Python SDK and creating multiple customProperties in one go

You’d have a couple different things going on. I suggest trying to break down your code into separate functions.

1. Bring the data into a dictionary from your input CSV - several examples on the internet. Reply here if your google-fu turns up nothing useful
2. Convert the dictionary to a list of dictionaries - example below
3. POST/PUT/PATCH the object in LM using the SDK (I only mention that because it’s in the title of this post) - Documentation here.

Example to translate properties into something that can be patched in:

myDictionary = {"corp.building":"Some Building Name","corp.department":"Some Department Name","corp.role":"Some Role"}
myListofDicts = [{"name":k,"value":v} for k,v in myDictionary.items()]
pprint(myListofDicts)
[{'name': 'corp.building', 'value': 'Some Building Name'},
 {'name': 'corp.department', 'value': 'Some Department Name'},
 {'name': 'corp.role', 'value': 'Some Role'}]

It’s not actually a Python problem It’s an LM problem. LM is the one that decided to use a list of dictionaries instead of a dictionary.

Thanks Stuart. One of the libraries I was using is “argparse” from the CLI.

My test script right now looks something like:
 

./script.py --primary ”Folder Name” --location ”123 somewhere ln, here, there 11111” --custom “building:Some Building” --custom “department:Some Department” --custom “role:Some Role”

When I push via the SDK to the api.add_device_group() function, I would basically take the --location and all the --custom ones set via the CLI and “build” the customProperties that I want to POST in the body of the api call (along with some more common , one off variables I scrape).

I’ll mess with the listofDicts as you illustrated and go from there. 

Thanks for your help.

Yep, I do something very similar, except that I’ve started using environment variables for the credentials (see input_method below). This function should facilitate creating your groups. Actually, it’s idempotent, so if you call it once, it’ll create the group. The second time you call it, it should just update the existing one:

import logicmonitor_sdk, argparse, os
from datetime import datetime

input_method = "envs"

if input_method == "args":
    parser = argparse.ArgumentParser()
    for arg,help in {
            "id":"LM API Access ID",
            "key": "LM API Access Key",
            "company": "LM API Company",
    }.items():
        parser.add_argument(arg, help=help)
    args = parser.parse_args()

    lm_creds = {
        "AccessId": args.id,
        "AccessKey": args.key,
        "Company": args.company
    }
elif input_method == "envs":
    lm_creds = {
        "AccessId": os.environ['accessid'],
        "AccessKey": os.environ['accesskey'],
        "Company": os.environ['company']
    }
else:
    print(f"Invalid input method chosen: {input_method}")
    quit()

configuration = logicmonitor_sdk.Configuration()
configuration.access_id = lm_creds['AccessId']
configuration.access_key = lm_creds['AccessKey']
configuration.company = lm_creds['Company']
lm = logicmonitor_sdk.LMApi(logicmonitor_sdk.ApiClient(configuration))

debug = False
info = False

def log_msg(msg, severity="INFO", end="\n"):
  if (severity == "DEBUG" and debug) or (severity == "INFO" and info) or severity not in ("DEBUG", "INFO"):
    print(f"{datetime.now().strftime('[%Y-%m-%d %H:%M:%S]')} {severity}: {msg}", end=end)

def create_group(name,parent,customProperties={},appliesTo=""): # shortcut function to the sdk that creates device groups with less fuss
    parsedProperties = [{"name": k, "value": v} for k,v in customProperties.items()]
    device_groups = []
    end_found = False
    offset = 0
    size = 1000
    while not end_found:
        current = lm.get_device_group_list(size=size,offset=offset,filter=f"parentId:\"{parent}\",name:\"{name}\"").items
        device_groups += current
        offset += len(current)
        end_found = len(current) != size
    if len(device_groups) == 0:
        try:
            result = lm.add_device_group({"name":name, "parentId":parent, "customProperties":parsedProperties, "appliesTo":appliesTo})
            log_msg(f"  Created resource group {result.full_path} ({result.id})", "API SDK POST")
            return result
        except lm.ApiException as e:
            log_msg("Exception when calling LMApi->addDeviceGroup: %s\n" % e)
    else:
        existing_group = device_groups[0]
        mismatches = 0
        log_msg(f"  Group {name} exists in LM ({existing_group.id}), checking if attributes match...", "DEBUG")
        existing_properties = {x.name:x.value for x in existing_group.custom_properties}
        for k,v in customProperties.items(): #check each of the custom attributes we want to pass in to see if they match
            if k in existing_properties.keys() and v==existing_properties.get(k,None):
                log_msg(f"    Property {k} matches attribute already in LM", "DEBUG")
            else:
                log_msg(f"    Property {k} doesn't match attribute already in LM")
                mismatches += 1
            log_msg(f"    {'='*80}", "DEBUG")
            log_msg(f"    {k}: {v}", "DEBUG")
            log_msg(f"    {k}: {existing_properties.get(k)}", "DEBUG")
            log_msg(f"    {'='*80}", "DEBUG")
        if len(appliesTo) > 0:
            if appliesTo == existing_group.applies_to:
                log_msg(f"    AppliesTo we want to patch in matches the AppliesTo already on the device", "DEBUG")
            else:
                log_msg(f"    AppliesTo we want to patch in doesn't match the AppliesTo already on the device")
                mismatches += 1
            log_msg(f"      {'='*80}", "DEBUG")
            formatted_appliesTo = appliesTo.replace('\n','\n      ')
            log_msg(f"      {formatted_appliesTo}", "DEBUG")
            log_msg(f"      {'='*80}", "DEBUG")
            formatted_appliesTo = existing_group.applies_to.replace('\n','\n      ')
            log_msg(f"      {formatted_appliesTo}", "DEBUG")
            log_msg(f"      {'='*80}", "DEBUG")
        else:
            log_msg(f"    There is no AppliesTo that we want to patch in.", "DEBUG")
        if mismatches > 0:
            log_msg(f"    Number of attribute mismatches: {mismatches}")
            existing_group.applies_to = appliesTo
            existing_properties.update(customProperties)
            existing_group.custom_properties = [{"name":k,"value":v} for k,v in existing_properties.items()]
            result = lm.patch_device_group_by_id(id=existing_group.id, body=existing_group) # , op_type="replace")
            log_msg(f"    Patched device group {existing_group.name}", "API SDK PATCH")
        else:
            log_msg(f"    {existing_group.full_path} already exists in LM with all the correct properties, moving on...", "SUMMARY")
            result = existing_group
    log_msg(f"    {result.id}: {result.full_path}: {result.custom_properties}", "DEBUG")
    return result

