* lang/qt/tests/Makefile.am (AM_CPPFLAGS): Include cpp before
gpgme src directory.
--
This fixes a problem where context.h would be picked up from gpgme/src
instead of including the context.h from gpgme++.
* lang/qt/src/threadedjobmixin.h: Revert using full path
for context.h
--
This reverts commit 47bfbc9026 as
it causes problems depending on the include path. The proper
fix will be to ensure that cpp/src is included before gpgme/src.
* lang/qt/src/threadedjobmixin.h: When building qgpgme look for
context.h in the full cpp subdirectory.
--
Gpgme core also contains a context.h that can lead to confusion
otherwise.
* src/lang/qt/tests/t-tofuinfo.cpp(testTofuSignCount): New.
(initTestCase): Set gpg-agent loopback pinentry config.
(signAndVerify): Helper for tofuTestSignCount.
--
Also needs the wait code because of GnuPG-Bug-Id: 2405
* lang/qt/tests/t-tofuinfo.cpp: New.
* lang/qt/tests/Makefile.am: Update accordingly.
--
The test currently contains a workaround for GnuPG-Bug-Id 2405
* lang/cpp/src/editinteractor.cpp (status_to_error): No error
for KEYEXPIRED.
--
As keyexpired status is sent even if a subkey is expired
we can not treat it as a global error.
Results returned by the GPGME are fragile, i.e. they are only valid
until the next operation is performed in the context.
We cannot arbitrarily constrain the lifetime of Python objects, we
therefore create deep copies of the results.
* lang/python/gpgme.i (gpgme_tofu_info_t): Turn these into a list.
(gpgme_*_result_t): Create deep copies of these objects.
* lang/python/helpers.c (pygpgme_wrap_fragile_result): New function.
* lang/python/helpers.h (pygpgme_wrap_fragile_result): New prototype.
* lang/python/pyme/results.py: New file.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/Makefile.am (gpgme_wrap.c): Use '-builtin' to make SWIG
generate builtin types for c types.
* lang/python/gpgme.i (pygpgme_wrap_gpgme_data_t): Adapt slightly.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/pyme/core.py: Rename '_getctype' to '_ctype' and turn it
into a string. Likewise rename '_getnameprepend' to '_cprefix'.
* lang/python/helpers.c: Adapt accordingly.
Signed-off-by: Justus Winter <justus@g10code.com>
Simplify how the lifetime of callback arguments is managed.
* lang/python/gpgme.i (gpgme_edit_cb_t): Check arguments.
(PyObject_p_p, void_p_p): Drop rather dangerous interface.
(pygpgme_unwrap_gpgme_ctx_t): New function.
* lang/python/helpers.c (pygpgme_clear_generic_cb): Drop dangerous
function.
(pyPassphraseCb): Assert contract.
(pygpgme_set_passphrase_cb): Use Python's calling convention so that
we can raise exceptions. Hand in 'self', get the wrapped object, and
simply store the hook data as attribute of the wrapper object.
(pyProgressCb, pygpgme_set_progress_cb): Likewise.
(pygpgme_set_status_cb): Likewise.
(pygpgme_data_new_from_cbs): Likewise.
* lang/python/helpers.h (pygpgme_clear_generic_cb): Drop prototype.
(pygpgme_set_passphrase_cb): Update prototype.
(pygpgme_set_progress_cb): Likewise.
(pygpgme_set_status_cb): Likewise.
(pygpgme_data_new_from_cbs): Likewise.
(pygpgme_unwrap_gpgme_ctx_t): New prottotype.
* lang/python/pyme/core.py (Context, Data): Update callsites.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/Makefile.am: Add the toplevel source directory to CFLAGS
when compiling the bindings so that we can use private header files.
* lang/python/gpgme.i (gpgme_data_t): Rework the object wrapping. Do
not create a Python wrapper object, merely a gpgme_data_t object, and
keep references to buffer objects, if any. If necessary, update the
buffer after the function call.
(pygpgme_wrap_gpgme_data_t): New function.
* lang/python/helpers.c (object_to_gpgme_data_t): Rework object
wrapping. Also wrap objects implementing the buffer protocol.
