GitLab

This is followup of the commit 72372764. The extraction to root can still
lead to crashes because the `parent` variable is `NULL` and is consequently
passed into the `g_object_unref` function. Let's finally fix this corner
case properly.

Almost all error messages are currently prefixed by an archive
basename. This is redundant. The caller knows what file is being
extracted. Let's drop the basename from all the error messages...

!34

Add comment explaining why password is requested for ZIP only to make
that obvious.

!34

Ignacy Kuchciński authored Jan 27, 2022 and Ondrej Holy committed Feb 07, 2022

Currently, when trying to extract password protected 7z archives that
don't have encrypted file lists, password prompt will be shown, even
though the extraction itself will fail.

To fix this situation, ensure that the archive format is supported ZIP
before showing the password prompt.

Ignacy Kuchciński authored Jan 27, 2022 and Ondrej Holy committed Feb 07, 2022

Currently, when trying to extract password protected 7z/RAR archives
that don't have encrypted file lists, no error will be reported, even
though the extraction itself fails creating empty files.

To fix this situation, propagate libarchive errors by checking if
archive_read_data_block return value is different from ARCHIVE_OK.

Currently, when trying to extract password protected 7z/RAR archives
with encrypted file lists, the reported error message is "empty
archive".

This message is not helpful and may be confusing for users.

To fix this situation, propagate libarchive errors by swapping the
"empty archive" and "autoar_common_g_error_new_a" statements.

Fixes #26

When building Fedora package, I see many "may be used uninitialzed"
warnings. I am not sure why but I don't see those warnings locally even
when `-Wmaybe-uninitialized` is used. Anyway, this is because the
`g_autofree` and `g_autoptr` variables are not explicitely initialized
as mentioned in the documentation. Let's initialize them to get rid of
those warnings.

The raw format archives are not tested currently. Let's two test cass
to reduce change that their support will be broken again.

Relates: #38

An extraction to the root of an archive which contains the "/" path
leads to crashes. Let's handle this rare corner case.

Relates: #38

An extraction of raw format archives leads to crashes currently.
This is because the generic "data" string is returned from libarchive
instead of the real pathname, which is not expected currently. Let's
handle this case properly and fallback to the source basename.

Fixes: #38

Currently, various tools fail to remove the read-only directory, which
is created as an output from the test suite. This for example breaks
package building when tests are enabled. Let's make it writable again
when test is done to fix the issue.

Fixes: #34

The "The file is not a directory" error is returned in case the archive has
common prefix, which have not been changed over the `decide-destination`
signal and the destination is a symlink. This is regression caused by the
commit b9590ab7. It causes that the security checks, which was added by the
commit 8109c368, verifies also the destination directory by mistake. Let's
partially rework the problematic commit to avoid this regression.

Fixes: #36

Currently, a misleading error appears when the passphrase prompt in Nautilus
is cancelled. This is because the `scanned` signal is emitted regardless the
`cancellable` state. This happens only for archives with one file. Let's
return immediately to prevent this problematic behavior.

Fixes: #35

This is follow up for commit ac21bd0c
as not all the build.meson files was actually rewritten.

Format strings got added in meson 0.58. This allows the meson
dependency to be downgraded to 0.56.

See https://mesonbuild.com/Syntax.html#string-formatting

In commit bcf17c9f the schema got
removed because it wasn't used. The meson.build file was still
referencing it.

The `filename` tag in enum files shows the whole file path. This has
been replaced with the `basename` tag just to show the file name.

Ondrej Holy authored Aug 05, 2021 and Iñigo Martínez committed Aug 09, 2021

The `test-extract-unit` suite expects that the source directory is the
same as the build directory. Consequently, it fails over `ninja test`
as it doesn't see the test files. Let's adapt the codes to use the
`G_TEST_SRCDIR`  so it can be used from the build dir.

The gschema file is not used so it has been removed.

To avoid the burden of maintaining multiple build systems, autotools
support has been removed.

meson is a build system focused on speed an ease of use, which
helps speeding up the software development.
Co-authored-by: Peter Keresztes Schmidt <carbenium@outlook.com>

The test utils doesn't support encrypted archives as it is not possible
to specify password currently. Let's simply add an password argument to
have a simple way to test this feature.

`AutoarExtractor` supports extraction of encrypted archives. It would be nice
to have support for encrypted archives also in `AutoarCompressor`. Though the
libarchive library doesn't provide a much. It supports encryption only for the
ZIP format. There is a choice between `zipcrypt`, `aes128` and `aes256`. Given
the phact that `zipcrypt` is insecure and not recommended, I think it is fine
to hardcode `aes256`. It is probably a bit slower then `aes128`, but more secure.
So the only thing which needs to be set is a passphrase. Let's add
`autoar_compressor_set_passphrase` to allow creating encrypted archives.

Relates: nautilus#822

Let's add a check for potential ABI breaks to not break ABI by mistake,
or to not forget to update the library version.

There are various problems with the rawhide image currently and I don't
know how to workaround them. Let's use fedora:latest (temporarily) to fix
our pipeline.

The meson package is not part of the CI image as Autotools are used
currently. Let's add meson as a preparation for the upcoming meson port.

António Fernandes authored Jul 15, 2021 and Ondrej Holy committed Jul 19, 2021

The AutoarCompressor object returns the format when asked for a filter.

Fix the obvious oversight from commit 2901c81d.

Currently, archive entries with absolute paths lose their paths when
extracting, or even cause crashes. The code expects that the
`autoar_extractor_do_sanitize_pathname` function will always return paths
inside, or equal to the destination directory, but this is not true. The
problem is that the `g_file_get_child` resp. `g_file_resolve_relative_path`
functions expect relative paths, but are used with absolute paths also.
Let's convert absolute paths to relative ones to ensure that they will
be exctracted as relative to the destination dir. Also, update the
corresponding test case to confirm this.

Fixes: #32

Matthias Clasen authored Feb 16, 2017 and Ondrej Holy committed Jun 14, 2021

Valgrind complained that the g_list_copy_deep call here leaks
memory. In fact, the set_property implementation already takes
care of copying the list and reffing the objects, so don't
do it here.

https://bugzilla.gnome.org/show_bug.cgi?id=778804

By default, the conflicting files are skipped and this default behavior is
also used by Nautilus and other clients. However, the `test-extract` util
overwrites the files instead. Let's honor the default behavior here as well
to make debugging easier.

The debug output doesn't contain info when conflict happens. This makes
hard to debug issues which are caused by them. Let's print one when
a conflict happens to make it obvious.

Currently, redundant path name handling happens when archive files have
a common prefix and an equal destination is returned from the `decide-destination`
signal. In this case, the old prefix is removed and the new one (the equal one)
is added again. Let's detect this case and prevent the redundancy.

Currently, an empty folder is created when extracting an archive over
Nautilus which contains one file with the same name as the archive.
This is because gnome-autoar always creates the top-level folder, but
the one file archive with the same name should be extracted directly
without creating additional folders. Otherwise the `conflict` signal
is emitted and Nautilus by default skips conflicting files, so only
empty folder remains after extraction. This problem have not been
unfortunately noticed by the `test-one-file-same-name` unit test due to
differences in the `decide-destination` signal handling. Let's do not
create the top-level folder at all as `g_file_make_directory_with_parents`
is used later anyway.

Fixes: #28