Previously, event codes without a kernel define would resolve into properly
named objects, e.g. libevdev.EV_REL.REL_0B

This breaks the API whenever the kernel introduces a new define because that
named object will just disappear. e.g. the above REL_0B is now
REL_WHEEL_HI_RES.

Since the undefined names aren't supposed to be used by callers directly
anyway, rename them to be underscored instead: libevdev.EV_REL._REL_0B
This means we can still return them from evbit() and events() but the
underscore signals a private API.

The actual name remains the same, so print(_REL_0B.name) will not have an
underscore. This is to avoid bugs/breakage where the caller uses the 3-lettter
prefix for other purposes.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

A large percentage of real-world use-cases are tied into some external
mainloop and require non-blocking operation. It's too easy to forget about
that, so let's add the O_NONBLOCK call to the example documentation here.
Those that don't need it can remove it easily anyway.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Filipe Laíns <lains@archlinux.org>

Signed-off-by: Filipe Laíns <lains@archlinux.org>

test/test_device.py:434: DeprecationWarning: Please use assertEqual instead.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fixes #5

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix #4 intercept error returns from libevdev_next_event()

Closes #4

See merge request !6

explicitly check for error codes returned by libevdev_next_event
in order to avoid returning bogus events in next_event generator.

Not needed, it's obvious anyway and we have the property documentation.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add missing import to example.

See merge request !5

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Set up the PYTHONPATH for pytest automatically

See merge request !3

Add EventCode.is_defined to determine which event codes are known

See merge request !2

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

This makes it easier to determine which one of the codes actually exist and
which ones are just made up because they fit below the respective REL_MAX,
KEY_MAX, ...

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

This actually pointed to the python-evdev docs. Oops

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

The previous one was technically correct, but a device with just EV_REL won't
get tagged by udev as mouse and won't get picked up by anything proper. And
likewise, without a delay before sending the event, the events won't get
picked up by userspace. Change that, so the example moves the pointer a bit
which is what a user would expect.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add support for LED writing

See merge request !1

The clib part was already mostly there, the frontend wasn't. This requires the
fd to be open in rw mode which implies buffering=0 (otherwise you get Illegal
Seek errors).

Add all this into an example for a simple LED toggle script.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Leftover from earlier versions

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>