--inplace] [--no-sandbox]
+            [-v] [-l] [--check-dependencies] [-L | -s]
             [files [files ...]]
 
 Metadata anonymisation toolkit 2
@@ -56,32 +71,61 @@ positional arguments:
 
 optional arguments:
   -h, --help            show this help message and exit
-  -v, --version         show program's version number and exit
-  -l, --list            list all supported fileformats
-  --check-dependencies  check if MAT2 has all the dependencies it needs
   -V, --verbose         show more verbose status information
   --unknown-members policy
                         how to handle unknown members of archive-style files
-                        (policy should be one of: abort, omit, keep)
-  -s, --show            list harmful metadata detectable by MAT2 without
-                        removing them
+                        (policy should be one of: abort, omit, keep) [Default:
+                        abort]
+  --inplace             clean in place, without backup
+  --no-sandbox          Disable bubblewrap's sandboxing
+  -v, --version         show program's version number and exit
+  -l, --list            list all supported fileformats
+  --check-dependencies  check if mat2 has all the dependencies it needs
   -L, --lightweight     remove SOME metadata
+  -s, --show            list harmful metadata detectable by mat2 without
+                        removing them
 ```
 
-Note that MAT2 **will not** clean files in-place, but will produce, for
+Note that mat2 **will not** clean files in-place, but will produce, for
 example, with a file named "myfile.png" a cleaned version named
 "myfile.cleaned.png".
 
+## Web interface
+
+It's possible to run mat2 as a web service, via
+[mat2-web](https://0xacab.org/jvoisin/mat2-web).
+
+## Desktop GUI
+
+For GNU/Linux desktops, it's possible to use the
+[Metadata Cleaner](https://gitlab.com/rmnvgr/metadata-cleaner) GTK application.
+
+# Supported formats
+
+The following formats are supported: avi, bmp, css, epub/ncx, flac, gif, jpeg,
+m4a/mp2/mp3/…, mp4, odc/odf/odg/odi/odp/ods/odt/…, off/opus/oga/spx/…, pdf,
+png, ppm, pptx/xlsx/docx/…, svg/svgz/…, tar/tar.gz/tar.bz2/tar.xz/…, tiff,
+torrent, wav, wmv, zip, …
+  
 # Notes about detecting metadata
 
-While MAT2 is doing its very best to display metadata when the `--show` flag is
-passed, it doesn't mean that a file is clean from any metadata if MAT2 doesn't
+While mat2 is doing its very best to display metadata when the `--show` flag is
+passed, it doesn't mean that a file is clean from any metadata if mat2 doesn't
 show any. There is no reliable way to detect every single possible metadata for
 complex file formats.
 
 This is why you shouldn't rely on metadata's presence to decide if your file must
 be cleaned or not.
 
+# Notes about the lightweight mode
+
+By default, mat2 might alter a bit the data of your files, in order to remove
+as much metadata as possible. For example, texts in PDF might not be selectable anymore,
+compressed images might get compressed again, …
+Since some users might be willing to trade some metadata's presence in exchange
+of the guarantee that mat2 won't modify the data of their files, there is the
+`-L` flag that precisely does that.
+
 # Related software
 
 - The first iteration of [MAT](https://mat.boum.org)
@@ -92,15 +136,21 @@ be cleaned or not.
 	watermarks from PDF.
 - [Scrambled Exif](https://f-droid.org/packages/com.jarsilio.android.scrambledeggsif/),
 	an open-source Android application to remove metadata from pictures.
+- [Dangerzone](https://dangerzone.rocks/), designed to sanitize harmful documents
+  into harmless ones.
 
 # Contact
 
 If possible, use the [issues system](https://0xacab.org/jvoisin/mat2/issues)
-or the [mailing list](https://mailman.boum.org/listinfo/mat-dev)
+or the [mailing list](https://www.autistici.org/mailman/listinfo/mat-dev)
 Should a more private contact be needed (eg. for reporting security issues),
 you can email Julien (jvoisin) Voisin at `julien.voisin+mat2@dustri.org`,
 using the gpg key `9FCDEE9E1A381F311EA62A7404D041E8171901CC`.
 
+# Donations
+
+If you want to donate some money, please give it to [Tails]( https://tails.boum.org/donate/?r=contribute ).
+
 # License
 
 This program is free software: you can redistribute it and/or modify
@@ -116,16 +166,20 @@ GNU General Public License for more details.
 You should have received a copy of the GNU Lesser General Public License
 along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-Copyright 2018 Julien (jvoisin) Voisin <julien.voisin+mat2@dustri.org>
-Copyright 2016 Marie Rose for MAT2's logo
+Copyright 2018 Julien (jvoisin) Voisin <julien.voisin+mat2@dustri.org>  
+Copyright 2016 Marie-Rose for mat2's logo
+
+The `tests/data/dirty_with_nsid.docx` file is licensed under GPLv3,
+and was borrowed from the Calibre project: https://calibre-ebook.com/downloads/demos/demo.docx
+
+The `narrated_powerpoint_presentation.pptx` file is in the public domain.
 
 # Thanks
 
-MAT2 wouldn't exist without:
+mat2 wouldn't exist without:
 
 - the [Google Summer of Code](https://summerofcode.withgoogle.com/);
 - the fine people from [Tails]( https://tails.boum.org);
 - friends
 
 Many thanks to them!
-

doc/comparison_to_others.md

@@ -0,0 +1,51 @@
+# Exiftool
+
+mat2 is in fact using exiftool to extract metadata from files,
+but not to remove them. The previous iteration of mat2, MAT,
+was using exiftool to remove metadata, which lead to several cases where
+they weren't correctly removed, if at all.
+For example, [Exiftool's documentation](https://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/PDF.html)
+states the following with regard to PDF:
+
+> All metadata edits are reversible. While this would normally be considered an
+advantage, it is a potential security problem because old information is never
+actually deleted from the file.
+
+To remove metadata, mat2 usually re-render the file completely, eliminating
+all possible original metadata. See the `implementation_notes.md` file for
+details.
+
+
+# jpegoptim, optipng, …
+
+While designed to reduce as much as possible the size of pictures,
+those software can be used to remove metadata. They usually have very good
+support for a single picture format, and can be used in place of mat2 for them.
+
+
+# PDF Redact Tools
+
+[PDF Redact Tools](https://github.com/firstlookmedia/pdf-redact-tools) is
+a software developed by the people from [First Look
+Media](https://firstlook.media/), the entity behind, amongst other things, 
+[The Intercept](https://theintercept.com/).
+
+The tool uses roughly the same approach than mat2 to deal with PDF,
+which is unfortunately the only fileformat that it does support.
+It's interesting to note that it has counter-measures against
+[yellow dots](https://en.wikipedia.org/wiki/Machine_Identification_Code),
+a capacity that mat2 [doesn't possess yet](https://0xacab.org/jvoisin/mat2/issues/43).
+
+
+# Exiv2
+
+[Exiv2](https://www.exiv2.org/) was considered for mat2,
+but it currently [misses a lot of metadata](https://0xacab.org/jvoisin/mat2/issues/85)
+
+
+# Others non open source software/online service
+
+There are a lot of closed-source software and online service claiming to remove
+metadata from your files, but since there is no way to actually verify that
+they're effectively removing them, let alone adding unique markers, they
+shouldn't be used.

doc/implementation_notes.md

@@ -4,7 +4,7 @@ Implementation notes
 Lightweight cleaning mode
 -------------------------
 
-Due to *popular* request, MAT2 is providing a *lightweight* cleaning mode,
+Due to *popular* request, mat2 is providing a *lightweight* cleaning mode,
 that only cleans the superficial metadata of your file, but not
 the ones that might be in **embedded** resources. Like for example,
 images in a PDF or an office document.
@@ -12,33 +12,49 @@ images in a PDF or an office document.
 Revisions handling
 ------------------
 
-Revisions are handled according to the principle of least astonishment: they are entirely removed.
+Revisions are handled according to the principle of least astonishment: they
+are entirely removed.
 
-- Either the users aren't aware of the revisions, are thus they should be deleted. For example journalists that are editing a document to erase mentions sources mentions.
+- Either the users aren't aware of the revisions, are thus they should be
+	deleted. For example journalists that are editing a document to erase
+	mentions sources mentions.
 
-- Or they are aware of it, and will likely not expect MAT2 to be able to keep the revisions, that are basically traces about how, when and who edited the document.
+- Or they are aware of it, and will likely not expect mat2 to be able to keep
+	the revisions, that are basically traces about how, when and who edited the
+	document.
 
 
 Race conditions
 ---------------
 
-MAT2 does its very best to avoid crashing at runtime. This is why it's checking
-if the file is valid __at parser creation__. MAT2 doesn't take any measure to
+mat2 does its very best to avoid crashing at runtime. This is why it's checking
+if the file is valid __at parser creation__. mat2 doesn't take any measure to
 ensure that the file is not changed between the time the parser is
 instantiated, and the call to clean or show the metadata.
 
 Symlink attacks
 ---------------
 
-MAT2 output predictable filenames (like yourfile.jpg.cleaned).
+mat2 output predictable filenames (like yourfile.jpg.cleaned).
 This may lead to symlink attack. Please check if you OS prevent
 against them
 
 Archives handling
 -----------------
 
-MAT2 doesn't support archives yet, because we haven't found an usable way to ask the user
-what to do when a non-supported files are encountered.
+By default, when cleaning a non-support file format in an archive,
+mat2 will abort with a detailed error message.
+While strongly discouraged, it's possible to override this behaviour to force
+the exclusion, or inclusion of unknown files into the cleaned archive.
+
+While Python's [zipfile](https://docs.python.org/3/library/zipfile.html) module
+provides *safe* way to extract members of a zip archive, the 
+[tarfile](https://docs.python.org/3/library/tarfile.html) one doesn't,
+meaning that it's up to mat2 to implement safety checks. Currently,
+it defends against path-traversal, both relative and absolute,
+symlink-related attacks, setuid/setgid attacks, duplicate members, block and
+char devices, … but there might still be dragons lurking there.
+
 
 PDF handling
 ------------
@@ -49,10 +65,10 @@ didn't remove any *deep metadata*, like the ones in embedded pictures. This was
 on of the reason MAT was abandoned: the absence of satisfying solution to
 handle PDF. But apparently, people are ok with [pdf redact
 tools](https://github.com/firstlookmedia/pdf-redact-tools), that simply
-transform the PDF into images. So this is what's MAT2 is doing too.
+transform the PDF into images. So this is what's mat2 is doing too.
 
 Of course, it would be possible to detect images in PDf file, and process them
-with MAT2, but since a PDF can contain a lot of things, like images, videos,
+with mat2, but since a PDF can contain a lot of things, like images, videos,
 javascript, pdf, blobs, … this is the easiest and safest way to clean them.
 
 Images handling
@@ -65,7 +81,7 @@ XML attacks
 -----------
 
 Since our threat model conveniently excludes files crafted to specifically
-bypass MAT2, fileformats containing harmful XML are out of our scope.
-But since MAT2 is using [etree](https://docs.python.org/3/library/xml.html#xml-vulnerabilities)
+bypass mat2, fileformats containing harmful XML are out of our scope.
+But since mat2 is using [etree](https://docs.python.org/3/library/xml.html#xml-vulnerabilities)
 to process XML, it's "only" vulnerable to DoS, and not memory corruption:
 odds are that the user will notice that the cleaning didn't succeed.

doc/mat2.1

@@ -1,4 +1,4 @@
-.TH MAT2 "1" "October 2018" "MAT2 0.5.0" "User Commands"
+.TH mat2 "1" "July 2022" "mat2 0.13.0" "User Commands"
 
 .SH NAME
 mat2 \- the metadata anonymisation toolkit 2
@@ -8,7 +8,7 @@ mat2 \- the metadata anonymisation toolkit 2
 
 .SH DESCRIPTION
 .B mat2
-removes metadata from various fileformats. It supports a wide variety of file 
+removes metadata from various fileformats. It supports a wide variety of file
 formats, audio, office, images, …
 
 Careful, mat2 does not clean files in-place, instead, it will produce a file with the word
@@ -32,7 +32,7 @@ show program's version number and exit
 list all supported fileformats
 .TP
 \fB\-\-check\-dependencies\fR
-check if MAT2 has all the dependencies it needs
+check if mat2 has all the dependencies it needs
 .TP
 \fB\-V\fR, \fB\-\-verbose\fR
 show more verbose status information
@@ -41,11 +41,16 @@ show more verbose status information
 how to handle unknown members of archive-style files (policy should be one of: abort, omit, keep)
 .TP
 \fB\-s\fR, \fB\-\-show\fR
-list harmful metadata detectable by MAT2 without
-removing them
+list harmful metadata detectable by mat2 without removing them
 .TP
 \fB\-L\fR, \fB\-\-lightweight\fR
 remove SOME metadata
+.TP
+\fB\--no-sandbox\fR
+disable bubblewrap's sandboxing
+.TP
+\fB\--inplace\fR
+clean in place, without backup
 
 .SH EXAMPLES
 To remove all the metadata from a PDF file:
@@ -57,6 +62,24 @@ mat2 ./myfile.pdf
 .fi
 .PP
 
+.SH NOTES ABOUT METADATA
+
+While mat2 is doing its very best to display metadata when the --show flag is
+passed, it doesn't mean that a file is clean from any metadata if mat2 doesn't
+show any. There is no reliable way to detect every single possible metadata for
+complex file formats.
+.PP
+This is why you shouldn't rely on metadata's presence to decide if your file must
+be cleaned or not.
+.PP
+Moreover, mat2 goes to great lengths to make sure that as much metadata as
+possible are removed. This might sometimes result in a loss of quality of the
+processed files. For example, textual based pdf file converted into image based
+one means that it'll be no longer possible to select text in them. If you're
+experiencing this, you might want to give the lightweight cleaning mode a try,
+but keep in mind by doing so, some metadata \fBwon't be cleaned\fR.
+
+
 .SH BUGS
 
 While mat2 does its very best to remove every single metadata,

doc/threat_model.md

@@ -3,7 +3,7 @@ Threat Model
 
 The Metadata Anonymisation Toolkit 2 adversary has a number
 of goals, capabilities, and counter-attack types that can be
-used to guide us towards a set of requirements for the MAT2.
+used to guide us towards a set of requirements for the mat2.
 
 This is an overhaul of MAT's (the first iteration of the software) one.
 
@@ -53,7 +53,7 @@ Adversary
     user. This is the strongest position for the adversary to
     have. In this case, the adversary is capable of inserting
     arbitrary, custom watermarks specifically for tracking
-    the user. In general, MAT2 cannot defend against this
+    the user. In general, mat2 cannot defend against this
     adversary, but we list it for completeness' sake.
 
     - The adversary created the document for a group of users.
@@ -65,7 +65,7 @@ Adversary
 		- The adversary did not create the document, the weakest
 		position for the adversary to have. The file format is
 		(most of the time) standard, nothing custom is added:
-		MAT2 must be able to remove all metadata from the file.
+		mat2 must be able to remove all metadata from the file.
 
 
 Requirements
@@ -73,28 +73,28 @@ Requirements
 
 * Processing
 
-    - MAT2 *should* avoid interactions with information.
+    - mat2 *should* avoid interactions with information.
     Its goal is to remove metadata, and the user is solely
     responsible for the information of the file.
 
-    - MAT2 *must* warn when encountering an unknown
-    format. For example, in a zipfile, if MAT encounters an
+    - mat2 *must* warn when encountering an unknown
+    format. For example, in a zipfile, if mat2 encounters an
     unknown format, it should warn the user, and ask if the
     file should be added to the anonymised archive that is
     produced.
 
-    - MAT2 *must* not add metadata, since its purpose is to
+    - mat2 *must* not add metadata, since its purpose is to
     anonymise files: every added items of metadata decreases
     anonymity.
 
-    - MAT2 *should* handle unknown/hidden metadata fields,
+    - mat2 *should* handle unknown/hidden metadata fields,
     like proprietary extensions of open formats.
 
-		- MAT2 *must not* fail silently. Upon failure,
-		MAT2 *must not* modify the file in any way.
+		- mat2 *must not* fail silently. Upon failure,
+		mat2 *must not* modify the file in any way.
 
-		- MAT2 *might* leak the fact that MAT2 was used on the file,
+		- mat2 *might* leak the fact that mat2 was used on the file,
 		since it might be uncommon for some file formats to come
 		without any kind of metadata, an adversary might suspect that
-		the user used MAT2 on certain files.
+		the user used mat2 on certain files.
 

dolphin/README.md

@@ -0,0 +1,14 @@
+Dolphin integration
+===================
+
+Thanks to [Miguel Marco](https://riemann.unizar.es/~mmarco/), here is an neat
+integration for [Dolphin](https://kde.org/applications/system/org.kde.dolphin),
+the KDE file manager:
+
+1. Add the `mat2.desktop` file either in
+	- `/usr/share/kservices5/ServiceMenus/` to install it globally
+	- `~/.local/share/kservices5/ServiceMenus/` for a specific user
+2. Run `kbuildsycoca5` to update the corresponding database
+3. Enjoy your new contextual menu to remove metadata from your files!
+
+

dolphin/mat2.desktop

@@ -0,0 +1,13 @@
+[Desktop Entry]
+X-KDE-ServiceTypes=KonqPopupMenu/Plugin
+MimeType=application/pdf;application/vnd.oasis.opendocument.chart;application/vnd.oasis.opendocument.formula;application/vnd.oasis.opendocument.graphics;application/vnd.oasis.opendocument.image;application/vnd.oasis.opendocument.presentation;application/vnd.oasis.opendocument.spreadsheet;application/vnd.oasis.opendocument.text;application/vnd.openxmlformats-officedocument.presentationml.presentation;application/vnd.openxmlformats-officedocument.spreadsheetml.sheet;application/vnd.openxmlformats-officedocument.wordprocessingml.document;application/x-bittorrent;application/zip;audio/flac;audio/mpeg;audio/ogg;audio/x-flac;image/jpeg;image/png;image/tiff;image/x-ms-bmp;text/plain;video/mp4;video/x-msvideo;
+Actions=cleanMetadata;
+Type=Service
+
+[Desktop Action cleanMetadata]
+Name=Clean metadata
+Name[de]=Metadaten löschen
+Name[es]=Limpiar metadatos
+Icon=/usr/share/icons/hicolor/scalable/apps/mat2.svg
+Exec=kdialog --yesno  "$( mat2 -s %F )" --title "Clean Metadata?" && mat2 %U
+Exec[de]=kdialog --yesno  "$( mat2 -s %F )" --title "Metadaten löschen?" && mat2 %U

libmat2/__init__.py

@@ -1,15 +1,15 @@
 #!/usr/bin/env python3
 
-import collections
 import enum
 import importlib
-from typing import Dict, Optional
+from typing import Dict, Optional, Union
 
 from . import exiftool, video
 
 # make pyflakes happy
 assert Dict
 assert Optional
+assert Union
 
 # A set of extension that aren't supported, despite matching a supported mimetype
 UNSUPPORTED_EXTENSIONS = {
@@ -30,31 +30,69 @@ UNSUPPORTED_EXTENSIONS = {
     }
 
 DEPENDENCIES = {
-    'cairo': 'Cairo',
-    'gi': 'PyGobject',
-    'gi.repository.GdkPixbuf': 'GdkPixbuf from PyGobject',
-    'gi.repository.Poppler': 'Poppler from PyGobject',
-    'gi.repository.GLib': 'GLib from PyGobject',
-    'mutagen': 'Mutagen',
-    }
+    'Cairo': {
+        'module': 'cairo',
+        'required': True,
+    },
+    'PyGobject': {
+        'module': 'gi',
+        'required': True,
+    },
+    'GdkPixbuf from PyGobject': {
+        'module': 'gi.repository.GdkPixbuf',
+        'required': True,
+    },
+    'Poppler from PyGobject': {
+        'module': 'gi.repository.Poppler',
+        'required': True,
+    },
+    'GLib from PyGobject': {
+        'module': 'gi.repository.GLib',
+        'required': True,
+    },
+    'Mutagen': {
+        'module': 'mutagen',
+        'required': True,
+    },
+}
 
+CMD_DEPENDENCIES = {
+    'Exiftool': {
+        'cmd': exiftool._get_exiftool_path,
+        'required': False,
+    },
+    'Ffmpeg': {
+        'cmd': video._get_ffmpeg_path,
+        'required': False,
+    },
+}
 
-
-def check_dependencies() -> Dict[str, bool]:
-    ret = collections.defaultdict(bool)  # type: Dict[str, bool]
-
-    ret['Exiftool'] = True if exiftool._get_exiftool_path() else False
-    ret['Ffmpeg'] = True if video._get_ffmpeg_path() else False
+def check_dependencies() -> Dict[str, Dict[str, bool]]:
+    ret = dict()  # type: Dict[str, dict]
 
     for key, value in DEPENDENCIES.items():
-        ret[value] = True
+        ret[key] = {
+            'found': True,
+            'required': value['required'],
+        }
         try:
-            importlib.import_module(key)
+            importlib.import_module(value['module'])  # type: ignore
         except ImportError:  # pragma: no cover
-            ret[value] = False  # pragma: no cover
+            ret[key]['found'] = False
+
+    for k, v in CMD_DEPENDENCIES.items():
+        ret[k] = {
+            'found': True,
+            'required': v['required'],
+        }
+        try:
+            v['cmd']()  # type: ignore
+        except RuntimeError:  # pragma: no cover
+            ret[k]['found'] = False
 
     return ret
 
+
 @enum.unique
 class UnknownMemberPolicy(enum.Enum):
     ABORT = 'abort'

libmat2/abstract.py

@@ -25,16 +25,23 @@ class AbstractParser(abc.ABC):
 
         self.filename = filename
         fname, extension = os.path.splitext(filename)
+
+        # Special case for tar.gz, tar.bz2, … files
+        if fname.endswith('.tar') and len(fname) > 4:
+            fname, extension = fname[:-4], '.tar' + extension
+
         self.output_filename = fname + '.cleaned' + extension
         self.lightweight_cleaning = False
+        self.sandbox = True
 
     @abc.abstractmethod
     def get_meta(self) -> Dict[str, Union[str, dict]]:
-        pass  # pragma: no cover
+        """Return all the metadata of the current file"""
 
     @abc.abstractmethod
     def remove_all(self) -> bool:
         """
+        Remove all the metadata of the current file
+
         :raises RuntimeError: Raised if the cleaning process went wrong.
         """
-        pass  # pragma: no cover

libmat2/archive.py

@@ -1,26 +1,58 @@
+import abc
+import stat
 import zipfile
 import datetime
+import tarfile
 import tempfile
 import os
 import logging
 import shutil
-from typing import Dict, Set, Pattern, Union
+from typing import Dict, Set, Pattern, Union, Any, List
 
 from . import abstract, UnknownMemberPolicy, parser_factory
 
 # Make pyflakes happy
 assert Set
 assert Pattern
-assert Union
+
+# pylint: disable=not-callable,assignment-from-no-return,too-many-branches
+
+# An ArchiveClass is a class representing an archive,
+# while an ArchiveMember is a class representing an element
+# (usually a file) of an archive.
+ArchiveClass = Union[zipfile.ZipFile, tarfile.TarFile]
+ArchiveMember = Union[zipfile.ZipInfo, tarfile.TarInfo]
 
 
 class ArchiveBasedAbstractParser(abstract.AbstractParser):
-    """ Office files (.docx, .odt, …) are zipped files. """
+    """Base class for all archive-based formats.
+
+    Welcome to a world of frustrating complexity and tediouness:
+        - A lot of file formats (docx, odt, epubs, …) are archive-based,
+          so we need to add callbacks erverywhere to allow their respective
+          parsers to apply specific cleanup to the required files.
+        - Python has two different modules to deal with .tar and .zip files,
+          with similar-but-yet-o-so-different API, so we need to write
+          a ghetto-wrapper to avoid duplicating everything
+        - The combination of @staticmethod and @abstractstaticmethod is
+          required because for now, mypy doesn't know that
+          @abstractstaticmethod is, indeed, a static method.
+        - Mypy is too dumb (yet) to realise that a type A is valid under
+          the Union[A, B] constrain, hence the weird `#  type: ignore`
+          annotations.
+    """
+    # Tarfiles can optionally support compression
+    # https://docs.python.org/3/library/tarfile.html#tarfile.open
+    compression = ''
+
     def __init__(self, filename):
         super().__init__(filename)
+        # We ignore typing here because mypy is too stupid
+        self.archive_class = None  #  type: ignore
+        self.member_class = None  #  type: ignore
 
         # Those are the files that have a format that _isn't_
-        # supported by MAT2, but that we want to keep anyway.
+        # supported by mat2, but that we want to keep anyway.
         self.files_to_keep = set()  # type: Set[Pattern]
 
         # Those are the files that we _do not_ want to keep,
@@ -31,10 +63,12 @@ class ArchiveBasedAbstractParser(abstract.AbstractParser):
         # the archive?
         self.unknown_member_policy = UnknownMemberPolicy.ABORT  # type: UnknownMemberPolicy
 
-        try:  # better fail here than later
-            zipfile.ZipFile(self.filename)
-        except zipfile.BadZipFile:
-            raise ValueError
+        # The LGTM comment is to mask a false-positive,
+        # see https://lgtm.com/projects/g/jvoisin/mat2/
+        self.is_archive_valid()  # lgtm [py/init-calls-subclass]
+
+    def is_archive_valid(self):
+        """Raise a ValueError is the current archive isn't a valid one."""
 
     def _specific_cleanup(self, full_path: str) -> bool:
         """ This method can be used to apply specific treatment
@@ -42,87 +76,389 @@ class ArchiveBasedAbstractParser(abstract.AbstractParser):
         # pylint: disable=unused-argument,no-self-use
         return True  # pragma: no cover
 
-    @staticmethod
-    def _clean_zipinfo(zipinfo: zipfile.ZipInfo) -> zipfile.ZipInfo:
-        zipinfo.create_system = 3  # Linux
-        zipinfo.comment = b''
-        zipinfo.date_time = (1980, 1, 1, 0, 0, 0)  # this is as early as a zipfile can be
-        return zipinfo
+    def _specific_get_meta(self, full_path: str, file_path: str) -> Dict[str, Any]:
+        """ This method can be used to extract specific metadata
+        from files present in the archive."""
+        # pylint: disable=unused-argument,no-self-use
+        return {}  # pragma: no cover
+
+    def _final_checks(self) -> bool:
+        """ This method is invoked after the file has been cleaned,
+        allowing to run final verifications.
+        """
+        # pylint: disable=unused-argument,no-self-use
+        return True
 
     @staticmethod
-    def _get_zipinfo_meta(zipinfo: zipfile.ZipInfo) -> Dict[str, str]:
-        metadata = {}
-        if zipinfo.create_system == 3:  # this is Linux
-            pass
-        elif zipinfo.create_system == 2:
-            metadata['create_system'] = 'Windows'
-        else:
-            metadata['create_system'] = 'Weird'
+    @abc.abstractmethod
+    def _get_all_members(archive: ArchiveClass) -> List[ArchiveMember]:
+        """Return all the members of the archive."""
 
-        if zipinfo.comment:
-            metadata['comment'] = zipinfo.comment  # type: ignore
+    @staticmethod
+    @abc.abstractmethod
+    def _clean_member(member: ArchiveMember) -> ArchiveMember:
+        """Remove all the metadata for a given member."""
 
-        if zipinfo.date_time != (1980, 1, 1, 0, 0, 0):
-            metadata['date_time'] = str(datetime.datetime(*zipinfo.date_time))
+    @staticmethod
+    @abc.abstractmethod
+    def _get_member_meta(member: ArchiveMember) -> Dict[str, str]:
+        """Return all the metadata of a given member."""
 
-        return metadata
+    @staticmethod
+    @abc.abstractmethod
+    def _get_member_name(member: ArchiveMember) -> str:
+        """Return the name of the given member."""
+
+    @abc.abstractmethod
+    def _add_file_to_archive(self, archive: ArchiveClass, member: ArchiveMember,
+                             full_path: str):
+        """Add the file at full_path to the archive, via the given member."""
+
+    @staticmethod
+    def _set_member_permissions(member: ArchiveMember, permissions: int) -> ArchiveMember:
+        """Set the permission of the archive member."""
+        # pylint: disable=unused-argument
+        return member
+
+    @staticmethod
+    def _get_member_compression(member: ArchiveMember):
+        """Get the compression of the archive member."""
+        # pylint: disable=unused-argument
+        return None
+
+    @staticmethod
+    def _set_member_compression(member: ArchiveMember, compression) -> ArchiveMember:
+        """Set the compression of the archive member."""
+        # pylint: disable=unused-argument
+        return member
+
+    def get_meta(self) -> Dict[str, Union[str, dict]]:
+        meta = dict()  # type: Dict[str, Union[str, dict]]
+
+        with self.archive_class(self.filename) as zin:
+            temp_folder = tempfile.mkdtemp()
+
+            for item in self._get_all_members(zin):
+                local_meta = self._get_member_meta(item)
+                member_name = self._get_member_name(item)
+
+                if member_name[-1] == '/':  # pragma: no cover
+                    # `is_dir` is added in Python3.6
+                    continue  # don't keep empty folders
+
+                zin.extract(member=item, path=temp_folder)
+                full_path = os.path.join(temp_folder, member_name)
+                os.chmod(full_path, stat.S_IRUSR)
+
+                specific_meta = self._specific_get_meta(full_path, member_name)
+                local_meta = {**local_meta, **specific_meta}
+
+                member_parser, _ = parser_factory.get_parser(full_path)  # type: ignore
+                if member_parser:
+                    local_meta = {**local_meta, **member_parser.get_meta()}
+
+                if local_meta:
+                    meta[member_name] = local_meta
+
+        shutil.rmtree(temp_folder)
+        return meta
 
     def remove_all(self) -> bool:
         # pylint: disable=too-many-branches
 
-        with zipfile.ZipFile(self.filename) as zin,\
-             zipfile.ZipFile(self.output_filename, 'w') as zout:
+        with self.archive_class(self.filename) as zin,\
+             self.archive_class(self.output_filename, 'w' + self.compression) as zout:
 
             temp_folder = tempfile.mkdtemp()
             abort = False
 
+            # Sort the items to process, to reduce fingerprinting,
+            # and keep them in the `items` variable.
