As you've mentioned it and in case it's useful.
When I was working on this I was guided by Michael De Simone of Zenoss. I asked an enormous number of questions at the start and he patiently helped me until I got my head around the API and the relationship between the object types. He was the one  who drew up something like the pseudo-code you posted Jay.

He has a Postman collection that he pointed me at to get me started and I've checked and he has said it's OK to share it:

https://www.getpostman.com/collections/75b6b0f1712aade35d1b

Looking up the API, "keys. What does that do?"
Oh, OK, good point.

I made so much use use of the full object structure to explore what I could get at that I didn't think to limit what I accessed.
I do write a few other parameters out to my export file, but it's easy to include them too.

Doing that is a LOT faster!

So that makes:

def get_devices(UID):
    '''
    Get a list of all devices
    Optionally, specify an object organiser UID, eg device class, group, system, or location
    If an organiser is not specified, all devices are returned
    '''
    print('Listing all devices')
    api_endpoint = url + '/device_router'
    router = 'DeviceRouter'
    method = 'getDevices'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + UID + '","limit":200000,"keys": ["name", "uid","productionState"]}]}'
    # print(payload)
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            # print(response.text)
            resp_data = response.json()['result']['devices']
            return resp_data
        else:
            print('No devices found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))

def get_device_components(uid):
    '''
    Devices can have associated components, which can also have templates attached, which can have thresholds set
    Get the Components associated with this device
    '''
    print('Getting device components for ' + uid)
    api_endpoint = url + '/device_router'
    router = 'DeviceRouter'
    method = 'getComponents'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + uid + '","limit":200000,"keys": ["name", "uid"]}]}'
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            # print(response.text)
            resp_data = response.json()['result']['data']
            return resp_data
        else:
            print('No components found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))

As I recall zenoss.py is Python 2.7?

That's really the only reason that I didn't use it and if I'd realised that so much of Zenoss was 2.7 I would have gone that way.
Also I guess I was at the bottom of my learning curve with the API when I was looking at the options for existing tooling and I struggled at the time to translate what I was doing in Postman to that module.
Now, it would be trivial, hindsight is awonderful thing.

That's almost exactly the logic I walked through, which I can't take any credit for, I was given pretty much the same pseudo-code.
The exception being that I didn't extract the DataSources.
I was pointed at them, but when I looked at what they contained there weren't any thresholds, which is what I was worried about and I suspect, what you want too.

I will warn you that if you have a large or complex config, this is going to be slow via the JSON API.

There's nothing wrong with the API, it just takes a few seconds to return each item and when you have to walk a tree of configuration for a large list of devices, it adds up, to hours. I added primitive resume support by dumping the device list to json and working off that. Then when I was in a hurry I copied the json file several times, trimmed each to contain a subset of the device list and ran several instances in parallel.

A  modified version of the reporting Zenpack is the logical way to do this in a more effective time. It's on my ToDo list

Using Python 3 (because I didn't know that Zenoss was built on 2.7 and I just picked the latest, um, because), the main functions in my script are:

import sys
import os
import simplejson  # only used to manage a config file
import requests
import re

def get_devices(UID):
    '''
    Get a list of all devices
    Optionally, specify an object organiser UID, eg device class, group, system, or location
    If an organiser is not specified, all devices are returned
    '''
    print('Listing all devices')
    api_endpoint = url + '/device_router'
    router = 'DeviceRouter'
    method = 'getDevices'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + UID + '","limit":200000}]}'
    # print(payload)
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            # print(response.text)
            resp_data = response.json()['result']['devices']
            return resp_data
        else:
            print('No devices found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))


def get_templates(uid):
    '''
    Devices have templates attached, which can have thresholds set
    Get the templates associated with this device
    '''
    print('Getting device templates for ' + uid)
    api_endpoint = url + '/template_router'
    router = 'TemplateRouter'
    method = 'getObjTemplates'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + uid + '"}]}'
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            # print(response.text)
            resp_data = {}
            for template in response.json()['result']['data']:
                # print(template['uid'])
                resp_data[template['uid']] = template
            return resp_data
        else:
            print('No templates found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))


def get_device_components(uid):
    '''
    Devices can have associated components, which can also have templates attached, which can have thresholds set
    Get the Components associated with this device
    '''
    print('Getting device components for ' + uid)
    api_endpoint = url + '/device_router'
    router = 'DeviceRouter'
    method = 'getComponents'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + uid + '","limit":200000}]}'
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            # print(response.text)
            resp_data = response.json()['result']['data']
            return resp_data
        else:
            print('No components found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))


def get_thresholds(UID):
    '''
    Get the thresholds contained in this template
    '''
    print('Getting thresholds for ' + UID)
    api_endpoint = url + '/template_router'
    router = 'TemplateRouter'
    method = 'getThresholds'
    payload = '{"action": "' + router + '", "tid": 1, "method": "' + method + '", "data":[{"uid":"' + UID + '"}]}'
    response = requests.post(api_endpoint, payload, headers=headers, auth=auth, verify=False)
    if response.status_code == 200:
        if response.json()['result']['success']:
            resp_data = {}
            # print(response.text)
            for threshold in response.json()['result']['data']:
                resp_data[threshold['uid']] = threshold
            return resp_data
        else:
            print('No thresholds found')
            resp_data = ()
    else:
        print('HTTP Status: %s' % (response.status_code))

Then you use this to do what Jay says.

In my approach I added items to the device list dictionary to build a dictionary of dictionaries a few levels deep, eg.

# for each template
#     get the thresholds
#     store in a dictionary with template uid as the key
#     add to a list in the device
#     print the output line for each threshold

# for each component_template
# get the thresholds
# if thresholds are defined
#   for each threshold
#       see if is is already in the dictionary
#       if not
#           get them
#           add them
# print the output line for each threshold

Then when I had gathered all of the data for a device, walked my dictionary for that device appending the data to a tab delimited text file so I would have something to look at if the script hit an error.

Kicked the whole thing off with:

if __name__ == '__main__':
    auth = ('username', 'password')
    headers = {'Content-Type': 'application/json'}
    url = 'https://zenoss_url/zport/dmd'

    main()

That main() was a bit of a hairy monster and I'm not proud of it, it should have been broken down a lot smaller but I ran out of time and it worked so ....

But it's where the real hard work was and where I had to accommodate some interesting quirks. What I've posted is really just examples of translating the JSON API to Python. I'm sure better Python could be written.


Oh, I was also pointed at Postman to initially test the JSON API and get my head around the data that was being returned.

Given the functions above you probably don't need that, you can use Python debugging to see the data structures.

https://www.zenoss.com/product/zenpacks/installed-templates-report

If my code can be useful I can provide extracts from it to help you. The whole thing had a specialised purpose and was only for internal use, but I can share that too if you want.