fix(postgrest): handle maybe_single zero-row responses correctly

[AhmedAli58]

Summary

Fixes #1207.

maybe_single() now treats zero-row responses returned as HTTP 204 (No Content) as an empty-success case and returns SingleAPIResponse(data=None) instead of raising APIError.

What changed

• Updated SyncMaybeSingleRequestBuilder.execute() and AsyncMaybeSingleRequestBuilder.execute() to treat 204/no-row outcomes as data=None.
• Added a small helper in both builders to recognize no-row conditions from either:
  • explicit 204 errors
  • legacy "The result contains 0 rows" error details
• Normalized the parsed empty-body [] path in maybe_single() to data=None.

Behavior guarantees

• maybe_single() on zero rows: returns response object with data=None.
• single() behavior is unchanged: zero-row error responses still raise APIError.
• Non-zero-row error handling remains unchanged.

Tests

Added regression coverage in both sync and async client tests:

• test_maybe_single_returns_none_on_no_rows_204
• test_single_raises_on_no_rows

Verification

• make ruff passed
• uv run --package postgrest mypy src/postgrest tests passed
• uv run --package postgrest pytest src/postgrest/tests/_sync/test_client.py src/postgrest/tests/_async/test_client.py -vv passed
• uv run --package postgrest pytest src/postgrest/tests -vv -k "not integration" passed

Note: make postgrest.tests could not run integration tests in this environment because docker is not installed (/bin/sh: docker: command not found).

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

Configuration used: Central YAML (base), Organization UI (inherited)

Review profile: CHILL

Plan: Pro

Cache: Disabled due to Reviews > Disable Cache setting

Disabled knowledge base sources:

• Linear integration is disabled

You can enable these sources in your CodeRabbit configuration.

📥 Commits

Reviewing files that changed from the base of the PR and between 59e3384 and 79a64a8.

📒 Files selected for processing (4)

• src/postgrest/src/postgrest/_async/request_builder.py
• src/postgrest/src/postgrest/_sync/request_builder.py
• src/postgrest/tests/_async/test_client.py
• src/postgrest/tests/_sync/test_client.py

---

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes

  • Fixed maybe_single() queries to gracefully handle no-row responses by returning None for the data field instead of raising an error.
  • Single queries maintain their strict behavior by raising errors when no matching rows are found.

• Tests

  • Added comprehensive test coverage for no-row edge cases in both single and maybe_single query types.

Walkthrough

This pull request fixes a bug where maybe_single() incorrectly raises an APIError when a query returns zero rows. The fix introduces private helper functions and constants to detect "no rows" error responses in both async and sync request builders. When a "no rows" error is detected, instead of propagating the error, the code now returns a SingleAPIResponse with data=None. Additionally, when a successful single request returns an empty list, it is converted to return data=None while preserving the count. Corresponding test cases are added to verify the corrected behavior for both async and sync implementations.

Assessment against linked issues

Objective	Addressed	Explanation
maybe_single() should return APIResponse(data=None) when 0 rows are returned instead of raising APIError [#1207]	✅

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[o-santi]

This would be a breaking change, so we won't be doing it for v2. I'll be fixing this in the next major version myself, you can follow it's development here #1370.