+            items = list()  # type: List[ArchiveMember]
+            for item in sorted(self._get_all_members(zin), key=self._get_member_name):
+                # Some fileformats do require to have the `mimetype` file
+                # as the first file in the archive.
+                if self._get_member_name(item) == 'mimetype':
+                    items.insert(0, item)
+                else:
+                    items.append(item)
+
             # Since files order is a fingerprint factor,
             # we're iterating (and thus inserting) them in lexicographic order.
-            for item in sorted(zin.infolist(), key=lambda z: z.filename):
-                if item.filename[-1] == '/':  # `is_dir` is added in Python3.6
+            for item in items:
+                member_name = self._get_member_name(item)
+                if member_name[-1] == '/':  # `is_dir` is added in Python3.6
                     continue  # don't keep empty folders
 
+                full_path = os.path.join(temp_folder, member_name)
+                if not os.path.abspath(full_path).startswith(temp_folder):
+                    logging.error("%s contains a file (%s) pointing outside (%s) of its root.",
+                            self.filename, member_name, full_path)
+                    abort = True
+                    break
+
                 zin.extract(member=item, path=temp_folder)
-                full_path = os.path.join(temp_folder, item.filename)
+
+                try:
+                    original_permissions = os.stat(full_path).st_mode
+                except FileNotFoundError:
+                    logging.error("Something went wrong during processing of "
+                            "%s in %s, likely a path traversal attack.",
+                            member_name, self.filename)
+                    abort = True
+                    # we're breaking instead of continuing, because this exception
+                    # is raised in case of weird path-traversal-like atttacks.
+                    break
+
+                os.chmod(full_path, original_permissions | stat.S_IWUSR | stat.S_IRUSR)
+
+                original_compression = self._get_member_compression(item)
 
                 if self._specific_cleanup(full_path) is False:
-                    logging.warning("Something went wrong during deep cleaning of %s",
-                                    item.filename)
+                    logging.warning("Something went wrong during deep cleaning of %s in %s",
+                                    member_name, self.filename)
                     abort = True
                     continue
 
-                if any(map(lambda r: r.search(item.filename), self.files_to_keep)):
+                if any(map(lambda r: r.search(member_name), self.files_to_keep)):
                     # those files aren't supported, but we want to add them anyway
                     pass
-                elif any(map(lambda r: r.search(item.filename), self.files_to_omit)):
+                elif any(map(lambda r: r.search(member_name), self.files_to_omit)):
                     continue
                 else:  # supported files that we want to first clean, then add
-                    tmp_parser, mtype = parser_factory.get_parser(full_path)  # type: ignore
-                    if not tmp_parser:
+                    member_parser, mtype = parser_factory.get_parser(full_path)  # type: ignore
+                    if not member_parser:
                         if self.unknown_member_policy == UnknownMemberPolicy.OMIT:
                             logging.warning("In file %s, omitting unknown element %s (format: %s)",
-                                            self.filename, item.filename, mtype)
+                                            self.filename, member_name, mtype)
                             continue
                         elif self.unknown_member_policy == UnknownMemberPolicy.KEEP:
                             logging.warning("In file %s, keeping unknown element %s (format: %s)",
-                                            self.filename, item.filename, mtype)
+                                            self.filename, member_name, mtype)
                         else:
-                            logging.error("In file %s, element %s's format (%s) " +
+                            logging.error("In file %s, element %s's format (%s) "
                                           "isn't supported",
-                                          self.filename, item.filename, mtype)
+                                          self.filename, member_name, mtype)
                             abort = True
                             continue
-                    if tmp_parser:
-                        tmp_parser.remove_all()
-                        os.rename(tmp_parser.output_filename, full_path)
+                    else:
+                        if member_parser.remove_all() is False:
+                            logging.warning("In file %s, something went wrong \
+                                             with the cleaning of %s \
+                                             (format: %s)",
+                                            self.filename, member_name, mtype)
+                            abort = True
+                            continue
+                        os.rename(member_parser.output_filename, full_path)
 
-                zinfo = zipfile.ZipInfo(item.filename)  # type: ignore
-                clean_zinfo = self._clean_zipinfo(zinfo)
-                with open(full_path, 'rb') as f:
-                    zout.writestr(clean_zinfo, f.read())
+                zinfo = self.member_class(member_name)  # type: ignore
+                zinfo = self._set_member_permissions(zinfo, original_permissions)
+                zinfo = self._set_member_compression(zinfo, original_compression)
+                clean_zinfo = self._clean_member(zinfo)
+                self._add_file_to_archive(zout, clean_zinfo, full_path)
 
         shutil.rmtree(temp_folder)
         if abort:
             os.remove(self.output_filename)
             return False
+        if not self._final_checks():
+            return False  # pragma: no cover
         return True
+
+
+class TarParser(ArchiveBasedAbstractParser):
+    mimetypes = {'application/x-tar'}
+    def __init__(self, filename):
+        super().__init__(filename)
+        # yes, it's tarfile.open and not tarfile.TarFile,
+        # as stated in the documentation:
+        # https://docs.python.org/3/library/tarfile.html#tarfile.TarFile
+        # This is required to support compressed archives.
+        self.archive_class = tarfile.open
+        self.member_class = tarfile.TarInfo
+
+    def is_archive_valid(self):
+        if tarfile.is_tarfile(self.filename) is False:
+            raise ValueError
+        self.__check_tarfile_safety()
+
+    def __check_tarfile_safety(self):
+        """Checks if the tarfile doesn't have any "suspicious" members.
+
+        This is a rewrite of this patch: https://bugs.python.org/file47826/safetarfile-4.diff
+        inspired by this bug from 2014: https://bugs.python.org/issue21109
+        because Python's stdlib doesn't provide a way to "safely" extract
+        things from a tar file.
+        """
+        names = set()
+        with tarfile.open(self.filename) as f:
+            members = f.getmembers()
+        for member in members:
+            name = member.name
+            if os.path.isabs(name):
+                raise ValueError("The archive %s contains a file with an " \
+                        "absolute path: %s" % (self.filename, name))
+            elif os.path.normpath(name).startswith('../') or '/../' in name:
+                raise ValueError("The archive %s contains a file with an " \
+                        "path traversal attack: %s" % (self.filename, name))
+
+            if name in names:
+                raise ValueError("The archive %s contains two times the same " \
+                        "file: %s" % (self.filename, name))
+            else:
+                names.add(name)
+
+            if member.isfile():
+                if member.mode & stat.S_ISUID:
+                    raise ValueError("The archive %s contains a setuid file: %s" % \
+                        (self.filename, name))
+                elif member.mode & stat.S_ISGID:
+                    raise ValueError("The archive %s contains a setgid file: %s" % \
+                            (self.filename, name))
+            elif member.issym():
+                linkname = member.linkname
+                if os.path.normpath(linkname).startswith('..'):
+                    raise ValueError('The archive %s contains a symlink pointing' \
+                            'outside of the archive via a path traversal: %s -> %s' % \
+                            (self.filename, name, linkname))
+                if os.path.isabs(linkname):
+                    raise ValueError('The archive %s contains a symlink pointing' \
+                            'outside of the archive: %s -> %s' % \
+                            (self.filename, name, linkname))
+            elif member.isdev():
+                raise ValueError("The archive %s contains a non-regular " \
+                        "file: %s" % (self.filename, name))
+            elif member.islnk():
+                raise ValueError("The archive %s contains a hardlink: %s" \
+                        % (self.filename, name))
+
+    @staticmethod
+    def _clean_member(member: ArchiveMember) -> ArchiveMember:
+        assert isinstance(member, tarfile.TarInfo)  # please mypy
+        member.mtime = member.uid = member.gid = 0
+        member.uname = member.gname = ''
+        return member
+
+    @staticmethod
+    def _get_member_meta(member: ArchiveMember) -> Dict[str, str]:
+        assert isinstance(member, tarfile.TarInfo)  # please mypy
+        metadata = {}
+        if member.mtime != 0:
+            metadata['mtime'] = str(datetime.datetime.fromtimestamp(member.mtime))
+        if member.uid != 0:
+            metadata['uid'] = str(member.uid)
+        if member.gid != 0:
+            metadata['gid'] = str(member.gid)
+        if member.uname != '':
+            metadata['uname'] = member.uname
+        if member.gname != '':
+            metadata['gname'] = member.gname
+        return metadata
+
+    def _add_file_to_archive(self, archive: ArchiveClass, member: ArchiveMember,
+                             full_path: str):
+        assert isinstance(member, tarfile.TarInfo)  # please mypy
+        assert isinstance(archive, tarfile.TarFile)  # please mypy
+        archive.add(full_path, member.name, filter=TarParser._clean_member)  # type: ignore
+
+    @staticmethod
+    def _get_all_members(archive: ArchiveClass) -> List[ArchiveMember]:
+        assert isinstance(archive, tarfile.TarFile)  # please mypy
+        return archive.getmembers()  # type: ignore
+
+    @staticmethod
+    def _get_member_name(member: ArchiveMember) -> str:
+        assert isinstance(member, tarfile.TarInfo)  # please mypy
+        return member.name
+
+    @staticmethod
+    def _set_member_permissions(member: ArchiveMember, permissions: int) -> ArchiveMember:
+        assert isinstance(member, tarfile.TarInfo)  # please mypy
+        member.mode = permissions
+        return member
+
+
+class TarGzParser(TarParser):
+    compression = ':gz'
+    mimetypes = {'application/x-tar+gz'}
+
+
+class TarBz2Parser(TarParser):
+    compression = ':bz2'
+    mimetypes = {'application/x-tar+bz2'}
+
+
+class TarXzParser(TarParser):
+    compression = ':xz'
+    mimetypes = {'application/x-tar+xz'}
+
+
+class ZipParser(ArchiveBasedAbstractParser):
+    mimetypes = {'application/zip'}
+    def __init__(self, filename):
+        super().__init__(filename)
+        self.archive_class = zipfile.ZipFile
+        self.member_class = zipfile.ZipInfo
+
+    def is_archive_valid(self):
+        try:
+            with zipfile.ZipFile(self.filename):
+                pass
+        except (zipfile.BadZipFile, OSError):
+            raise ValueError
+
+    @staticmethod
+    def _clean_member(member: ArchiveMember) -> ArchiveMember:
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        member.create_system = 3  # Linux
+        member.comment = b''
+        member.date_time = (1980, 1, 1, 0, 0, 0)  # this is as early as a zipfile can be
+        return member
+
+    @staticmethod
+    def _get_member_meta(member: ArchiveMember) -> Dict[str, str]:
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        metadata = {}
+        if member.create_system == 3:  # this is Linux
+            pass
+        elif member.create_system == 2:
+            metadata['create_system'] = 'Windows'
+        else:
+            metadata['create_system'] = 'Weird'
+
+        if member.comment:
+            metadata['comment'] = member.comment  # type: ignore
+
+        if member.date_time != (1980, 1, 1, 0, 0, 0):
+            metadata['date_time'] = str(datetime.datetime(*member.date_time))
+
+        return metadata
+
+    def _add_file_to_archive(self, archive: ArchiveClass, member: ArchiveMember,
+                             full_path: str):
+        assert isinstance(archive, zipfile.ZipFile)  # please mypy
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        with open(full_path, 'rb') as f:
+            archive.writestr(member, f.read(),
+                             compress_type=member.compress_type)
+
+    @staticmethod
+    def _get_all_members(archive: ArchiveClass) -> List[ArchiveMember]:
+        assert isinstance(archive, zipfile.ZipFile)  # please mypy
+        return archive.infolist()  # type: ignore
+
+    @staticmethod
+    def _get_member_name(member: ArchiveMember) -> str:
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        return member.filename
+
+    @staticmethod
+    def _get_member_compression(member: ArchiveMember):
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        return member.compress_type
+
+    @staticmethod
+    def _set_member_compression(member: ArchiveMember, compression) -> ArchiveMember:
+        assert isinstance(member, zipfile.ZipInfo)  # please mypy
+        member.compress_type = compression
+        return member

libmat2/audio.py

@@ -6,28 +6,32 @@ from typing import Dict, Union
 
 import mutagen
 
-from . import abstract, parser_factory
+from . import abstract, parser_factory, video
 
 
 class MutagenParser(abstract.AbstractParser):
     def __init__(self, filename):
         super().__init__(filename)
         try:
-            mutagen.File(self.filename)
+            if mutagen.File(self.filename) is None:
+                raise ValueError
         except mutagen.MutagenError:
             raise ValueError
 
     def get_meta(self) -> Dict[str, Union[str, dict]]:
         f = mutagen.File(self.filename)
         if f.tags:
-            return {k:', '.join(v) for k, v in f.tags.items()}
+            return {k:', '.join(map(str, v)) for k, v in f.tags.items()}
         return {}
 
     def remove_all(self) -> bool:
         shutil.copy(self.filename, self.output_filename)
         f = mutagen.File(self.output_filename)
-        f.delete()
-        f.save()
+        try:
+            f.delete()
+            f.save()
+        except mutagen.MutagenError:
+            raise ValueError
         return True
 
 
@@ -37,7 +41,14 @@ class MP3Parser(MutagenParser):
     def get_meta(self) -> Dict[str, Union[str, dict]]:
         metadata = {}  # type: Dict[str, Union[str, dict]]
         meta = mutagen.File(self.filename).tags
+        if not meta:
+            return metadata
         for key in meta:
+            if isinstance(key, tuple):
+                metadata[key[0]] = key[1]
+                continue
+            if not hasattr(meta[key], 'text'):  # pragma: no cover
+                continue
             metadata[key.rstrip(' \t\r\n\0')] = ', '.join(map(str, meta[key].text))
         return metadata
 
@@ -75,3 +86,25 @@ class FLACParser(MutagenParser):
             meta[name] = p.get_meta() if p else 'harmful data'  # type: ignore
             os.remove(fname)
         return meta
+
+
+class WAVParser(video.AbstractFFmpegParser):
+    mimetypes = {'audio/x-wav', }
+    meta_allowlist = {'AvgBytesPerSec', 'BitsPerSample', 'Directory',
+                      'Duration', 'Encoding', 'ExifToolVersion',
+                      'FileAccessDate', 'FileInodeChangeDate',
+                      'FileModifyDate', 'FileName', 'FilePermissions',
+                      'FileSize', 'FileType', 'FileTypeExtension',
+                      'MIMEType', 'NumChannels', 'SampleRate', 'SourceFile',
+                     }
+
+class AIFFParser(video.AbstractFFmpegParser):
+    mimetypes = {'audio/aiff', 'audio/x-aiff'}
+    meta_allowlist = {'AvgBytesPerSec', 'BitsPerSample', 'Directory',
+                      'Duration', 'Encoding', 'ExifToolVersion',
+                      'FileAccessDate', 'FileInodeChangeDate',
+                      'FileModifyDate', 'FileName', 'FilePermissions',
+                      'FileSize', 'FileType', 'FileTypeExtension',
+                      'MIMEType', 'NumChannels', 'SampleRate', 'SourceFile',
+                      'NumSampleFrames', 'SampleSize',
+                     }

libmat2/bubblewrap.py

@@ -0,0 +1,111 @@
+"""
+Wrapper around a subset of the subprocess module,
+that uses bwrap (bubblewrap) when it is available.
+
+Instead of importing subprocess, other modules should use this as follows:
+
+  from . import subprocess
+"""
+
+import os
+import shutil
+import subprocess
+import tempfile
+from typing import List, Optional
+
+
+__all__ = ['PIPE', 'run', 'CalledProcessError']
+PIPE = subprocess.PIPE
+CalledProcessError = subprocess.CalledProcessError
+
+# pylint: disable=subprocess-run-check
+
+
+def _get_bwrap_path() -> str:
+    which_path = shutil.which('bwrap')
+    if which_path:
+        return which_path
+
+    raise RuntimeError("Unable to find bwrap")  # pragma: no cover
+
+
+def _get_bwrap_args(tempdir: str,
+                    input_filename: str,
+                    output_filename: Optional[str] = None) -> List[str]:
+    ro_bind_args = []
+    cwd = os.getcwd()
+
+    # XXX: use --ro-bind-try once all supported platforms
+    # have a bubblewrap recent enough to support it.
+    ro_bind_dirs = ['/usr', '/lib', '/lib64', '/bin', '/sbin', '/etc/alternatives', cwd]
+    for bind_dir in ro_bind_dirs:
+        if os.path.isdir(bind_dir):  # pragma: no cover
+            ro_bind_args.extend(['--ro-bind', bind_dir, bind_dir])
+
+    ro_bind_files = ['/etc/ld.so.cache']
+    for bind_file in ro_bind_files:
+        if os.path.isfile(bind_file):  # pragma: no cover
+            ro_bind_args.extend(['--ro-bind', bind_file, bind_file])
+
+    args = ro_bind_args + \
+        ['--dev', '/dev',
+         '--proc', '/proc',
+         '--chdir', cwd,
+         '--unshare-user-try',
+         '--unshare-ipc',
+         '--unshare-pid',
+         '--unshare-net',
+         '--unshare-uts',
+         '--unshare-cgroup-try',
+         '--new-session',
+         '--cap-drop', 'all',
+         # XXX: enable --die-with-parent once all supported platforms have
+         # a bubblewrap recent enough to support it.
+         # '--die-with-parent',
+        ]
+
+    if output_filename:
+        # Mount an empty temporary directory where the sandboxed
+        # process will create its output file
+        output_dirname = os.path.dirname(os.path.abspath(output_filename))
+        args.extend(['--bind', tempdir, output_dirname])
+
+    absolute_input_filename = os.path.abspath(input_filename)
+    args.extend(['--ro-bind', absolute_input_filename, absolute_input_filename])
+
+    return args
+
+
+def run(args: List[str],
+        input_filename: str,
+        output_filename: Optional[str] = None,
+        **kwargs) -> subprocess.CompletedProcess:
+    """Wrapper around `subprocess.run`, that uses bwrap (bubblewrap) if it
+    is available.
+
+    Extra supported keyword arguments:
+
+     - `input_filename`, made available read-only in the sandbox
+     - `output_filename`, where the file created by the sandboxed process
+       is copied upon successful completion; an empty temporary directory
+       is made visible as the parent directory of this file in the sandbox.
+       Optional: one valid use case is to invoke an external process
+       to inspect metadata present in a file.
+    """
+    try:
+        bwrap_path = _get_bwrap_path()
+    except RuntimeError:  # pragma: no cover
+        # bubblewrap is not installed ⇒ short-circuit
+        return subprocess.run(args, **kwargs)
+
+    with tempfile.TemporaryDirectory() as tempdir:
+        prefix_args = [bwrap_path] + \
+            _get_bwrap_args(input_filename=input_filename,
+                            output_filename=output_filename,
+                            tempdir=tempdir)
+        completed_process = subprocess.run(prefix_args + args, **kwargs)
+        if output_filename and completed_process.returncode == 0:
+            shutil.copy(os.path.join(tempdir, os.path.basename(output_filename)),
+                        output_filename)
+
+        return completed_process

libmat2/epub.py

@@ -0,0 +1,116 @@
+import logging
+import re
+import uuid
+import zipfile
+import xml.etree.ElementTree as ET  # type: ignore
+from typing import Dict, Any
+
+from . import archive, office
+
+class EPUBParser(archive.ZipParser):
+    mimetypes = {'application/epub+zip', }
+    metadata_namespace = '{http://purl.org/dc/elements/1.1/}'
+
+    def __init__(self, filename):
+        super().__init__(filename)
+        self.files_to_keep = set(map(re.compile, {  # type: ignore
+            'META-INF/container.xml',
+            'mimetype',
+            'OEBPS/content.opf',
+            'content.opf',
+            'hmh.opf',
+            'OPS/.+.xml'
+            }))
+        self.files_to_omit = set(map(re.compile, {  # type: ignore
+            'iTunesMetadata.plist',
+            'META-INF/calibre_bookmarks.txt',
+            'OEBPS/package.opf',
+             }))
+        self.uniqid = uuid.uuid4()
+
+
+    def is_archive_valid(self):
+        super().is_archive_valid()
+        with zipfile.ZipFile(self.filename) as zin:
+            for item in self._get_all_members(zin):
+                member_name = self._get_member_name(item)
+                if member_name.endswith('META-INF/encryption.xml'):
+                    raise ValueError('the file contains encrypted fonts')
+
+    def _specific_get_meta(self, full_path, file_path) -> Dict[str, Any]:
+        if not file_path.endswith('.opf'):
+            return {}
+
+        with open(full_path, encoding='utf-8') as f:
+            try:
+                results = re.findall(r"<((?:meta|dc|cp).+?)[^>]*>(.+)</\1>",
+                                     f.read(), re.I|re.M)
+                return {k:v for (k, v) in results}
+            except (TypeError, UnicodeDecodeError):
+                return {file_path: 'harmful content', }
+
+    def _specific_cleanup(self, full_path: str) -> bool:
+        if full_path.endswith('hmh.opf') or full_path.endswith('content.opf'):
+            return self.__handle_contentopf(full_path)
+        elif full_path.endswith('OEBPS/toc.ncx'):
+            return self.__handle_tocncx(full_path)
+        elif re.search('/OPS/[^/]+.xml$', full_path):
+            return self.__handle_ops_xml(full_path)
+        return True
+
+    def __handle_ops_xml(self, full_path: str) -> bool:
+        try:
+            tree, namespace = office._parse_xml(full_path)
+        except ET.ParseError:  # pragma: nocover
+            logging.error("Unable to parse %s in %s.", full_path, self.filename)
+            return False
+
+        for item in tree.iterfind('.//', namespace):  # pragma: nocover
+            if item.tag.strip().lower().endswith('head'):
+                item.clear()
+                break
+        tree.write(full_path, xml_declaration=True, encoding='utf-8',
+                   short_empty_elements=False)
+        return True
+
+
+    def __handle_tocncx(self, full_path: str) -> bool:
+        try:
+            tree, namespace = office._parse_xml(full_path)
+        except ET.ParseError:  # pragma: nocover
+            logging.error("Unable to parse %s in %s.", full_path, self.filename)
+            return False
+
+        for item in tree.iterfind('.//', namespace):  # pragma: nocover
+            if item.tag.strip().lower().endswith('head'):
+                item.clear()
+                ET.SubElement(item, 'meta', attrib={'name': '', 'content': ''})
+                break
+        tree.write(full_path, xml_declaration=True, encoding='utf-8',
+                   short_empty_elements=False)
+        return True
+
+    def __handle_contentopf(self, full_path: str) -> bool:
+        try:
+            tree, namespace = office._parse_xml(full_path)
+        except ET.ParseError:
+            logging.error("Unable to parse %s in %s.", full_path, self.filename)
+            return False
+
+        for item in tree.iterfind('.//', namespace):  # pragma: nocover
+            if item.tag.strip().lower().endswith('metadata'):
+                item.clear()
+
+                # item with mandatory content
+                uniqid = ET.Element(self.metadata_namespace + 'identifier')
+                uniqid.text = str(self.uniqid)
+                uniqid.set('id', 'id')
+                item.append(uniqid)
+
+                # items without mandatory content
+                for name in ['language', 'title']:
+                    uniqid = ET.Element(self.metadata_namespace + name)
+                    item.append(uniqid)
+                break  # there is only a single <metadata> block
+        tree.write(full_path, xml_declaration=True, encoding='utf-8')
+        return True

libmat2/exiftool.py

@@ -1,10 +1,13 @@
+import functools
 import json
 import logging
 import os
+import shutil
 import subprocess
 from typing import Dict, Union, Set
 
 from . import abstract
+from . import bubblewrap
 
 # Make pyflakes happy
 assert Set
@@ -15,19 +18,29 @@ class ExiftoolParser(abstract.AbstractParser):
     from a import file, hence why several parsers are re-using its `get_meta`
     method.
     """
-    meta_whitelist = set()  # type: Set[str]
+    meta_allowlist = set()  # type: Set[str]
 
     def get_meta(self) -> Dict[str, Union[str, dict]]:
-        out = subprocess.check_output([_get_exiftool_path(), '-json', self.filename])
+        try:
+            if self.sandbox:
+                out = bubblewrap.run([_get_exiftool_path(), '-json',
+                                      self.filename],
+                                     input_filename=self.filename,
+                                     check=True, stdout=subprocess.PIPE).stdout
+            else:
+                out = subprocess.run([_get_exiftool_path(), '-json',
+                                      self.filename],
+                                     check=True, stdout=subprocess.PIPE).stdout
+        except subprocess.CalledProcessError:  # pragma: no cover
+            raise ValueError
         meta = json.loads(out.decode('utf-8'))[0]
-        for key in self.meta_whitelist:
+        for key in self.meta_allowlist:
             meta.pop(key, None)
         return meta
 
     def _lightweight_cleanup(self) -> bool:
         if os.path.exists(self.output_filename):
-            try:
-                # exiftool can't force output to existing files
+            try:  # exiftool can't force output to existing files
                 os.remove(self.output_filename)
             except OSError as e:  # pragma: no cover
                 logging.error("The output file %s is already existing and \
@@ -46,22 +59,25 @@ class ExiftoolParser(abstract.AbstractParser):
                '-o', self.output_filename,
                self.filename]
         try:
-            subprocess.check_call(cmd)
+            if self.sandbox:
+                bubblewrap.run(cmd, check=True,
+                               input_filename=self.filename,
+                               output_filename=self.output_filename)
+            else:
+                subprocess.run(cmd, check=True)
         except subprocess.CalledProcessError as e:  # pragma: no cover
             logging.error("Something went wrong during the processing of %s: %s", self.filename, e)
             return False
         return True
 
+@functools.lru_cache()
 def _get_exiftool_path() -> str:  # pragma: no cover
-    exiftool_path = '/usr/bin/exiftool'
-    if os.path.isfile(exiftool_path):
-        if os.access(exiftool_path, os.X_OK):
-            return exiftool_path
+    which_path = shutil.which('exiftool')
+    if which_path:
+        return which_path
 
-    # ArchLinux
-    exiftool_path = '/usr/bin/vendor_perl/exiftool'
-    if os.path.isfile(exiftool_path):
-        if os.access(exiftool_path, os.X_OK):
-            return exiftool_path
+    # Exiftool on Arch Linux has a weird path
+    if os.access('/usr/bin/vendor_perl/exiftool', os.X_OK):
+        return '/usr/bin/vendor_perl/exiftool'
 
     raise RuntimeError("Unable to find exiftool")

libmat2/images.py

@@ -1,21 +1,67 @@
 import imghdr
 import os
-from typing import Set
+import re
+from typing import Set, Dict, Union, Any
 
 import cairo
 
 import gi
 gi.require_version('GdkPixbuf', '2.0')
-from gi.repository import GdkPixbuf
+gi.require_version('Rsvg', '2.0')
+from gi.repository import GdkPixbuf, GLib, Rsvg
 
-from . import exiftool
+from . import exiftool, abstract
 
 # Make pyflakes happy
 assert Set
+assert Any
+
+class SVGParser(exiftool.ExiftoolParser):
+    mimetypes = {'image/svg+xml', }
+    meta_allowlist = {'Directory', 'ExifToolVersion', 'FileAccessDate',
+                      'FileInodeChangeDate', 'FileModifyDate', 'FileName',
+                      'FilePermissions', 'FileSize', 'FileType',
+                      'FileTypeExtension', 'ImageHeight', 'ImageWidth',
+                      'MIMEType', 'SVGVersion', 'SourceFile', 'ViewBox'
+                      }
+
+    def remove_all(self) -> bool:
+        try:
+            svg = Rsvg.Handle.new_from_file(self.filename)
+        except GLib.GError:
+            raise ValueError
+
+        try:
+            _, _, _, _, has_viewbox, viewbox = svg.get_intrinsic_dimensions()
+            if has_viewbox is False:
+                raise ValueError
+            _, width, height = svg.get_intrinsic_size_in_pixels()
+        except AttributeError:
+            dimensions = svg.get_dimensions()
+            height, width = dimensions.height, dimensions.width
+
+        surface = cairo.SVGSurface(self.output_filename, height, width)
+        context = cairo.Context(surface)
+        try:
+            svg.render_document(context, viewbox)
+        except AttributeError:
+            svg.render_cairo(context)
+
+        surface.finish()
+        return True
+
+    def get_meta(self) -> Dict[str, Union[str, dict]]:
+        meta = super().get_meta()
+
+        # The namespace is mandatory, but only the …/2000/svg is valid.
+        ns = 'http://www.w3.org/2000/svg'
+        if meta.get('Xmlns') == ns:
+            meta.pop('Xmlns')
+        return meta
 
 class PNGParser(exiftool.ExiftoolParser):
     mimetypes = {'image/png', }
-    meta_whitelist = {'SourceFile', 'ExifToolVersion', 'FileName',
+    meta_allowlist = {'SourceFile', 'ExifToolVersion', 'FileName',
                       'Directory', 'FileSize', 'FileModifyDate',
                       'FileAccessDate', 'FileInodeChangeDate',
                       'FilePermissions', 'FileType', 'FileTypeExtension',
@@ -31,7 +77,8 @@ class PNGParser(exiftool.ExiftoolParser):
 
         try:  # better fail here than later
             cairo.ImageSurface.create_from_png(self.filename)
-        except MemoryError:  # pragma: no cover
+        except:  # pragma: no cover
+            # Cairo is returning some weird exceptions :/
             raise ValueError
 
     def remove_all(self) -> bool:
@@ -42,6 +89,21 @@ class PNGParser(exiftool.ExiftoolParser):
         return True
 
 
+class GIFParser(exiftool.ExiftoolParser):
+    mimetypes = {'image/gif'}
+    meta_allowlist = {'AnimationIterations', 'BackgroundColor', 'BitsPerPixel',
+                      'ColorResolutionDepth', 'Directory', 'Duration',
+                      'ExifToolVersion', 'FileAccessDate',
+                      'FileInodeChangeDate', 'FileModifyDate', 'FileName',
+                      'FilePermissions', 'FileSize', 'FileType',
+                      'FileTypeExtension', 'FrameCount', 'GIFVersion',
+                      'HasColorMap', 'ImageHeight', 'ImageSize', 'ImageWidth',
+                      'MIMEType', 'Megapixels', 'SourceFile',}
+
+    def remove_all(self) -> bool:
+        return self._lightweight_cleanup()
+
+
 class GdkPixbufAbstractParser(exiftool.ExiftoolParser):
     """ GdkPixbuf can handle a lot of surfaces, so we're rending images on it,
         this has the side-effect of completely removing metadata.
