--
* Connection.js: In some cases, the browser disconnect does not
happen inmediately (e.g. wrong extension for the app). I added a
delay of 25 ms to see if the connection was closed by the browser.
Also, I tried to make the checkConnection more readable.
--
* Connection.js
- Add some meaningful nativeMessaging feedback for failing
communication due to misconfiguration or other browser-originated
fails
- add an "isDisconnected" property
- "isNativeHostUnknown" tries to match browser's feedback string if
the browser does not find gpgme-json
* init.js
- initialization will now reject with a more meaningful error if the
configuration is not set up or other browser-based errors
(chrome.runtime.lastError) are present. This should speed up
the normal initialization (not having to waiting for a timeout
any more in case of improper setup)
* errors.js
- CONN_NATIVEMESSAGE: New error that passes the browser's
nativeMessaging error
- CONN_NO_CONFIG: native messaging error indicating that the
nativeMessaging host was not set up properly
* unittests.js:
- added the "isDisconnected" property to the startup tests
- added tests for proper behavior of connection checks
* lang/python/tests/Makefile.am (gpg.conf): Set a default key.
--
It seems we need to set a default key because at least t-sign.py does
not specify the key to use and we do not want to rely on the order of
keys in the keyring.
Also
Fixes-commit: f3ca2c9ce9
gpg 2.0 is end-of-life and we don't need these extra options anymore.
In fact they for the use of some gpg version and don't use the version
gpgconf knows about. This also aligns the python tests to what we use
for the C test suite.
Signed-off-by: Werner Koch <wk@gnupg.org>
* configure.ac: Configure new Makefiles.
* lang/python/Makefile.am: Remove dirs from extra dist and use
subdirs.
* lang/python/examples/Makefile.am, lang/python/src/Makefile.am,
lang/python/doc/Makefile.am: New. Files that list EXTRA_DIST files.
--
This is similar to what lang/js does by explicitly listing the
files. This ensures that we have clean distribution tarballs
without accidentall additions that just lay in the directory.
GnuPG-Bug-Id: T4481
* src/core.py (decrypt): filter out signatures with errors from the
returned verify_result, but avoid raising BadSignatures
* tests/t-decrypt-verify.py: ensure that only a single signature is
returned when evaluating cipher-3.asc, since the other signature is
unknown.
--
This change preserves the invariant that decrypt() only ever returns
valid signatures in the verify_result, but it avoids unnecessary
errors in the face of the presence of an additional bad signature.
GnuPG-bug-id: 4276
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* lang/python/tests/t-decrypt.py: test decryption of cipher-3.asc and
cipher-no-sig.asc
* lang/python/tests/t-decrypt-verify.py: test decryption and
verification of cipher-3.asc and cipher-no-sig.asc
--
note that this introduces a failed test -- decrypt-verify.py
misbehaves on cipher-3.asc by throwing a BadSignature even though
GnuPG-bug-id: 4276
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* lang/python/tests/Makefile.am: prefer py_tests from the environment
if present.
--
I'm trying to make it nicer/quicker to hack on the testsuite for
python bindings. With this change, if you're improving the python
bindings test suite, you can selectively run only a few specific tests
like so:
lang/python$ make check py_tests='t-decrypt.py t-decrypt-verify.py'
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* NEWS: Mention interface change.
* lang/cpp/src/gpggencardkeyinteractor.cpp
(GpgGenCardKeyInteractor::setAlgo): New.
(GpgGenCardKeyInteractor::action),
(GpgGenCardKeyInteractor::nextState: Handle new interface.
--
Tested that this workes with the old interface of GnuPG 2.2.5
and the new interface since GnuPG 2.2.6
GnuPG-Bug-Id: T4428
* lang/cpp/src/context.cpp (Error::hasSystemError): Invert logic to
do what it says.
--
This is safe because according to codesearch it is only used
in QGpgME
* lang/cpp/src/data.cpp (GpgME::Data::toKeys): Rewind afterards.
--
This fixes unexpected behavior that the seek pointer is changed
after calling the const toKeys.
* A rather obvious variant of the existing key import examples, except
directed at Mailvelope's keyserver.
