Scheduled weekly dependency update for week 38

[pyup-bot]

Update prompt-toolkit from 1.0.15 to 2.0.4.

Changelog

2.0.4

-----------------

Bug fixes:
- Fix render height for rendering full screen applications in Windows.
- Fix in `TextArea`. Set `accept_handler` to `None` if not given.
- Go to the beginning of the next line when enter is pressed in Vi navigation
mode, and the buffer doesn't have an accept handler.
- Fix the `default` argument of the `prompt` function when called multiple
times.
- Display decomposed multiwidth characters correctly.
- Accept `history` in `prompt()` function again.

Backwards incompatible changes:
- Renamed `PipeInput` to `PosixPipeInput`. Added `Win32PipeInput` and
`create_input_pipe`.
- Pass `buffer` argument to the `accept_handler` of `TextArea`.

New features:
- Added `accept_default` argument to `prompt()`.
- Make it easier to change the body/title of a Frame/Dialog.
- Added `DynamicContainer`.
- Added `merge_completers` for merging multiple completers together.
- Add vt100 data to key presses in Windows.
- Handle left/right key bindings in Vi block insert mode.

2.0.3

-----------------

Bug fixes:
- Fix in 'x' and 'X' Vi key bindings. Correctly handle line endings and args.
- Fixed off by one error in Vi line selection.
- Fixed bugs in Vi block selection. Correctly handle lines that the selection
doesn't cross.
- Python 2 bugfix. Handle str/unicode correctly.
- Handle option+left/right in iTerm.

2.0.2

-----------------

Bug fixes:
- Python 3.7 support: correctly handle StopIteration in asynchronous generator.
- Fixed off-by-one bug in Vi visual block mode.
- Bugfix in TabsProcessor: handle situations when the cursor is at the end of
the line.

2.0.1

-----------------

2.0

includes the merge of the CommandLineInterface and the Application object, a
rewrite of how user controls are focused, a rewrite of how event loops work
and the removal of the buffers dictionary. This introduces many backwards
incompatible changes, but the result is a very nice and powerful architecture.

Most architectural changes effect full screen applications. For applications
that use `prompt_toolkit.shortcuts` for simple prompts, there are fewer
incompatibilities.

Changes:

- No automatic translation from \r into \n during the input processing. These
are two different keys that can be handled independently. This is a big
backward-incompatibility, because the `Enter` key is `ControlM`, not
`ControlJ`. So, now that we stopped translating \r into \n, it could be that
custom key bindings for `Enter` don't work anymore. Make sure to bind
`Keys.Enter` instead of `Keys.ControlJ` for handling the `Enter` key.

- The `CommandLineInterface` and the `Application` classes are merged. First,
`CommandLineInterface` contained all the I/O objects (like the input, output
and event loop), while the `Application` contained everything else. There was
no practical reason to keep this separation. (`CommandLineInterface` was
mostly a proxy to `Application`.)

A consequence is that almost all code which used to receive a
`CommandLineInterface`, will now use an `Application`. Usually, where we
had an attribute `cli`, we'll now have an attribute `app`.

Secondly, the `Application` object is no longer passed around. The `get_app`
function can be used at any time to acquire the active application.

(For backwards-compatibility, we have aliases to the old names, whenever
possible.)

- prompt_toolkit no longer depends on Pygments, but it can still use Pygments
for its color schemes and lexers. In many places we used Pygments "Tokens",
this has been replaced by the concept of class names, somewhat similar to
HTML and CSS.

* `PygmentsStyle` and `PygmentsLexer` adaptors are available for
  plugging in Pygments styles and lexers.

* Wherever we had a list of `(Token, text)` tuples, we now have lists of
 `(style_string, text)` tuples. The style string can contain both inline
 styling as well as refer to a class from the style sheet. `PygmentsTokens`
 is an adaptor that converts a list of Pygments tokens into a list of
 `(style_string, text)` tuples.

- Changes in the `Style` classes.

* `style.from_dict` does not exist anymore. Instantiate the ``Style`` class
 directory to create a new style. ``Style.from_dict`` can be used to create
 a style from a dictionary, where the dictionary keys are a space separated
 list of class names, and the values, style strings (like before).

* `print_tokens` was renamed to `print_formatted_text`.

* In many places in the layout, we accept a parameter named `style`. All the
 styles from the layout hierarchy are combined to decide what style to be
 used.

* The ANSI color names were confusing and inconsistent with common naming
 conventions. This has been fixed, but aliases for the original names were
 kept.