@@ -50,22 +112,34 @@ class GdkPixbufAbstractParser(exiftool.ExiftoolParser):
 
     def __init__(self, filename):
         super().__init__(filename)
-        if imghdr.what(filename) != self._type:  # better safe than sorry
+        # we can't use imghdr here because of https://bugs.python.org/issue28591
+        try:
+            GdkPixbuf.Pixbuf.new_from_file(self.filename)
+        except GLib.GError:
             raise ValueError
 
     def remove_all(self) -> bool:
+        if self.lightweight_cleaning:
+            return self._lightweight_cleanup()
+
         _, extension = os.path.splitext(self.filename)
         pixbuf = GdkPixbuf.Pixbuf.new_from_file(self.filename)
         if extension.lower() == '.jpg':
             extension = '.jpeg'  # gdk is picky
-        pixbuf.savev(self.output_filename, extension[1:], [], [])
+        elif extension.lower() == '.tif':
+            extension = '.tiff'  # gdk is picky
+        try:
+            pixbuf.savev(self.output_filename, type=extension[1:],
+                         option_keys=[], option_values=[])
+        except GLib.GError:  # pragma: no cover
+            return False
         return True
 
 
 class JPGParser(GdkPixbufAbstractParser):
     _type = 'jpeg'
     mimetypes = {'image/jpeg'}
-    meta_whitelist = {'SourceFile', 'ExifToolVersion', 'FileName',
+    meta_allowlist = {'SourceFile', 'ExifToolVersion', 'FileName',
                       'Directory', 'FileSize', 'FileModifyDate',
                       'FileAccessDate', "FileInodeChangeDate",
                       'FilePermissions', 'FileType', 'FileTypeExtension',
@@ -78,7 +152,7 @@ class JPGParser(GdkPixbufAbstractParser):
 class TiffParser(GdkPixbufAbstractParser):
     _type = 'tiff'
     mimetypes = {'image/tiff'}
-    meta_whitelist = {'Compression', 'ExifByteOrder', 'ExtraSamples',
+    meta_allowlist = {'Compression', 'ExifByteOrder', 'ExtraSamples',
                       'FillOrder', 'PhotometricInterpretation',
                       'PlanarConfiguration', 'RowsPerStrip', 'SamplesPerPixel',
                       'StripByteCounts', 'StripOffsets', 'BitsPerSample',
@@ -87,3 +161,43 @@ class TiffParser(GdkPixbufAbstractParser):
                       'FilePermissions', 'FileSize', 'FileType',
                       'FileTypeExtension', 'ImageHeight', 'ImageSize',
                       'ImageWidth', 'MIMEType', 'Megapixels', 'SourceFile'}
+
+class PPMParser(abstract.AbstractParser):
+    mimetypes = {'image/x-portable-pixmap'}
+
+    def get_meta(self) -> Dict[str, Union[str, dict]]:
+        meta = {}  # type: Dict[str, Union[str, Dict[Any, Any]]]
+        with open(self.filename) as f:
+            for idx, line in enumerate(f):
+                if line.lstrip().startswith('#'):
+                    meta[str(idx)] = line.lstrip().rstrip()
+        return meta
+
+    def remove_all(self) -> bool:
+        with open(self.filename) as fin:
+            with open(self.output_filename, 'w') as fout:
+                for line in fin:
+                    if not line.lstrip().startswith('#'):
+                        line = re.sub(r"\s+", "", line, flags=re.UNICODE)
+                        fout.write(line)
+        return True
+
+class HEICParser(exiftool.ExiftoolParser):
+    mimetypes = {'image/heic'}
+    meta_allowlist = {'SourceFile', 'ExifToolVersion', 'FileName','Directory',
+            'FileSize', 'FileModifyDate', 'FileAccessDate',
+            'FileInodeChangeDate', 'FilePermissions', 'FileType',
+            'FileTypeExtension', 'MIMEType', 'MajorBrand', 'MinorVersion',
+            'CompatibleBrands','HandlerType', 'PrimaryItemReference',
+            'HEVCConfigurationVersion', 'GeneralProfileSpace',
+            'GeneralTierFlag', 'GeneralProfileIDC',
+            'GenProfileCompatibilityFlags', 'ConstraintIndicatorFlags',
+            'GeneralLevelIDC', 'MinSpatialSegmentationIDC',
+            'ParallelismType','ChromaFormat', 'BitDepthLuma', 'BitDepthChroma',
+            'NumTemporalLayers', 'TemporalIDNested', 'ImageWidth',
+            'ImageHeight', 'ImageSpatialExtent', 'ImagePixelDepth',
+            'AverageFrameRate', 'ConstantFrameRate', 'MediaDataSize',
+            'MediaDataOffset','ImageSize', 'Megapixels'}
+
+    def remove_all(self) -> bool:
+        return self._lightweight_cleanup()

libmat2/office.py

@@ -1,12 +1,14 @@
+import random
+import uuid
 import logging
 import os
 import re
 import zipfile
-from typing import Dict, Set, Pattern, Tuple, Union
+from typing import Dict, Set, Pattern, Tuple, Any
 
 import xml.etree.ElementTree as ET  # type: ignore
 
-from .archive import ArchiveBasedAbstractParser
+from .archive import ZipParser
 
 # pylint: disable=line-too-long
 
@@ -43,7 +45,13 @@ def _sort_xml_attributes(full_path: str) -> bool:
     return True
 
 
-class MSOfficeParser(ArchiveBasedAbstractParser):
+class MSOfficeParser(ZipParser):
+    """
+    The methods modifying XML documents are usually doing so in two loops:
+        1. finding the tag/attributes to remove;
+        2. actually editing the document
+    since it's tricky to modify the XML while iterating on it.
+    """
     mimetypes = {
         'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
         'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
@@ -62,36 +70,68 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
 
         # 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
     }
 
 
     def __init__(self, filename):
         super().__init__(filename)
 
+        # MSOffice documents are using various counters for cross-references,
+        # we collect them all, to make sure that they're effectively counters,
+        # and not unique id used for fingerprinting.
+        self.__counters = {
+            'cNvPr': set(),
+            'rid': set(),
+            }
+
         self.files_to_keep = set(map(re.compile, {  # type: ignore
             r'^\[Content_Types\]\.xml$',
             r'^_rels/\.rels$',
-            r'^word/_rels/document\.xml\.rels$',
-            r'^word/_rels/footer[0-9]*\.xml\.rels$',
-            r'^word/_rels/header[0-9]*\.xml\.rels$',
-
+            r'^xl/sharedStrings\.xml$',  # https://docs.microsoft.com/en-us/office/open-xml/working-with-the-shared-string-table
+            r'^xl/calcChain\.xml$',
+            r'^(?:word|ppt|xl)/_rels/document\.xml\.rels$',
+            r'^(?:word|ppt|xl)/_rels/footer[0-9]*\.xml\.rels$',
+            r'^(?:word|ppt|xl)/_rels/header[0-9]*\.xml\.rels$',
+            r'^(?:word|ppt|xl)/styles\.xml$',
+            # TODO: randomize axId ( https://docs.microsoft.com/en-us/openspecs/office_standards/ms-oi29500/089f849f-fcd6-4fa0-a281-35aa6a432a16 )
+            r'^(?:word|ppt|xl)/charts/chart[0-9]*\.xml$',
+            r'^xl/workbook\.xml$',
+            r'^xl/worksheets/sheet[0-9]+\.xml$',
+            r'^ppt/slideLayouts/_rels/slideLayout[0-9]+\.xml\.rels$',
+            r'^ppt/slideLayouts/slideLayout[0-9]+\.xml$',
+            r'^(?:word|ppt|xl)/tableStyles\.xml$',
+            r'^ppt/slides/_rels/slide[0-9]*\.xml\.rels$',
+            r'^ppt/slides/slide[0-9]*\.xml$',
             # https://msdn.microsoft.com/en-us/library/dd908153(v=office.12).aspx
-            r'^word/stylesWithEffects\.xml$',
+            r'^(?:word|ppt|xl)/stylesWithEffects\.xml$',
+            r'^ppt/presentation\.xml$',
+            # TODO: check if p:bgRef can be randomized
+            r'^ppt/slideMasters/slideMaster[0-9]+\.xml',
+            r'^ppt/slideMasters/_rels/slideMaster[0-9]+\.xml\.rels',
+            r'^xl/worksheets/_rels/sheet[0-9]+\.xml\.rels',
+            r'^xl/drawings/vmlDrawing[0-9]+\.vml',
+            r'^xl/drawings/drawing[0-9]+\.xml',
         }))
         self.files_to_omit = set(map(re.compile, {  # type: ignore
+            r'^\[trash\]/',
             r'^customXml/',
             r'webSettings\.xml$',
             r'^docProps/custom\.xml$',
-            r'^word/printerSettings/',
-            r'^word/theme',
-            r'^word/people\.xml$',
+            r'^(?:word|ppt|xl)/printerSettings/',
+            r'^(?:word|ppt|xl)/theme',
+            r'^(?:word|ppt|xl)/people\.xml$',
+            r'^(?:word|ppt|xl)/numbering\.xml$',
+            r'^(?:word|ppt|xl)/tags/',
+            # View properties like view mode, last viewed slide etc
+            r'^(?:word|ppt|xl)/viewProps\.xml$',
+            # Additional presentation-wide properties like printing properties,
+            # presentation show properties etc.
+            r'^(?:word|ppt|xl)/presProps\.xml$',
+            r'^(?:word|ppt|xl)/comments[0-9]+\.xml$',
 
-            # we have a whitelist in self.files_to_keep,
+            # we have an allowlist in self.files_to_keep,
             # so we can trash everything else
-            r'^word/_rels/',
+            r'^(?:word|ppt|xl)/_rels/',
         }))
 
         if self.__fill_files_to_keep_via_content_types() is False:
@@ -100,7 +140,7 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
     def __fill_files_to_keep_via_content_types(self) -> bool:
         """ There is a suer-handy `[Content_Types].xml` file
         in MS Office archives, describing what each other file contains.
-        The self.content_types_to_keep member contains a type whitelist,
+        The self.content_types_to_keep member contains a type allowlist,
         so we're using it to fill the self.files_to_keep one.
         """
         with zipfile.ZipFile(self.filename) as zin:
@@ -114,7 +154,7 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         except ET.ParseError:
             return False
         for c in tree:
-            if 'PartName' not in c.attrib or 'ContentType' not in c.attrib:
+            if 'PartName' not in c.attrib or 'ContentType' not in c.attrib:  # pragma: no cover
                 continue
             elif c.attrib['ContentType'] in self.content_types_to_keep:
                 fname = c.attrib['PartName'][1:]  # remove leading `/`
@@ -124,24 +164,22 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
 
     @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`, …
 
-        We're removing rsid tags in two times, because we can't modify
-        the xml while we're iterating on it.
-
         For more details, see
         - 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/
         """
         try:
             tree, namespace = _parse_xml(full_path)
-        except ET.ParseError:
+        except ET.ParseError as e:  # pragma: no cover
+            logging.error("Unable to parse %s: %s", full_path, e)
             return False
 
         # rsid, tags or attributes, are always under the `w` namespace
-        if 'w' not in namespace.keys():
+        if 'w' not in namespace:
             return True
 
         parent_map = {c:p for p in tree.iter() for c in p}
@@ -162,14 +200,41 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         return True
 
     @staticmethod
-    def __remove_revisions(full_path: str) -> bool:
-        """ 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.
+    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
         """
         try:
             tree, namespace = _parse_xml(full_path)
-        except ET.ParseError as e:
+        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:
+            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:
+        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
 
@@ -208,11 +273,12 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         """
         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:
-            return False  # there should be only one namespace for Types
+        if len(namespace.items()) != 1:  # pragma: no cover
+            logging.debug("Got several namespaces for Types: %s", namespace.items())
 
         removed_fnames = set()
         with zipfile.ZipFile(self.filename) as zin:
@@ -220,7 +286,7 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
                 for file_to_omit in self.files_to_omit:
                     if file_to_omit.search(fname):
                         matches = map(lambda r: r.search(fname), self.files_to_keep)
-                        if any(matches):  # the file is whitelisted
+                        if any(matches):  # the file is in the allowlist
                             continue
                         removed_fnames.add(fname)
                         break
@@ -234,23 +300,80 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         tree.write(full_path, xml_declaration=True)
         return True
 
+    def _final_checks(self) -> bool:
+        for k, v in self.__counters.items():
+            if v and len(v) != max(v):
+                # TODO: make this an error and return False
+                # once the ability to correct the counters is implemented
+                logging.warning("%s contains invalid %s: %s", self.filename, k, v)
+                return True
+        return True
+
+    def __collect_counters(self, full_path: str):
+        with open(full_path, encoding='utf-8') as f:
+            content = f.read()
+            # "relationship Id"
+            for i in re.findall(r'(?:\s|r:)[iI][dD]="rId([0-9]+)"(?:\s|/)', content):
+                self.__counters['rid'].add(int(i))
+            # "connector for Non-visual property"
+            for i in re.findall(r'<p:cNvPr id="([0-9]+)"', content):
+                self.__counters['cNvPr'].add(int(i))
+
+
+    @staticmethod
+    def __randomize_creationId(full_path: str) -> bool:
+        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
+
+        if 'p14' not in namespace:
+            return True  # pragma: no cover
+
+        for item in tree.iterfind('.//p14:creationId', namespace):
+            item.set('val', '%s' % random.randint(0, 2**32))
+        tree.write(full_path, xml_declaration=True)
+        return True
+
+    @staticmethod
+    def __randomize_sldMasterId(full_path: str) -> bool:
+        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
+
+        if 'p' not in namespace:
+            return True  # pragma: no cover
+
+        for item in tree.iterfind('.//p:sldMasterId', namespace):
+            item.set('id', '%s' % random.randint(0, 2**32))
+        tree.write(full_path, xml_declaration=True)
+        return True
+
     def _specific_cleanup(self, full_path: str) -> bool:
-        # pylint: disable=too-many-return-statements
+        # pylint: disable=too-many-return-statements,too-many-branches
         if os.stat(full_path).st_size == 0:  # Don't process empty files
             return True
 
         if not full_path.endswith('.xml'):
             return True
 
+        if self.__randomize_creationId(full_path) is False:
+            return False
+
+        self.__collect_counters(full_path)
+
         if full_path.endswith('/[Content_Types].xml'):
             # this file contains references to files that we might
             # remove, and MS Office doesn't like dangling references
-            if self.__remove_content_type_members(full_path) is False:
+            if self.__remove_content_type_members(full_path) is False:  # pragma: no cover
                 return False
         elif full_path.endswith('/word/document.xml'):
             # this file contains the revisions
             if self.__remove_revisions(full_path) is False:
-                return False
+                return False  # pragma: no cover
         elif full_path.endswith('/docProps/app.xml'):
             # This file must be present and valid,
             # so we're removing as much as we can.
@@ -265,10 +388,22 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
                 f.write(b'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>')
                 f.write(b'<cp:coreProperties xmlns:cp="http://schemas.openxmlformats.org/package/2006/metadata/core-properties">')
                 f.write(b'</cp:coreProperties>')
-
+        elif full_path.endswith('/ppt/tableStyles.xml'):  # pragma: no cover
+            # This file must be present and valid,
+            # so we're removing as much as we can.
+            with open(full_path, 'wb') as f:
+                f.write(b'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>')
+                uid = str(uuid.uuid4()).encode('utf-8')
+                f.write(b'<a:tblStyleLst def="{%s}" xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"/>' % uid)
+        elif full_path.endswith('ppt/presentation.xml'):
+            if self.__randomize_sldMasterId(full_path) is False:
+                return False  # pragma: no cover
 
         if self.__remove_rsid(full_path) is False:
-            return False
+            return False  # pragma: no cover
+
+        if self.__remove_nsid(full_path) is False:
+            return False  # pragma: no cover
 
         try:
             _sort_xml_attributes(full_path)
@@ -296,29 +431,24 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
 
         return True
 
-    def get_meta(self) -> Dict[str, Union[str, dict]]:
+    def _specific_get_meta(self, full_path: str, file_path: str) -> Dict[str, Any]:
         """
         Yes, I know that parsing xml with regexp ain't pretty,
         be my guest and fix it if you want.
         """
-        metadata = {}
-        zipin = zipfile.ZipFile(self.filename)
-        for item in zipin.infolist():
-            if item.filename.startswith('docProps/') and item.filename.endswith('.xml'):
-                try:
-                    content = zipin.read(item).decode('utf-8')
-                    results = re.findall(r"<(.+)>(.+)</\1>", content, re.I|re.M)
-                    for (key, value) in results:
-                        metadata[key] = value
-                except (TypeError, UnicodeDecodeError):  # We didn't manage to parse the xml file
-                    metadata[item.filename] = 'harmful content'
-            for key, value in self._get_zipinfo_meta(item).items():
-                metadata[key] = value
-        zipin.close()
-        return metadata
+        if not file_path.startswith('docProps/') or not file_path.endswith('.xml'):
+            return {}
+
+        with open(full_path, encoding='utf-8') as f:
+            try:
+                results = re.findall(r"<(.+)>(.+)</\1>", f.read(), re.I|re.M)
+                return {k:v for (k, v) in results}
+            except (TypeError, UnicodeDecodeError):
+                # We didn't manage to parse the xml file
+                return {file_path: 'harmful content', }
 
 
-class LibreOfficeParser(ArchiveBasedAbstractParser):
+class LibreOfficeParser(ZipParser):
     mimetypes = {
         'application/vnd.oasis.opendocument.text',
         'application/vnd.oasis.opendocument.spreadsheet',
@@ -343,6 +473,7 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
         }))
         self.files_to_omit = set(map(re.compile, {  # type: ignore
             r'^meta\.xml$',
+            r'^layout-cache$',
             r'^Configurations2/',
             r'^Thumbnails/',
         }))
@@ -355,7 +486,7 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
             logging.error("Unable to parse %s: %s", full_path, e)
             return False
 
-        if 'office' not in namespace.keys():  # no revisions in the current file
+        if 'office' not in namespace:  # no revisions in the current file
             return True
 
         for text in tree.getroot().iterfind('.//office:text', namespace):
@@ -381,23 +512,17 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
                 return False
         return True
 
-    def get_meta(self) -> Dict[str, Union[str, dict]]:
+    def _specific_get_meta(self, full_path: str, file_path: str) -> Dict[str, Any]:
         """
         Yes, I know that parsing xml with regexp ain't pretty,
         be my guest and fix it if you want.
         """
-        metadata = {}
-        zipin = zipfile.ZipFile(self.filename)
-        for item in zipin.infolist():
-            if item.filename == 'meta.xml':
-                try:
-                    content = zipin.read(item).decode('utf-8')
-                    results = re.findall(r"<((?:meta|dc|cp).+?)>(.+)</\1>", content, re.I|re.M)
-                    for (key, value) in results:
-                        metadata[key] = value
-                except (TypeError, UnicodeDecodeError):  # We didn't manage to parse the xml file
-                    metadata[item.filename] = 'harmful content'
-            for key, value in self._get_zipinfo_meta(item).items():
-                metadata[key] = value
-        zipin.close()
-        return metadata
+        if file_path != 'meta.xml':
+            return {}
+        with open(full_path, encoding='utf-8') as f:
+            try:
+                results = re.findall(r"<((?:meta|dc|cp).+?)[^>]*>(.+)</\1>", f.read(), re.I|re.M)
+                return {k:v for (k, v) in results}
+            except (TypeError, UnicodeDecodeError):  # We didn't manage to parse the xml file
+                # We didn't manage to parse the xml file
+                return {file_path: 'harmful content', }

libmat2/parser_factory.py

@@ -6,10 +6,16 @@ from typing import TypeVar, List, Tuple, Optional
 
 from . import abstract, UNSUPPORTED_EXTENSIONS
 
-assert Tuple  # make pyflakes happy
-
 T = TypeVar('T', bound='abstract.AbstractParser')
 
+mimetypes.add_type('application/epub+zip', '.epub')
+mimetypes.add_type('application/x-dtbncx+xml', '.ncx')  # EPUB Navigation Control XML File
+
+# This should be removed after we move to python3.10
+# https://github.com/python/cpython/commit/20a5b7e986377bdfd929d7e8c4e3db5847dfdb2d
+mimetypes.add_type('image/heic', '.heic')
+
+
 def __load_all_parsers():
     """ Loads every parser in a dynamic way """
     current_dir = os.path.dirname(__file__)
@@ -24,8 +30,10 @@ def __load_all_parsers():
         name, _ = os.path.splitext(basename)
         importlib.import_module('.' + name, package='libmat2')
 
+
 __load_all_parsers()
 
+
 def _get_parsers() -> List[T]:
     """ Get all our parsers!"""
     def __get_parsers(cls):
@@ -35,17 +43,22 @@ def _get_parsers() -> List[T]:
 
 
 def get_parser(filename: str) -> Tuple[Optional[T], Optional[str]]:
-    """ Return the appropriate parser for a giver filename. """
+    """ Return the appropriate parser for a given filename.
+
+        :raises ValueError: Raised if the instantiation of the parser went wrong.
+    """
     mtype, _ = mimetypes.guess_type(filename)
 
     _, extension = os.path.splitext(filename)
     if extension.lower() in UNSUPPORTED_EXTENSIONS:
         return None, mtype
 
+    if mtype == 'application/x-tar':
+        if extension[1:] in ('bz2', 'gz', 'xz'):
+            mtype = mtype + '+' + extension[1:]
+
     for parser_class in _get_parsers():  # type: ignore
         if mtype in parser_class.mimetypes:
-            try:
-                return parser_class(filename), mtype
-            except ValueError:
-                return None, mtype
+            # This instantiation might raise a ValueError on malformed files
+            return parser_class(filename), mtype
     return None, mtype

libmat2/pdf.py

@@ -19,9 +19,10 @@ from . import abstract
 
 poppler_version = Poppler.get_version()
 if LooseVersion(poppler_version) < LooseVersion('0.46'):  # pragma: no cover
-    raise ValueError("MAT2 needs at least Poppler version 0.46 to work. \
+    raise ValueError("mat2 needs at least Poppler version 0.46 to work. \
 The installed version is %s." % poppler_version)  # pragma: no cover
 
+FIXED_PDF_VERSION = cairo.PDFVersion.VERSION_1_5
 
 class PDFParser(abstract.AbstractParser):
     mimetypes = {'application/pdf', }
@@ -32,7 +33,7 @@ class PDFParser(abstract.AbstractParser):
     def __init__(self, filename):
         super().__init__(filename)
         self.uri = 'file://' + os.path.abspath(self.filename)
-        self.__scale = 2  # how much precision do we want for the render
+        self.__scale = 200 / 72.0  # how much precision do we want for the render
         try:  # Check now that the file is valid, to avoid surprises later
             Poppler.Document.new_from_file(self.uri, None)
         except GLib.GError:  # Invalid PDF
@@ -52,6 +53,7 @@ class PDFParser(abstract.AbstractParser):
 
         tmp_path = tempfile.mkstemp()[1]
         pdf_surface = cairo.PDFSurface(tmp_path, 10, 10)  # resized later anyway
+        pdf_surface.restrict_to_version(FIXED_PDF_VERSION)
         pdf_context = cairo.Context(pdf_surface)  # context draws on the surface
 
         for pagenum in range(pages_count):
@@ -80,15 +82,19 @@ class PDFParser(abstract.AbstractParser):
 
         _, tmp_path = tempfile.mkstemp()
         pdf_surface = cairo.PDFSurface(tmp_path, 32, 32)  # resized later anyway
+        pdf_surface.restrict_to_version(FIXED_PDF_VERSION)
         pdf_context = cairo.Context(pdf_surface)
 
         for pagenum in range(pages_count):
             page = document.get_page(pagenum)
+            if page is None:  # pragma: no cover
+                logging.error("Unable to get PDF pages")
+                return False
             page_width, page_height = page.get_size()
             logging.info("Rendering page %d/%d", pagenum + 1, pages_count)
 
-            width = int(page_width) * self.__scale
-            height = int(page_height) * self.__scale
+            width = int(page_width * self.__scale)
+            height = int(page_height * self.__scale)
             img_surface = cairo.ImageSurface(cairo.FORMAT_ARGB32, width, height)
             img_context = cairo.Context(img_surface)
 
@@ -102,7 +108,11 @@ class PDFParser(abstract.AbstractParser):
             buf.seek(0)
 
             img = cairo.ImageSurface.create_from_png(buf)
-            pdf_surface.set_size(page_width*self.__scale, page_height*self.__scale)
+            if cairo.version_info < (1, 12, 0):
+                pdf_surface.set_size(width, height)
+            else:
+                pdf_surface.set_size(page_width, page_height)
+                pdf_surface.set_device_scale(1 / self.__scale, 1 / self.__scale)
             pdf_context.set_source_surface(img, 0, 0)
             pdf_context.paint()
             pdf_context.show_page()  # draw pdf_context on pdf_surface
@@ -122,6 +132,17 @@ class PDFParser(abstract.AbstractParser):
         document.set_creator('')
         document.set_creation_date(-1)
         document.save('file://' + os.path.abspath(out_file))
+
+        # Cairo adds "/Producer" and "/CreationDate", and Poppler sometimes
+        # fails to remove them, we have to use this terrible regex.
+        # It should(tm) be alright though, because cairo's output format
+        # for metadata is fixed.
+        with open(out_file, 'rb') as f:
+            out = re.sub(rb'<<[\s\n]*/Producer.*?>>', b' << >>', f.read(), 0,
+                         re.DOTALL | re.IGNORECASE)
+        with open(out_file, 'wb') as f:
+            f.write(out)
+
         return True
 
     @staticmethod

libmat2/torrent.py

@@ -3,9 +3,10 @@ from typing import Union, Tuple, Dict
 
 from . import abstract
 
+
 class TorrentParser(abstract.AbstractParser):
     mimetypes = {'application/x-bittorrent', }
-    whitelist = {b'announce', b'announce-list', b'info'}
+    allowlist = {b'announce', b'announce-list', b'info'}
 
     def __init__(self, filename):
         super().__init__(filename)
@@ -17,14 +18,14 @@ class TorrentParser(abstract.AbstractParser):
     def get_meta(self) -> Dict[str, Union[str, dict]]:
         metadata = {}
         for key, value in self.dict_repr.items():
-            if key not in self.whitelist:
+            if key not in self.allowlist:
                 metadata[key.decode('utf-8')] = value
         return metadata
 
     def remove_all(self) -> bool:
         cleaned = dict()
         for key, value in self.dict_repr.items():
-            if key in self.whitelist:
+            if key in self.allowlist:
                 cleaned[key] = value
         with open(self.output_filename, 'wb') as f:
             f.write(_BencodeHandler().bencode(cleaned))
@@ -32,10 +33,10 @@ class TorrentParser(abstract.AbstractParser):
         return True
 
 
-class _BencodeHandler(object):
+class _BencodeHandler:
     """
     Since bencode isn't that hard to parse,
-    MAT2 comes with its own parser, based on the spec
+    mat2 comes with its own parser, based on the spec
     https://wiki.theory.org/index.php/BitTorrentSpecification#Bencoding
     """
     def __init__(self):

libmat2/video.py

@@ -1,13 +1,89 @@
-import os
 import subprocess
+import functools
+import shutil
 import logging
 
+from typing import Dict, Union
+
 from . import exiftool
+from . import bubblewrap
 
 
-class AVIParser(exiftool.ExiftoolParser):
+class AbstractFFmpegParser(exiftool.ExiftoolParser):
+    """ Abstract parser for all FFmpeg-based ones, mainly for video. """
+    # Some fileformats have mandatory metadata fields
+    meta_key_value_allowlist = {}  # type: Dict[str, Union[str, int]]
+
+    def remove_all(self) -> bool:
+        if self.meta_key_value_allowlist:
+            logging.warning('The format of "%s" (%s) has some mandatory '
+                            'metadata fields; mat2 filled them with standard '
+                            'data.', self.filename, ', '.join(self.mimetypes))
+        cmd = [_get_ffmpeg_path(),
+               '-i', self.filename,      # input file
+               '-y',                     # overwrite existing output file
+               '-map', '0',              # copy everything all streams from input to output
+               '-codec', 'copy',         # don't decode anything, just copy (speed!)
+               '-loglevel', 'panic',     # Don't show log
+               '-hide_banner',           # hide the banner
+               '-map_metadata', '-1',    # remove supperficial metadata
+               '-map_chapters', '-1',    # remove chapters
+               '-disposition', '0',      # Remove dispositions (check ffmpeg's manpage)
+               '-fflags', '+bitexact',   # don't add any metadata
+               '-flags:v', '+bitexact',  # don't add any metadata
+               '-flags:a', '+bitexact',  # don't add any metadata
+               self.output_filename]
+        try:
+            if self.sandbox:
+                bubblewrap.run(cmd, check=True,
+                               input_filename=self.filename,
+                               output_filename=self.output_filename)
+            else:
+                subprocess.run(cmd, check=True)
+        except subprocess.CalledProcessError as e:
+            logging.error("Something went wrong during the processing of %s: %s", self.filename, e)
+            return False
+        return True
+
+    def get_meta(self) -> Dict[str, Union[str, dict]]:
+        meta = super().get_meta()
+
+        ret = dict()  # type: Dict[str, Union[str, dict]]
+        for key, value in meta.items():
+            if key in self.meta_key_value_allowlist:
+                if value == self.meta_key_value_allowlist[key]:
+                    continue
+            ret[key] = value
+        return ret
+
+
+class WMVParser(AbstractFFmpegParser):
+    mimetypes = {'video/x-ms-wmv', }
+    meta_allowlist = {'AudioChannels', 'AudioCodecID', 'AudioCodecName',
+                      'ErrorCorrectionType', 'AudioSampleRate', 'DataPackets',
+                      'Directory', 'Duration', 'ExifToolVersion',
+                      'FileAccessDate', 'FileInodeChangeDate', 'FileLength',
+                      'FileModifyDate', 'FileName', 'FilePermissions',
+                      'FileSize', 'FileType', 'FileTypeExtension',
+                      'FrameCount', 'FrameRate', 'ImageHeight', 'ImageSize',
+                      'ImageWidth', 'MIMEType', 'MaxBitrate', 'MaxPacketSize',
+                      'Megapixels', 'MinPacketSize', 'Preroll', 'SendDuration',
+                      'SourceFile', 'StreamNumber', 'VideoCodecName', }
+    meta_key_value_allowlist = {  # some metadata are mandatory :/
+        'AudioCodecDescription': '',
+        'CreationDate': '0000:00:00 00:00:00Z',
+        'FileID': '00000000-0000-0000-0000-000000000000',
+        'Flags': 2,  # FIXME: What is this? Why 2?
