Skip to main content

Simple python event signaling

Project description

# smokesignal - simple python signaling

[![Build Status](https://travis-ci.org/shaunduncan/smokesignal.png)](https://travis-ci.org/shaunduncan/smokesignal)

`smokesignal` is a simple python library for sending and receiving signals.
It draws some inspiration from the django signal framework but is meant as a
general purpose variant.


## Requirements & Compatibility

`smokesignal` requires no dependencies outside of the python standard library.
It has been tested on and is compatible with python versions 2.6, 2.7, 3.2, and 3.3.


## How To Use

Most uses of `smokesignal` involve registering a single callable to respond to a signal
using `on` and sending signals using `emit`.

### Registering Callbacks

Most callback registration happens using `on`, which can be used either as a decorator
or a direct function call. This method also accepts an optional argument `max_calls` which
indicates the maximum number of times a callback should respond to an emitted signal:

```python
import smokesignal

@smokesignal.on('foo')
def my_callback():
pass

@smokesignal.on('foo', max_calls=2)
def my_callback():
pass

smokesignal.on('foo', my_callback, max_calls=5)
```

If you intend a callback to respond to a signal at most one time, rather than indicating
`max_calls=1`, you can use `once` as a convenience:

```python
import smokesignal

@smokesignal.once('foo')
def my_callback():
pass
```

### Sending Signals

Signals are sent to all registered callbacks using `emit`. This method optionally accepts
argument and keyword argument lists that are passed directly to each callback:

```python
import smokesignal

# Each callback responding to 'foo' is called
smokesignal.emit('foo')

# Each callback responding to 'foo' is called with arguments
smokesignal.emit('foo', 1, 2, 3, four=4)
```

You can also send signals with the included context manager `emitting`. By default, this context
manager accepts one argument, which is a signal to send once the context manager exits. However,
you can supply keyword arguments for `enter` and `exit` that will be sent at those points of the
context manager:

```python
import smokesignal

# 'foo' emitted on exit
with smokesignal.emitting('foo'):
pass

# 'foo' emitting on enter, 'bar' emitted on exit
with smokesignal.emitting(enter='foo', exit='bar'):
pass
```

### Disconnecting Callbacks

If you no longer wish for a callback to respond to any signals, you can use either
`disconnect_from`, if you intend on removing specific signals, or `disconnect` if you intend
to remove all of them:

```python
import smokesignal

# my_callback will no longer respond to signals
smokesignal.disconnect(my_callback)

# my_callback will no longer respond to 'foo', but may repond to others
smokesignal.disconnect_from(my_callback, 'foo')
```

### Other Batteries Included

```python
# Remove all callbacks responding to a specific signal
smokesignal.clear('foo')

# Remove all callbacks responding to all signals
smokesignal.clear_all()

# Get a tuple of all signals a callback responds to
smokesignal.signals(my_callback)

# Check if a callback responds to a signal
smokesignal.responds_to(my_callback, 'foo')
```


## Caveats

What would be great is if you could decorate instance methods using `on` or `once`. However,
that doesn't work because there is no knowledge of the class instance at the time the callback
is registered to respond to signals:

```python
import smokesignal

class Foo(object):

# THIS DOES NOT WORK
@smokesignal.on('bar')
def callback(self):
pass
```

There is a workaround if you would like instance methods to respond to callbacks:

```python
import smokesignal

class Foo(object):
def __init__(self, *args, **kwargs):
@smokesignal.on('bar')
def _callback():
self.callback()
super(Foo, self).__init__(*args, **kwargs)

def callback(self):
pass
```

The above will register the callback without any argument requirements but will
also ensure that the _intended_ callback method is called correctly.


## Changelog

### 0.2
- Added `emitting` context manager
- Updated internals no longer require decorator magic for enforcing maximum call counts

### 0.1
- Initial Version


## Contribution and License

Developed by [Shaun Duncan](mailto:shaun.duncan@gmail.com) and is
licensed under the terms of a MIT license.

Project details


Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page