std::bad cast in python binding when using nested structure about reelay HOT 3 CLOSED

Yes, I have removed nested structures in v2008 as they were half-baked. Currently monitors work on flat json/dict objects.

I would love to hear about your particular use case. Could you elaborate the case and the structure of json/python objects that you would like to monitor?

Thanks.

from reelay.

Dogan, first of all, thanks for making public this software. I stated using it only a short time ago.

I am monitoring MQTT messages which are composed of JSON objects. Convert MQTT topics into objects (e.g. "a/b/c" goes to {"a": {"b": {"c":{}}}) and put the payload object into the innermost sub-objects.

On a second thought, I could ditch this (admittedly complicated) logic and flatten all using name mangling (e.g. "a__b__c__etc"). In fact (correct me if I'm wrong), all of v2005's nested structured features can be emulated using flat, mangled names, since Rye-2005 only allowed operating on leaf values.

from reelay.

Yes, the workaround is flattennig the JSON objects. Two popular choices are using dots ("a.b.c") as in Javascript or using slashes ("/a/b/c") as in JSON pointer standard.

The other alternative is to use a python type that supports nested keys. Below is an example code that uses pyhocon library.

from reelay import discrete_timed_monitor
from pyhocon import ConfigFactory

my_monitor = discrete_timed_monitor(
    pattern=r"{'root.speed' > 13.0} since[:3] {'root.lights_on'}",  #notice the quotes on nested keys
    condense=False
)

input_sequence = [
    dict(root=dict(speed=3.3,  lights_on=False)),
    dict(root=dict(speed=6.3,  lights_on=False)),
    dict(root=dict(speed=9.3,  lights_on=True)),
    dict(root=dict(speed=13.3, lights_on=False)),
    dict(root=dict(speed=13.4, lights_on=False)),
    dict(root=dict(speed=13.3, lights_on=False)),
    dict(root=dict(speed=13.2, lights_on=False)),
    dict(root=dict(speed=13.2, lights_on=False)),
]

result = []
for x in input_sequence:
    y = my_monitor.update(ConfigFactory.from_dict(x))  #converting to hocon type
    result.append(y)

expected = [
    {'value': False},
    {'value': False},
    {'value': True},
    {'value': True},
    {'value': True},
    {'value': True},
    {'value': False},
    {'value': False}]

print(result == expected)

This works because Reelay takes the string in the specification, and use it directly to access object fields.

from reelay.