+        'ModifyDate': '0000:00:00 00:00:00',
+        'TimeOffset': '0 s',
+        'VideoCodecDescription': '',
+        'StreamType': 'Audio',
+        }
+
+
+class AVIParser(AbstractFFmpegParser):
     mimetypes = {'video/x-msvideo', }
-    meta_whitelist = {'SourceFile', 'ExifToolVersion', 'FileName', 'Directory',
+    meta_allowlist = {'SourceFile', 'ExifToolVersion', 'FileName', 'Directory',
                       'FileSize', 'FileModifyDate', 'FileAccessDate',
                       'FileInodeChangeDate', 'FilePermissions', 'FileType',
                       'FileTypeExtension', 'MIMEType', 'FrameRate', 'MaxDataRate',
@@ -24,31 +100,45 @@ class AVIParser(exiftool.ExiftoolParser):
                       'SampleRate', 'AvgBytesPerSec', 'BitsPerSample',
                       'Duration', 'ImageSize', 'Megapixels'}
 
-    def remove_all(self) -> bool:
-        cmd = [_get_ffmpeg_path(),
-               '-i', self.filename,      # input file
-               '-y',                     # overwrite existing output file
-               '-loglevel', 'panic',     # Don't show log
-               '-hide_banner',           # hide the banner
-               '-codec', 'copy',         # don't decode anything, just copy (speed!)
-               '-map_metadata', '-1',    # remove supperficial metadata
-               '-map_chapters', '-1',    # remove chapters
-               '-fflags', '+bitexact',   # don't add any metadata
-               '-flags:v', '+bitexact',  # don't add any metadata
-               '-flags:a', '+bitexact',  # don't add any metadata
-               self.output_filename]
-        try:
-            subprocess.check_call(cmd)
-        except subprocess.CalledProcessError as e:
-            logging.error("Something went wrong during the processing of %s: %s", self.filename, e)
-            return False
-        return True
+
+class MP4Parser(AbstractFFmpegParser):
+    mimetypes = {'video/mp4', }
+    meta_allowlist = {'AudioFormat', 'AvgBitrate', 'Balance', 'TrackDuration',
+                      'XResolution', 'YResolution', 'ExifToolVersion',
+                      'FileAccessDate', 'FileInodeChangeDate', 'FileModifyDate',
+                      'FileName', 'FilePermissions', 'MIMEType', 'FileType',
+                      'FileTypeExtension', 'Directory', 'ImageWidth',
+                      'ImageSize', 'ImageHeight', 'FileSize', 'SourceFile',
+                      'BitDepth', 'Duration', 'AudioChannels',
+                      'AudioBitsPerSample', 'AudioSampleRate', 'Megapixels',
+                      'MovieDataSize', 'VideoFrameRate', 'MediaTimeScale',
+                      'SourceImageHeight', 'SourceImageWidth',
+                      'MatrixStructure', 'MediaDuration'}
+    meta_key_value_allowlist = {  # some metadata are mandatory :/
+        'CreateDate': '0000:00:00 00:00:00',
+        'CurrentTime': '0 s',
+        'MediaCreateDate': '0000:00:00 00:00:00',
+        'MediaLanguageCode': 'und',
+        'MediaModifyDate': '0000:00:00 00:00:00',
+        'ModifyDate': '0000:00:00 00:00:00',
+        'OpColor': '0 0 0',
+        'PosterTime': '0 s',
+        'PreferredRate': '1',
+        'PreferredVolume': '100.00%',
+        'PreviewDuration': '0 s',
+        'PreviewTime': '0 s',
+        'SelectionDuration': '0 s',
+        'SelectionTime': '0 s',
+        'TrackCreateDate': '0000:00:00 00:00:00',
+        'TrackModifyDate': '0000:00:00 00:00:00',
+        'TrackVolume': '0.00%',
+    }
 
 
+@functools.lru_cache()
 def _get_ffmpeg_path() -> str:  # pragma: no cover
-    ffmpeg_path = '/usr/bin/ffmpeg'
-    if os.path.isfile(ffmpeg_path):
-        if os.access(ffmpeg_path, os.X_OK):
-            return ffmpeg_path
+    which_path = shutil.which('ffmpeg')
+    if which_path:
+        return which_path
 
     raise RuntimeError("Unable to find ffmpeg")

libmat2/web.py

@@ -0,0 +1,194 @@
+from html import parser, escape
+from typing import Dict, Any, List, Tuple, Set, Optional
+import re
+import string
+
+from . import abstract
+
+assert Set
+
+# pylint: disable=too-many-instance-attributes
+
+class CSSParser(abstract.AbstractParser):
+    """There is no such things as metadata in CSS files,
+    only comments of the form `/* … */`, so we're removing the laters."""
+    mimetypes = {'text/css', }
+    flags = re.MULTILINE | re.DOTALL
+
+    def remove_all(self) -> bool:
+        with open(self.filename, encoding='utf-8') as f:
+            try:
+                content = f.read()
+            except UnicodeDecodeError:  # pragma: no cover
+                raise ValueError
+            cleaned = re.sub(r'/\*.*?\*/', '', content, 0, self.flags)
+        with open(self.output_filename, 'w', encoding='utf-8') as f:
+            f.write(cleaned)
+        return True
+
+    def get_meta(self) -> Dict[str, Any]:
+        metadata = {}
+        with open(self.filename, encoding='utf-8') as f:
+            try:
+                content = f.read()
+            except UnicodeDecodeError:  # pragma: no cover
+                raise ValueError
+        cssdoc = re.findall(r'/\*(.*?)\*/', content, self.flags)
+        for match in cssdoc:
+            for line in match.splitlines():
+                try:
+                    k, v = line.split(':')
+                    metadata[k.strip(string.whitespace + '*')] = v.strip()
+                except ValueError:
+                    metadata['harmful data'] = line.strip()
+        return metadata
+
+
+class AbstractHTMLParser(abstract.AbstractParser):
+    tags_blocklist = set()  # type: Set[str]
+    # In some html/xml-based formats some tags are mandatory,
+    # so we're keeping them, but are discarding their content
+    tags_required_blocklist = set()  # type: Set[str]
+
+    def __init__(self, filename):
+        super().__init__(filename)
+        self.__parser = _HTMLParser(self.filename, self.tags_blocklist,
+                                    self.tags_required_blocklist)
+        with open(filename, encoding='utf-8') as f:
+            self.__parser.feed(f.read())
+        self.__parser.close()
+
+    def get_meta(self) -> Dict[str, Any]:
+        return self.__parser.get_meta()
+
+    def remove_all(self) -> bool:
+        return self.__parser.remove_all(self.output_filename)
+
+
+class HTMLParser(AbstractHTMLParser):
+    mimetypes = {'text/html', 'application/xhtml+xml'}
+    tags_blocklist = {'meta', }
+    tags_required_blocklist = {'title', }
+
+
+class DTBNCXParser(AbstractHTMLParser):
+    mimetypes = {'application/x-dtbncx+xml', }
+    tags_required_blocklist = {'title', 'doctitle', 'meta'}
+
+
+class _HTMLParser(parser.HTMLParser):
+    """Python doesn't have a validating html parser in its stdlib, so
+    we're using an internal queue to track all the opening/closing tags,
+    and hoping for the best.
+
+    Moreover, the parser.HTMLParser call doesn't provide a get_endtag_text
+    method, so we have to use get_starttag_text instead, put its result in a
+    LIFO, and transform it in a closing tag when needed.
+
+    Also, gotcha: the `tag` parameters are always in lowercase.
+    """
+    def __init__(self, filename, blocklisted_tags, required_blocklisted_tags):
+        super().__init__()
+        self.filename = filename
+        self.__textrepr = ''
+        self.__meta = {}
+        self.__validation_queue = []  # type: List[str]
+
+        # We're using counters instead of booleans, to handle nested tags
+        self.__in_dangerous_but_required_tag = 0
+        self.__in_dangerous_tag = 0
+
+        if required_blocklisted_tags & blocklisted_tags:  # pragma: nocover
+            raise ValueError("There is an overlap between %s and %s" % (
+                required_blocklisted_tags, blocklisted_tags))
+        self.tag_required_blocklist = required_blocklisted_tags
+        self.tag_blocklist = blocklisted_tags
+
+    # pylint: disable=R0201
+    def error(self, message):  # pragma: no cover
+        """ Amusingly, Python's documentation doesn't mention that this
+        function needs to be implemented in subclasses of the parent class
+        of parser.HTMLParser. This was found by fuzzing,
+        triggering the following exception:
+            NotImplementedError: subclasses of ParserBase must override error()
+        """
+        raise ValueError(message)
+
+    def handle_starttag(self, tag: str, attrs: List[Tuple[str, Optional[str]]]):
+        # Ignore the type, because mypy is too stupid to infer
+        # that get_starttag_text() can't return None.
+        original_tag = self.get_starttag_text()  # type: ignore
+        self.__validation_queue.append(original_tag)  # type: ignore
+
+        if tag in self.tag_blocklist:
+            self.__in_dangerous_tag += 1
+
+        if self.__in_dangerous_tag == 0:
+            if self.__in_dangerous_but_required_tag == 0:
+                self.__textrepr += original_tag
+
+        if tag in self.tag_required_blocklist:
+            self.__in_dangerous_but_required_tag += 1
+
+    def handle_endtag(self, tag: str):
+        if not self.__validation_queue:
+            raise ValueError("The closing tag %s doesn't have a corresponding "
+                             "opening one in %s." % (tag, self.filename))
+
+        previous_tag = self.__validation_queue.pop()
+        previous_tag = previous_tag[1:-1]  # remove < and >
+        previous_tag = previous_tag.split(' ')[0]  # remove attributes
+        if tag != previous_tag.lower():
+            raise ValueError("The closing tag %s doesn't match the previous "
+                             "tag %s in %s" %
+                             (tag, previous_tag, self.filename))
+
+        if tag in self.tag_required_blocklist:
+            self.__in_dangerous_but_required_tag -= 1
+
+        if self.__in_dangerous_tag == 0:
+            if self.__in_dangerous_but_required_tag == 0:
+                # There is no `get_endtag_text()` method :/
+                self.__textrepr += '</' + previous_tag + '>'
+
+        if tag in self.tag_blocklist:
+            self.__in_dangerous_tag -= 1
+
+    def handle_data(self, data: str):
+        if self.__in_dangerous_but_required_tag == 0:
+            if self.__in_dangerous_tag == 0:
+                if data.strip():
+                    self.__textrepr += escape(data)
+
+    def handle_startendtag(self, tag: str,
+                           attrs: List[Tuple[str, Optional[str]]]):
+        if tag in self.tag_required_blocklist | self.tag_blocklist:
+            meta = {k:v for k, v in attrs}
+            name = meta.get('name', 'harmful metadata')
+            content = meta.get('content', 'harmful data')
+            self.__meta[name] = content
+
+            if self.__in_dangerous_tag == 0:
+                if tag in self.tag_required_blocklist:
+                    self.__textrepr += '<' + tag + ' />'
+                return
+
+        if self.__in_dangerous_tag == 0:
+            if self.__in_dangerous_but_required_tag == 0:
+                self.__textrepr += self.get_starttag_text()
+
+    def remove_all(self, output_filename: str) -> bool:
+        if self.__validation_queue:
+            raise ValueError("Some tags (%s) were left unclosed in %s" % (
+                ', '.join(self.__validation_queue),
+                self.filename))
+        with open(output_filename, 'w', encoding='utf-8') as f:
+            f.write(self.__textrepr)
+        return True
+
+    def get_meta(self) -> Dict[str, Any]:
+        if self.__validation_queue:
+            raise ValueError("Some tags (%s) were left unclosed in %s" % (
+                ', '.join(self.__validation_queue),
+                self.filename))
+        return self.__meta

mat2

@@ -1,112 +1,161 @@
 #!/usr/bin/env python3
 
 import os
-from typing import Tuple, Generator, List, Union
+import shutil
+from typing import Tuple, List, Union, Set
 import sys
 import mimetypes
 import argparse
 import logging
+import unicodedata
+import concurrent.futures
 
 try:
     from libmat2 import parser_factory, UNSUPPORTED_EXTENSIONS
     from libmat2 import check_dependencies, UnknownMemberPolicy
-except ValueError as e:
-    print(e)
+except ValueError as ex:
+    print(ex)
     sys.exit(1)
 
-__version__ = '0.5.0'
+__version__ = '0.13.0'
 
 # Make pyflakes happy
+assert Set
 assert Tuple
 assert Union
 
+logging.basicConfig(format='%(levelname)s: %(message)s', level=logging.WARNING)
 
-def __check_file(filename: str, mode: int=os.R_OK) -> bool:
+def __print_without_chars(s: str):
+    """ Remove control characters
+    We might use 'Cc' instead of 'C', but better safe than sorry
+    https://www.unicode.org/reports/tr44/#GC_Values_Table
+    """
+    print(''.join(ch for ch in s if not unicodedata.category(ch).startswith('C')))
+
+def __check_file(filename: str, mode: int = os.R_OK) -> bool:
     if not os.path.exists(filename):
-        print("[-] %s is doesn't exist." % filename)
+        __print_without_chars("[-] %s doesn't exist." % filename)
         return False
     elif not os.path.isfile(filename):
-        print("[-] %s is not a regular file." % filename)
+        __print_without_chars("[-] %s is not a regular file." % filename)
         return False
     elif not os.access(filename, mode):
-        print("[-] %s is not readable and writeable." % filename)
+        mode_str = []  # type: List[str]
+        if mode & os.R_OK:
+            mode_str += 'readable'
+        if mode & os.W_OK:
+            mode_str += 'writeable'
+        __print_without_chars("[-] %s is not %s." % (filename, 'nor '.join(mode_str)))
         return False
     return True
 
 
 def create_arg_parser() -> argparse.ArgumentParser:
     parser = argparse.ArgumentParser(description='Metadata anonymisation toolkit 2')
-    parser.add_argument('files', nargs='*', help='the files to process')
-    parser.add_argument('-v', '--version', action='version',
-                        version='MAT2 %s' % __version__)
-    parser.add_argument('-l', '--list', action='store_true',
-                        help='list all supported fileformats')
-    parser.add_argument('--check-dependencies', action='store_true',
-                        help='check if MAT2 has all the dependencies it needs')
+
     parser.add_argument('-V', '--verbose', action='store_true',
                         help='show more verbose status information')
     parser.add_argument('--unknown-members', metavar='policy', default='abort',
-                        help='how to handle unknown members of archive-style files (policy should' +
-                        ' be one of: %s)' % ', '.join(p.value for p in UnknownMemberPolicy))
+                        help='how to handle unknown members of archive-style '
+                        'files (policy should be one of: %s) [Default: abort]' %
+                        ', '.join(p.value for p in UnknownMemberPolicy))
+    parser.add_argument('--inplace', action='store_true',
+                        help='clean in place, without backup')
+    parser.add_argument('--no-sandbox', dest='sandbox', action='store_true',
+                        default=False, help='Disable bubblewrap\'s sandboxing')
 
+    excl_group = parser.add_mutually_exclusive_group()
+    excl_group.add_argument('files', nargs='*', help='the files to process',
+                            default=[])
+    excl_group.add_argument('-v', '--version', action='version',
+                            version='mat2 %s' % __version__)
+    excl_group.add_argument('-l', '--list', action='store_true', default=False,
+                            help='list all supported fileformats')
+    excl_group.add_argument('--check-dependencies', action='store_true',
+                            default=False,
+                            help='check if mat2 has all the dependencies it '
+                            'needs')
+
+    excl_group = parser.add_mutually_exclusive_group()
+    excl_group.add_argument('-L', '--lightweight', action='store_true',
+                            help='remove SOME metadata')
+    excl_group.add_argument('-s', '--show', action='store_true',
+                            help='list harmful metadata detectable by mat2 '
+                            'without removing them')
 
-    info = parser.add_mutually_exclusive_group()
-    info.add_argument('-s', '--show', action='store_true',
-                      help='list harmful metadata detectable by MAT2 without removing them')
-    info.add_argument('-L', '--lightweight', action='store_true',
-                      help='remove SOME metadata')
     return parser
 
 
-def show_meta(filename: str):
+def show_meta(filename: str, sandbox: bool):
     if not __check_file(filename):
         return
 
-    p, mtype = parser_factory.get_parser(filename)  # type: ignore
-    if p is None:
-        print("[-] %s's format (%s) is not supported" % (filename, mtype))
+    try:
+        p, mtype = parser_factory.get_parser(filename)  # type: ignore
+    except ValueError as e:
+        __print_without_chars("[-] something went wrong when processing %s: %s" % (filename, e))
         return
+    if p is None:
+        __print_without_chars("[-] %s's format (%s) is not supported" % (filename, mtype))
+        return
+    p.sandbox = sandbox
     __print_meta(filename, p.get_meta())
 
 
-def __print_meta(filename: str, metadata: dict, depth: int=1):
+def __print_meta(filename: str, metadata: dict, depth: int = 1):
     padding = " " * depth*2
     if not metadata:
-        print(padding + "No metadata found")
+        __print_without_chars(padding + "No metadata found in %s." % filename)
         return
 
-    print("[%s] Metadata for %s:" % ('+'*depth, filename))
+    __print_without_chars("[%s] Metadata for %s:" % ('+'*depth, filename))
 
     for (k, v) in sorted(metadata.items()):
         if isinstance(v, dict):
             __print_meta(k, v, depth+1)
             continue
+
         try:  # FIXME this is ugly.
-            print(padding + "  %s: %s" % (k, v))
+            __print_without_chars(padding + "  %s: %s" % (k, v))
         except UnicodeEncodeError:
-            print(padding + "  %s: harmful content" % k)
+            __print_without_chars(padding + "  %s: harmful content" % k)
+        except TypeError:
+            pass  # for things that aren't iterable
 
 
-def clean_meta(filename: str, is_lightweight: bool, policy: UnknownMemberPolicy) -> bool:
-    if not __check_file(filename, os.R_OK|os.W_OK):
+def clean_meta(filename: str, is_lightweight: bool, inplace: bool, sandbox: bool,
+               policy: UnknownMemberPolicy) -> bool:
+    mode = (os.R_OK | os.W_OK) if inplace else os.R_OK
+    if not __check_file(filename, mode):
         return False
 
-    p, mtype = parser_factory.get_parser(filename)  # type: ignore
+    try:
+        p, mtype = parser_factory.get_parser(filename)  # type: ignore
+    except ValueError as e:
+        __print_without_chars("[-] something went wrong when cleaning %s: %s" % (filename, e))
+        return False
     if p is None:
-        print("[-] %s's format (%s) is not supported" % (filename, mtype))
+        __print_without_chars("[-] %s's format (%s) is not supported" % (filename, mtype))
         return False
     p.unknown_member_policy = policy
     p.lightweight_cleaning = is_lightweight
+    p.sandbox = sandbox
 
     try:
-        return p.remove_all()
+        logging.debug('Cleaning %s…', filename)
+        ret = p.remove_all()
+        if ret is True:
+            shutil.copymode(filename, p.output_filename)
+            if inplace is True:
+                os.rename(p.output_filename, filename)
+        return ret
     except RuntimeError as e:
-        print("[-] %s can't be cleaned: %s" % (filename, e))
+        __print_without_chars("[-] %s can't be cleaned: %s" % (filename, e))
     return False
 
 
-
-def show_parsers() -> bool:
+def show_parsers():
     print('[+] Supported formats:')
     formats = set()  # Set[str]
     for parser in parser_factory._get_parsers():  # type: ignore
@@ -120,53 +169,66 @@ def show_parsers() -> bool:
                 # mimetype, so there is not point in showing the mimetype at all
                 continue
             formats.add('  - %s (%s)' % (mtype, ', '.join(extensions)))
-    print('\n'.join(sorted(formats)))
-    return True
+    __print_without_chars('\n'.join(sorted(formats)))
 
 
-def __get_files_recursively(files: List[str]) -> Generator[str, None, None]:
+def __get_files_recursively(files: List[str]) -> List[str]:
+    ret = set()  # type: Set[str]
     for f in files:
         if os.path.isdir(f):
             for path, _, _files in os.walk(f):
                 for _f in _files:
                     fname = os.path.join(path, _f)
                     if __check_file(fname):
-                        yield fname
+                        ret.add(fname)
         elif __check_file(f):
-            yield f
+            ret.add(f)
+    return list(ret)
+
 
 def main() -> int:
     arg_parser = create_arg_parser()
     args = arg_parser.parse_args()
 
     if args.verbose:
-        logging.basicConfig(level=logging.INFO)
+        logging.getLogger().setLevel(logging.DEBUG)
 
     if not args.files:
         if args.list:
-            return show_parsers()
+            show_parsers()
+            return 0
         elif args.check_dependencies:
-            print("Dependencies required for MAT2 %s:" % __version__)
+            __print_without_chars("Dependencies for mat2 %s:" % __version__)
             for key, value in sorted(check_dependencies().items()):
-                print('- %s: %s' % (key, 'yes' if value else 'no'))
+                __print_without_chars('- %s: %s %s' % (key, 'yes' if value['found'] else 'no',
+                                       '(optional)' if not value['required'] else ''))
         else:
             arg_parser.print_help()
         return 0
 
     elif args.show:
         for f in __get_files_recursively(args.files):
-            show_meta(f)
+            show_meta(f, args.sandbox)
         return 0
 
     else:
+        inplace = args.inplace
         policy = UnknownMemberPolicy(args.unknown_members)
         if policy == UnknownMemberPolicy.KEEP:
             logging.warning('Keeping unknown member files may leak metadata in the resulting file!')
 
         no_failure = True
-        for f in __get_files_recursively(args.files):
-            if clean_meta(f, args.lightweight, policy) is False:
-                no_failure = False
+        files = __get_files_recursively(args.files)
+        # We have to use Processes instead of Threads, since
+        # we're using tempfile.mkdtemp, which isn't thread-safe.
+        futures = list()
+        with concurrent.futures.ProcessPoolExecutor() as executor:
+            for f in files:
+                future = executor.submit(clean_meta, f, args.lightweight,
+                                         inplace, args.sandbox, policy)
+                futures.append(future)
+        for future in concurrent.futures.as_completed(futures):
+            no_failure &= future.result()
         return 0 if no_failure is True else -1
 
 

nautilus/README.md

@@ -0,0 +1,15 @@
+# mat2's Nautilus extension
+
+# Dependencies
+
+- Nautilus (now known as [Files](https://wiki.gnome.org/action/show/Apps/Files))
+- [nautilus-python](https://gitlab.gnome.org/GNOME/nautilus-python) >= 2.10
+
+# Installation
+
+Simply copy the `mat2.py` file to `~/.local/share/nautilus-python/extensions`,
+and launch Nautilus; you should now have a "Remove metadata" item in the
+right-click menu on supported files.
+
+Please note: This is not needed if using a distribution provided package. It
+only applies if installing from source.

nautilus/mat2.py

@@ -2,7 +2,7 @@
 
 """
 Because writing GUI is non-trivial (cf. https://0xacab.org/jvoisin/mat2/issues/3),
-we decided to write a Nautilus extensions instead
+we decided to write a Nautilus extension instead
 (cf. https://0xacab.org/jvoisin/mat2/issues/2).
 
 The code is a little bit convoluted because Gtk isn't thread-safe,
@@ -14,8 +14,9 @@ thread, so we'll have to resort to using a `queue` to pass "messages" around.
 
 import queue
 import threading
-from typing import Tuple
+from typing import Tuple, Optional, List
 from urllib.parse import unquote
+import gettext
 
 import gi
 gi.require_version('Nautilus', '3.0')
@@ -25,10 +26,10 @@ from gi.repository import Nautilus, GObject, Gtk, Gio, GLib, GdkPixbuf
 
 from libmat2 import parser_factory
 
-# make pyflakes happy
-assert Tuple
+_ = gettext.gettext
 
-def _remove_metadata(fpath):
+
+def _remove_metadata(fpath) -> Tuple[bool, Optional[str]]:
     """ This is a simple wrapper around libmat2, because it's
     easier and cleaner this way.
     """
@@ -37,8 +38,8 @@ def _remove_metadata(fpath):
         return False, mtype
     return parser.remove_all(), mtype
 
-class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationWidgetProvider):
-    """ This class adds an item to the right-clic menu in Nautilus. """
+class Mat2Extension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationWidgetProvider):
+    """ This class adds an item to the right-click menu in Nautilus. """
 
     def __init__(self):
         super().__init__()
@@ -53,17 +54,17 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         self.infobar.set_show_close_button(True)
         self.infobar_hbox = Gtk.Box(orientation=Gtk.Orientation.HORIZONTAL)
 
-        btn = Gtk.Button("Show")
+        btn = Gtk.Button(_("Show"))
         btn.connect("clicked", self.__cb_show_failed)
         self.infobar_hbox.pack_end(btn, False, False, 0)
 
-        infobar_msg = Gtk.Label("Failed to clean some items")
+        infobar_msg = Gtk.Label(_("Failed to clean some items"))
         self.infobar_hbox.pack_start(infobar_msg, False, False, 0)
 
         self.infobar.get_content_area().pack_start(self.infobar_hbox, True, True, 0)
         self.infobar.show_all()
 
-    def get_widget(self, uri, window):
+    def get_widget(self, uri, window) -> Gtk.Widget:
         """ This is the method that we have to implement (because we're
         a LocationWidgetProvider) in order to show our infobar.
         """
@@ -92,9 +93,9 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         window = Gtk.Window()
         headerbar = Gtk.HeaderBar()
         window.set_titlebar(headerbar)
-        headerbar.props.title = "Metadata removal failed"
+        headerbar.props.title = _("Metadata removal failed")
 
-        close_buton = Gtk.Button("Close")
+        close_buton = Gtk.Button(_("Close"))
         close_buton.connect("clicked", lambda _: window.close())
         headerbar.pack_end(close_buton)
 
@@ -109,9 +110,9 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         """ Validate if a given file FileInfo `fileinfo` can be processed.
         Returns a boolean, and a textreason why"""
         if fileinfo.get_uri_scheme() != "file" or fileinfo.is_directory():
-            return False, "Not a file"
+            return False, _("Not a file")
         elif not fileinfo.can_write():
-            return False, "Not writeable"
+            return False, _("Not writeable")
         return True, ""
 
     def __create_treeview(self) -> Gtk.TreeView:
@@ -122,7 +123,7 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         column_pixbuf = Gtk.TreeViewColumn("Icon", renderer_pixbuf, pixbuf=0)
         treeview.append_column(column_pixbuf)
 
-        for idx, name in enumerate(['File', 'Reason']):
+        for idx, name in enumerate([_('File'), _('Reason')]):
             renderer_text = Gtk.CellRendererText()
             column_text = Gtk.TreeViewColumn(name, renderer_text, text=idx+1)
             treeview.append_column(column_text)
@@ -175,14 +176,14 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         if fname is None:
             self.infobar_hbox.destroy()
             self.infobar.hide()
-            if len(self.failed_items):
+            if self.failed_items:
                 self.__infobar_failure()
             if not processing_queue.empty():
                 print("Something went wrong, the queue isn't empty :/")
             return False
 
         progressbar.pulse()
-        progressbar.set_text("Cleaning %s" % fname)
+        progressbar.set_text(_("Cleaning %s") % fname)
         progressbar.show_all()
         self.infobar_hbox.show_all()
         self.infobar.show_all()
@@ -204,7 +205,7 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
             fpath = unquote(fileinfo.get_uri()[7:])  # `len('file://') = 7`
             success, mtype = _remove_metadata(fpath)
             if not success:
-                self.failed_items.append((fname, mtype, 'Unsupported/invalid'))
+                self.failed_items.append((fname, mtype, _('Unsupported/invalid')))
         processing_queue.put(None)  # signal that we processed all the files
         return True
 
@@ -228,18 +229,18 @@ class ColumnExtension(GObject.GObject, Nautilus.MenuProvider, Nautilus.LocationW
         """ https://bugzilla.gnome.org/show_bug.cgi?id=784278 """
         return None
 
-    def get_file_items(self, window, files):
+    def get_file_items(self, window, files) -> Optional[List[Nautilus.MenuItem]]:
         """ This method is the one allowing us to create a menu item.
         """
         # Do not show the menu item if not a single file has a chance to be
         # processed by mat2.
-        if not any([is_valid for (is_valid, _) in map(self.__validate, files)]):
+        if not any((is_valid for (is_valid, _) in map(self.__validate, files))):
             return None
 
         item = Nautilus.MenuItem(
-            name="MAT2::Remove_metadata",
-            label="Remove metadata",
-            tip="Remove metadata"
+            name="mat2::Remove_metadata",
+            label=_("Remove metadata"),
+            tip=_("Remove metadata")
         )
         item.connect('activate', self.__cb_menu_activate, files)
 

setup.py

@@ -1,11 +1,11 @@
 import setuptools
 
-with open("README.md", "r") as fh:
+with open("README.md", encoding='utf-8') as fh:
     long_description = fh.read()
 
 setuptools.setup(
     name="mat2",
-    version='0.5.0',
+    version='0.13.0',
     author="Julien (jvoisin) Voisin",
     author_email="julien.voisin+mat2@dustri.org",
     description="A handy tool to trash your metadata",

tests/data/dirty.css

@@ -0,0 +1,14 @@
+/**
+ * This is my super css framework
+ * version: 1.0
+ * author : jvoisin
+ */
+
+body {
+	color: red;
+	background-color: blue;
+}
+
+.underline {
+	text-decoration: underline; /* underline is cool */	
+}

tests/data/dirty.html

@@ -0,0 +1,14 @@
+<html>
+	<head>
+		<meta content="vim" name="generator"/>
+		<meta content="jvoisin" name="author"/>
+</head>
+<body>
+	<p>
+		<h1>Hello</h1>
+		I am a web page.