- The way focusing works is different. Before it was always a `Buffer` that
was focused, and because of that, any visible `BufferControl` that contained
this `Buffer` would be focused. Now, any user control can be focused. All
of this is handled in the `Application.layout` object.

- The `buffers` dictionary (`CommandLineInterface.buffers`) does not exist
anymore. Further, `buffers` was a `BufferMapping` that keeps track of which
buffer has the focus. This significantly reduces the freedom for creating
complex applications. We wanted to move toward a layout that can be defined
as a (hierarchical) collection of user widgets. A user widget does not need
to have a `Buffer` underneath and any widget should be focusable.

* `layout.Layout` was introduced to contain the root layout widget and keep
 track of the focus.

- The key bindings were refactored. It became much more flexible to combine
sets of key bindings.

* `Registry` has been renamed to `KeyBindings`.
* The `add_binding` function has been renamed to simply `add`.
* Every `load_*` function returns one `KeyBindings` objects, instead of
 populating an existing one, like before.
* `ConditionalKeyBindings` was added. This can be used to enable/disable
 all the key bindings from a given `Registry`.
* A function named `merge_key_bindings` was added. This takes a list of
 `KeyBindings` and merges them into one.
* `key_binding.defaults.load_key_bindings` was added to load all the key
 bindings.
* `KeyBindingManager` has been removed completely.
* `input_processor` was renamed to `key_processor`.

Further:

* The `Key` class does not exist anymore. Every key is a string and it's
 considered fine to use string literals in the key bindings. This is more
 readable, but we still have run-time validation. The `Keys` enum still
 exist (for backwards-compatibility, but also to have an overview of which
 keys are supported.)
* 'enter' and 'tab' are key aliases for 'c-m' and 'c-i'.

- User controls can define key bindings, which are active when the user control
is focused.

* `UIControl` got a `get_key_bindings` (abstract) method.

- Changes in the layout engine:

* `LayoutDimension` was renamed to `Dimension`.
* `VSplit` and `HSplit` now take a `padding` argument.
* `VSplit` and `HSplit` now take an `align` argument.
 (TOP/CENTER/BOTTOM/JUSTIFY) or (LEFT/CENTER/RIGHT/JUSTIFY).
* `Float` now takes `allow_cover_cursor` and `attach_to_window` arguments.
* `Window` got an `WindowAlign` argument. This can be used for the alignment
 of the content. `TokenListControl` (renamed to `FormattedTextControl`) does
 not have an alignment argument anymore.
* All container objects, like `Window`, got a `style` argument. The style for
 parent containers propagate to child containers, but can be overriden.
 This is in particular useful for setting a background color.
* `FillControl` does not exist anymore. Use the `style` and `char` arguments
 of the `Window` class instead.
* `DummyControl` was added.
* The continuation function of `PromptMargin` now takes `line_number` and
 `is_soft_wrap` as input.

- Changes to `BufferControl`:

* The `InputProcessor` class has been refactored. The `apply_transformation`
 method should now takes a `TransformationInput` object as input.

* The text `(reverse-i-search)` is now displayed through a processor. (See
 the `shortcuts` module for an example of its usage.)

- `widgets` and `dialogs` modules:

* A small collection of widgets was added. These are more complex collections
 of user controls that are ready to embed in a layout. A `shortcuts.dialogs`
 module was added as a high level API for displaying input, confirmation and
 message dialogs.

* Every class that exposes a ``__pt_container__`` method (which is supposed
 to return a ``Container`` instance) is considered a widget. The
 ``to_container`` shortcut will call this method in situations where a
 ``Container`` object is expected. This avoids inheritance from other
 ``Container`` types, but also having to unpack the container object from
 the widget, in case we would have used composition.

* Warning: The API of the widgets module is not considered stable yet, and
 can change is the future, if needed.

- Changes to `Buffer`:

* A `Buffer` no longer takes an `accept_action`.  Both `AcceptAction` and
 `AbortAction` have been removed. Instead it takes an `accept_handler`.

- Changes regarding auto completion:

* The left and right arrows now work in the multi-column auto completion
 menu.
* By default, autocompletion is synchronous. The completer needs to be
 wrapped in `ThreadedCompleter` in order to get asynchronous autocompletion.
* When the completer runs in a background thread, completions will be
 displayed as soon as they are generated. This means that we don't have to
 wait for all the completions to be generated, before displaying the first
 one. The completion menus are updated as soon as new completions arrive.

