CamelCase vs snake_case - Python SDK schema

[aarfe]

Hello All,

We are using the Python SDK to query some of the APIs and store the responses in Google BigQuery and having some problems using the response schema matching the API response.

We can see that all the APIs responses' schemas are documented under each API method --> API Responses --> 200 successful operation, both with the Schema definition and and a JSON example response.
For example, see: POST /api/v2/analytics/conversations/details/query .
We also got hold of the Swagger Schema definition file

We have two problems:
1 - We need an API response schema definition with fields' names and types to be able to create the target BigQuery table. We also need the response schema to schema-validate incoming payloads before storing them in our tables.
A JSON example response (like the one we find in the documentation, left side in the pic) could also work, since we could use BigQuery schema generator tools to create the target BigQuery schema.
The API response schema OR the JSON example response need to match the field's names and types of the json payload we receive from the Python SDK.
Problem: The Schemas in the documentation and in the Swagger Schema definition file, as well as the JSON example response in the online docs are CamelCase while the response from the Python SDK comes as snake_case. Thus, the fields are not named the same (see pic).

2 - The JSON example response (very useful to autogenerate a schema for almost any database) does not differ all the atomic data types, for example, the numeric fields (both float and integers) are represented with a 0, zero digit. Same for the timestamps fields which are represented as a string (conversationEnd, conversationStart), giving little option to infer the type correctly.
Problem : we cannot programmatically generate the correct field types for all the fields.
We can manually read the docs and try to figure out the specific types of the ambiguous fields, but it proves hard with so many fields

We are looking at some workarounds, like creating the schemas we need from the Swagger Schema definition file (A: dereference the objects we need, b: do a CamelCase to snake_case conversion and C: apply some kind of JSON-Schema to BigQuery schema transformer) but do you know if there's any easy way to get hold of the actual response schema used by the Python SDK responses ?

Thank you

[tim.smith]

@aarfe

The Schemas in the documentation and in the Swagger Schema definition file are CamelCase while the response from the Python SDK comes as snake_case. Thus, the fields are not named the same (see pic).

It's hard to say what you're doing without seeing any code, but it looks like maybe you're dumping the python object instead of serializing it to JSON. You can use ApiClient's built-in function for this purpose: https://github.com/MyPureCloud/platform-client-sdk-python/blob/4d4604cd0fbcd93011dddb93400de60e648882b3/build/PureCloudPlatformClientV2/api_client.py#L386

@aarfe

The JSON example response (very useful to autogenerate a schema for almost any database) does not differ all the atomic data types, for example, the numeric fields (both float and integers) are represented with a 0, zero digit. Same for the timestamps fields which are represented as a string (conversationEnd, conversationStart), giving little option to infer the type correctly.

The property descriptions, shown both in API Explorer and the SDK docs, should generally describe the format of the data if the type is ambiguous. If you have any specific questions, please ask.

[aarfe]

Hi Tim,
Thanks for the reply.

I added some more description on what we are trying to do.

Basically the problem goes around the fact that to insert the whole JSON payload in a BigQuery table, we need a schema with fields' names and types from which we can generate the table structure.
Once the table is created we can easily insert data into it.
With the same schema, we can also validate future incoming payloads.

We have already managed to do this with the AnalyticsConversationQueryResponse, just that we had to manually adjust the schema, both to fix the CamelCase vs snake_case problem and for ambiguous types.
The pic shows the AnalyticsConversationQueryResponse BigQuery schema we created semi-manually using the JSON API response example from the docs and this tool: BigQuery schema generator tools .

We are now using a similar procedure for other APIs responses we need, but it proves to be quite tricky and time consuming, since there's hundreds of fields in the responses.
We have already created some schemas that failed due to some actual data types not matching our manually generated schema

That's why we are asking if there is a way to get these schema matching the python SDK responses.