+		Please <b>love</b> me.
+		Here, have a pretty picture: <img src='dirty.jpg' alt='a pretty picture'/>
+	</p>
+</body>
+</html>

tests/data/dirty.ppm

@@ -0,0 +1,8 @@
+P3
+# A metadata
+3 2 1
+1 0 1   0 1 0   0 0 1
+# And an other one
+1 1 0   1 0 1   1 0 0
+  # and a final one here
+

tests/data/dirty.svg

@@ -0,0 +1,636 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!-- Created with Inkscape (http://www.inkscape.org/) -->
+
+<svg
+   xmlns:dc="http://purl.org/dc/elements/1.1/"
+   xmlns:cc="http://creativecommons.org/ns#"
+   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
+   xmlns:svg="http://www.w3.org/2000/svg"
+   xmlns="http://www.w3.org/2000/svg"
+   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
+   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
+   inkscape:export-ydpi="384"
+   inkscape:export-xdpi="384"
+   inkscape:export-filename="mat2.png"
+   width="128"
+   height="128"
+   id="svg11300"
+   sodipodi:version="0.32"
+   inkscape:version="0.92.3 (2405546, 2018-03-11)"
+   sodipodi:docname="dirty.svg"
+   inkscape:output_extension="org.inkscape.output.svg.inkscape"
+   version="1.0"
+   style="display:inline;enable-background:new"
+   viewBox="0 0 128 128">
+  <script
+     id="script4600" />
+  <title
+     id="title4162">Adwaita Icon Template</title>
+  <defs
+     id="defs3" />
+  <sodipodi:namedview
+     stroke="#ef2929"
+     fill="#f57900"
+     id="base"
+     pagecolor="#ffffff"
+     bordercolor="#666666"
+     borderopacity="0.25490196"
+     inkscape:pageopacity="0.0"
+     inkscape:pageshadow="2"
+     inkscape:zoom="4.0446508"
+     inkscape:cx="61.536232"
+     inkscape:cy="41.548134"
+     inkscape:current-layer="layer1"
+     showgrid="true"
+     inkscape:grid-bbox="true"
+     inkscape:document-units="px"
+     inkscape:showpageshadow="false"
+     inkscape:window-width="1366"
+     inkscape:window-height="747"
+     inkscape:window-x="0"
+     inkscape:window-y="21"
+     width="400px"
+     height="300px"
+     inkscape:snap-nodes="true"
+     inkscape:snap-bbox="false"
+     objecttolerance="7"
+     gridtolerance="12"
+     guidetolerance="13"
+     inkscape:window-maximized="1"
+     inkscape:pagecheckerboard="false"
+     showguides="true"
+     inkscape:guide-bbox="true"
+     inkscape:locked="false"
+     inkscape:measure-start="0,0"
+     inkscape:measure-end="0,0"
+     inkscape:object-nodes="true"
+     inkscape:bbox-nodes="true"
+     inkscape:snap-global="true"
+     inkscape:object-paths="true"
+     inkscape:snap-intersection-paths="true"
+     inkscape:snap-bbox-edge-midpoints="true"
+     inkscape:snap-bbox-midpoints="true"
+     showborder="false"
+     inkscape:snap-center="true"
+     inkscape:snap-object-midpoints="true"
+     inkscape:snap-midpoints="true"
+     inkscape:snap-smooth-nodes="true">
+    <inkscape:grid
+       type="xygrid"
+       id="grid5883"
+       spacingx="2"
+       spacingy="2"
+       enabled="true"
+       visible="true"
+       empspacing="4"
+       originx="0"
+       originy="0" />
+    <sodipodi:guide
+       position="64,8"
+       orientation="0,1"
+       id="guide1073"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="12,64"
+       orientation="1,0"
+       id="guide1075"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,104"
+       orientation="0,1"
+       id="guide1099"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,128"
+       orientation="0,1"
+       id="guide993"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="104,64"
+       orientation="1,0"
+       id="guide995"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="9.2651362e-08,64"
+       orientation="1,0"
+       id="guide867"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="120,64"
+       orientation="1,0"
+       id="guide869"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,116"
+       orientation="0,1"
+       id="guide871"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <inkscape:grid
+       type="xygrid"
+       id="grid873"
+       spacingx="1"
+       spacingy="1"
+       empspacing="8"
+       color="#000000"
+       opacity="0.49019608"
+       empcolor="#000000"
+       empopacity="0.08627451"
+       dotted="true" />
+    <sodipodi:guide
+       position="24,64"
+       orientation="1,0"
+       id="guide877"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="116,64"
+       orientation="1,0"
+       id="guide879"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,120"
+       orientation="0,1"
+       id="guide881"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,12"
+       orientation="0,1"
+       id="guide883"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="8,64"
+       orientation="1,0"
+       id="guide885"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="128,64"
+       orientation="1,0"
+       id="guide887"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,0"
+       orientation="0,1"
+       id="guide897"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,24"
+       orientation="0,1"
+       id="guide899"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="256,256"
+       orientation="-0.70710678,0.70710678"
+       id="guide950"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,64"
+       orientation="0.70710678,0.70710678"
+       id="guide952"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+  </sodipodi:namedview>
+  <metadata
+     id="metadata4">
+    <rdf:RDF>
+      <cc:Work
+         rdf:about="">
+        <dc:format>image/svg+xml</dc:format>
+        <dc:type
+           rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
+        <dc:creator>
+          <cc:Agent>
+            <dc:title>GNOME Design Team</dc:title>
+          </cc:Agent>
+        </dc:creator>
+        <dc:source>mat2's source code</dc:source>
+        <cc:license
+           rdf:resource="http://creativecommons.org/licenses/by-sa/4.0/" />
+        <dc:title>Adwaita Icon Template</dc:title>
+        <dc:subject>
+          <rdf:Bag>
+            <rdf:li>mat2</rdf:li>
+            <rdf:li>logo</rdf:li>
+            <rdf:li>metadata</rdf:li>
+          </rdf:Bag>
+        </dc:subject>
+        <dc:date>2019 07 13</dc:date>
+        <dc:rights>
+          <cc:Agent>
+            <dc:title>LGPL</dc:title>
+          </cc:Agent>
+        </dc:rights>
+        <dc:publisher>
+          <cc:Agent>
+            <dc:title>jvoisin</dc:title>
+          </cc:Agent>
+        </dc:publisher>
+        <dc:identifier>mat2-testdata-svg</dc:identifier>
+        <dc:relation />
+        <dc:language>English</dc:language>
+        <dc:coverage />
+        <dc:description>This is a test svg image for mat2's testsuite</dc:description>
+        <dc:contributor>
+          <cc:Agent>
+            <dc:title>jvoisin, and Rose for the design</dc:title>
+          </cc:Agent>
+        </dc:contributor>
+      </cc:Work>
+      <cc:License
+         rdf:about="http://creativecommons.org/licenses/by-sa/4.0/">
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#Reproduction" />
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#Distribution" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#Notice" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#Attribution" />
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#ShareAlike" />
+      </cc:License>
+    </rdf:RDF>
+  </metadata>
+  <g
+     id="layer1"
+     inkscape:label="Icon"
+     inkscape:groupmode="layer"
+     style="display:inline"
+     transform="translate(0,-172)">
+    <g
+       inkscape:groupmode="layer"
+       id="layer2"
+       inkscape:label="baseplate"
+       style="display:none">
+      <text
+         xml:space="preserve"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:5.33333349px;line-height:125%;font-family:Cantarell;-inkscape-font-specification:'Cantarell, Normal';text-align:start;writing-mode:lr-tb;text-anchor:start;display:inline;fill:#000000;fill-opacity:1;stroke:none;stroke-width:0.33264872;enable-background:new"
+         x="7.9499588"
+         y="148.65199"
+         id="context"
+         inkscape:label="context"><tspan
+           sodipodi:role="line"
+           id="tspan2716"
+           x="7.9499588"
+           y="148.65199"
+           style="font-size:5.33333349px;stroke-width:0.33264872">apps</tspan></text>
+      <text
+         inkscape:label="icon-name"
+         id="text3021"
+         y="157.23398"
+         x="7.7533054"
+         style="font-style:normal;font-variant:normal;font-weight:bold;font-stretch:normal;font-size:5.33333349px;line-height:125%;font-family:Cantarell;-inkscape-font-specification:'Cantarell, Bold';text-align:start;writing-mode:lr-tb;text-anchor:start;display:inline;fill:#000000;fill-opacity:1;stroke:none;stroke-width:0.33264872;enable-background:new"
+         xml:space="preserve"><tspan
+           y="157.23398"
+           x="7.7533054"
+           id="tspan3023"
+           sodipodi:role="line"
+           style="font-size:5.33333349px;stroke-width:0.33264872">org.gnome.</tspan></text>
+      <g
+         style="display:inline;fill:#000000;enable-background:new"
+         transform="matrix(7.9911709,0,0,8.0036407,-167.7909,-4846.0776)"
+         id="g12027"
+         inkscape:export-xdpi="12"
+         inkscape:export-ydpi="12" />
+      <rect
+         style="display:inline;overflow:visible;visibility:visible;fill:#f0f0f0;fill-opacity:1;fill-rule:nonzero;stroke:none;stroke-width:0.5;marker:none;enable-background:accumulate"
+         id="rect13805"
+         width="128"
+         height="128"
+         x="9.2651362e-08"
+         y="172"
+         inkscape:label="512x512" />
+      <g
+         id="g883"
+         style="fill:none;fill-opacity:0.25098039;stroke:#a579b3;stroke-opacity:1"
+         transform="translate(-24,24)" />
+      <g
+         id="g900"
+         style="fill:none;fill-opacity:0.25098039;stroke:#a579b3;stroke-opacity:1"
+         transform="translate(-24,24)" />
+      <g
+         id="g1168"
+         transform="matrix(0.25,0,0,0.25,6.9488522e-8,225)">
+        <circle
+           cx="256"
+           cy="44"
+           r="240"
+           id="path1142"
+           style="opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-180"
+           x="96"
+           height="448"
+           width="319.99979"
+           id="rect1110"
+           style="opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-164"
+           x="48"
+           height="416"
+           width="416"
+           id="rect1110-8"
+           style="display:inline;opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal;enable-background:new" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-116"
+           x="32"
+           height="320"
+           width="448"
+           id="rect1110-8-9"
+           style="display:inline;opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal;enable-background:new" />
+      </g>
+    </g>
+    <g
+       inkscape:groupmode="layer"
+       id="layer9"
+       inkscape:label="hires"
+       style="display:none" />
+    <g
+       id="g944"
+       transform="matrix(1,0,0,0.93868822,0,14.545966)">
+      <path
+         style="fill:#99c1f1;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.41013032;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1"
+         d="m 50.53899,195.25817 6.396029,-11.43484 1.082405,-0.87215 4.821622,-10.46578 0.885604,-0.38763 2.558412,4.74837 2.755213,9.59364 1.672808,1.35667 3.542417,-0.87215 5.707227,12.59771 12.988859,9.59364 3.050415,3.87621 v 2.71335 l -16.334476,-1.25977 -7.084833,1.45359 -4.428021,-0.38763 -7.084833,0.29072 -11.414452,-0.58143 -3.640817,0.96905 -9.052843,-1.64739 -2.066409,0.0969 -1.476008,-0.48452 1.377607,-1.45358 1.869609,-1.06596 6.002428,-11.04722 1.279206,0.48453 5.412025,-6.49267 z"
+         id="path3455"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#241f31;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 49.358184,215.31759 -3.444016,0.9206 -9.003641,-1.74429 -1.918809,0.24226 -1.623608,-0.58143 1.574407,-1.50204 1.722008,-0.96905 5.953228,-11.09567 1.279205,0.53298 5.510426,-6.54112 0.344401,0.29072 -4.969223,10.27197 2.214011,1.93811 -0.246001,4.45765 z"
+         id="path3459"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#241f31;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 50.438601,195.22912 6.470906,-11.5803 1.113274,-0.6167 4.870575,-10.62099 0.904535,-0.41113 -0.417479,3.3576 0.626218,0.89079 0.834954,15.89722 1.391594,3.70021 -3.687722,5.34476 0.208739,1.37044 -0.347898,5.68737 1.87865,3.28908 7.375442,2.19272 1.252433,2.19272 -0.487057,0.13704 -4.244358,-0.54818 -6.540486,0.41114 -2.435287,-2.19272 -0.626216,-4.24839 -2.087389,-6.16703 -4.035619,-3.42612 -2.087388,-4.38544"
+         id="path3461"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 32.069579,218.11563 c -0.06958,-0.27409 0.695796,-1.23341 0.695796,-1.23341 l 2.783185,-0.0685 1.739491,2.26124 4.661836,5.13919 0.139158,1.57602 -4.174778,5.96145 -0.487057,6.16703 -2.922344,2.26124 -0.06958,1.57601 h -1.113274 l -1.322013,-3.08351 2.017809,-14.86938 z"
+         id="path3400"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 48.83827,222.43255 1.600331,-3.01499 -0.695796,-0.75375 -5.635951,-1.16488 -3.200663,0.82227 -0.06958,1.50749 1.53075,0.75375 1.461174,2.67237 -0.208739,1.71307 1.739489,1.02783 2.296129,-0.54818 z"
+         id="path3402"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 51.203977,217.70449 1.113274,-0.68522 2.365707,1.02784 1.322013,2.67237 -2.226548,2.26125 -1.322013,-0.82227 -1.322013,-0.61671 0.834956,-1.71306 z"
+         id="path3404"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 43.758957,226.61242 1.948228,0.68522 0.417479,1.91863 -0.626216,1.30193 -1.182854,0.34261 -1.113275,1.02784 -0.765376,3.63169 0.626218,3.01499 -1.252435,0.68522 -0.487057,-0.41113 -0.278319,-1.5075 -1.80907,-1.37045 -0.765376,-3.49464 3.618141,-3.42613 1.669912,-2.67237"
+         id="path3406"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 50.57776,223.25481 0.13916,0.68523 -2.783187,3.83726 0.06958,1.64454 -0.626218,1.50749 -1.60033,1.43897 -0.06958,0.75375 1.600333,1.91863 1.182854,3.08351 0.974114,0.68523 1.669911,-2.80942 -0.278318,-3.22056 3.966039,-3.3576 0.695796,-1.09636 -3.270243,-4.45396 z"
+         id="path3408"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 51.839954,236.39856 -0.834826,1.58948 0.166966,1.26061 1.057445,1.97315 0.500896,-0.32886 0.389584,-1.7539 1.447031,-1.151 2.337512,-4.0559 -0.22262,-1.04138 -1.947927,-1.69909 -2.114892,1.31542 0.278276,3.39819 z"
+         id="path3410"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 57.593778,229.84236 -1.043694,1.09636 0.765375,0.89079 1.043695,-0.20556 v -1.43898 z"
+         id="path3412"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 59.344793,218.25267 -0.765376,2.19272 -0.695796,0.27409 -0.695796,1.91863 -2.226548,2.26124 2.574446,3.56317 h 1.182854 l 0.487057,0.75375 0.626217,1.09636 1.948229,1.30193 2.922346,-0.6167 1.53075,-2.26125 -1.043694,-3.3576 -1.043693,-1.64454 1.322011,-2.60385 -0.904535,-1.37045 -2.226548,0.0685 z"
+         id="path3416"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.150522,238.17554 -0.518261,1.78635 1.036524,2.16915 1.684349,-2.04155 -0.647826,-2.16915 z"
+         id="path3418"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 66.789813,223.66595 1.600333,-0.75375 1.739489,-4.11135 2.922346,0.75375 1.322013,0.41114 0.139159,6.7152 -1.461172,1.02784 -2.226548,4.17987 -0.834956,-0.41114 -0.626216,0.95932 -2.574448,-0.61671 0.904537,-3.08351 z"
+         id="path3422"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 77.505077,218.59529 1.182854,-0.20557 2.435287,1.30193 -0.974115,1.02783 -2.087389,3.63169 -1.391593,0.0685 -1.113274,-0.61671 1.043695,-2.19271 z"
+         id="path3426"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 73.539038,231.06638 1.043695,-1.30193 1.043694,-2.80942 4.522676,1.71306 -0.974115,2.87795 -1.94823,-0.41114 -1.80907,1.09636 z"
+         id="path3428"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 78.200873,225.6531 7.932079,-7.94861 3.339822,1.09636 0.974115,0.13705 1.600331,-1.02784 3.339822,0.0685 -5.079314,12.81371 -3.200663,-1.98715 0.139161,-1.16489 -0.695798,-0.6167 -0.208737,-1.16488 -1.043696,0.27409 -3.200663,2.39829 z"
+         id="path3430"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 81.401536,230.99786 c 0,-0.2741 2.156968,-1.98716 2.156968,-1.98716 l 2.017811,1.30193 -0.904535,2.32976 -1.182855,0.75375 z"
+         id="path3432"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 81.679855,237.8501 0.765375,-1.91863 0.208739,-1.2334 2.156969,0.20557 2.156968,-2.87795 3.409403,1.02784 -0.904535,2.80942 -0.904535,0.34261 -0.626218,2.80943 1.043694,4.72805 -0.904535,1.09636 -1.80907,-2.19272 -0.626217,-1.37045 z"
+         id="path3434"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 78.131294,238.60385 0.626216,3.08351 -0.626216,3.22056 0.765375,0.95931 -0.626216,5.68737 2.504866,2.32976 1.87865,-0.47965 0.417478,-3.35761 1.669911,-0.0685 3.757301,-1.8501 -0.20874,-1.98716 -2.226548,-0.20556 -1.182854,-3.01499 -3.200662,-2.05568 -1.252434,-2.39828 z"
+         id="path3436"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 84.532619,251.41755 -0.278318,1.43898 -0.695797,0.6167 1.322013,2.67238 2.365709,-0.20557 1.53075,-2.94647 -2.365707,-1.98715 z"
+         id="path3438"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 64.993183,249.51154 -1.14959,2.51583 0.766392,1.69818 2.618509,0.25159 0.702526,1.19502 1.021857,2.39003 -0.574794,2.32714 3.89583,1.88688 0.95799,-1.06923 0.510928,-4.59139 -4.023561,-2.70451 -0.127732,-4.21402 z"
+         id="path3440"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.912822,251.00642 h 1.391592 l 2.574446,0.75375 1.391593,1.98715 1.461172,1.30193 -0.139159,3.42612 -3.409402,1.57602 -0.974115,-1.85011 0.626217,-3.3576 -3.270243,-1.85011 z"
+         id="path3442"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.147446,264.77944 1.80907,-1.98715 3.339822,-1.85011 1.322013,-0.0685 4.661835,-3.63169 1.391594,0.34261 0.556637,4.52248 -3.200664,4.04283 -2.852765,-0.82227 -1.80907,0.54818 -0.765376,1.43897 -2.087389,0.68522 z"
+         id="path3444"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 75.06979,272.93361 0.765376,-1.30192 1.252433,-0.41114 0.904535,-2.87794 1.94823,-0.61671 0.556637,2.60386 -3.339822,6.0985 -1.391593,-0.0685 z"
+         id="path3446"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.451649,268.20556 -1.252433,1.85011 2.504867,1.98715 0.765376,0.82227 1.73949,-2.39829 -2.296127,-2.80942 -1.461173,0.27409 z"
+         id="path3448"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.24531,254.0948 1.461172,1.02784 1.948229,0.54818 0.487058,1.64454 -1.461173,2.67237 -0.06958,1.78159 -1.669911,1.85011 -1.252433,-2.05568 0.487057,-2.80942 -1.391593,-0.34261 -0.904535,-2.80942 z"
+         id="path3450"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 47.585836,246.55246 -0.695796,3.70021 -0.139159,1.37045 1.87865,0.68523 1.391592,0.95931 1.809071,-1.64454 -0.417478,-0.95931 z"
+         id="path3452"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 54.682958,247.78586 -1.043694,1.02784 0.208739,1.98715 1.600331,0.89079 0.626217,-0.47965 0.06958,-2.26125 z"
+         id="path3454"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 48.629531,258.95503 4.800994,-6.16703 3.409402,0.82227 0.556637,1.78159 3.131083,4.79657 -1.669911,5.82441 -3.200663,-1.37045 -0.417478,-3.49464 -2.087388,1.30192 z"
+         id="path3456"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 45.915924,252.71948 -0.487056,1.98715 1.60033,1.57602 1.461174,-0.20557 -0.347899,-2.19272 z"
+         id="path3458"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 67.555189,261.6274 -1.80907,2.80943 -2.435287,8.42826 2.783185,3.76874 1.461172,-0.0685 1.113274,-2.12419 1.043696,-0.20557 0.487057,-1.09636 -1.043694,-4.45396 1.182853,-4.31692 z"
+         id="path3460"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.718577,267.79443 1.600331,-1.23341 2.017809,1.71306 -0.904535,1.85011 z"
+         id="path3462"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.509838,276.49678 2.156968,-4.591 1.391593,-0.27409 0.834955,1.50749 -2.017809,5.13919 z"
+         id="path3464"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.242911,274.02997 1.391592,0.20557 1.043694,3.01499 2.01781,0.68522 1.530751,1.57602 -0.904535,2.87795 -2.365707,2.32976 -0.139159,3.56317 -1.322013,1.98715 -2.504867,-1.85011 -0.278318,-2.67237 -1.530752,-1.78159 -1.113274,-3.08351 3.61814,-4.17987 z"
+         id="path3466"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.893354,276.5653 3.270244,1.16489 0.06958,3.70021 -0.556637,0.68523 0.974115,3.70021 1.252433,1.64454 0.06958,3.08351 -2.017809,1.37045 -2.574447,8.08566 -2.574447,-1.30193 -1.948229,-9.79872 z"
+         id="path3468"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.440258,283.5546 h 0.556637 l 0.417478,0.95931 -0.208739,1.30193 -1.461172,0.13704 z"
+         id="path3472"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 56.700767,279.16916 -1.113274,0.95931 0.834956,2.80943 1.600331,0.20556 0.487058,-2.05567 -0.695796,-1.91863 z"
+         id="path3474"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 53.152207,272.17987 0.139159,5.13918 1.87865,1.23341 0.834955,-0.54818 0.904535,-3.63169 1.530752,-1.57602 -1.669911,-3.97431 -3.548561,3.08352 z"
+         id="path3476"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 45.915924,258.33832 -0.208739,3.83726 -4.731414,3.97431 1.948229,2.80942 8.488716,0.82227 0.417478,1.98715 1.043694,-0.75375 0.487057,-2.19272 1.182854,-1.64454 -0.417478,-1.09635 -1.87865,-2.60386 -3.757299,-1.37045 -1.461174,-3.22056 z"
+         id="path3480"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 40.279975,263.68308 1.669912,0.6167 3.061502,-6.37259 -0.904535,-5.61884 -2.504867,-0.34262 -1.391592,-1.2334 2.156968,-7.606 -2.087388,-4.45396 -3.409402,1.57602 -0.834956,3.42612 -1.87865,0.20557 -0.347898,2.1242 1.530752,1.64454 h 1.322013 l 0.626217,3.90578 2.296127,5.61884 -0.347898,2.19272 z"
+         id="path3482"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 66.531337,247.61066 -0.590018,-0.31657 -0.420783,-1.71262 0.427793,-0.66945 1.306823,-1.13114 2.316342,-1.38746 1.06612,0.23465 -0.01701,2.21105 -2.36166,3.35302 z"
+         id="path4284"
+         inkscape:connector-curvature="0"
+         inkscape:transform-center-x="4.9927099"
+         inkscape:transform-center-y="-9.3161687" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.373733,232.22199 -0.815102,1.03206 4.017286,4.12827 1.571981,0.17201 1.339096,-0.86006 0.931544,0.63071 2.387083,-2.98152 -2.794634,-0.91739 -3.027519,0.22934 z"
+         id="path3601"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 57.407878,237.1102 -1.301737,2.34289 -1.301738,0.61888 -0.17955,1.45878 -4.488748,1.54719 -0.403989,1.50299 0.314213,0.30944 1.032412,0.0884 v 1.41457 l 1.660839,1.50299 2.154598,-1.94504 1.571064,0.35364 2.738136,-1.94504 -1.436399,-2.56392 0.987525,-3.44803 -0.583538,-1.37037 z"
+         id="path3603"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.104217,246.96003 5.843936,-6.55723 0.659867,-2.66044 2.221783,-0.40757 -0.386451,-3.39556 -2.000988,-0.60704 -6.246127,-0.36572 -2.624948,2.5137 1.519708,2.75102 -0.347742,5.51876 z"
+         id="path3605"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.024647,249.63275 5.822153,1.31875 1.047988,-3.89891 -1.280874,-1.43343 0.523995,-6.02038 -3.551515,5.275 0.34933,2.06413 -2.037753,0.80272 -1.164431,0.45869 z"
+         id="path3607"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 59.099222,247.24427 -2.095974,1.72011 -0.05822,1.60543 0.465772,1.72011 1.455539,0.97473 -0.407551,0.97473 2.328861,-0.34402 2.27064,-2.86685 -1.571981,-0.57337 -0.640437,-2.86685 -1.51376,-0.40136 z"
+         id="path3609"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 44.078067,234.34346 0.291107,4.47228 -1.863089,1.43342 2.095976,3.72691 2.037753,0.0573 2.27064,-3.55489 -2.969297,-4.98831 z"
+         id="path3611"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 44.660282,245.46683 -3.318627,4.30027 1.339096,1.26141 2.561747,-0.28668 1.222652,-3.15354 z"
+         id="path3613"
+         inkscape:connector-curvature="0" />
+    </g>
+  </g>
+</svg>

tests/data/weird.svg

@@ -0,0 +1,636 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!-- Created with Inkscape (http://www.inkscape.org/) -->
+
+<svg
+   xmlns:dc="http://purl.org/dc/elements/1.1/"
+   xmlns:cc="http://creativecommons.org/ns#"
+   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
+   xmlns:svg="http://www.w3.org/2000/svg"
+   xmlns="http://www.w3.org/1337/svg"
+   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
+   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
+   inkscape:export-ydpi="384"
+   inkscape:export-xdpi="384"
+   inkscape:export-filename="mat2.png"
+   width="128"
+   height="128"
+   id="svg11300"
+   sodipodi:version="0.32"
+   inkscape:version="0.92.3 (2405546, 2018-03-11)"
+   sodipodi:docname="dirty.svg"
+   inkscape:output_extension="org.inkscape.output.svg.inkscape"
+   version="1.0"
+   style="display:inline;enable-background:new"
+   viewBox="0 0 128 128">
+  <script
+     id="script4600" />
+  <title
+     id="title4162">Adwaita Icon Template</title>
+  <defs
+     id="defs3" />
+  <sodipodi:namedview
+     stroke="#ef2929"
+     fill="#f57900"
+     id="base"
+     pagecolor="#ffffff"
+     bordercolor="#666666"
+     borderopacity="0.25490196"
+     inkscape:pageopacity="0.0"
+     inkscape:pageshadow="2"
+     inkscape:zoom="4.0446508"
+     inkscape:cx="61.536232"
+     inkscape:cy="41.548134"
+     inkscape:current-layer="layer1"
+     showgrid="true"
+     inkscape:grid-bbox="true"
+     inkscape:document-units="px"
+     inkscape:showpageshadow="false"
+     inkscape:window-width="1366"
+     inkscape:window-height="747"
+     inkscape:window-x="0"
+     inkscape:window-y="21"
+     width="400px"
+     height="300px"
+     inkscape:snap-nodes="true"
+     inkscape:snap-bbox="false"
+     objecttolerance="7"
+     gridtolerance="12"
+     guidetolerance="13"
+     inkscape:window-maximized="1"
+     inkscape:pagecheckerboard="false"
+     showguides="true"
+     inkscape:guide-bbox="true"
+     inkscape:locked="false"
+     inkscape:measure-start="0,0"
+     inkscape:measure-end="0,0"
+     inkscape:object-nodes="true"
+     inkscape:bbox-nodes="true"
+     inkscape:snap-global="true"
+     inkscape:object-paths="true"
+     inkscape:snap-intersection-paths="true"
+     inkscape:snap-bbox-edge-midpoints="true"
+     inkscape:snap-bbox-midpoints="true"
+     showborder="false"
+     inkscape:snap-center="true"
+     inkscape:snap-object-midpoints="true"
+     inkscape:snap-midpoints="true"
+     inkscape:snap-smooth-nodes="true">
+    <inkscape:grid
+       type="xygrid"
+       id="grid5883"
+       spacingx="2"
+       spacingy="2"
+       enabled="true"
+       visible="true"
+       empspacing="4"
+       originx="0"
+       originy="0" />
+    <sodipodi:guide
+       position="64,8"
+       orientation="0,1"
+       id="guide1073"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="12,64"
+       orientation="1,0"
+       id="guide1075"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,104"
+       orientation="0,1"
+       id="guide1099"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,128"
+       orientation="0,1"
+       id="guide993"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="104,64"
+       orientation="1,0"
+       id="guide995"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="9.2651362e-08,64"
+       orientation="1,0"
+       id="guide867"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="120,64"
+       orientation="1,0"
+       id="guide869"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,116"
+       orientation="0,1"
+       id="guide871"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <inkscape:grid
+       type="xygrid"
+       id="grid873"
+       spacingx="1"
+       spacingy="1"
+       empspacing="8"
+       color="#000000"
+       opacity="0.49019608"
+       empcolor="#000000"
+       empopacity="0.08627451"
+       dotted="true" />
+    <sodipodi:guide
+       position="24,64"
+       orientation="1,0"
+       id="guide877"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="116,64"
+       orientation="1,0"
+       id="guide879"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,120"
+       orientation="0,1"
+       id="guide881"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,12"
+       orientation="0,1"
+       id="guide883"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="8,64"
+       orientation="1,0"
+       id="guide885"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="128,64"
+       orientation="1,0"
+       id="guide887"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,0"
+       orientation="0,1"
+       id="guide897"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,24"
+       orientation="0,1"
+       id="guide899"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="256,256"
+       orientation="-0.70710678,0.70710678"
+       id="guide950"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+    <sodipodi:guide
+       position="64,64"
+       orientation="0.70710678,0.70710678"
+       id="guide952"
+       inkscape:locked="false"
+       inkscape:label=""
+       inkscape:color="rgb(0,0,255)" />
+  </sodipodi:namedview>
+  <metadata