- Changes regarding input validation:

* Added the `Validator.from_callable` class method for easy creation of
 new validators.

- Changes regarding the `History` classes:

* The `History` base class has a different interface. This was needed for
 asynchronous loading of the history. `ThreadedHistory` was added for this.

- Changes related to `shortcuts.prompt`:

* There is now a class `PromptSession` which also has a method `prompt`. Both
 the class and the method take about the same arguments. This can be used to
 create a session. Every `prompt` call of the same instance will reuse all
 the arguments given to the class itself.

 The input history is always shared during the entire session.

 Of course, it's still possible to call the global `prompt` function. This
 will create a new `PromptSession` every time when it's called.

* The `prompt` function now takes a `key_bindings` argument instead of
 `key_bindings_registry`. This should only contain the additional bindings.
 (The default bindings are always included.)

- Changes to the event loops:

* The event loop API is now closer to how asyncio works. A prompt_toolkit
 `Application` now has a `Future` object. Calling the `.run_async()` method
 creates and returns that `Future`. An event loop has a `run_until_complete`
 method that takes a future and runs the event loop until the Future is set.

 The idea is to be able to transition easily to asyncio when Python 2
 support can be dropped in the future.

* `Application` still has a method `run()` that underneath still runs the
 event loop until the `Future` is set and returns that result.

* The asyncio adaptors (like the asyncio event loop integration) now require
 Python 3.5. (We use the async/await syntax internally.)

* The `Input` and `Output` classes have some changes. (Not really important.)

* `Application.run_sub_applications` has been removed. The alternative is to
 call `run_coroutine_in_terminal` which returns a `Future`.

- Changes to the `filters` module:

* The `Application` is no longer passed around, so both `CLIFilter` and
 `SimpleFilter` were merged into `Filter`. `to_cli_filter` and
 `to_simple_filter` became `to_filter`.

* All filters have been turned into functions. For instance, `IsDone`
 became `is_done` and `HasCompletions` became `has_completions`.

 This was done because almost all classes were called without any arguments
 in the `__init__` causing additional braces everywhere. This means that
 `HasCompletions()` has to be replaced by `has_completions` (without
 parenthesis).

 The few filters that took arguments as input, became functions, but still
 have to be called with the given arguments.

 For new filters, it is recommended to use the `Condition` decorator,
 rather then inheriting from `Filter`.

- Other renames:

* `IncrementalSearchDirection` was renamed to `SearchDirection`.
* The `use_alternate_screen` parameter has been renamed to `full_screen`.
* `Buffer.initial_document` was renamed to `Buffer.document`.
* `TokenListControl` has been renamed to `FormattedTextControl`.
* `Application.set_return_value` has been renamed to `Application.set_result`.

- Other new features:

* `DummyAutoSuggest` and `DynamicAutoSuggest` were added.
* `DummyClipboard` and `DynamicClipboard` were added.
* `DummyCompleter` and `DynamicCompleter` were added.
* `DummyHistory` and `DynamicHistory` was added.

* `to_container` and `to_window` utilities were added.

Links

• PyPI: https://pypi.org/project/prompt-toolkit
• Changelog: https://pyup.io/changelogs/prompt-toolkit/
• Repo: https://github.com/jonathanslenders/python-prompt-toolkit

Update pycodestyle from 2.3.1 to 2.4.0.

Changelog

2.4.0

------------------

New checks:

* Add W504 warning for checking that a break doesn't happen after a binary
operator. This check is ignored by default. PR 502.
* Add W605 warning for invalid escape sequences in string literals. PR 676.
* Add W606 warning for 'async' and 'await' reserved keywords being introduced
in Python 3.7. PR 684.
* Add E252 error for missing whitespace around equal sign in type annotated
function arguments with defaults values. PR 717.

Changes:

* An internal bisect search has replaced a linear search in order to improve
efficiency. PR 648.
* pycodestyle now uses PyPI trove classifiers in order to document supported
python versions on PyPI. PR 654.
* 'setup.cfg' '[wheel]' section has been renamed to '[bdist_wheel]', as
the former is legacy. PR 653.
* pycodestyle now handles very long lines much more efficiently for python
3.2+. Fixes 643. PR 644.
* You can now write 'pycodestyle.StyleGuide(verbose=True)' instead of
'pycodestyle.StyleGuide(verbose=True, paths=['-v'])' in order to achieve
verbosity. PR 663.
* The distribution of pycodestyle now includes the license text in order to
comply with open source licenses which require this. PR 694.
* 'maximum_line_length' now ignores shebang ('!') lines. PR 736.
* Add configuration option for the allowed number of blank lines. It is
implemented as a top level dictionary which can be easily overwritten. Fixes
732. PR 733.