* lang/python/helpers.h (object_to_gpgme_data_t): Update prototype.
(pygpgme_wrap_gpgme_data_t): New prototype.
* lang/python/tests/t-idiomatic.py: Demonstrate this.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/Makefile.am: Copy the README file.
* lang/python/README: Rename, convert to org, and update.
* lang/python/pyme/__init__.py: Move license out of the docstring,
update docstring.
* lang/python/pyme/core.py: Add and update docstrings.
Signed-off-by: Justus Winter <justus@g10code.com>
* configure.ac: Generate 'setup.py' and 'version.py'.
* lang/python/Makefile.am: Use generated setup script.
* lang/python/pyme/version.py: Turn it into a template, and get
version information from the build system. Also drop some variables.
* lang/python/setup.py: Likewise. This way we can avoid importing the
version module, which is frowned upon and actually caused a problem.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/pyme/core.py (Context.__del__): Make function
idemptotent.
(Context.{__enter__,__exit__}): Implement the context manager
protocol.
(Data.__del__): Make function idemptotent, drop debug print.
(Data.{__enter__,__exit__}): Implement the context manager
protocol.
* lang/python/tests/t-idiomatic.py: Demonstrate this.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/gpgme.i: Fix freeing an uninitialized pointer in the
error handling of generated wrapper functions by explicitly storing
the pointer in a local variable which can be initialized.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/gpgme.i (gpgme_data_t): Use new function to create
wrapper objects if necessary, and deallocate them after the function
call.
* lang/python/helpers.c (object_to_gpgme_data_t): New function.
* lang/python/helpers.h (object_to_gpgme_data_t): New prototype.
* lang/python/tests/Makefile.am (pytests): Add new test.
* lang/python/tests/t-idiomatic.py: New file.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/python/Makefile.am: Pass CFLAGS to python build system.
* lang/python/helpers.c (pyPassphraseCb): Use correct type for length.
(pygpgme_data_new_from_cbs): Drop unused variable.
Signed-off-by: Justus Winter <justus@g10code.com>
* lang/cpp/src/editinteractor.cpp (EditInteractor::needsNoResponse):
Use whitelist instead of blacklist.
--
This should be more robust when new status messages are added.
The whitelist is the same GPA uses. Fixes Qt's t-ownertrust.
instead of calling gpg-error-config.
This is useful when configuring with --with-gpgerror-prefix
because then GPG_ERROR_CONFIG contains the correct
/path/to/bin/gpg-error-config whereas calling gpg-error-config
directly would look in the user's PATH (and not in the prefix
where gpg-error-config was installed).
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/python/pyme/core.py (Data._error_check): Add
'gpgme_data_get_file_name' to the list of functions not returning an
error code.
* lang/python/tests/Makefile.am (pytests): Add new tests.
* lang/python/tests/support.py (verbose): New variable.
* lang/python/tests/t-data.py: Test setting and getting the filename.
* lang/python/tests/t-encrypt-large.py: New file.
* lang/python/tests/t-file-name.py: Likewise.
* lang/python/tests/t-trustlist.py: Likewise.
Signed-off-by: Justus Winter <justus@gnupg.org>
Handing a reference to the wrapper object created a non-trivial
circular reference that Pythons garbage collector is unable to break.
Explicitly break it by using a weak reference.
* lang/python/helpers.c (pygpgme_stash_callback_exception): Retrieve
object from weak reference.
* lang/python/pyme/core.py (Context.__del__): Free status callback.
(Context.set_passphrase_cb): Use a weak reference.
(Context.set_progress_cb): Likewise.
(Context.set_status_cb): Likewise.
(Context.op_edit): Likewise.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/python/pyme/core.py (Data.__init__): Add 'copy' kwargument, and
pass it to functions supporting it. PEP8 fix.
(Data.new_from_fd): PEP8 fix.
(Data.new_from_file): Give a more helpful error message if copy is
False. PEP8 fix.
(Data.new_from_fd): Hand the file descriptor to
'gpgme_data_new_from_fd', not a stream. Fix docstring.
* lang/python/tests/t-data.py: Add tests for this.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/python/pyme/core.py (Context.wait): Improve docstring. As the
context passed to 'gpgme_wait' is never NULL, it is pointless to look
at the returned context. Always raise exceptions.
