TypeError: __init__() takes at least 2 arguments (2 given)

[jinnatar]

Running latest HEAD the api.pushes() interface crashes:

Traceback (most recent call last):
  File "./receiver.py", line 59, in <module>
    for push in api.pushes():
  File "/usr/local/lib/python2.7/dist-packages/pushybullet.py", line 1221, in <genexpr>
    return (self.make_push(o) for o in it if bool(o.get('type', None)))
  File "/usr/local/lib/python2.7/dist-packages/pushybullet.py", line 1100, in make_push
    push = pushcls(pusharg, **pushargs) if pusharg else pushcls(**pushargs)
TypeError: __init__() takes at least 2 arguments (2 given)

This is my "backlog" loop so the relevant code is very simple:

api = pb.PushBullet(API_KEY)
receiver = api['receiver'].iden
try:
    for push in api.pushes():
        target = getattr(push, 'target_device_iden', 'broadcast')
        if target == receiver: # only act on ones directed at us
            acquirePush(push)
except Exception as inst:
    logger.warn('Failed to pull from pushbullet: %s' % inst)
    raise

Normally the loop handles url and file type pushes and even does so successfully if I push new items into the queue. But eventually it always crashes and burns like this. So almost 100% sure there's something funny in the queue that kills processing, I just can't see it from anywhere.

The same issue repeats with the pb client:

 %> pb --apikey $(cat apikey) pushes
[..snip a ton of pushes..]
Traceback (most recent call last):
  File "/usr/local/bin/pb", line 147, in <module>
    main()
  File "/usr/local/bin/pb", line 144, in main
    command(api, args)
  File "/usr/local/bin/pb", line 71, in command_pushes
    for push in pushes:
  File "/usr/local/lib/python2.7/dist-packages/pushybullet.py", line 1221, in <genexpr>
    return (self.make_push(o) for o in it if bool(o.get('type', None)))
  File "/usr/local/lib/python2.7/dist-packages/pushybullet.py", line 1100, in make_push
    push = pushcls(pusharg, **pushargs) if pusharg else pushcls(**pushargs)
TypeError: __init__() takes at least 2 arguments (2 given)

So, at a minimum I'd say some basic exception handling needs to be implemented to catch some whiff of the push causing the meltdown, and then can think about working around said meltdown.

[jinnatar]

I tracked down the push that makes it die, it's a link push. Comparing it to a working one there's a few peculiarities about it:

 %> comm broken working
                u'active'
        u'awake_app_guids'
                u'created'
                u'direction'
                u'dismissed'
        u'guid'
                u'iden'
                u'modified'
                u'receiver_email'
                u'receiver_email_normalized'
                u'receiver_iden'
                u'sender_email'
                u'sender_email_normalized'
                u'sender_iden'
                u'sender_name'
        u'source_device_iden'
                u'target_device_iden'
u'title'
                u'type'
        u'url'

So the broken one has only these fields set:
u'active'
u'created'
u'direction'
u'dismissed'
u'iden'
u'modified'
u'receiver_email'
u'receiver_email_normalized'
u'receiver_iden'
u'sender_email'
u'sender_email_normalized'
u'sender_iden'
u'sender_name'
u'target_device_iden'
u'title'
u'type'

So missing:

• awake_app_guids (probably not an issue)
• guid (seems like an issue)
• source_device_iden (could be an issue?)
• url (pretty sure this is what kills it, the url is for some reason set as the title field)

I'm not too good with Python but seems to me like this bit here:

pushybullet/pushybullet.py

Line 1100 in e457bc9

push = pushcls(pusharg, **pushargs) if pusharg else pushcls(**pushargs)

could be a bit more robust to catch this kind of brokenness.