+     id="metadata4">
+    <rdf:RDF>
+      <cc:Work
+         rdf:about="">
+        <dc:format>image/svg+xml</dc:format>
+        <dc:type
+           rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
+        <dc:creator>
+          <cc:Agent>
+            <dc:title>GNOME Design Team</dc:title>
+          </cc:Agent>
+        </dc:creator>
+        <dc:source>mat2's source code</dc:source>
+        <cc:license
+           rdf:resource="http://creativecommons.org/licenses/by-sa/4.0/" />
+        <dc:title>Adwaita Icon Template</dc:title>
+        <dc:subject>
+          <rdf:Bag>
+            <rdf:li>mat2</rdf:li>
+            <rdf:li>logo</rdf:li>
+            <rdf:li>metadata</rdf:li>
+          </rdf:Bag>
+        </dc:subject>
+        <dc:date>2019 07 13</dc:date>
+        <dc:rights>
+          <cc:Agent>
+            <dc:title>LGPL</dc:title>
+          </cc:Agent>
+        </dc:rights>
+        <dc:publisher>
+          <cc:Agent>
+            <dc:title>jvoisin</dc:title>
+          </cc:Agent>
+        </dc:publisher>
+        <dc:identifier>mat2-testdata-svg</dc:identifier>
+        <dc:relation />
+        <dc:language>English</dc:language>
+        <dc:coverage />
+        <dc:description>This is a test svg image for mat2's testsuite</dc:description>
+        <dc:contributor>
+          <cc:Agent>
+            <dc:title>jvoisin, and Rose for the design</dc:title>
+          </cc:Agent>
+        </dc:contributor>
+      </cc:Work>
+      <cc:License
+         rdf:about="http://creativecommons.org/licenses/by-sa/4.0/">
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#Reproduction" />
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#Distribution" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#Notice" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#Attribution" />
+        <cc:permits
+           rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
+        <cc:requires
+           rdf:resource="http://creativecommons.org/ns#ShareAlike" />
+      </cc:License>
+    </rdf:RDF>
+  </metadata>
+  <g
+     id="layer1"
+     inkscape:label="Icon"
+     inkscape:groupmode="layer"
+     style="display:inline"
+     transform="translate(0,-172)">
+    <g
+       inkscape:groupmode="layer"
+       id="layer2"
+       inkscape:label="baseplate"
+       style="display:none">
+      <text
+         xml:space="preserve"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:5.33333349px;line-height:125%;font-family:Cantarell;-inkscape-font-specification:'Cantarell, Normal';text-align:start;writing-mode:lr-tb;text-anchor:start;display:inline;fill:#000000;fill-opacity:1;stroke:none;stroke-width:0.33264872;enable-background:new"
+         x="7.9499588"
+         y="148.65199"
+         id="context"
+         inkscape:label="context"><tspan
+           sodipodi:role="line"
+           id="tspan2716"
+           x="7.9499588"
+           y="148.65199"
+           style="font-size:5.33333349px;stroke-width:0.33264872">apps</tspan></text>
+      <text
+         inkscape:label="icon-name"
+         id="text3021"
+         y="157.23398"
+         x="7.7533054"
+         style="font-style:normal;font-variant:normal;font-weight:bold;font-stretch:normal;font-size:5.33333349px;line-height:125%;font-family:Cantarell;-inkscape-font-specification:'Cantarell, Bold';text-align:start;writing-mode:lr-tb;text-anchor:start;display:inline;fill:#000000;fill-opacity:1;stroke:none;stroke-width:0.33264872;enable-background:new"
+         xml:space="preserve"><tspan
+           y="157.23398"
+           x="7.7533054"
+           id="tspan3023"
+           sodipodi:role="line"
+           style="font-size:5.33333349px;stroke-width:0.33264872">org.gnome.</tspan></text>
+      <g
+         style="display:inline;fill:#000000;enable-background:new"
+         transform="matrix(7.9911709,0,0,8.0036407,-167.7909,-4846.0776)"
+         id="g12027"
+         inkscape:export-xdpi="12"
+         inkscape:export-ydpi="12" />
+      <rect
+         style="display:inline;overflow:visible;visibility:visible;fill:#f0f0f0;fill-opacity:1;fill-rule:nonzero;stroke:none;stroke-width:0.5;marker:none;enable-background:accumulate"
+         id="rect13805"
+         width="128"
+         height="128"
+         x="9.2651362e-08"
+         y="172"
+         inkscape:label="512x512" />
+      <g
+         id="g883"
+         style="fill:none;fill-opacity:0.25098039;stroke:#a579b3;stroke-opacity:1"
+         transform="translate(-24,24)" />
+      <g
+         id="g900"
+         style="fill:none;fill-opacity:0.25098039;stroke:#a579b3;stroke-opacity:1"
+         transform="translate(-24,24)" />
+      <g
+         id="g1168"
+         transform="matrix(0.25,0,0,0.25,6.9488522e-8,225)">
+        <circle
+           cx="256"
+           cy="44"
+           r="240"
+           id="path1142"
+           style="opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-180"
+           x="96"
+           height="448"
+           width="319.99979"
+           id="rect1110"
+           style="opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-164"
+           x="48"
+           height="416"
+           width="416"
+           id="rect1110-8"
+           style="display:inline;opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal;enable-background:new" />
+        <rect
+           ry="32"
+           rx="32"
+           y="-116"
+           x="32"
+           height="320"
+           width="448"
+           id="rect1110-8-9"
+           style="display:inline;opacity:0.1;fill:#2864b0;fill-opacity:1;stroke:none;stroke-width:1;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1;marker:none;marker-start:none;marker-mid:none;marker-end:none;paint-order:normal;enable-background:new" />
+      </g>
+    </g>
+    <g
+       inkscape:groupmode="layer"
+       id="layer9"
+       inkscape:label="hires"
+       style="display:none" />
+    <g
+       id="g944"
+       transform="matrix(1,0,0,0.93868822,0,14.545966)">
+      <path
+         style="fill:#99c1f1;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.41013032;stroke-linecap:butt;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1"
+         d="m 50.53899,195.25817 6.396029,-11.43484 1.082405,-0.87215 4.821622,-10.46578 0.885604,-0.38763 2.558412,4.74837 2.755213,9.59364 1.672808,1.35667 3.542417,-0.87215 5.707227,12.59771 12.988859,9.59364 3.050415,3.87621 v 2.71335 l -16.334476,-1.25977 -7.084833,1.45359 -4.428021,-0.38763 -7.084833,0.29072 -11.414452,-0.58143 -3.640817,0.96905 -9.052843,-1.64739 -2.066409,0.0969 -1.476008,-0.48452 1.377607,-1.45358 1.869609,-1.06596 6.002428,-11.04722 1.279206,0.48453 5.412025,-6.49267 z"
+         id="path3455"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#241f31;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 49.358184,215.31759 -3.444016,0.9206 -9.003641,-1.74429 -1.918809,0.24226 -1.623608,-0.58143 1.574407,-1.50204 1.722008,-0.96905 5.953228,-11.09567 1.279205,0.53298 5.510426,-6.54112 0.344401,0.29072 -4.969223,10.27197 2.214011,1.93811 -0.246001,4.45765 z"
+         id="path3459"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#241f31;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 50.438601,195.22912 6.470906,-11.5803 1.113274,-0.6167 4.870575,-10.62099 0.904535,-0.41113 -0.417479,3.3576 0.626218,0.89079 0.834954,15.89722 1.391594,3.70021 -3.687722,5.34476 0.208739,1.37044 -0.347898,5.68737 1.87865,3.28908 7.375442,2.19272 1.252433,2.19272 -0.487057,0.13704 -4.244358,-0.54818 -6.540486,0.41114 -2.435287,-2.19272 -0.626216,-4.24839 -2.087389,-6.16703 -4.035619,-3.42612 -2.087388,-4.38544"
+         id="path3461"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 32.069579,218.11563 c -0.06958,-0.27409 0.695796,-1.23341 0.695796,-1.23341 l 2.783185,-0.0685 1.739491,2.26124 4.661836,5.13919 0.139158,1.57602 -4.174778,5.96145 -0.487057,6.16703 -2.922344,2.26124 -0.06958,1.57601 h -1.113274 l -1.322013,-3.08351 2.017809,-14.86938 z"
+         id="path3400"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 48.83827,222.43255 1.600331,-3.01499 -0.695796,-0.75375 -5.635951,-1.16488 -3.200663,0.82227 -0.06958,1.50749 1.53075,0.75375 1.461174,2.67237 -0.208739,1.71307 1.739489,1.02783 2.296129,-0.54818 z"
+         id="path3402"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 51.203977,217.70449 1.113274,-0.68522 2.365707,1.02784 1.322013,2.67237 -2.226548,2.26125 -1.322013,-0.82227 -1.322013,-0.61671 0.834956,-1.71306 z"
+         id="path3404"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 43.758957,226.61242 1.948228,0.68522 0.417479,1.91863 -0.626216,1.30193 -1.182854,0.34261 -1.113275,1.02784 -0.765376,3.63169 0.626218,3.01499 -1.252435,0.68522 -0.487057,-0.41113 -0.278319,-1.5075 -1.80907,-1.37045 -0.765376,-3.49464 3.618141,-3.42613 1.669912,-2.67237"
+         id="path3406"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 50.57776,223.25481 0.13916,0.68523 -2.783187,3.83726 0.06958,1.64454 -0.626218,1.50749 -1.60033,1.43897 -0.06958,0.75375 1.600333,1.91863 1.182854,3.08351 0.974114,0.68523 1.669911,-2.80942 -0.278318,-3.22056 3.966039,-3.3576 0.695796,-1.09636 -3.270243,-4.45396 z"
+         id="path3408"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 51.839954,236.39856 -0.834826,1.58948 0.166966,1.26061 1.057445,1.97315 0.500896,-0.32886 0.389584,-1.7539 1.447031,-1.151 2.337512,-4.0559 -0.22262,-1.04138 -1.947927,-1.69909 -2.114892,1.31542 0.278276,3.39819 z"
+         id="path3410"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 57.593778,229.84236 -1.043694,1.09636 0.765375,0.89079 1.043695,-0.20556 v -1.43898 z"
+         id="path3412"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 59.344793,218.25267 -0.765376,2.19272 -0.695796,0.27409 -0.695796,1.91863 -2.226548,2.26124 2.574446,3.56317 h 1.182854 l 0.487057,0.75375 0.626217,1.09636 1.948229,1.30193 2.922346,-0.6167 1.53075,-2.26125 -1.043694,-3.3576 -1.043693,-1.64454 1.322011,-2.60385 -0.904535,-1.37045 -2.226548,0.0685 z"
+         id="path3416"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.150522,238.17554 -0.518261,1.78635 1.036524,2.16915 1.684349,-2.04155 -0.647826,-2.16915 z"
+         id="path3418"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 66.789813,223.66595 1.600333,-0.75375 1.739489,-4.11135 2.922346,0.75375 1.322013,0.41114 0.139159,6.7152 -1.461172,1.02784 -2.226548,4.17987 -0.834956,-0.41114 -0.626216,0.95932 -2.574448,-0.61671 0.904537,-3.08351 z"
+         id="path3422"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 77.505077,218.59529 1.182854,-0.20557 2.435287,1.30193 -0.974115,1.02783 -2.087389,3.63169 -1.391593,0.0685 -1.113274,-0.61671 1.043695,-2.19271 z"
+         id="path3426"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 73.539038,231.06638 1.043695,-1.30193 1.043694,-2.80942 4.522676,1.71306 -0.974115,2.87795 -1.94823,-0.41114 -1.80907,1.09636 z"
+         id="path3428"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 78.200873,225.6531 7.932079,-7.94861 3.339822,1.09636 0.974115,0.13705 1.600331,-1.02784 3.339822,0.0685 -5.079314,12.81371 -3.200663,-1.98715 0.139161,-1.16489 -0.695798,-0.6167 -0.208737,-1.16488 -1.043696,0.27409 -3.200663,2.39829 z"
+         id="path3430"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 81.401536,230.99786 c 0,-0.2741 2.156968,-1.98716 2.156968,-1.98716 l 2.017811,1.30193 -0.904535,2.32976 -1.182855,0.75375 z"
+         id="path3432"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 81.679855,237.8501 0.765375,-1.91863 0.208739,-1.2334 2.156969,0.20557 2.156968,-2.87795 3.409403,1.02784 -0.904535,2.80942 -0.904535,0.34261 -0.626218,2.80943 1.043694,4.72805 -0.904535,1.09636 -1.80907,-2.19272 -0.626217,-1.37045 z"
+         id="path3434"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 78.131294,238.60385 0.626216,3.08351 -0.626216,3.22056 0.765375,0.95931 -0.626216,5.68737 2.504866,2.32976 1.87865,-0.47965 0.417478,-3.35761 1.669911,-0.0685 3.757301,-1.8501 -0.20874,-1.98716 -2.226548,-0.20556 -1.182854,-3.01499 -3.200662,-2.05568 -1.252434,-2.39828 z"
+         id="path3436"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 84.532619,251.41755 -0.278318,1.43898 -0.695797,0.6167 1.322013,2.67238 2.365709,-0.20557 1.53075,-2.94647 -2.365707,-1.98715 z"
+         id="path3438"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 64.993183,249.51154 -1.14959,2.51583 0.766392,1.69818 2.618509,0.25159 0.702526,1.19502 1.021857,2.39003 -0.574794,2.32714 3.89583,1.88688 0.95799,-1.06923 0.510928,-4.59139 -4.023561,-2.70451 -0.127732,-4.21402 z"
+         id="path3440"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.912822,251.00642 h 1.391592 l 2.574446,0.75375 1.391593,1.98715 1.461172,1.30193 -0.139159,3.42612 -3.409402,1.57602 -0.974115,-1.85011 0.626217,-3.3576 -3.270243,-1.85011 z"
+         id="path3442"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.147446,264.77944 1.80907,-1.98715 3.339822,-1.85011 1.322013,-0.0685 4.661835,-3.63169 1.391594,0.34261 0.556637,4.52248 -3.200664,4.04283 -2.852765,-0.82227 -1.80907,0.54818 -0.765376,1.43897 -2.087389,0.68522 z"
+         id="path3444"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 75.06979,272.93361 0.765376,-1.30192 1.252433,-0.41114 0.904535,-2.87794 1.94823,-0.61671 0.556637,2.60386 -3.339822,6.0985 -1.391593,-0.0685 z"
+         id="path3446"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.451649,268.20556 -1.252433,1.85011 2.504867,1.98715 0.765376,0.82227 1.73949,-2.39829 -2.296127,-2.80942 -1.461173,0.27409 z"
+         id="path3448"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.24531,254.0948 1.461172,1.02784 1.948229,0.54818 0.487058,1.64454 -1.461173,2.67237 -0.06958,1.78159 -1.669911,1.85011 -1.252433,-2.05568 0.487057,-2.80942 -1.391593,-0.34261 -0.904535,-2.80942 z"
+         id="path3450"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 47.585836,246.55246 -0.695796,3.70021 -0.139159,1.37045 1.87865,0.68523 1.391592,0.95931 1.809071,-1.64454 -0.417478,-0.95931 z"
+         id="path3452"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 54.682958,247.78586 -1.043694,1.02784 0.208739,1.98715 1.600331,0.89079 0.626217,-0.47965 0.06958,-2.26125 z"
+         id="path3454"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 48.629531,258.95503 4.800994,-6.16703 3.409402,0.82227 0.556637,1.78159 3.131083,4.79657 -1.669911,5.82441 -3.200663,-1.37045 -0.417478,-3.49464 -2.087388,1.30192 z"
+         id="path3456"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 45.915924,252.71948 -0.487056,1.98715 1.60033,1.57602 1.461174,-0.20557 -0.347899,-2.19272 z"
+         id="path3458"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 67.555189,261.6274 -1.80907,2.80943 -2.435287,8.42826 2.783185,3.76874 1.461172,-0.0685 1.113274,-2.12419 1.043696,-0.20557 0.487057,-1.09636 -1.043694,-4.45396 1.182853,-4.31692 z"
+         id="path3460"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.718577,267.79443 1.600331,-1.23341 2.017809,1.71306 -0.904535,1.85011 z"
+         id="path3462"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.509838,276.49678 2.156968,-4.591 1.391593,-0.27409 0.834955,1.50749 -2.017809,5.13919 z"
+         id="path3464"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.242911,274.02997 1.391592,0.20557 1.043694,3.01499 2.01781,0.68522 1.530751,1.57602 -0.904535,2.87795 -2.365707,2.32976 -0.139159,3.56317 -1.322013,1.98715 -2.504867,-1.85011 -0.278318,-2.67237 -1.530752,-1.78159 -1.113274,-3.08351 3.61814,-4.17987 z"
+         id="path3466"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.893354,276.5653 3.270244,1.16489 0.06958,3.70021 -0.556637,0.68523 0.974115,3.70021 1.252433,1.64454 0.06958,3.08351 -2.017809,1.37045 -2.574447,8.08566 -2.574447,-1.30193 -1.948229,-9.79872 z"
+         id="path3468"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 58.440258,283.5546 h 0.556637 l 0.417478,0.95931 -0.208739,1.30193 -1.461172,0.13704 z"
+         id="path3472"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 56.700767,279.16916 -1.113274,0.95931 0.834956,2.80943 1.600331,0.20556 0.487058,-2.05567 -0.695796,-1.91863 z"
+         id="path3474"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 53.152207,272.17987 0.139159,5.13918 1.87865,1.23341 0.834955,-0.54818 0.904535,-3.63169 1.530752,-1.57602 -1.669911,-3.97431 -3.548561,3.08352 z"
+         id="path3476"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 45.915924,258.33832 -0.208739,3.83726 -4.731414,3.97431 1.948229,2.80942 8.488716,0.82227 0.417478,1.98715 1.043694,-0.75375 0.487057,-2.19272 1.182854,-1.64454 -0.417478,-1.09635 -1.87865,-2.60386 -3.757299,-1.37045 -1.461174,-3.22056 z"
+         id="path3480"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 40.279975,263.68308 1.669912,0.6167 3.061502,-6.37259 -0.904535,-5.61884 -2.504867,-0.34262 -1.391592,-1.2334 2.156968,-7.606 -2.087388,-4.45396 -3.409402,1.57602 -0.834956,3.42612 -1.87865,0.20557 -0.347898,2.1242 1.530752,1.64454 h 1.322013 l 0.626217,3.90578 2.296127,5.61884 -0.347898,2.19272 z"
+         id="path3482"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 66.531337,247.61066 -0.590018,-0.31657 -0.420783,-1.71262 0.427793,-0.66945 1.306823,-1.13114 2.316342,-1.38746 1.06612,0.23465 -0.01701,2.21105 -2.36166,3.35302 z"
+         id="path4284"
+         inkscape:connector-curvature="0"
+         inkscape:transform-center-x="4.9927099"
+         inkscape:transform-center-y="-9.3161687" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 72.373733,232.22199 -0.815102,1.03206 4.017286,4.12827 1.571981,0.17201 1.339096,-0.86006 0.931544,0.63071 2.387083,-2.98152 -2.794634,-0.91739 -3.027519,0.22934 z"
+         id="path3601"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 57.407878,237.1102 -1.301737,2.34289 -1.301738,0.61888 -0.17955,1.45878 -4.488748,1.54719 -0.403989,1.50299 0.314213,0.30944 1.032412,0.0884 v 1.41457 l 1.660839,1.50299 2.154598,-1.94504 1.571064,0.35364 2.738136,-1.94504 -1.436399,-2.56392 0.987525,-3.44803 -0.583538,-1.37037 z"
+         id="path3603"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 62.104217,246.96003 5.843936,-6.55723 0.659867,-2.66044 2.221783,-0.40757 -0.386451,-3.39556 -2.000988,-0.60704 -6.246127,-0.36572 -2.624948,2.5137 1.519708,2.75102 -0.347742,5.51876 z"
+         id="path3605"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 71.024647,249.63275 5.822153,1.31875 1.047988,-3.89891 -1.280874,-1.43343 0.523995,-6.02038 -3.551515,5.275 0.34933,2.06413 -2.037753,0.80272 -1.164431,0.45869 z"
+         id="path3607"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 59.099222,247.24427 -2.095974,1.72011 -0.05822,1.60543 0.465772,1.72011 1.455539,0.97473 -0.407551,0.97473 2.328861,-0.34402 2.27064,-2.86685 -1.571981,-0.57337 -0.640437,-2.86685 -1.51376,-0.40136 z"
+         id="path3609"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 44.078067,234.34346 0.291107,4.47228 -1.863089,1.43342 2.095976,3.72691 2.037753,0.0573 2.27064,-3.55489 -2.969297,-4.98831 z"
+         id="path3611"
+         inkscape:connector-curvature="0" />
+      <path
+         style="fill:#1a5fb4;fill-opacity:1;fill-rule:evenodd;stroke:none;stroke-width:0.13671011px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+         d="m 44.660282,245.46683 -3.318627,4.30027 1.339096,1.26141 2.561747,-0.28668 1.222652,-3.15354 z"
+         id="path3613"
+         inkscape:connector-curvature="0" />
+    </g>
+  </g>
+</svg>

tests/fuzz.py

@@ -0,0 +1,54 @@
+import mimetypes
+import os
+import sys
+
+sys.path.append('..')
+
+import atheris
+
+with atheris.instrument_imports(enable_loader_override=False):
+    from libmat2 import parser_factory, UNSUPPORTED_EXTENSIONS
+
+extensions = set()
+for parser in parser_factory._get_parsers():  # type: ignore
+    for mtype in parser.mimetypes:
+        if mtype.startswith('video'):
+            continue
+        if 'aif' in mtype:
+            continue
+        if 'wav' in mtype:
+            continue
+        if 'gif' in mtype:
+            continue
+        if 'aifc' in mtype:
+            continue
+        for extension in mimetypes.guess_all_extensions(mtype):
+            if extension not in UNSUPPORTED_EXTENSIONS:
+                extensions.add(extension)
+extensions = list(extensions)
+
+
+
+def TestOneInput(data):
+    fdp = atheris.FuzzedDataProvider(data)
+    extension = fdp.PickValueInList(extensions)
+    data = fdp.ConsumeBytes(sys.maxsize)
+
+    fname = '/tmp/mat2_fuzz' + extension
+
+    with open(fname, 'wb') as f:
+        f.write(data)
+    try:
+        p, _ = parser_factory.get_parser(fname)
+        if p:
+            p.sandbox = False
+            p.get_meta()
+            p.remove_all()
+            p, _ = parser_factory.get_parser(fname)
+            p.get_meta()
+    except ValueError:
+        pass
+    os.remove(fname)
+
+atheris.Setup(sys.argv, TestOneInput)
+atheris.Fuzz()

tests/test_climat2.py

@@ -1,7 +1,12 @@
+import random
 import os
 import shutil
+import stat
 import subprocess
 import unittest
+import glob
+
+from libmat2 import images, parser_factory
 
 
 mat2_binary = ['./mat2']
@@ -16,29 +21,39 @@ class TestHelp(unittest.TestCase):
     def test_help(self):
         proc = subprocess.Popen(mat2_binary + ['--help'], stdout=subprocess.PIPE)
         stdout, _ = proc.communicate()
-        self.assertIn(b'usage: mat2 [-h] [-v] [-l] [--check-dependencies] [-V]',
-                      stdout)
-        self.assertIn(b'[--unknown-members policy] [-s | -L]', stdout)
+        self.assertIn(b'mat2 [-h] [-V]', stdout)
+        self.assertIn(b'[--unknown-members policy]', stdout)
+        self.assertIn(b'[--inplace]', stdout)
+        self.assertIn(b'[--no-sandbox]', stdout)
+        self.assertIn(b' [-v] [-l]', stdout)
+        self.assertIn(b'[--check-dependencies]', stdout)
+        self.assertIn(b'[-L | -s]', stdout)
+        self.assertIn(b'[files ...]', stdout)
 
     def test_no_arg(self):
         proc = subprocess.Popen(mat2_binary, stdout=subprocess.PIPE)
         stdout, _ = proc.communicate()
-        self.assertIn(b'usage: mat2 [-h] [-v] [-l] [--check-dependencies] [-V]',
-                      stdout)
-        self.assertIn(b'[--unknown-members policy] [-s | -L]', stdout)
+        self.assertIn(b'mat2 [-h] [-V]', stdout)
+        self.assertIn(b'[--unknown-members policy]', stdout)
+        self.assertIn(b'[--inplace]', stdout)
+        self.assertIn(b'[--no-sandbox]', stdout)
+        self.assertIn(b' [-v] [-l] [--check-dependencies] [-L | -s]', stdout)
+        self.assertIn(b'[files ...]', stdout)
 
 
 class TestVersion(unittest.TestCase):
     def test_version(self):
         proc = subprocess.Popen(mat2_binary + ['--version'], stdout=subprocess.PIPE)
         stdout, _ = proc.communicate()
-        self.assertTrue(stdout.startswith(b'MAT2 '))
+        self.assertTrue(stdout.startswith(b'mat2 '))
+
 
 class TestDependencies(unittest.TestCase):
     def test_dependencies(self):
         proc = subprocess.Popen(mat2_binary + ['--check-dependencies'], stdout=subprocess.PIPE)
         stdout, _ = proc.communicate()
-        self.assertTrue(b'MAT2' in stdout)
+        self.assertTrue(b'mat2' in stdout)
+
 
 class TestReturnValue(unittest.TestCase):
     def test_nonzero(self):
@@ -106,6 +121,52 @@ class TestCleanMeta(unittest.TestCase):
 
         os.remove('./tests/data/clean.jpg')
 
+    def test_jpg_nosandbox(self):
+        shutil.copy('./tests/data/dirty.jpg', './tests/data/clean.jpg')
+
+        proc = subprocess.Popen(mat2_binary + ['--show', '--no-sandbox', './tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertIn(b'Comment: Created with GIMP', stdout)
+
+        proc = subprocess.Popen(mat2_binary + ['./tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/clean.cleaned.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertNotIn(b'Comment: Created with GIMP', stdout)
+
+        os.remove('./tests/data/clean.jpg')
+        os.remove('./tests/data/clean.cleaned.jpg')
+
+
+class TestCopyPermissions(unittest.TestCase):
+    def test_jpg_777(self):
+        shutil.copy('./tests/data/dirty.jpg', './tests/data/clean.jpg')
+        os.chmod('./tests/data/clean.jpg', 0o777)
+
+        proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertIn(b'Comment: Created with GIMP', stdout)
+
+        proc = subprocess.Popen(mat2_binary + ['./tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/clean.cleaned.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertNotIn(b'Comment: Created with GIMP', stdout)
+
+        permissions = os.stat('./tests/data/clean.cleaned.jpg')[stat.ST_MODE]
+        self.assertEqual(permissions, 0o100777)
+
+        os.remove('./tests/data/clean.jpg')
+        os.remove('./tests/data/clean.cleaned.jpg')
+
 
 class TestIsSupported(unittest.TestCase):
     def test_pdf(self):
@@ -121,7 +182,7 @@ class TestGetMeta(unittest.TestCase):
         proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/dirty.pdf'],
                 stdout=subprocess.PIPE)
         stdout, _ = proc.communicate()
-        self.assertIn(b'producer: pdfTeX-1.40.14', stdout)
+        self.assertIn(b'Producer: pdfTeX-1.40.14', stdout)
 
     def test_png(self):
         proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/dirty.png'],
@@ -174,3 +235,99 @@ class TestGetMeta(unittest.TestCase):
         self.assertIn(b'genre: Python', stdout)
         self.assertIn(b'i am a : various comment', stdout)
         self.assertIn(b'artist: jvoisin', stdout)
+
+
+class TestControlCharInjection(unittest.TestCase):
+    def test_jpg(self):
+        proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/control_chars.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertIn(b'Comment: GQ\n', stdout)
+
+
+class TestCommandLineParallel(unittest.TestCase):
+    iterations = 24
+
+    def test_same(self):
+        for i in range(self.iterations):
+            shutil.copy('./tests/data/dirty.jpg', './tests/data/dirty_%d.jpg' % i)
+
+        proc = subprocess.Popen(mat2_binary + ['./tests/data/dirty_%d.jpg' % i for i in range(self.iterations)],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        for i in range(self.iterations):
+            path = './tests/data/dirty_%d.jpg' % i
+            p = images.JPGParser('./tests/data/dirty_%d.cleaned.jpg' % i)
+            self.assertEqual(p.get_meta(), {})
+            os.remove('./tests/data/dirty_%d.cleaned.jpg' % i)
+            os.remove(path)
+
+    def test_different(self):
+        src = './tests/data/'
+        dst = './tests/data/parallel'
+        shutil.copytree(src, dst)
+
+        proc = subprocess.Popen(mat2_binary + glob.glob('./tests/data/parallel/dirty.*'),
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        for i in glob.glob('./test/data/parallel/dirty.cleaned.*'):
+            p, mime = parser_factory.get_parser(i)
+            self.assertIsNotNone(mime)
+            self.assertIsNotNone(p)
+            p = parser_factory.get_parser(p.output_filename)
+            self.assertEqual(p.get_meta(), {})
+        shutil.rmtree('./tests/data/parallel/')
+
+    def test_faulty(self):
+        for i in range(self.iterations):
+            shutil.copy('./tests/data/dirty.jpg', './tests/data/dirty_%d.jpg' % i)
+            shutil.copy('./tests/data/dirty.torrent', './tests/data/dirty_%d.docx' % i)
+
+        to_process = ['./tests/data/dirty_%d.jpg' % i for i in range(self.iterations)]
+        to_process.extend(['./tests/data/dirty_%d.docx' % i for i in range(self.iterations)])
+        random.shuffle(to_process)
+        proc = subprocess.Popen(mat2_binary + to_process,
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        for i in range(self.iterations):
+            path = './tests/data/dirty_%d.jpg' % i
+            p = images.JPGParser('./tests/data/dirty_%d.cleaned.jpg' % i)
+            self.assertEqual(p.get_meta(), {})
+            os.remove('./tests/data/dirty_%d.cleaned.jpg' % i)
+            os.remove(path)
+            os.remove('./tests/data/dirty_%d.docx' % i)
+
+
+class TestInplaceCleaning(unittest.TestCase):
+    def test_cleaning(self):
+        shutil.copy('./tests/data/dirty.jpg', './tests/data/clean.jpg')
+        proc = subprocess.Popen(mat2_binary + ['--inplace', './tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        proc = subprocess.Popen(mat2_binary + ['--show', './tests/data/clean.jpg'],
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+        self.assertIn(b'  No metadata found in ./tests/data/clean.jpg.\n', stdout)
+        os.remove('./tests/data/clean.jpg')
+
+    def test_cleaning_multiple_one_fails(self):
+        files = ['./tests/data/clean_%d.jpg' % i for i in range(9)]
+        for f in files:
+            shutil.copy('./tests/data/dirty.jpg', f)
+        shutil.copy('./tests/data/dirty.torrent', './tests/data/clean_9.jpg')
+
+        proc = subprocess.Popen(mat2_binary + ['--inplace'] + files,
+                stdout=subprocess.PIPE)
+        stdout, _ = proc.communicate()
+
+        for f in files:
+            p = images.JPGParser(f)
+            meta = p.get_meta()
+            self.assertEqual(meta, {})
+
+        for i in range(10):
+            os.remove('./tests/data/clean_%d.jpg' % i)
+

tests/test_corrupted_files.py

@@ -1,12 +1,16 @@
 #!/usr/bin/env python3
 
 import unittest
+import stat
+import time
 import shutil
 import os
 import logging
+import zipfile
+import tarfile
 
 from libmat2 import pdf, images, audio, office, parser_factory, torrent
-from libmat2 import harmless, video
+from libmat2 import harmless, video, web, archive
 
