:sparkles: Source Zendesk Support: Migrate to low code

[tolik0]

What

Migrate to low code. Add new stream Ticket Activities: https://github.com/airbytehq/product-request-backlog/issues/72 This PR adds many changes to different streams. For simplicity, all errors from the previous version are described in the file: https://docs.google.com/document/d/1JeC4fPl9zqgPJZLJs7KVRnCk--v8cO8hKyJHAeKoXVw/edit

Here is a short description of the changes:

• Articles - filtering with start_time works, but sorting doesn't work ( no error, but results are not sorted) - preserved sorting params as they are described in the API docs. - returned back to Python implementation.
• Audit Logs - start_time is not applicable, no sorting - Changed to filter[created_at][] and sort parameters described in the API docs.
• Group Memberships - start_time is not applicable, sorting doesn't work - Changed to semi incremental with local filtering.
• Groups - start_time is not applicable - Changed to semi incremental with local filtering.
• Macros - start_time is not applicable - Changed to semi incremental with local filtering.
• Organization Fields - start_time is not applicable, no sorting - Changed to semi incremental with local filtering.
• Organization Memberships - start_time is not applicable, no sorting - Changed to semi incremental with local filtering.
• Posts - start_time is not applicable - Changed to semi incremental with local filtering, add sorting. - returned back to Python implementation
• Satisfaction Ratings - sorting is not working - Changed sorting to sort=created_at, updated_at is not supported.
• Ticket Audits - sorting is not working, start_time was not used but it is not applicable - Data is sorted in descending order by default, but preserving sorting parameters as data feed is used to stop iterating when records are earlier than start date or state.
• Ticket Fields - start_time is not applicable - Changed to semi incremental with local filtering
• Ticket Forms - start_time is not applicable - Changed to semi incremental with local filtering
• Ticket Metrics - start_time is not applicable - Data is sorted by created at, so semi-incremental with local filtering is used - reverted to Pyton code
• Ticket Skips - start_time is not applicable - Changed to semi incrementla with local filtering, add sorting. Didn't add data feed as not sure if sorted by created or updated.
• Topic - nor start_time (no error but results are not filtered), nor sorting is not working - Changed to semi-incremental

Substreams are in Python as currently, low code doesn't support reading parent streams with a state.

🚨 User Impact 🚨

No breaking changes. A new stream added: Ticket Activities.

[vercel[bot]]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Comments	Updated (UTC)
airbyte-docs	⬜️ Ignored (Inspect)	Visit Preview		Apr 29, 2024 0:25am

[tolik0]

Regression Tool results:

catalog: A new stream was added - Ticket Activities.

test_record_schema_match_with_state: brands, topics - normalization ticket_fields - records were filtered by time

Read:

• custom_roles - control version doesnt filter records by time
• ticket_fields - control version doesnt filter records by time
• organization_memberships - control version doesnt filter records by time
• group_memberships - control version doesnt filter records by time
• ticket_metric_events - difference in records are either due to cursor comparison changed from > to >=, or records that were extracted after the state of the first read.
• satisfaction_rating - +1 record - cursor comparison changed from > to >=
• ticket_audits - new records were added
• groups - +1 record - cursor comparison changed from > to >=
• ticket_forms - +1 record - cursor comparison changed from > to >=
• ticket_fields - +1 record - cursor comparison changed from > to >=
• organizations - control version doesn't use start date
• groups - control version doesnt filter records by time

[bnchrch]

Hey just running the regression test for the first time.

A few things stood out to me I wondered if you could explain

1. We saw significantly more record messages (+262) specifically for ticket_audits.

Is that expected? If so why?

2. Type changes

SLA policy had a type change int -> string Is there any chance of a down stream implication because of the change from int to string?

source_id and actor_id from int -> float Is there any chance of a down stream implication because of the change int to float?

---

Im sure these are obvious answers, just want to sharpen my understanding.

[tolik0]

Hi, @bnchrch,

1. Great catch! You identified an issue where records in the macros and ticket_audits streams weren't being filtered by time. I've already pushed a fix for that. For other streams, the behaviour is expected: they will either include new records or records with timestamps matching the cursor.
2. Additionally, the data types have been corrected in the normalization process to align with the defined schemas. Here is the correct type of source_id field: https://github.com/airbytehq/airbyte/blob/e14841284cba6104d0a6c2c82fa7dffc70ddfb33/airbyte-integrations/connectors/source-zendesk-support/source_zendesk_support/schemas/audit_logs.json#L32.