fix: handle non-JSON-serializable types in serialization fallback (AI-Triage PR)

[devin-ai-integration]

Summary

Adds a second fallback in airbyte_message_to_string() to prevent unhandled exceptions (and consequent deadlocks in the concurrent source pipeline) when record data contains types that neither orjson nor stdlib json can serialize.

Problem: The existing fallback chain is orjson.dumps() → json.dumps(). But json.dumps() can also raise TypeError for types like Python complex numbers. When this unhandled exception occurs in a worker thread of the concurrent source pipeline, the main thread deadlocks waiting on queue.get() because the worker silently dies.

Fix: Wrap the json.dumps() fallback in its own try/except, falling back to json.dumps(serialized_message, default=str) which converts any remaining non-serializable values to their string representation.

This is a last-resort path — it only activates when both orjson and stdlib json fail on the same message.

Resolves https://github.com/airbytehq/oncall/issues/11654:

• https://github.com/airbytehq/oncall/issues/11654

Related: airbytehq/airbyte#74883

Review & Testing Checklist for Human

• Silent data coercion is acceptable: The default=str fallback silently converts non-serializable values to string representations (e.g., complex(1,2) → "(1+2j)"). No additional warning is logged when this second fallback fires — only the original orjson warning is emitted. Decide if this is acceptable or if an extra log line should be added.
• Broad except Exception is appropriate here: The inner catch is intentionally broad since this is a last-resort to prevent deadlocks. Confirm this doesn't mask errors that should propagate.
• Verify the fix addresses the reported deadlock scenario: The original issue reports deadlocks in source-google-search-console with complex type values in ctr/position fields. The test covers this path but an integration-level verification with the actual connector would provide higher confidence.

Suggested test plan: Run source-google-search-console with the search_analytics_by_query stream against a real account. Confirm the sync completes without deadlock and that ctr/position values are present (even if stringified as a fallback).

Notes

• The root cause of complex values entering the data flow was separately fixed in CDK PR airbytehq/airbyte-python-cdk#579 (Jinja interpolation rejecting complex types, merged in v6.54.5). This PR hardens the serialization layer as defense-in-depth against any future non-serializable type reaching this code path.
• Prior art: The original orjson → json fallback was added in airbytehq/airbyte-python-cdk#210 (v6.16.1, by Artem Inzhyyants (@artem1205) / Maxime Carbonneau-Leclerc (@maxi297)).

Link to Devin session: https://app.devin.ai/sessions/d8317c1f4ce64f70b5e807425b72aca2

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring

[github-actions]

👋 Greetings, Airbyte Team Member!

Here are some helpful tips and reminders for your convenience.

💡 Show Tips and Tricks

Testing This CDK Version

You can test this version of the CDK using the following:

# Run the CLI from this branch:
uvx 'git+https://github.com/airbytehq/airbyte-python-cdk.git@devin/1773523110-fix-serialization-fallback-complex-types#egg=airbyte-python-cdk[dev]' --help

# Update a connector to use the CDK from this branch ref:
cd airbyte-integrations/connectors/source-example
poe use-cdk-branch devin/1773523110-fix-serialization-fallback-complex-types

PR Slash Commands

Airbyte Maintainers can execute the following slash commands on your PR:

• /autofix - Fixes most formatting and linting issues
• /poetry-lock - Updates poetry.lock file
• /test - Runs connector tests with the updated CDK
• /prerelease - Triggers a prerelease publish with default arguments
• /poe build - Regenerate git-committed build artifacts, such as the pydantic models which are generated from the manifest JSON schema in YAML.
• /poe <command> - Runs any poe command in the CDK environment

📚 Show Repo Guidance

Helpful Resources

• CDK API Reference

📝 Edit this welcome message.

[github-actions]

PyTest Results (Fast)

3 935 tests  +1   3 923 ✅ +1   6m 42s ⏱️ -27s
    1 suites ±0      12 💤 ±0 
    1 files   ±0       0 ❌ ±0 

Results for commit 8320500. ± Comparison against base commit 0e57414.

♻️ This comment has been updated with latest results.

[github-actions]

PyTest Results (Full)

3 938 tests  +1   3 926 ✅ +1   11m 37s ⏱️ +23s
    1 suites ±0      12 💤 ±0 
    1 files   ±0       0 ❌ ±0 

Results for commit 8320500. ± Comparison against base commit 0e57414.

♻️ This comment has been updated with latest results.