 # No need to logging messages, should something go wrong,
 # the testsuite _will_ fail.
@@ -61,20 +65,17 @@ class TestCorruptedEmbedded(unittest.TestCase):
     def test_docx(self):
         shutil.copy('./tests/data/embedded_corrupted.docx', './tests/data/clean.docx')
         parser, _ = parser_factory.get_parser('./tests/data/clean.docx')
-        self.assertFalse(parser.remove_all())
-        self.assertIsNotNone(parser.get_meta())
+        with self.assertRaises(ValueError):
+            parser.remove_all()
+        with self.assertRaises(ValueError):
+            self.assertIsNotNone(parser.get_meta())
         os.remove('./tests/data/clean.docx')
 
     def test_odt(self):
-        expected = {
-                'create_system': 'Weird',
-                'date_time': '2018-06-10 17:18:18',
-                'meta.xml': 'harmful content'
-                }
         shutil.copy('./tests/data/embedded_corrupted.odt', './tests/data/clean.odt')
         parser, _ = parser_factory.get_parser('./tests/data/clean.odt')
         self.assertFalse(parser.remove_all())
-        self.assertEqual(parser.get_meta(), expected)
+        self.assertTrue(parser.get_meta())
         os.remove('./tests/data/clean.odt')
 
 
@@ -90,9 +91,8 @@ class TestExplicitelyUnsupportedFiles(unittest.TestCase):
 class TestWrongContentTypesFileOffice(unittest.TestCase):
     def test_office_incomplete(self):
         shutil.copy('./tests/data/malformed_content_types.docx', './tests/data/clean.docx')
-        p = office.MSOfficeParser('./tests/data/clean.docx')
-        self.assertIsNotNone(p)
-        self.assertFalse(p.remove_all())
+        with self.assertRaises(ValueError):
+            office.MSOfficeParser('./tests/data/clean.docx')
         os.remove('./tests/data/clean.docx')
 
     def test_office_broken(self):
@@ -122,8 +122,8 @@ class TestCorruptedFiles(unittest.TestCase):
 
     def test_png2(self):
         shutil.copy('./tests/test_libmat2.py', './tests/clean.png')
-        parser, _ = parser_factory.get_parser('./tests/clean.png')
-        self.assertIsNone(parser)
+        with self.assertRaises(ValueError):
+            parser_factory.get_parser('./tests/clean.png')
         os.remove('./tests/clean.png')
 
     def test_torrent(self):
@@ -188,6 +188,15 @@ class TestCorruptedFiles(unittest.TestCase):
             audio.MP3Parser('./tests/data/clean.mp3')
         os.remove('./tests/data/clean.mp3')
 
+    def test_wrong_tif(self):
+        shutil.copy('./tests/data/dirty.tiff', './tests/data/clean.tif')
+        p = images.TiffParser('./tests/data/clean.tif')
+        p.remove_all()
+        p = images.TiffParser('./tests/data/clean.cleaned.tif')
+        self.assertEqual(p.get_meta(), {})
+        os.remove('./tests/data/clean.tif')
+        os.remove('./tests/data/clean.cleaned.tif')
+
     def test_jpg(self):
         shutil.copy('./tests/data/dirty.mp3', './tests/data/clean.jpg')
         with self.assertRaises(ValueError):
@@ -195,10 +204,9 @@ class TestCorruptedFiles(unittest.TestCase):
         os.remove('./tests/data/clean.jpg')
 
     def test_png_lightweight(self):
-        return
         shutil.copy('./tests/data/dirty.torrent', './tests/data/clean.png')
-        p = images.PNGParser('./tests/data/clean.png')
-        self.assertTrue(p.remove_all())
+        with self.assertRaises(ValueError):
+            images.PNGParser('./tests/data/clean.png')
         os.remove('./tests/data/clean.png')
 
     def test_avi(self):
@@ -222,3 +230,242 @@ class TestCorruptedFiles(unittest.TestCase):
         p = video.AVIParser('./tests/data/--output.avi')
         self.assertFalse(p.remove_all())
         os.remove('./tests/data/--output.avi')
+
+    def test_zip(self):
+        with zipfile.ZipFile('./tests/data/clean.zip', 'w') as zout:
+            zout.write('./tests/data/dirty.flac')
+            zout.write('./tests/data/dirty.docx')
+            zout.write('./tests/data/dirty.jpg')
+            zout.write('./tests/data/embedded_corrupted.docx')
+        p, mimetype = parser_factory.get_parser('./tests/data/clean.zip')
+        self.assertEqual(mimetype, 'application/zip')
+        with self.assertRaises(ValueError):
+            p.get_meta()
+        with self.assertRaises(ValueError):
+            self.assertFalse(p.remove_all())
+        os.remove('./tests/data/clean.zip')
+
+    def test_html(self):
+        shutil.copy('./tests/data/dirty.html', './tests/data/clean.html')
+        with open('./tests/data/clean.html', 'a') as f:
+            f.write('<open>but not</closed>')
+        with self.assertRaises(ValueError):
+            web.HTMLParser('./tests/data/clean.html')
+        os.remove('./tests/data/clean.html')
+
+        # Yes, we're able to deal with malformed html :/
+        shutil.copy('./tests/data/dirty.html', './tests/data/clean.html')
+        with open('./tests/data/clean.html', 'a') as f:
+            f.write('<meta name=\'this" is="weird"/>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        self.assertTrue(p.remove_all())
+        p = web.HTMLParser('./tests/data/clean.cleaned.html')
+        self.assertEqual(p.get_meta(), {})
+        os.remove('./tests/data/clean.html')
+        os.remove('./tests/data/clean.cleaned.html')
+
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('</meta>')
+        with self.assertRaises(ValueError):
+            web.HTMLParser('./tests/data/clean.html')
+        os.remove('./tests/data/clean.html')
+
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('<meta><a>test</a><set/></meta><title></title><meta>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        with self.assertRaises(ValueError):
+            p.get_meta()
+        p = web.HTMLParser('./tests/data/clean.html')
+        with self.assertRaises(ValueError):
+            p.remove_all()
+        os.remove('./tests/data/clean.html')
+
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('<doctitle><br/></doctitle><br/><notclosed>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        with self.assertRaises(ValueError):
+            p.get_meta()
+        p = web.HTMLParser('./tests/data/clean.html')
+        with self.assertRaises(ValueError):
+            p.remove_all()
+        os.remove('./tests/data/clean.html')
+
+    def test_epub(self):
+        with zipfile.ZipFile('./tests/data/clean.epub', 'w') as zout:
+            zout.write('./tests/data/dirty.jpg', 'OEBPS/content.opf')
+        p, mimetype = parser_factory.get_parser('./tests/data/clean.epub')
+        self.assertEqual(mimetype, 'application/epub+zip')
+        meta = p.get_meta()
+        self.assertEqual(meta['OEBPS/content.opf']['OEBPS/content.opf'],
+                'harmful content')
+
+        self.assertFalse(p.remove_all())
+        os.remove('./tests/data/clean.epub')
+
+    def test_tar(self):
+        with tarfile.TarFile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.flac')
+            zout.add('./tests/data/dirty.docx')
+            zout.add('./tests/data/dirty.jpg')
+            zout.add('./tests/data/embedded_corrupted.docx')
+            tarinfo = tarfile.TarInfo(name='./tests/data/dirty.png')
+            tarinfo.mtime = time.time()
+            tarinfo.uid = 1337
+            tarinfo.gid = 1338
+            tarinfo.size = os.stat('./tests/data/dirty.png').st_size
+            with open('./tests/data/dirty.png', 'rb') as f:
+                zout.addfile(tarinfo, f)
+        p, mimetype = parser_factory.get_parser('./tests/data/clean.tar')
+        self.assertEqual(mimetype, 'application/x-tar')
+        with self.assertRaises(ValueError):
+            p.get_meta()
+        with self.assertRaises(ValueError):
+            self.assertFalse(p.remove_all())
+        os.remove('./tests/data/clean.tar')
+
+        shutil.copy('./tests/data/dirty.png', './tests/data/clean.tar')
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+class TestReadOnlyArchiveMembers(unittest.TestCase):
+    def test_onlymember_tar(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+            tarinfo.mtime = time.time()
+            tarinfo.uid = 1337
+            tarinfo.gid = 0
+            tarinfo.mode = 0o000
+            tarinfo.size = os.stat('./tests/data/dirty.jpg').st_size
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        p, mimetype = parser_factory.get_parser('./tests/data/clean.tar')
+        self.assertEqual(mimetype, 'application/x-tar')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.jpg']['uid'], '1337')
+        self.assertTrue(p.remove_all())
+
+        p = archive.TarParser('./tests/data/clean.cleaned.tar')
+        self.assertEqual(p.get_meta(), {})
+        os.remove('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.cleaned.tar')
+
+
+class TestPathTraversalArchiveMembers(unittest.TestCase):
+    def test_tar_traversal(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+            tarinfo.name = '../../../../../../../../../../tmp/mat2_test.png'
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_absolute_path(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+            tarinfo.name = '/etc/passwd'
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_duplicate_file(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            for _ in range(3):
+                zout.add('./tests/data/dirty.png')
+                tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+                with open('./tests/data/dirty.jpg', 'rb') as f:
+                    zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_setuid(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+            tarinfo.mode |= stat.S_ISUID
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_setgid(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            tarinfo = tarfile.TarInfo('./tests/data/dirty.jpg')
+            tarinfo.mode |= stat.S_ISGID
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_symlink_absolute(self):
+        os.symlink('/etc/passwd', './tests/data/symlink')
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/symlink')
+            tarinfo = tarfile.TarInfo('./tests/data/symlink')
+            tarinfo.linkname = '/etc/passwd'
+            tarinfo.type = tarfile.SYMTYPE
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+        os.remove('./tests/data/symlink')
+
+    def test_tar_symlink_ok(self):
+        shutil.copy('./tests/data/dirty.png', './tests/data/clean.png')
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.png')
+            t = tarfile.TarInfo('mydir')
+            t.type = tarfile.DIRTYPE
+            zout.addfile(t)
+            zout.add('./tests/data/clean.png')
+            t = tarfile.TarInfo('mylink')
+            t.type = tarfile.SYMTYPE
+            t.linkname = './tests/data/clean.png'
+            zout.addfile(t)
+            zout.add('./tests/data/dirty.jpg')
+        archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.png')
+
+    def test_tar_symlink_relative(self):
+        os.symlink('../../../etc/passwd', './tests/data/symlink')
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('./tests/data/symlink')
+            tarinfo = tarfile.TarInfo('./tests/data/symlink')
+            with open('./tests/data/dirty.jpg', 'rb') as f:
+                zout.addfile(tarinfo=tarinfo, fileobj=f)
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+        os.remove('./tests/data/symlink')
+
+    def test_tar_device_file(self):
+        with tarfile.open('./tests/data/clean.tar', 'w') as zout:
+            zout.add('/dev/null')
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/clean.tar')
+        os.remove('./tests/data/clean.tar')
+
+    def test_tar_hardlink(self):
+        shutil.copy('./tests/data/dirty.png', './tests/data/clean.png')
+        os.link('./tests/data/clean.png', './tests/data/hardlink.png')
+        with tarfile.open('./tests/data/cleaner.tar', 'w') as zout:
+            zout.add('tests/data/clean.png')
+            zout.add('tests/data/hardlink.png')
+        with self.assertRaises(ValueError):
+            archive.TarParser('./tests/data/cleaner.tar')
+        os.remove('./tests/data/cleaner.tar')
+        os.remove('./tests/data/clean.png')
+        os.remove('./tests/data/hardlink.png')

tests/test_deep_cleaning.py

@@ -36,6 +36,7 @@ class TestZipMetadata(unittest.TestCase):
 
         meta = p.get_meta()
         self.assertIsNotNone(meta)
+        self.assertEqual(meta['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
 
         ret = p.remove_all()
         self.assertTrue(ret)
@@ -82,6 +83,8 @@ class TestZipOrder(unittest.TestCase):
             previous_name = ''
             for item in zin.infolist():
                 if previous_name == '':
+                    if item.filename == 'mimetype':
+                        continue
                     previous_name = item.filename
                     continue
                 elif item.filename < previous_name:
@@ -96,6 +99,8 @@ class TestZipOrder(unittest.TestCase):
             previous_name = ''
             for item in zin.infolist():
                 if previous_name == '':
+                    if item.filename == 'mimetype':
+                        continue
                     previous_name = item.filename
                     continue
                 self.assertGreaterEqual(item.filename, previous_name)
@@ -132,3 +137,34 @@ class TestRsidRemoval(unittest.TestCase):
 
         os.remove('./tests/data/clean.docx')
         os.remove('./tests/data/clean.cleaned.docx')
+
+
+class TestNsidRemoval(unittest.TestCase):
+    def test_office(self):
+        shutil.copy('./tests/data/dirty_with_nsid.docx', './tests/data/clean.docx')
+        p = office.MSOfficeParser('./tests/data/clean.docx')
+
+        meta = p.get_meta()
+        self.assertIsNotNone(meta)
+
+        how_many_rsid = False
+        with zipfile.ZipFile('./tests/data/clean.docx') as zin:
+            for item in zin.infolist():
+                if not item.filename.endswith('.xml'):
+                    continue
+                num = zin.read(item).decode('utf-8').lower().count('w:rsid')
+                how_many_rsid += num
+        self.assertEqual(how_many_rsid, 1190)
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        with zipfile.ZipFile('./tests/data/clean.cleaned.docx') as zin:
+            for item in zin.infolist():
+                if not item.filename.endswith('.xml'):
+                    continue
+                num = zin.read(item).decode('utf-8').lower().count('w:nsid')
+                self.assertEqual(num, 0)
+
+        os.remove('./tests/data/clean.docx')
+        os.remove('./tests/data/clean.cleaned.docx')

tests/test_libmat2.py

@@ -3,21 +3,21 @@
 import unittest
 import shutil
 import os
+import re
+import tarfile
+import tempfile
 import zipfile
 
 from libmat2 import pdf, images, audio, office, parser_factory, torrent, harmless
-from libmat2 import check_dependencies, video
+from libmat2 import check_dependencies, video, archive, web, epub
 
 
 class TestCheckDependencies(unittest.TestCase):
     def test_deps(self):
-        try:
-            ret = check_dependencies()
-        except RuntimeError:
-            return   # this happens if not every dependency is installed
-
-        for value in ret.values():
-            self.assertTrue(value)
+        ret = check_dependencies()
+        for key, value in ret.items():
+            if value['required']:
+                self.assertTrue(value['found'], "The value for %s is False" % key)
 
 
 class TestParserFactory(unittest.TestCase):
@@ -27,6 +27,14 @@ class TestParserFactory(unittest.TestCase):
         self.assertEqual(mimetype, 'audio/mpeg')
         self.assertEqual(parser.__class__, audio.MP3Parser)
 
+    def test_tarfile_double_extension_handling(self):
+        """ Test that our module auto-detection is handling sub-sub-classes """
+        with tarfile.TarFile.open('./tests/data/dirty.tar.bz2', 'w:bz2') as zout:
+            zout.add('./tests/data/dirty.jpg')
+        parser, mimetype = parser_factory.get_parser('./tests/data/dirty.tar.bz2')
+        self.assertEqual(mimetype, 'application/x-tar+bz2')
+        os.remove('./tests/data/dirty.tar.bz2')
+
 
 class TestParameterInjection(unittest.TestCase):
     def test_ver_injection(self):
@@ -65,13 +73,13 @@ class TestParameterInjection(unittest.TestCase):
 
 
 class TestUnsupportedEmbeddedFiles(unittest.TestCase):
-    def test_odt_with_svg(self):
+    def test_odt_with_py(self):
         shutil.copy('./tests/data/embedded.odt', './tests/data/clean.odt')
         p = office.LibreOfficeParser('./tests/data/clean.odt')
         self.assertFalse(p.remove_all())
         os.remove('./tests/data/clean.odt')
 
-    def test_docx_with_svg(self):
+    def test_docx_with_py(self):
         shutil.copy('./tests/data/embedded.docx', './tests/data/clean.docx')
         p = office.MSOfficeParser('./tests/data/clean.docx')
         self.assertFalse(p.remove_all())
@@ -105,6 +113,13 @@ class TestGetMeta(unittest.TestCase):
         meta = p.get_meta()
         self.assertEqual(meta['Comment'], 'Created with GIMP')
 
+    def test_ppm(self):
+        p = images.PPMParser('./tests/data/dirty.ppm')
+        meta = p.get_meta()
+        self.assertEqual(meta['1'], '# A metadata')
+        self.assertEqual(meta['4'], '# And an other one')
+        self.assertEqual(meta['6'], '# and a final one here')
+
     def test_tiff(self):
         p = images.TiffParser('./tests/data/dirty.tiff')
         meta = p.get_meta()
@@ -112,6 +127,11 @@ class TestGetMeta(unittest.TestCase):
         self.assertEqual(meta['Model'], 'C7070WZ')
         self.assertEqual(meta['ModifyDate'], '2005:12:26 17:09:35')
 
+    def test_wav(self):
+        p = audio.WAVParser('./tests/data/dirty.wav')
+        meta = p.get_meta()
+        self.assertEqual(meta['Artist'], 'jvoisin')
+
     def test_mp3(self):
         p = audio.MP3Parser('./tests/data/dirty.mp3')
         meta = p.get_meta()
@@ -131,21 +151,21 @@ class TestGetMeta(unittest.TestCase):
     def test_docx(self):
         p = office.MSOfficeParser('./tests/data/dirty.docx')
         meta = p.get_meta()
-        self.assertEqual(meta['cp:lastModifiedBy'], 'Julien Voisin')
-        self.assertEqual(meta['dc:creator'], 'julien voisin')
-        self.assertEqual(meta['Application'], 'LibreOffice/5.4.5.1$Linux_X86_64 LibreOffice_project/40m0$Build-1')
+        self.assertEqual(meta['docProps/core.xml']['cp:lastModifiedBy'], 'Julien Voisin')
+        self.assertEqual(meta['docProps/core.xml']['dc:creator'], 'julien voisin')
+        self.assertEqual(meta['docProps/app.xml']['Application'], 'LibreOffice/5.4.5.1$Linux_X86_64 LibreOffice_project/40m0$Build-1')
 
     def test_libreoffice(self):
         p = office.LibreOfficeParser('./tests/data/dirty.odt')
         meta = p.get_meta()
-        self.assertEqual(meta['meta:initial-creator'], 'jvoisin ')
-        self.assertEqual(meta['meta:creation-date'], '2011-07-26T03:27:48')
-        self.assertEqual(meta['meta:generator'], 'LibreOffice/3.3$Unix LibreOffice_project/330m19$Build-202')
+        self.assertEqual(meta['meta.xml']['meta:initial-creator'], 'jvoisin ')
+        self.assertEqual(meta['meta.xml']['meta:creation-date'], '2011-07-26T03:27:48')
+        self.assertEqual(meta['meta.xml']['meta:generator'], 'LibreOffice/3.3$Unix LibreOffice_project/330m19$Build-202')
 
         p = office.LibreOfficeParser('./tests/data/weird_producer.odt')
         meta = p.get_meta()
-        self.assertEqual(meta['create_system'], 'Windows')
-        self.assertEqual(meta['comment'], b'YAY FOR COMMENTS')
+        self.assertEqual(meta['mimetype']['create_system'], 'Windows')
+        self.assertEqual(meta['mimetype']['comment'], b'YAY FOR COMMENTS')
 
     def test_txt(self):
         p, mimetype = parser_factory.get_parser('./tests/data/dirty.txt')
@@ -153,6 +173,90 @@ class TestGetMeta(unittest.TestCase):
         meta = p.get_meta()
         self.assertEqual(meta, {})
 
+    def test_zip(self):
+        with zipfile.ZipFile('./tests/data/dirty.zip', 'w') as zout:
+            zout.write('./tests/data/dirty.flac',
+                       compress_type = zipfile.ZIP_STORED)
+            zout.write('./tests/data/dirty.docx',
+                       compress_type = zipfile.ZIP_DEFLATED)
+            zout.write('./tests/data/dirty.jpg',
+                       compress_type = zipfile.ZIP_BZIP2)
+            zout.write('./tests/data/dirty.txt',
+                       compress_type = zipfile.ZIP_LZMA)
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.zip')
+        self.assertEqual(mimetype, 'application/zip')
+        meta = p.get_meta()
+        self.assertEqual(meta['tests/data/dirty.flac']['comments'], 'Thank you for using MAT !')
+        self.assertEqual(meta['tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        with zipfile.ZipFile('./tests/data/dirty.zip') as zipin:
+            members = {
+                'tests/data/dirty.flac' : zipfile.ZIP_STORED,
+                'tests/data/dirty.docx': zipfile.ZIP_DEFLATED,
+                'tests/data/dirty.jpg' : zipfile.ZIP_BZIP2,
+                'tests/data/dirty.txt' : zipfile.ZIP_LZMA,
+            }
+            for k, v in members.items():
+                self.assertEqual(zipin.getinfo(k).compress_type, v)
+
+        os.remove('./tests/data/dirty.zip')
+
+    def test_wmv(self):
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.wmv')
+        self.assertEqual(mimetype, 'video/x-ms-wmv')
+        meta = p.get_meta()
+        self.assertEqual(meta['EncodingSettings'], 'Lavf52.103.0')
+
+    def test_gif(self):
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.gif')
+        self.assertEqual(mimetype, 'image/gif')
+        meta = p.get_meta()
+        self.assertEqual(meta['Comment'], 'this is a test comment')
+
+    def test_epub(self):
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.epub')
+        self.assertEqual(mimetype, 'application/epub+zip')
+        meta = p.get_meta()
+        self.assertEqual(meta['OEBPS/content.opf']['dc:creator'], 'Dorothy L. Sayers')
+        self.assertEqual(meta['OEBPS/toc.ncx']['dtb:generator'], 'Ebookmaker 0.4.0a5 by Marcello Perathoner <webmaster@gutenberg.org>')
+        self.assertEqual(meta['OEBPS/@public@vhost@g@gutenberg@html@files@58820@58820-h@images@shield25.jpg']['CreatorTool'], 'Adobe Photoshop CS5 Macintosh')
+        self.assertEqual(meta['OEBPS/@public@vhost@g@gutenberg@html@files@58820@58820-h@58820-h-2.htm.html']['generator'], 'Ebookmaker 0.4.0a5 by Marcello Perathoner <webmaster@gutenberg.org>')
+
+    def test_css(self):
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.css')
+        self.assertEqual(mimetype, 'text/css')
+        meta = p.get_meta()
+        self.assertEqual(meta['author'], 'jvoisin')
+        self.assertEqual(meta['version'], '1.0')
+        self.assertEqual(meta['harmful data'], 'underline is cool')
+
+    def test_tar(self):
+        with tarfile.TarFile('./tests/data/dirty.tar', 'w') as tout:
+            tout.add('./tests/data/dirty.flac')
+            tout.add('./tests/data/dirty.docx')
+            tout.add('./tests/data/dirty.jpg')
+        p, mimetype = parser_factory.get_parser('./tests/data/dirty.tar')
+        self.assertEqual(mimetype, 'application/x-tar')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.flac']['comments'], 'Thank you for using MAT !')
