feat(gong): Convert to Checkpoint Connector

[justin-tahara]

Description

How Has This Been Tested?

Additional Options

• [Optional] Please cherry-pick this PR to the latest release version.
• [Optional] Override Linear Check

---

Summary by cubic

Converted the Gong connector to a checkpointed connector with cursor-based pagination and multi-workspace support. Adds reliable resuming, better error handling, and comprehensive tests.

• New Features

  • Switched to CheckpointedConnector with GongConnectorCheckpoint (cursor, workspace_index).
  • Added load_from_checkpoint and build_dummy_checkpoint; removed legacy polling path.
  • Handles pagination per workspace; resumes via API cursor; skips 404 workspaces.
  • Retries fetching call details with exponential backoff to avoid API race conditions.
  • Applies a 1‑day lookback and honors GONG_CONNECTOR_START_TIME.
  • Emits ConnectorFailure on missing call details when continue_on_fail=True.
  • Respects hide_user_info to anonymize speaker names.
  • Updated tests: daily test uses checkpoint loader; added thorough unit tests for paging, workspaces, retries, and flags.

• Migration

  • Replace poll_source/load_from_state with load_from_checkpoint in a checkpoint loop (e.g., via connector_runner CheckpointOutputWrapper).
  • No credential changes required; GONG_CONNECTOR_START_TIME continues to be supported.

^{Written for commit 26a35eb. Summary will update on new commits.}

[justin-tahara]

• feat(gong): Convert to Checkpoint Connector #8827 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cubic-dev-ai]

No issues found across 4 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

[greptile-apps]

Greptile Summary

Successfully converts the Gong connector from the legacy LoadConnector/PollConnector interfaces to the modern CheckpointedConnector pattern. This enables more granular state management and better resumability for large data syncs.

Key improvements:

• Implements checkpoint-based pagination with GongConnectorCheckpoint tracking cursor position and workspace index
• Extracts retry logic into dedicated _fetch_call_details_with_retry() method
• Refactors time range computation into reusable _compute_time_range() helper
• Adds proper ConnectorFailure support for better error tracking

Minor note:
The batch_size parameter is initialized but never used in the new implementation. The checkpointed architecture processes one API page at a time rather than batching internally. Consider removing this parameter if it's no longer needed, or documenting why it's retained for backward compatibility.

Confidence Score: 5/5

• This PR is safe to merge with high confidence
• The checkpoint conversion is well-implemented with comprehensive test coverage. The core logic correctly handles pagination, multi-workspace scenarios, error cases, and state management. All edge cases (404 responses, invalid workspaces, missing call details) are properly tested. The only minor issue is an unused batch_size parameter which doesn't affect functionality.
• No files require special attention

Important Files Changed

Filename	Overview
backend/onyx/connectors/gong/connector.py	Converted from LoadConnector/PollConnector to CheckpointedConnector with proper pagination and state management. Refactored retry logic into separate method, added checkpoint model, and improved error handling with ConnectorFailure support.
backend/tests/unit/onyx/connectors/gong/test_gong_checkpointing.py	Comprehensive unit tests covering checkpoint serialization, pagination, multi-workspace handling, 404 responses, error handling, and hide_user_info flag. All test scenarios appear well-designed.
backend/tests/daily/connectors/gong/test_gong.py	Updated integration test to use new checkpoint-based loading mechanism. Test validates basic functionality with real Gong API credentials.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    Start([load_from_checkpoint called]) --> Init[Initialize from checkpoint<br/>workspace_idx and cursor]
    Init --> LoopStart{workspace_idx less than<br/>workspace count?}
    
    LoopStart -->|No| AllDone[Set has_more=False<br/>Return checkpoint]
    LoopStart -->|Yes| CheckCursor{Has cursor from<br/>checkpoint?}
    
    CheckCursor -->|Yes| AddCursor[Add cursor to request body]
    CheckCursor -->|No| FetchTranscripts[Fetch transcripts for workspace]
    AddCursor --> FetchTranscripts
    
    FetchTranscripts --> Check404{Status 404?}
    Check404 -->|Yes| IncrementWS[Increment workspace_idx<br/>Continue to next workspace]
    IncrementWS --> LoopStart
    
    Check404 -->|No| FetchDetails[Fetch call details with retry]
    FetchDetails --> YieldDocs[Yield documents/failures]
    YieldDocs --> CheckNextCursor{Response has<br/>cursor?}
    
    CheckNextCursor -->|Yes| SaveCursor[Save cursor<br/>Keep same workspace_idx<br/>Set has_more=True]
    CheckNextCursor -->|No| NextWS[Clear cursor<br/>Increment workspace_idx<br/>Set has_more accordingly]
    
    SaveCursor --> Return[Return checkpoint]
    NextWS --> Return
    Return --> End([Checkpoint returned])
    AllDone --> End

Loading

_{Last reviewed commit: 26a35eb}

[github-actions]

🖼️ Visual Regression Report

Project	Changed	Added	Removed	Unchanged	Report
admin	0	0	0	109	✅ No changes
exclusive	0	0	0	8	✅ No changes