Bugs:

* Prevent a 'DeprecationWarning', and a 'SyntaxError' in future python, caused
by an invalid escape sequence. PR 625.
* Correctly report E501 when the first line of a docstring is too long.
Resolves 622. PR 630.
* Support variable annotation when variable start by a keyword, such as class
variable type annotations in python 3.6. PR 640.
* pycodestyle internals have been changed in order to allow 'python3 -m
cProfile' to report correct metrics. PR 647.
* Fix a spelling mistake in the description of E722. PR 697.
* 'pycodestyle --diff' now does not break if your 'gitconfig' enables
'mnemonicprefix'. PR 706.

Links

• PyPI: https://pypi.org/project/pycodestyle
• Changelog: https://pyup.io/changelogs/pycodestyle/
• Docs: https://pycodestyle.readthedocs.io/

Update pyflakes from 1.6.0 to 2.0.0.

Changelog

2.0.0

- Drop support for EOL Python <2.7 and 3.2-3.3
- Check for unused exception binding in `except:` block
- Handle string literal type annotations
- Ignore redefinitions of `_`, unless originally defined by import
- Support `__class__` without `self` in Python 3
- Issue an error for `raise NotImplemented(...)`

Links

• PyPI: https://pypi.org/project/pyflakes
• Changelog: https://pyup.io/changelogs/pyflakes/
• Repo: https://github.com/PyCQA/pyflakes

Update pytest from 3.8.0 to 3.8.1.

Changelog

3.8.1

=========================

Bug Fixes
---------

- `3286 <https://github.com/pytest-dev/pytest/issues/3286>`_: ``.pytest_cache`` directory is now automatically ignored by Git. Users who would like to contribute a solution for other SCMs please consult/comment on this issue.


- `3749 <https://github.com/pytest-dev/pytest/issues/3749>`_: Fix the following error during collection of tests inside packages::

   TypeError: object of type 'Package' has no len()


- `3941 <https://github.com/pytest-dev/pytest/issues/3941>`_: Fix bug where indirect parametrization would consider the scope of all fixtures used by the test function to determine the parametrization scope, and not only the scope of the fixtures being parametrized.


- `3973 <https://github.com/pytest-dev/pytest/issues/3973>`_: Fix crash of the assertion rewriter if a test changed the current working directory without restoring it afterwards.


- `3998 <https://github.com/pytest-dev/pytest/issues/3998>`_: Fix issue that prevented some caplog properties (for example ``record_tuples``) from being available when entering the debugger with ``--pdb``.


- `3999 <https://github.com/pytest-dev/pytest/issues/3999>`_: Fix ``UnicodeDecodeError`` in python2.x when a class returns a non-ascii binary ``__repr__`` in an assertion which also contains non-ascii text.



Improved Documentation
----------------------

- `3996 <https://github.com/pytest-dev/pytest/issues/3996>`_: New `Deprecations and Removals <https://docs.pytest.org/en/latest/deprecations.html>`_ page shows all currently
deprecated features, the rationale to do so, and alternatives to update your code. It also list features removed
from pytest in past major releases to help those with ancient pytest versions to upgrade.



Trivial/Internal Changes
------------------------

- `3955 <https://github.com/pytest-dev/pytest/issues/3955>`_: Improve pre-commit detection for changelog filenames


- `3975 <https://github.com/pytest-dev/pytest/issues/3975>`_: Remove legacy code around im_func as that was python2 only

Links

• PyPI: https://pypi.org/project/pytest
• Changelog: https://pyup.io/changelogs/pytest/
• Homepage: https://docs.pytest.org/en/latest/

Update twine from 1.11.0 to 1.12.0.

Changelog

1.12.0

* :feature:`395 major` Add ``twine check`` command to check long description
* :feature:`392 major` Drop support for Python 3.3
* :feature:`363` Empower ``--skip-existing`` for Artifactory repositories
* :bug:`367` Avoid MD5 when Python is compiled in FIPS mode

Links

• PyPI: https://pypi.org/project/twine
• Changelog: https://pyup.io/changelogs/twine/
• Docs: https://twine.readthedocs.io/

[coveralls]

Coverage remained the same at 92.788% when pulling dfce0bc on pyup-scheduled-update-2018-09-24 into b2736eb on master.