recordings to the stored timestamp and sync the rest to the main
recording file. When integrated with the likes of Jellyfin, a handful
of associated files are created for each recording, this makes all of
them have the same modification time. It is run on epg startup, so on
UNIX just by killing _epg, the modification times will be
synchronized. On Windows it'll be neccessary to close and open again
the proxy.

Signed-off-by: Javier Marcet <javier@marcet.info>

network to mu7d, it does not belong to epg anymore

Signed-off-by: Javier Marcet <javier@marcet.info>

WIN32. If the console is closed while a client is watching a stream,
if the last executed process is u7d, that's the one Windows kills, and
it cannot do its cleanup. For the client to not get hung when the
stream is not closed properly (like Kodi does), by launching epg last,
we make sure u7d is closed properly and it can close all the streams
before exiting.

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

and cleaning up the retorted code flow, now everything is
encapsulated and each task takes care of terminating its own
subprocesses when needed, instead of controlling them through globals

Signed-off-by: Javier Marcet <javier@marcet.info>

the proxy

Signed-off-by: Javier Marcet <javier@marcet.info>

on WIN32

Signed-off-by: Javier Marcet <javier@marcet.info>

are not enabled

Signed-off-by: Javier Marcet <javier@marcet.info>

embedded ones

Signed-off-by: Javier Marcet <javier@marcet.info>

the (broken) Windows world

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

proceeding with a timers_check()

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

send_prom_events() to prioritize the streaming loop

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

app.stop(), current Sanic is killing subprocesses on exit on WIN32
and VODs need to do their cleanup

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

otherwise there are ugly "Task cancelled but pending" messages when
aborted with ongoing recordings

Signed-off-by: Javier Marcet <javier@marcet.info>

maximum added to the requested time, always depending on the available
time recorded by Movistar

Signed-off-by: Javier Marcet <javier@marcet.info>

Signed-off-by: Javier Marcet <javier@marcet.info>

anymore, being used just for debugging, also drop recording_cleanup,
embed the logic within reap_vod_child

Signed-off-by: Javier Marcet <javier@marcet.info>

and exceptions following Sanic's own style

Signed-off-by: Javier Marcet <javier@marcet.info>

received

Signed-off-by: Javier Marcet <javier@marcet.info>

way

Signed-off-by: Javier Marcet <javier@marcet.info>

WIN32 by setting stdin=DEVNULL

if epg is killed

recording_cleanup() on WIN32, only stopping the proxy when the console
is closed while making sure no stale vod's subprocess remains after
vod dies abruptly

while several recordings are taking place in the same folder

cancellation

Signed-off-by: Javier Marcet <javier@marcet.info>

different timers yielding the same match.

We could avoid warnings about ongoing recordings entirely by checking
the filename instead of the tuple (channel, program), but this way we
can see how often the same program is aired more than once within the
same week, which is always nice and shows how crappy TV is nowadays.

Signed-off-by: Javier Marcet <javier@marcet.info>

nonlocal

Signed-off-by: Javier Marcet <javier@marcet.info>