* Yeah, Werner, I know ... but it exists because I used it and there's
no harm in sharing.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Version bump in preparation for whenever GPGME 1.13.0 happens.
* Ran the post_installer.py for docs preparation again.
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Removed auto-generated .texi files from doc/src/ so only the
corrected versions are left.
* Which means now it is complete, but with the initial work to expand
it with info file generation later.
* Moved post_installer.py into the examples/howto/ directory.
* Added instructions for its use to the Python Bindings HOWTO.
* Ran it as intended from the lang/python/ directory in order to both
prove it works and quickly and easily get the updated howto
replicated. Also to fix all those .texi files.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Fixed inter-edit.py so it will actually work now.
* made 3 others executable.
* Fixed the semantics of assuan.py's instructions.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Added some EPUB specific config options to the Sphinx config file
which might help reduce some of Sphinx's more stupid default errors
with EPUB validation.
* Added lang/python/post_installer.py script for automating the
generation of .texi and .rst "source" files from the real source
files written in Org mode. Includes recreating the Sphinx Makefile
which is excluded due to the m4 toolchain in parent directories, it
also handles the rewriting of the reST index file properly and
rewrites the .texi files so they don't impale themselves on Unicode.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Found a bug in org-mode's export to texinfo function which will
require either manual modification of each file or a customs sed run
over the generated files for all updates.
* Manually updated the current files for now, but will need to add
some post-install processing scripts for future use (I already have
some of these for my specific setup, they just need to be made a
little more generic and platform independent for here).
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Dropped the ASCII copyright line, since even MS have conceded their
battle against Unicode enough to load UTF-8.
* Fixed the drafts section since there will be less need of multiple
output format testing from next year.
* Tightening up both the documentation and some of the example code.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Fixed a bug in the ProtonMail importers (pmkey-*.py) where multiple
keys found for a username would always result in the last email
address checked being returned in the printed output for all located
keys.
* Though Keybase really should not be encouraged due to
disengenuosness and FUD emanating from that souce, this new script
will obtain a key hosted on that site and import it when supplied
with the keybase username.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Added small requirements.txt file for additional modules not in the
standard python library and not including the bindings themselves
and not including Cython, which is for more advanced examples.
* Tweaked it slightly to avoid repetition of key searches when there
is only one search pattern to check (i.e. usually a single key ID or
fingerprint).
* Tweaked the code again so that it can also handle the cases where
someone has included a hexadecimal string in their user ID.
* Updated the HOWTO to match.
* Exported to .rst and .texi.
* Fixed the logic used to search for any given pattern.
* Added a sensible method of checking whether a pattern is a key ID or
fingerprint.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* Multiple updates, expanding on the Windows installation issues.
* Also adding to the new maintenance mode reference document.
* Includes content relating to the resolution of T4271 and T4191.
Tested-by: Ben McGinnes <ben@adversary.org>
Signed-off-by: Ben McGinnes <ben@adversary.org>
* lang/python/src/core.py (Context.decrypt): simplify and clarify the
logic behind handling verify=False.
* lang/python/tests/t-decrypt.py: ensure that we test verify=False
--
The function-internal variables were pretty unclear to the reader, and
the logic caused pretty nasty breakage when verify=False.
GnuPG-Bug-Id: 4271
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* lang/python/src/core.py (Context.decrypt): document odd
error-handling behavior as a potential problem to be addressed.
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* lang/python/src/core.py (Context.decrypt): docstring clarification
of what it means to pass an empty list to the verify argument.
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
Both of these function-internal variables are never used for anything
other than a binary state. Implement them as the booleans they are.
Otherwise, casual readers of the code might think that they're
supposed to represent something other than a flag (e.g. "verify_sigs"
could mean "the signatures to verify", and "sink_result" could mean
"the place where we sink the result").
Signed-Off-By: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
It's easy to miss that verify can take a list of keys. Make it more
obvious to the average python dev who reads docstrings.
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
In the course of trying to address https://dev.gnupg.org/T4271, i
discovered that gpg.Context.decrypt() has a bit of superfluous code.
This changeset is intended to simplify the code without making any
functional changes.
Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
* added warning that he Windows C runtime issues will also affect use
with Cython and that relying on binary installers would remove that
possible use case.
