Because of the mechanism for catching exceptions in pubs, the
UnexpectedInput exception raised by FakeInput never reached the catch
statement in the CommandTestCase and raised a FakeSystemExit instead.
This commit monkey-patches the exception handler in the ui at the same
time as the patching of the input functions to ignore UnexptectedInput
at the ui level.
This involved many changes, some side effects of the change include:
- remove of all `u"abc"` forms, in favor of
`from __future__ import unicode_literals`. Their usage was
inconsistent anyway, leading to problems when mixing with
unicode content.
- improve the tests, to allow printing for usecase even when
crashing. Should make future test easier. This is done with a
rather hacky `StdIO` class in `p3`, but it works.
- for some reason, the skipped test for Python 2 seems to work
now. While the previous point might seem related, it is not clear
that this is actually the case.
For now, the editor_input and edit_file functions do not make sense
outside of the context of an ui. Having them in the content module
was confusing. (See issue fixed by a600855.)
Replace the in-house stubbing by the facilities offered
by pyfakefs for running tests. We don't use `io.open`
anymore, because pyfakefs does not stub it correctly.
We rely on a patched version of pyfakefs, as a few bug
remain in the official repo.
The implementation is designed to be robust to filesystems having integers
or nanosecond stat time.
Added a requirements.txt file for the tests:
`pip install -r tests/requirements.txt` to install the necessary packages
to run the tests.
Fix#13, Fix#14
1. Removed the unicode wrapper around fake_env io
2. Overrided pyfakefs cStringIO with StringIO (Python 2.x)
3. copy_dir will copy file in binary file to the fake filesystem
*unless* they are bib, yaml, txt or md files
4. ui.handle_exception() conserve traceback
Numerous unicode subtelties required diving into pyfakefs and configobj
source code. Unicode on Python 2.x is the worst.
The error was due to python 2 and 3 not playing nice when opening files.
This is probably linked to some botched unicode handling on our part somewhere
else is the package, and needs to be addressed when the `configs` module is
rewritten.
Also addresses some other minor unicode bugs.
Fix#27.
Related #18.
First this brings a change to the paper API: the terminology of bibdata
and bibentry is clarified: now bibentry is a dictionary of the form
{citekey: bibdata} where bibdata corresponds to the actual dictionary of
bibliographic fields and values {author: ..., year: ...}.
Now bibentry is an attribute of the paper object that is generated from
citekey and bibdata.
This commit fixes all tests in particular an issue with citekey in
bibentry not updated.
Also removes prints in tests and deprecated assertEquals.
Usecase tests now fail if the command ask for unexpected inputs.
Removes queries for user input in attach and add commands (for deletion
of a copied document file). The input was not coherent with tests and is
annoying.
Not so clean since trying to change stdout encoding requires accessing
sys.stdout.buffer, so fake_env has to mock this layer also. The basic
differences between p2 and p3 are handled in p3.py.
- Update fake_env for compatibility with io.
- Uniform open (through _my_open in content).
- Fix read issue for config (although still using SafeConfigParser that
is deprecated in py3).
Originally intended to fix a bug in edit openning fils with non-ascii
characters.
Now all data is assumed to be manipulated as unicode. Therefore all
values returned by functions from content are unicode. There are a few
exception in order to download non-unicode data without failing to
decode. These exception are marked by the 'byte_' prefix.
The io package is used instead of builtin open for all file
transactions.
The fake_env test helper has to be modified (hacked, to be honnest) since
fake_filesystem does not offer a mock of io.
This is still WIP. Two issues still have to be solved:
- first there is an UnicodeWarning raised by bibparser,
- also config is still directly using builtin open.
Paths should be normalized through content.system_path before any actual
operation on files. This is in particular taken care of in content
functions check_file, check_dir, write_file, and read_file.
* only bibtex format is supported
* all tests except test_repo.py and edit test pass
* edit and update commands were not updated
* removed --format argument from export, only bibtex is supported.
