Metadata-Version: 2.0
Name: watchgod
Version: 0.0.1
Summary: Simple, modern file watching and code reload in python.
Home-page: https://github.com/samuelcolvin/watchgod
Author: Samuel Colvin
Author-email: s@muelcolvin.com
License: MIT
Description-Content-Type: UNKNOWN
Platform: UNKNOWN
Classifier: Development Status :: 3 - Alpha
Classifier: Environment :: Console
Classifier: Programming Language :: Python
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3 :: Only
Classifier: Programming Language :: Python :: 3.5
Classifier: Programming Language :: Python :: 3.6
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Information Technology
Classifier: Intended Audience :: System Administrators
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: Unix
Classifier: Operating System :: POSIX :: Linux
Classifier: Environment :: MacOS X
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.5

watchgod
========

|BuildStatus| |Coverage| |pypi|

Simple, modern file watching and code reload in python.

Usage
-----

To watch for changes in a directory:

.. code:: python

   from watchgod import watch

   for changes in watch('./path/to/dir'):
       print(changes)


To run a function and restart it when code changes:

.. code:: python

   from watchgod import run_process

   def foobar(a, b, c):
       ...

   run_process('./path/to/dir', foobar, process_args=(1, 2, 3))

``run_process`` uses ``PythonWatcher`` so only changes to python files will prompt a
reload, see *custom watchers* below.

If you need notifications about change events as well as to restart a process you can
use the ``callback`` argument to pass a function will will be called on every file change
with one argument: the set of file changes.

Asynchronous Methods
....................

*watchgod* comes with an asynchronous equivalents of ``watch``: ``awatch`` which uses
a ``ThreadPoolExecutor`` to iterate over files.

.. code:: python

   import asyncio
   from watchgod import awatch

   async def main():
       async for changes in awatch('/path/to/dir'):
           print(changes)

   loop = asyncio.get_event_loop()
   loop.run_until_complete(main())


There's also an asynchronous equivalents of ``run_process``: ``arun_process`` which in turn
uses ``awatch``:

.. code:: python

   import asyncio
   from watchgod import arun_process

   def foobar(a, b, c):
       ...

   async def main():
       await arun_process('./path/to/dir', foobar, process_args=(1, 2, 3))

   loop = asyncio.get_event_loop()
   loop.run_until_complete(main())

``arun_process`` uses ``PythonWatcher`` so only changes to python files will prompt a
reload, see *custom watchers* below.

The signature of ``arun_process`` is almost identical to ``run_process`` except that
the ``callback`` argument if provide must be a coroutine, not a function.

Custom Watchers
...............

*watchgod* comes with the following watcher classes which can be used via the ``watcher_cls``
keyword argument to any of the methods above.

For more details, checkout
`watcher.py <https://github.com/samuelcolvin/watchgod/blob/master/watchgod/watcher.py>`_,
it's pretty simple.

**AllWatcher**
    The base watcher, all files are checked for changes.

**DefaultWatcher**
    The watcher used by default by ``watch`` and ``awatch``, commonly ignored files
    like ``*.swp``, ``*.pyc`` and ``*~`` are ignored along with directories like
    ``.git``.

**PythonWatcher**
    Specific to python files, only ``*.py``, ``*.pyx`` and ``*.pyd`` files are watched.

**DefaultDirWatcher**
    Is the base for ``DefaultWatcher`` and ``DefaultDirWatcher`` and takes care of ignoring
    some regular directories.


If these classes aren't sufficient you can define your own watcher, in particular
you will want to override ``should_watch_dir`` and ``should_watch_file``. Unless you're
doing something very odd you'll want to inherit from ``DefaultDirWatcher``.


Why no inotify / kqueue / fsevent / winapi support
--------------------------------------------------

*watchgod* (for now) uses file polling rather than the OS's built in file change notifications.

This is not an oversight, it's a decision with the following rationale:

1. Polling is "fast enough", particularly since PEP 471 introduced fast ``scandir``.

   With a reasonably large project like the TutorCruncher code base with 850 files and 300k lines
   of code *watchdog* can scan the entire tree in ~24ms. With a scan interval of 400ms that's roughly
   5% of one CPU - perfectly acceptable load during development.

2. The clue is in the title, there are at least 4 different file notification systems to integrate
   with, most of them not trivial. And that's before we get to changes between different OS versions.

3. Polling works well when you want to group or "debounce" changes.

   Let's say you're running a dev server and you change branch in git, 100 files change.
   Do you want to reload the dev server 100 times or once? Right.

   Polling periodically will likely group these changes into one event. If you're receiving a
   stream of events you need to delay execution of the reload when you receive the first event
   to see if it's part of a whole bunch of file changes, this is not completely trivial.


All that said, I might still implement ``inotify`` support. I don't use anything other
than Linux so I definitely won't be working on dedicated support for any other OS.


.. |BuildStatus| image:: https://travis-ci.org/samuelcolvin/watchgod.svg?branch=master
   :target: https://travis-ci.org/samuelcolvin/watchgod
.. |Coverage| image:: https://codecov.io/gh/samuelcolvin/watchgod/branch/master/graph/badge.svg
   :target: https://codecov.io/gh/samuelcolvin/watchgod
.. |pypi| image:: https://img.shields.io/pypi/v/watchgod.svg
   :target: https://pypi.python.org/pypi/watchgod


