On Python 3, evaluating a Vim expression will raise a unicode exception
if it contains an invalid sequence of bytes for the current encoding.
We can't really do anything about it because this is the way Vim and
Python 3 interact. However, we can prevent this situation to occur by
not evaluating Vim data that we have no control over: in particular,
the Vim globals. This is done by:
- adding one by one the YCM default options instead of extending the
Vim globals with them;
- only evaluating the Vim global variable names (and not their values)
when building the YCM options for the ycmd server.
Display an error message to the user depending on the status code
returned by the ycmd server.
Remove ycm_core checks in plugin/youcompleteme.vim. These checks are
now done by the ycmd server.
Do not start a separate process to check the core version but rely on
ycmd returning a specific exit code. This slightly improves the Vim
startup time.
vim.eval returns a str() object on py2, but our internal strings are all unicode().
We use vimsupport.VimExpressionToPythonType to wrap the conversion complexities.
Do not send an "event_notification" request in OnFileReadyToParse
function if server process is terminated. Otherwise, it blocks Vim
for one second or results in a traceback each time the InsertLeave,
CursorMoved, CursorHold, and BufferVisit events are triggered.
On Windows and Python 2, the full exception message from IOError
in CheckFilename will contain the filepath formatted as a unicode
string. Since the filepath is already added in the RuntimeError
message, use the strerror attribute to only display the error.
Move PostComplete tests inside a class that defines setUp and tearDown
methods. Clean YCM object in tearDown method. This fixes the error
"OSError: [WinError 6] The handle is invalid" on Windows with
Python 3.5.
Python 3 is much stricter around mixing bytes with unicode (and by
"stricter," I mean it doesn't allow it at all) so we're making
vimsupport only return `unicode` objects (`str` on py3). The idea is
that YCM (and ycmd) internals only ever deal with unicode.
We simply apply the changes to each file in turn. The existing replacement
logic is unchanged, except that it now no longer implicitly assumes we are
talking about the current buffer.
If a buffer is not visible for the requested file name, we open it in
a horizontal split, make the edits, then hide the window. Because this
can cause UI flickering, and leave hidden, modified buffers around, we
issue a warning to the user stating the number of files for which we are
going to do this. We pop up the quickfix list at the end of applying
the edits to allow the user to see what we changed.
If the user opts to abort due to, say, the file being open in another
window, we simply raise an error and give up, as undoing the changes
is too complex to do programatically, but trivial to do manually in such
a rare case.
Now 'GoTo' and 'FixIt' commands don't need to start with those
prefixes. For 'FixIt' we can detect the response type by looking for
the 'fixits' entry in the response.
For 'GoTo' this is a touch harder, as there is no completely obvious
way to tell. However it is unique in this respect, so we can simply
fall back to it.
Completers returning other types of response are not supported by
this client.
Moved File parse request handling and diagnostic extraction flow into
python to simplify flow and allow easier addition of new parse request
handlers such as semantic highlighter.
Refactored base_test to patch separate vimsupport functions instead of
the whole module, and interfering the test results afterwards.
Added new tests for diagnostic sign place/unplace and error/warning
count extraction API.
[READY] Fix issue in EventNotification tests
While I was reviewing PR #1905, I found an issue with the `EventNotification` tests. It's easy to reproduce. You just need to comment [this line in `python/ycm/youcompleteme.py`](https://github.com/Valloric/YouCompleteMe/blob/master/python/ycm/youcompleteme.py#L508) and run the tests. With this change, the `EventNotification` tests should fail since the second call of `ValidateParseRequest` re-raises the warning. However, tests are still passing because `assert_has_calls` does not check if a call was not made. For example, if `functionA` is called twice, both `assert_has_calls( [ functionA ] )` and `assert_has_calls( [ functionA, functionA ] )` are successful.
To fix this, we just need to check the number of calls using `call_count`. This is done by creating a subclass of `MagicMock` implementing the `assert_has_exact_calls` method and using it in tests.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/valloric/youcompleteme/1927)
<!-- Reviewable:end -->
[READY] Implement new strategy to find the Python interpreter path
See discussion in issue #1891 for details.
Implement a new strategy to find the Python interpreter path:
- if specified, use `g:ycm_path_to_python_interpreter` option.
- on UNIX platforms, use `sys.executable` as the path to Python interpreter;
- on Windows, deduce it from `os.__file__` path (it should always be in the parent folder).
In all cases, check the version (2.6 or 2.7) of the Python interpreter path by running it.
This PR may break things. It needs thorough testing.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/valloric/youcompleteme/1896)
<!-- Reviewable:end -->
assert_has_calls is not enough to check if a call didn't raise a
warning. We also need to check the number of calls. This is done by
creating a subclass of MagicMock implementing the
assert_has_exact_calls method.
OmniCompletionRequest is missing the RawResponse method, so any attempt to call
it calls the base class method instead. However, since the data structures of
this class and base class are different, this causes an error.