Example usage:

>>> response = create_group("SDK Test",812,{"prop1":"1","prop2":"2"})
[2023-02-21 16:30:11] API SDK POST:   Created resource group Customers/Longhorn Smokers/SDK Test (854)
>>> response = create_group("SDK Test",812,{"prop1":"1","prop2":"2"})
[2023-02-21 16:33:06] SUMMARY:     Customers/Longhorn Smokers/SDK Test already exists in LM with all the correct properties, moving on...
>>> response = create_group("SDK Test",812,{"prop1":"1","prop2":"2"})
[2023-02-21 16:33:43] API SDK PATCH:     Patched device group SDK Test
>>> debug = True
>>> info = True
>>> response = create_group("SDK Test",812,{"prop1":"1","prop2":"2"})
[2023-02-21 16:35:57] DEBUG:   Group SDK Test exists in LM (854), checking if attributes match...
[2023-02-21 16:35:57] DEBUG:     Property prop1 matches attribute already in LM
[2023-02-21 16:35:57] DEBUG:     ================================================================================
[2023-02-21 16:35:57] DEBUG:     prop1: 1
[2023-02-21 16:35:57] DEBUG:     prop1: 1
[2023-02-21 16:35:57] DEBUG:     ================================================================================
[2023-02-21 16:35:57] DEBUG:     Property prop2 matches attribute already in LM
[2023-02-21 16:35:57] DEBUG:     ================================================================================
[2023-02-21 16:35:57] DEBUG:     prop2: 2
[2023-02-21 16:35:57] DEBUG:     prop2: 2
[2023-02-21 16:35:57] DEBUG:     ================================================================================
[2023-02-21 16:35:57] DEBUG:     There is no AppliesTo that we want to patch in.
[2023-02-21 16:35:57] SUMMARY:     Customers/Longhorn Smokers/SDK Test already exists in LM with all the correct properties, moving on...
[2023-02-21 16:35:57] DEBUG:     854: Customers/Longhorn Smokers/SDK Test: [{'name': 'prop2', 'value': '2'}, {'name': 'prop1', 'value': '1'}]
>>>

FWIW, I wouldn’t use argparse to pull in the csv data. I’d actually use the built in CSV handling capabilities of Python:

def read_csv(filename):
    import csv
    result = []
    with open(filename, mode='r') as csv_file:
        csv_reader = csv.DictReader(csv_file)
        result = [
            {"name":row["name"], "department":row["department"], "bday":row["birthday month"]}
            for row
            in csv_reader
        ]
    return result

Given the following CSV file called “example.csv” in the current working directory:

name,department,birthday month
John Smith,Accounting,November
Erica Meyers,IT,March

The usage would look like this:

>>> from pprint import pprint
>>> def read_csv(filename):
...     import csv
...     result = []
...     with open(filename, mode='r') as csv_file:
...         csv_reader = csv.DictReader(csv_file)
...         result = [
...             {"name":row["name"], "department":row["department"], "bday":row["birthday month"]}
...             for row
...             in csv_reader
...         ]
...     return result
...
>>> data = read_csv("example.csv")
>>> pprint(data)
[{'bday': 'November', 'department': 'Accounting', 'name': 'John Smith'},
 {'bday': 'March', 'department': 'IT', 'name': 'Erica Meyers'}]
>>>

This should be helpful for me. Once I set up my customProps as a list of dictionaries it worked like a charm (this is where my python lacks, coming more from a PHP background). I was close, just slightly off ;) 

Thanks for the tips and snippets.