+        self.assertEqual(meta['./tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+        os.remove('./tests/data/dirty.tar')
+
+    def test_svg(self):
+        p = images.SVGParser('./tests/data/weird.svg')
+        self.assertEqual(p.get_meta()['Xmlns'], 'http://www.w3.org/1337/svg')
+
+    def test_aiff(self):
+        p = audio.AIFFParser('./tests/data/dirty.aiff')
+        meta = p.get_meta()
+        self.assertEqual(meta['Name'], 'I am so')
+
+    def test_heic(self):
+        p = images.HEICParser('./tests/data/dirty.heic')
+        meta = p.get_meta()
+        self.assertEqual(meta['ProfileCopyright'], 'Public Domain')
+        self.assertEqual(meta['ProfileDescription'], 'GIMP built-in sRGB')
+
 
 class TestRemovingThumbnails(unittest.TestCase):
     def test_odt(self):
@@ -212,29 +316,516 @@ class TestRevisionsCleaning(unittest.TestCase):
         os.remove('./tests/data/revision_clean.docx')
         os.remove('./tests/data/revision_clean.cleaned.docx')
 
+
 class TestCleaning(unittest.TestCase):
-    def test_pdf(self):
-        shutil.copy('./tests/data/dirty.pdf', './tests/data/clean.pdf')
-        p = pdf.PDFParser('./tests/data/clean.pdf')
+    data = [{
+        'name': 'pdf',
+        'parser': pdf.PDFParser,
+        'meta': {'producer': 'pdfTeX-1.40.14'},
+        'expected_meta': {'creation-date': -1, 'format': 'PDF-1.5', 'mod-date': -1},
+        }, {
+            'name': 'png',
+            'parser': images.PNGParser,
+            'meta': {'Comment': 'This is a comment, be careful!'},
+            'expected_meta': {},
+        }, {
+            'name': 'jpg',
+            'parser': images.JPGParser,
+            'meta': {'Comment': 'Created with GIMP'},
+            'expected_meta': {},
+        }, {
+            'name': 'wav',
+            'parser': audio.WAVParser,
+            'meta': {'Comment': 'Zomg, a comment!'},
+            'expected_meta': {},
+        }, {
+            'name': 'aiff',
+            'parser': audio.AIFFParser,
+            'meta': {'Annotation': 'Thank you for using MAT !'},
+            'expected_meta': {},
+        },
+        {
+            'name': 'mp3',
+            'parser': audio.MP3Parser,
+            'meta': {'TXXX:I am a': 'various comment'},
+            'expected_meta': {},
+        }, {
+            'name': 'ogg',
+            'parser': audio.OGGParser,
+            'meta': {'title': 'I am so'},
+            'expected_meta': {},
+        }, {
+            'name': 'flac',
+            'parser': audio.FLACParser,
+            'meta': {'title': 'I am so'},
+            'expected_meta': {},
+        }, {
+            'name': 'docx',
+            'parser': office.MSOfficeParser,
+            'meta': {'word/media/image1.png' :
+               {'Comment': 'This is a comment, be careful!',
+                'ModifyDate': '2018:03:20 21:59:25',
+                'PixelUnits': 'meters',
+                'PixelsPerUnitX': 2835,
+                'PixelsPerUnitY': 2835,
+                'create_system': 'Weird',
+                'date_time': '2018-03-31 13:15:38'} ,
+               },
+            'expected_meta': {},
+        }, {
+            'name': 'odt',
+            'parser': office.LibreOfficeParser,
+            'meta': {
+                'Pictures/1000000000000032000000311EC5314D.png': {
+                    'create_system': 'Weird',
+                    'date_time': '2011-07-26 02:40:16',
+                    'PixelsPerUnitX': 4847,
+                    'PixelsPerUnitY': 4760,
+                    'PixelUnits': 'meters',
+                    },
+                },
+            'expected_meta': {},
+        },{
+            'name': 'tiff',
+            'parser': images.TiffParser,
+            'meta': {'Model': 'C7070WZ'},
+            'expected_meta':
+             {'Orientation': 'Horizontal (normal)',
+                  'ResolutionUnit': 'inches',
+                  'XResolution': 72,
+                  'YResolution': 72}
+        },{
+            'name': 'bmp',
+            'parser': harmless.HarmlessParser,
+            'meta': {},
+            'expected_meta': {},
+        },{
+            'name': 'torrent',
+            'parser': torrent.TorrentParser,
+            'meta': {'created by': b'mktorrent 1.0', 'creation date': 1522397702},
+            'expected_meta': {},
+        }, {
+            'name': 'odf',
+            'parser': office.LibreOfficeParser,
+            'meta': {'meta.xml': {'create_system': 'Weird', 'date_time':
+                '2018-04-22 22:20:24', 'meta:initial-creator': 'Julien Voisin',
+                'meta:creation-date': '2018-04-23T00:18:59.438231281',
+                'dc:date': '2018-04-23T00:20:23.978564933', 'dc:creator':
+                'Julien Voisin', 'meta:editing-duration': 'PT1M24S',
+                'meta:editing-cycles': '1', 'meta:generator':
+                'LibreOffice/5.4.6.2$Linux_X86_64 LibreOffice_project/40m0$Build-2'}},
+            'expected_meta': {},
+        }, {
+            'name': 'odg',
+            'parser': office.LibreOfficeParser,
+            'meta': {'meta.xml': {'create_system': 'Weird', 'date_time':
+                '2018-04-22 22:26:58', 'meta:initial-creator': 'Julien Voisin',
+                'meta:creation-date': '2018-04-23T00:25:59.953271949',
+                'dc:date': '2018-04-23T00:26:59.385838550', 'dc:creator':
+                'Julien Voisin', 'meta:editing-duration': 'PT59S',
+                'meta:editing-cycles': '1', 'meta:generator':
+                'LibreOffice/5.4.6.2$Linux_X86_64 LibreOffice_project/40m0$Build-2'}},
+            'expected_meta': {},
+        }, {
+            'name': 'txt',
+            'parser': harmless.HarmlessParser,
+            'meta': {},
+            'expected_meta': {},
+        },{
+            'name': 'gif',
+            'parser': images.GIFParser,
+            'meta': {'Comment': 'this is a test comment'},
+            'expected_meta': {'TransparentColor': '5'},
+        },{
+            'name': 'css',
+            'parser': web.CSSParser,
+            'meta': {
+                'harmful data': 'underline is cool',
+                'version': '1.0',
+                'author': 'jvoisin'
+            },
+            'expected_meta': {},
+        },{
+            'name': 'svg',
+            'parser': images.SVGParser,
+            'meta': {
+                'WorkDescription': "This is a test svg image for mat2's testsuite",
+            },
+            'expected_meta': {
+                'ImageSize': '128x128',
+                'Megapixels': '0.016',
+            },
+        } ,{
+            'name': 'ppm',
+            'parser': images.PPMParser,
+            'meta': {
+                '1': '# A metadata',
+            },
+            'expected_meta': {},
+        } ,{
+            'name': 'avi',
+            'ffmpeg': 1,
+            'parser': video.AVIParser,
+            'meta': {
+                'Software': 'MEncoder SVN-r33148-4.0.1',
+            },
+            'expected_meta': {},
+        } ,{
+            'name': 'mp4',
+            'ffmpeg': 1,
+            'parser': video.MP4Parser,
+            'meta': {
+                'Encoder':  'HandBrake 0.9.4 2009112300',
+            },
+            'expected_meta': {
+                'AverageBitrate': 465641,
+                'BufferSize': 0,
+                'CompatibleBrands': ['isom', 'iso2', 'avc1', 'mp41'],
+                'ColorRepresentation': 'nclx 1 1 1',
+                'CompressorID': 'avc1',
+                'GraphicsMode': 'srcCopy',
+                'HandlerDescription': 'SoundHandler',
+                'HandlerType': 'Metadata',
+                'HandlerVendorID': 'Apple',
+                'MajorBrand': 'Base Media v1 [IS0 14496-12:2003]',
+                'MaxBitrate': 465641,
+                'MediaDataOffset': 48,
+                'MediaDataSize': 379872,
+                'MediaHeaderVersion': 0,
+                'MinorVersion': '0.2.0',
+                'MovieDataOffset': 48,
+                'MovieHeaderVersion': 0,
+                'NextTrackID': 3,
+                'PreferredRate': 1,
+                'Rotation': 0,
+                'TimeScale': 1000,
+                'TrackHeaderVersion': 0,
+                'TrackID': 1,
+                'TrackLayer': 0},
+        },{
+            'name': 'wmv',
+            'ffmpeg': 1,
+            'parser': video.WMVParser,
+            'meta': {
+                'EncodingSettings': 'Lavf52.103.0',
+            },
+            'expected_meta': {},
+        },{
+            'name': 'heic',
+            'parser': images.HEICParser,
+            'meta': {},
+            'expected_meta': {},
+        }
+        ]
+
+    def test_all_parametred(self):
+        for case in self.data:
+            with self.subTest(case=case):
+                if 'ffmpeg' in case:
+                    try:
+                        video._get_ffmpeg_path()
+                    except RuntimeError:
+                        raise unittest.SkipTest
+
+                print('[+] Testing %s' % case['name'])
+                target = './tests/data/clean.' + case['name']
+                shutil.copy('./tests/data/dirty.' + case['name'], target)
+                p1 = case['parser'](target)
+
+                for k, v in p1.get_meta().items():
+                    if k not in case['meta']:
+                        continue
+                    if isinstance(v, dict):
+                        for _k, _v in v.items():
+                            if _k in case['meta'][k]:
+                                self.assertEqual(_v, case['meta'][k][_k])
+                    else:
+                        self.assertEqual(v, case['meta'][k])
+
+                p1.lightweight_cleaning = True
+                self.assertTrue(p1.remove_all())
+
+                p2 = case['parser'](p1.output_filename)
+                meta = p2.get_meta()
+                if meta:
+                    for k, v in p2.get_meta().items():
+                        self.assertIn(k, case['expected_meta'], '"%s" is not in "%s" (%s)' % (k, case['expected_meta'], case['name']))
+                        self.assertIn(str(case['expected_meta'][k]), str(v))
+                self.assertTrue(p2.remove_all())
+
+                os.remove(target)
+                os.remove(p1.output_filename)
+                os.remove(p2.output_filename)
+
+
+    def test_html(self):
+        shutil.copy('./tests/data/dirty.html', './tests/data/clean.html')
+        p = web.HTMLParser('./tests/data/clean.html')
 
         meta = p.get_meta()
-        self.assertEqual(meta['producer'], 'pdfTeX-1.40.14')
+        self.assertEqual(meta['author'], 'jvoisin')
 
         ret = p.remove_all()
         self.assertTrue(ret)
 
-        p = pdf.PDFParser('./tests/data/clean.cleaned.pdf')
-        expected_meta = {'creation-date': -1, 'format': 'PDF-1.5', 'mod-date': -1}
-        self.assertEqual(p.get_meta(), expected_meta)
+        p = web.HTMLParser('./tests/data/clean.cleaned.html')
+        self.assertEqual(p.get_meta(), {})
         self.assertTrue(p.remove_all())
 
-        os.remove('./tests/data/clean.pdf')
-        os.remove('./tests/data/clean.cleaned.pdf')
-        os.remove('./tests/data/clean.cleaned.cleaned.pdf')
+        os.remove('./tests/data/clean.html')
+        os.remove('./tests/data/clean.cleaned.html')
+        os.remove('./tests/data/clean.cleaned.cleaned.html')
 
-    def test_png(self):
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('<title><title><pouet/><meta/></title></title><test/>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        self.assertTrue(p.remove_all())
+        with open('./tests/data/clean.cleaned.html', 'r') as f:
+            self.assertEqual(f.read(), '<title></title><test/>')
+        os.remove('./tests/data/clean.html')
+        os.remove('./tests/data/clean.cleaned.html')
+
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('<test><title>Some<b>metadata</b><br/></title></test>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        self.assertTrue(p.remove_all())
+        with open('./tests/data/clean.cleaned.html', 'r') as f:
+            self.assertEqual(f.read(), '<test><title></title></test>')
+        os.remove('./tests/data/clean.html')
+        os.remove('./tests/data/clean.cleaned.html')
+
+        with open('./tests/data/clean.html', 'w') as f:
+            f.write('<meta><meta/><!----><!-- test--></meta>')
+        p = web.HTMLParser('./tests/data/clean.html')
+        self.assertTrue(p.remove_all())
+        with open('./tests/data/clean.cleaned.html', 'r') as f:
+            self.assertEqual(f.read(), '')
+        os.remove('./tests/data/clean.html')
+        os.remove('./tests/data/clean.cleaned.html')
+
+    def test_epub(self):
+        shutil.copy('./tests/data/dirty.epub', './tests/data/clean.epub')
+        p = epub.EPUBParser('./tests/data/clean.epub')
+
+        meta = p.get_meta()
+        self.assertEqual(meta['OEBPS/content.opf']['dc:source'], 'http://www.gutenberg.org/files/58820/58820-h/58820-h.htm')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = epub.EPUBParser('./tests/data/clean.cleaned.epub')
+        meta = p.get_meta()
+        res = re.match(meta['OEBPS/content.opf']['metadata'], '^<dc:identifier>[0-9a-f-]+</dc:identifier><dc:title /><dc:language />$')
+        self.assertNotEqual(res, False)
+
+        self.assertTrue(p.remove_all())
+
+        os.remove('./tests/data/clean.epub')
+        os.remove('./tests/data/clean.cleaned.epub')
+        os.remove('./tests/data/clean.cleaned.cleaned.epub')
+
+
+class TestCleaningArchives(unittest.TestCase):
+    def test_zip(self):
+        with zipfile.ZipFile('./tests/data/dirty.zip', 'w') as zout:
+            zout.write('./tests/data/dirty.flac',
+                       compress_type = zipfile.ZIP_STORED)
+            zout.write('./tests/data/dirty.docx',
+                       compress_type = zipfile.ZIP_DEFLATED)
+            zout.write('./tests/data/dirty.jpg',
+                       compress_type = zipfile.ZIP_BZIP2)
+            zout.write('./tests/data/dirty.txt',
+                       compress_type = zipfile.ZIP_LZMA)
+        p = archive.ZipParser('./tests/data/dirty.zip')
+        meta = p.get_meta()
+        self.assertEqual(meta['tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = archive.ZipParser('./tests/data/dirty.cleaned.zip')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        with zipfile.ZipFile('./tests/data/dirty.zip') as zipin:
+            members = {
+                'tests/data/dirty.flac' : zipfile.ZIP_STORED,
+                'tests/data/dirty.docx': zipfile.ZIP_DEFLATED,
+                'tests/data/dirty.jpg' : zipfile.ZIP_BZIP2,
+                'tests/data/dirty.txt' : zipfile.ZIP_LZMA,
+            }
+            for k, v in members.items():
+                self.assertEqual(zipin.getinfo(k).compress_type, v)
+
+        os.remove('./tests/data/dirty.zip')
+        os.remove('./tests/data/dirty.cleaned.zip')
+        os.remove('./tests/data/dirty.cleaned.cleaned.zip')
+
+    def test_tar(self):
+        with tarfile.TarFile.open('./tests/data/dirty.tar', 'w') as zout:
+            zout.add('./tests/data/dirty.flac')
+            zout.add('./tests/data/dirty.docx')
+            zout.add('./tests/data/dirty.jpg')
+        p = archive.TarParser('./tests/data/dirty.tar')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = archive.TarParser('./tests/data/dirty.cleaned.tar')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        tmp_dir = tempfile.mkdtemp()
+        with tarfile.open('./tests/data/dirty.cleaned.tar') as zout:
+            zout.extractall(path=tmp_dir)
+            zout.close()
+
+        number_of_files = 0
+        for root, _, fnames in os.walk(tmp_dir):
+            for f in fnames:
+                complete_path = os.path.join(root, f)
+                p, _ = parser_factory.get_parser(complete_path)
+                self.assertIsNotNone(p)
+                self.assertEqual(p.get_meta(), {})
+                number_of_files += 1
+        self.assertEqual(number_of_files, 3)
+
+        os.remove('./tests/data/dirty.tar')
+        os.remove('./tests/data/dirty.cleaned.tar')
+        os.remove('./tests/data/dirty.cleaned.cleaned.tar')
+
+    def test_targz(self):
+        with tarfile.TarFile.open('./tests/data/dirty.tar.gz', 'w:gz') as zout:
+            zout.add('./tests/data/dirty.flac')
+            zout.add('./tests/data/dirty.docx')
+            zout.add('./tests/data/dirty.jpg')
+        p = archive.TarParser('./tests/data/dirty.tar.gz')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = archive.TarParser('./tests/data/dirty.cleaned.tar.gz')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        tmp_dir = tempfile.mkdtemp()
+        with tarfile.open('./tests/data/dirty.cleaned.tar.gz') as zout:
+            zout.extractall(path=tmp_dir)
+            zout.close()
+
+        number_of_files = 0
+        for root, _, fnames in os.walk(tmp_dir):
+            for f in fnames:
+                complete_path = os.path.join(root, f)
+                p, _ = parser_factory.get_parser(complete_path)
+                self.assertIsNotNone(p)
+                self.assertEqual(p.get_meta(), {})
+                number_of_files += 1
+        self.assertEqual(number_of_files, 3)
+
+        os.remove('./tests/data/dirty.tar.gz')
+        os.remove('./tests/data/dirty.cleaned.tar.gz')
+        os.remove('./tests/data/dirty.cleaned.cleaned.tar.gz')
+
+    def test_tarbz2(self):
+        with tarfile.TarFile.open('./tests/data/dirty.tar.bz2', 'w:bz2') as zout:
+            zout.add('./tests/data/dirty.flac')
+            zout.add('./tests/data/dirty.docx')
+            zout.add('./tests/data/dirty.jpg')
+        p = archive.TarParser('./tests/data/dirty.tar.bz2')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = archive.TarParser('./tests/data/dirty.cleaned.tar.bz2')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        tmp_dir = tempfile.mkdtemp()
+        with tarfile.open('./tests/data/dirty.cleaned.tar.bz2') as zout:
+            zout.extractall(path=tmp_dir)
+            zout.close()
+
+        number_of_files = 0
+        for root, _, fnames in os.walk(tmp_dir):
+            for f in fnames:
+                complete_path = os.path.join(root, f)
+                p, _ = parser_factory.get_parser(complete_path)
+                self.assertIsNotNone(p)
+                self.assertEqual(p.get_meta(), {})
+                number_of_files += 1
+        self.assertEqual(number_of_files, 3)
+
+        os.remove('./tests/data/dirty.tar.bz2')
+        os.remove('./tests/data/dirty.cleaned.tar.bz2')
+        os.remove('./tests/data/dirty.cleaned.cleaned.tar.bz2')
+
+    def test_tarxz(self):
+        with tarfile.TarFile.open('./tests/data/dirty.tar.xz', 'w:xz') as zout:
+            zout.add('./tests/data/dirty.flac')
+            zout.add('./tests/data/dirty.docx')
+            zout.add('./tests/data/dirty.jpg')
+        p = archive.TarParser('./tests/data/dirty.tar.xz')
+        meta = p.get_meta()
+        self.assertEqual(meta['./tests/data/dirty.docx']['word/media/image1.png']['Comment'], 'This is a comment, be careful!')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = archive.TarParser('./tests/data/dirty.cleaned.tar.xz')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        tmp_dir = tempfile.mkdtemp()
+        with tarfile.open('./tests/data/dirty.cleaned.tar.xz') as zout:
+            zout.extractall(path=tmp_dir)
+            zout.close()
+
+        number_of_files = 0
+        for root, _, fnames in os.walk(tmp_dir):
+            for f in fnames:
+                complete_path = os.path.join(root, f)
+                p, _ = parser_factory.get_parser(complete_path)
+                self.assertIsNotNone(p)
+                self.assertEqual(p.get_meta(), {})
+                number_of_files += 1
+        self.assertEqual(number_of_files, 3)
+
+        os.remove('./tests/data/dirty.tar.xz')
+        os.remove('./tests/data/dirty.cleaned.tar.xz')
+        os.remove('./tests/data/dirty.cleaned.cleaned.tar.xz')
+
+class TestNoSandbox(unittest.TestCase):
+    def test_avi_nosandbox(self):
+        shutil.copy('./tests/data/dirty.avi', './tests/data/clean.avi')
+        p = video.AVIParser('./tests/data/clean.avi')
+        p.sandbox = False
+
+        meta = p.get_meta()
+        self.assertEqual(meta['Software'], 'MEncoder SVN-r33148-4.0.1')
+
+        ret = p.remove_all()
+        self.assertTrue(ret)
+
+        p = video.AVIParser('./tests/data/clean.cleaned.avi')
+        self.assertEqual(p.get_meta(), {})
+        self.assertTrue(p.remove_all())
+
+        os.remove('./tests/data/clean.avi')
+        os.remove('./tests/data/clean.cleaned.avi')
+        os.remove('./tests/data/clean.cleaned.cleaned.avi')
+
+    def test_png_nosandbox(self):
         shutil.copy('./tests/data/dirty.png', './tests/data/clean.png')
         p = images.PNGParser('./tests/data/clean.png')
+        p.sandbox = False
+        p.lightweight_cleaning = True
 
         meta = p.get_meta()
         self.assertEqual(meta['Comment'], 'This is a comment, be careful!')
@@ -250,241 +841,12 @@ class TestCleaning(unittest.TestCase):
         os.remove('./tests/data/clean.cleaned.png')
         os.remove('./tests/data/clean.cleaned.cleaned.png')
 
-    def test_jpg(self):
-        shutil.copy('./tests/data/dirty.jpg', './tests/data/clean.jpg')
-        p = images.JPGParser('./tests/data/clean.jpg')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['Comment'], 'Created with GIMP')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = images.JPGParser('./tests/data/clean.cleaned.jpg')
-        self.assertEqual(p.get_meta(), {})
+class TestComplexOfficeFiles(unittest.TestCase):
+    def test_complex_pptx(self):
+        target = './tests/data/clean.pptx'
+        shutil.copy('./tests/data/narrated_powerpoint_presentation.pptx', target)
+        p = office.MSOfficeParser(target)
         self.assertTrue(p.remove_all())
 
-        os.remove('./tests/data/clean.jpg')
-        os.remove('./tests/data/clean.cleaned.jpg')
-        os.remove('./tests/data/clean.cleaned.cleaned.jpg')
-
-    def test_mp3(self):
-        shutil.copy('./tests/data/dirty.mp3', './tests/data/clean.mp3')
-        p = audio.MP3Parser('./tests/data/clean.mp3')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['TXXX:I am a'], 'various comment')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = audio.MP3Parser('./tests/data/clean.cleaned.mp3')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.mp3')
-        os.remove('./tests/data/clean.cleaned.mp3')
-        os.remove('./tests/data/clean.cleaned.cleaned.mp3')
-
-    def test_ogg(self):
-        shutil.copy('./tests/data/dirty.ogg', './tests/data/clean.ogg')
-        p = audio.OGGParser('./tests/data/clean.ogg')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['title'], 'I am so')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = audio.OGGParser('./tests/data/clean.cleaned.ogg')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.ogg')
-        os.remove('./tests/data/clean.cleaned.ogg')
-        os.remove('./tests/data/clean.cleaned.cleaned.ogg')
-
-    def test_flac(self):
-        shutil.copy('./tests/data/dirty.flac', './tests/data/clean.flac')
-        p = audio.FLACParser('./tests/data/clean.flac')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['title'], 'I am so')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = audio.FLACParser('./tests/data/clean.cleaned.flac')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.flac')
-        os.remove('./tests/data/clean.cleaned.flac')
-        os.remove('./tests/data/clean.cleaned.cleaned.flac')
-
-    def test_office(self):
-        shutil.copy('./tests/data/dirty.docx', './tests/data/clean.docx')
-        p = office.MSOfficeParser('./tests/data/clean.docx')
-
-        meta = p.get_meta()
-        self.assertIsNotNone(meta)
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = office.MSOfficeParser('./tests/data/clean.cleaned.docx')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.docx')
-        os.remove('./tests/data/clean.cleaned.docx')
-        os.remove('./tests/data/clean.cleaned.cleaned.docx')
-
-    def test_libreoffice(self):
-        shutil.copy('./tests/data/dirty.odt', './tests/data/clean.odt')
-        p = office.LibreOfficeParser('./tests/data/clean.odt')
-
-        meta = p.get_meta()
-        self.assertIsNotNone(meta)
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = office.LibreOfficeParser('./tests/data/clean.cleaned.odt')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.odt')
-        os.remove('./tests/data/clean.cleaned.odt')
-        os.remove('./tests/data/clean.cleaned.cleaned.odt')
-
-    def test_tiff(self):
-        shutil.copy('./tests/data/dirty.tiff', './tests/data/clean.tiff')
-        p = images.TiffParser('./tests/data/clean.tiff')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['Model'], 'C7070WZ')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = images.TiffParser('./tests/data/clean.cleaned.tiff')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.tiff')
-        os.remove('./tests/data/clean.cleaned.tiff')
-        os.remove('./tests/data/clean.cleaned.cleaned.tiff')
-
-    def test_bmp(self):
-        shutil.copy('./tests/data/dirty.bmp', './tests/data/clean.bmp')
-        p = harmless.HarmlessParser('./tests/data/clean.bmp')
-
-        meta = p.get_meta()
-        self.assertEqual(meta, {})  # bmp has no meta :)
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = harmless.HarmlessParser('./tests/data/clean.cleaned.bmp')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.bmp')
-        os.remove('./tests/data/clean.cleaned.bmp')
-        os.remove('./tests/data/clean.cleaned.cleaned.bmp')
-
-    def test_torrent(self):
-        shutil.copy('./tests/data/dirty.torrent', './tests/data/clean.torrent')
-        p = torrent.TorrentParser('./tests/data/clean.torrent')
-
-        meta = p.get_meta()
-        self.assertEqual(meta, {'created by': b'mktorrent 1.0', 'creation date': 1522397702})
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = torrent.TorrentParser('./tests/data/clean.cleaned.torrent')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.torrent')
-        os.remove('./tests/data/clean.cleaned.torrent')
-        os.remove('./tests/data/clean.cleaned.cleaned.torrent')
-
-    def test_odf(self):
-        shutil.copy('./tests/data/dirty.odf', './tests/data/clean.odf')
-        p = office.LibreOfficeParser('./tests/data/clean.odf')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['meta:creation-date'], '2018-04-23T00:18:59.438231281')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = office.LibreOfficeParser('./tests/data/clean.cleaned.odf')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.odf')
-        os.remove('./tests/data/clean.cleaned.odf')
-        os.remove('./tests/data/clean.cleaned.cleaned.odf')
-
-    def test_odg(self):
-        shutil.copy('./tests/data/dirty.odg', './tests/data/clean.odg')
-        p = office.LibreOfficeParser('./tests/data/clean.odg')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['dc:date'], '2018-04-23T00:26:59.385838550')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = office.LibreOfficeParser('./tests/data/clean.cleaned.odg')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.odg')
-        os.remove('./tests/data/clean.cleaned.odg')
-        os.remove('./tests/data/clean.cleaned.cleaned.odg')
-
-    def test_txt(self):
-        shutil.copy('./tests/data/dirty.txt', './tests/data/clean.txt')
-        p = harmless.HarmlessParser('./tests/data/clean.txt')
-
-        meta = p.get_meta()
-        self.assertEqual(meta, {})
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = harmless.HarmlessParser('./tests/data/clean.cleaned.txt')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.txt')
-        os.remove('./tests/data/clean.cleaned.txt')
-        os.remove('./tests/data/clean.cleaned.cleaned.txt')
-
-    def test_avi(self):
-        try:
-            video._get_ffmpeg_path()
-        except RuntimeError:
-            raise unittest.SkipTest
-
-        shutil.copy('./tests/data/dirty.avi', './tests/data/clean.avi')
-        p = video.AVIParser('./tests/data/clean.avi')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['Software'], 'MEncoder SVN-r33148-4.0.1')
-
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = video.AVIParser('./tests/data/clean.cleaned.avi')
-        self.assertEqual(p.get_meta(), {})
-        self.assertTrue(p.remove_all())
-
-        os.remove('./tests/data/clean.avi')
-        os.remove('./tests/data/clean.cleaned.avi')
-        os.remove('./tests/data/clean.cleaned.cleaned.avi')
+        os.remove(target)
+        os.remove(p.output_filename)

tests/test_lightweigh_cleaning.py

@@ -1,65 +0,0 @@
-#!/usr/bin/env python3
-
-import unittest
-import shutil
-import os
-
-from libmat2 import pdf, images
-
-class TestLightWeightCleaning(unittest.TestCase):
-    def test_pdf(self):
-        shutil.copy('./tests/data/dirty.pdf', './tests/data/clean.pdf')
-        p = pdf.PDFParser('./tests/data/clean.pdf')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['producer'], 'pdfTeX-1.40.14')
-
-        p.lightweight_cleaning = True
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = pdf.PDFParser('./tests/data/clean.cleaned.pdf')
-        expected_meta = {'creation-date': -1, 'format': 'PDF-1.5', 'mod-date': -1}
-        self.assertEqual(p.get_meta(), expected_meta)
-
-        os.remove('./tests/data/clean.pdf')
-        os.remove('./tests/data/clean.cleaned.pdf')
-
-    def test_png(self):
-        shutil.copy('./tests/data/dirty.png', './tests/data/clean.png')
-        p = images.PNGParser('./tests/data/clean.png')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['Comment'], 'This is a comment, be careful!')
-
-        p.lightweight_cleaning = True
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = images.PNGParser('./tests/data/clean.cleaned.png')
-        self.assertEqual(p.get_meta(), {})
-
-        p = images.PNGParser('./tests/data/clean.png')
-        p.lightweight_cleaning = True
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        os.remove('./tests/data/clean.png')
-        os.remove('./tests/data/clean.cleaned.png')
-
-    def test_jpg(self):
-        shutil.copy('./tests/data/dirty.jpg', './tests/data/clean.jpg')
-        p = images.JPGParser('./tests/data/clean.jpg')
-
-        meta = p.get_meta()
-        self.assertEqual(meta['Comment'], 'Created with GIMP')
-
-        p.lightweight_cleaning = True
-        ret = p.remove_all()
-        self.assertTrue(ret)
-
-        p = images.JPGParser('./tests/data/clean.cleaned.jpg')
-        self.assertEqual(p.get_meta(), {})
-
-        os.remove('./tests/data/clean.jpg')
-        os.remove('./tests/data/clean.cleaned.jpg')

tests/test_lightweight_cleaning.py

@@ -0,0 +1,76 @@
+#!/usr/bin/env python3
+
+import unittest
+import shutil
+import os
+
+from libmat2 import pdf, images, torrent
+
+
+class TestLightWeightCleaning(unittest.TestCase):
+    data = [{
+            'name': 'pdf',
+            'parser': pdf.PDFParser,
+            'meta': {'producer': 'pdfTeX-1.40.14'},
+            'expected_meta': {'creation-date': -1, 'format': 'PDF-1.5', 'mod-date': -1},
+            }, {
+                'name': 'png',
+                'parser': images.PNGParser,
+                'meta': {'Comment': 'This is a comment, be careful!'},
+                'expected_meta': {},
+            }, {
+                'name': 'jpg',
+                'parser': images.JPGParser,
+                'meta': {'Comment': 'Created with GIMP'},
+                'expected_meta': {},
+            }, {
+                'name': 'torrent',
+                'parser': torrent.TorrentParser,
+                'meta': {'created by': b'mktorrent 1.0'},
+                'expected_meta': {},
+            },{
+                'name': 'tiff',
+                'parser': images.TiffParser,
+                'meta': {'ImageDescription': 'OLYMPUS DIGITAL CAMERA         '},
+                'expected_meta': {
+                    'Orientation': 'Horizontal (normal)',
+                    'ResolutionUnit': 'inches',
+                    'XResolution': 72,
+                    'YResolution': 72
+                    }
+            },
+        ]
+
+    def test_all(self):
+        for case in self.data:
+            target = './tests/data/clean.' + case['name']
+            shutil.copy('./tests/data/dirty.' + case['name'], target)
+            p1 = case['parser'](target)
+
+            meta = p1.get_meta()
+            for k, v in case['meta'].items():
+                self.assertEqual(meta[k], v)
+
+            p1.lightweight_cleaning = True
+            self.assertTrue(p1.remove_all())
+
+            p2 = case['parser'](p1.output_filename)
+            self.assertEqual(p2.get_meta(), case['expected_meta'])
+
+            os.remove(target)
+            os.remove(p1.output_filename)
+
+    def test_exiftool_overwrite(self):
+        target = './tests/data/clean.png'
+        shutil.copy('./tests/data/dirty.png', target)
+
+        p1 = images.PNGParser(target)
+        p1.lightweight_cleaning = True
+        shutil.copy('./tests/data/dirty.png', p1.output_filename)
+        self.assertTrue(p1.remove_all())
+
+        p2 = images.PNGParser(p1.output_filename)
+        self.assertEqual(p2.get_meta(), {})
+
+        os.remove(target)
+        os.remove(p1.output_filename)

tests/test_policy.py

@@ -7,25 +7,26 @@ import os
 from libmat2 import office, UnknownMemberPolicy
 
 class TestPolicy(unittest.TestCase):
+    target = './tests/data/clean.docx'
+
     def test_policy_omit(self):
-        shutil.copy('./tests/data/embedded.docx', './tests/data/clean.docx')
-        p = office.MSOfficeParser('./tests/data/clean.docx')
+        shutil.copy('./tests/data/embedded.docx', self.target)
+        p = office.MSOfficeParser(self.target)
         p.unknown_member_policy = UnknownMemberPolicy.OMIT
         self.assertTrue(p.remove_all())
-        os.remove('./tests/data/clean.docx')
-        os.remove('./tests/data/clean.cleaned.docx')
+        os.remove(p.filename)
 
     def test_policy_keep(self):
-        shutil.copy('./tests/data/embedded.docx', './tests/data/clean.docx')
-        p = office.MSOfficeParser('./tests/data/clean.docx')
+        shutil.copy('./tests/data/embedded.docx', self.target)
+        p = office.MSOfficeParser(self.target)
         p.unknown_member_policy = UnknownMemberPolicy.KEEP
         self.assertTrue(p.remove_all())
-        os.remove('./tests/data/clean.docx')
-        os.remove('./tests/data/clean.cleaned.docx')
+        os.remove(p.filename)
+        os.remove(p.output_filename)
 
     def test_policy_unknown(self):
-        shutil.copy('./tests/data/embedded.docx', './tests/data/clean.docx')
-        p = office.MSOfficeParser('./tests/data/clean.docx')
+        shutil.copy('./tests/data/embedded.docx', self.target)
+        p = office.MSOfficeParser(self.target)
         with self.assertRaises(ValueError):
             p.unknown_member_policy = UnknownMemberPolicy('unknown_policy_name_totally_invalid')
-        os.remove('./tests/data/clean.docx')
+        os.remove(p.filename)

utils/ci/codespell/ignored_words.txt

@@ -0,0 +1,3 @@
+# Words to be ignored by codespell.
+# Put one word per line and sort alphabetically.
+process'