* lang/python/tests/Makefile.am (pytests): Add new test.
* lang/python/tests/t-wait.py: New file.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/qt/tests/t-ownertrust.cpp: New test.
* lang/qt/tests/Makefile.am: Add test.
--
This test would have catched the issue created by not handling
GPGME_STATUS_KEY_CONSIDERED.
* lang/python/helpers.c (pyPassphraseCb): Cope with 'uid_hint' being
NULL, convert it to an Unicode object, and cope with the callback
returning both Unicode and bytes objects.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/qt/tests/Makefile.am (AM_CPPFLAGS): Add -DBUILDING_QGPGME.
--
Without building QGpgME the headers of gpgme++ are not included
corretly and taken from an already installed gpgme++.
* lang/cpp/src/context.cpp,
lang/cpp/src/context.h,
lang/cpp/src/context_p.h (Context::createForEngine),
(Context::edit, Context::startEditing),
(Context::takeLastEditInteractor, Context::cardEdit),
(Context::startCardEditing, Context::takeLastCardEditInteractor),
(Context::assuanTransact, Context::startAssuanTransaction),
(Context::takeLastAssuanTransaction): Port to unique_ptr.
* lang/qt/src/qgpgmeadduseridjob.cpp,
lang/qt/src/qgpgmechangeexpiryjob.cpp,
lang/qt/src/qgpgmechangeownertrustjob.cpp,
lang/qt/src/qgpgmechangepasswdjob.cpp,
lang/qt/src/qgpgmesignkeyjob.cpp: Update accordingly.
--
This is another API break but as we already broke API with GpgME++
and QGpgME from KDE Frameworks this is a good time to do this
to avoid using a deprected C++ class in the API.
* lang/python/Makefile.am: Add bits from the c test suite.
* lang/python/support.py: New file.
* lang/python/t-decrypt.py: Likewise.
* lang/python/t-encrypt.py: Likewise.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/python/pyme/core.py: Avoid races at interpreter shutdown. This
silences the most annoying occurrences, however this problem also
affects the SWIG generated code, which might indicate that the real
problem is somewhere else. If so, this change can be easily reverted.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/python/gpgme.i: Adjust to Python3's string type being
'Unicode', not 'bytes'. Fix type checking.
* lang/python/core.py (Data.write): Add docstring mentioning the
expected type of parameter 'buffer'.
(Data.read): Adjust read loop. Also, use a saner chunk size, and join
all chunks at the end instead of adding them.
* lang/python/examples/simple.py: Adjust example.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/cpp/src/Makefile.am (AM_CPPFLAGS): Add gpgme.h location.
--
This fixes the build in case an incompatible gpgme.h is also
installed in other include locations.
* configure.ac: Make Python bindings configurable, add new Makefile.
* lang/python/Makefile.am: New file.
* lang/python/setup.py: Integrate into the build system.
* m4/ax_pkg_swig.m4: New file from the autoconf archive.
* m4/m4_ax_swig_python.m4: Likewise.
Signed-off-by: Justus Winter <justus@gnupg.org>
* lang/qt/src/Makefile.am (qgpgme_headers): Add protocol.h
(private_qgpgme_headers): Add protocol_p.h
* lang/qt/src/protocol.h: New. From QGpgMEBackend.
* lang/qt/src/protocol_p.h: New. From QGpgMEBackend.
* lang/qt/src/qgpgmebackend.h,
lang/qt/src/qgpgmebackend.cpp (Protocol): Removed.
--
The backend class does not make much sense anymore as we
only have the GpgME backend obviously. It's purpose was
for Libkleo's Backend abstraction.
* lang/qt/src/Makefile.am: Do not install all headers.
--
The qgpgme classes hide the implementation and thus make it
easier to change it without ABI breaks. They should not
be installed.
* lang/cpp/src/editinteractor.cpp (EditInteractor::needsNoResponse):
Add GPGME_STATUS_PINENTRY_LAUNCHED.
--
PINENTRY_LAUNCHED needs no response in any Job. This fixes
Jobs which errored out on unknown status lines.
