[v-pratap]

ReadPayload stores its data in an absl::Cord, but the only accessor (contents()) returns a std::vector. Applications that need to retain the data beyond the lifetime of the payload must copy it.

This PR adds ReadPayload::cord_contents(), which exposes the underlying absl::Cord:

absl::Cord const& cord_contents() const&;
absl::Cord cord_contents() &&;

Copying an absl::Cord is cheap (the buffers are reference counted), and the rvalue overload lets applications move the Cord out of the payload. Retaining the Cord keeps only the data buffers alive, not the RPC that received them (the content is stolen from the response proto before the payload is created).

Testing

• Unit tests in read_payload_impl_test.cc: contents, multi-chunk payloads, empty payloads, and that copies / moves share the underlying buffer (no data copy) and moving transfers ownership.
• Integration tests in async_client_integration_test.cc, for both AsyncClient::ReadObject() and AsyncClient::Open() + ObjectDescriptor::Read():
  • cord_contents() refers to the same buffers as contents(), and copies share them.
  • Cords retained from each payload remain valid and contain the full object after the payloads, reader, and client are destroyed.
  • StreamingReadCordContents uses InsertObject (works with standard test buckets and the testbench emulator), and OpenReadCordContents uses StartAppendableObjectUpload (supported on zonal buckets in production and the testbench emulator).

b/533700890

[gemini-code-assist]

Code Review

This pull request introduces cord_contents() to ReadPayload to allow retrieving payload contents as an absl::Cord, which is useful for sharing underlying buffers without copying. It also adds corresponding unit tests and integration tests to verify this behavior. The review feedback highlights several style guide violations: std::string_view and std::optional should be preferred over their absl counterparts, and the use of auto should be avoided when it obscures StatusOr or other complex return types.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.39%. Comparing base (61709d3) to head (db53a9f).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #16539      +/-   ##
==========================================
- Coverage   92.40%   92.39%   -0.01%     
==========================================
  Files        2265     2265              
  Lines      219433   219470      +37     
==========================================
+ Hits       202760   202781      +21     
- Misses      16673    16689      +16     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[kalragauri]

Please update the note in the PR description about regional buckets: StartAppendableObjectUpload (appendable: true) is only supported on zonal buckets in production (and the testbench emulator).

[kalragauri]

clang-tidy fails on this PR because of one of the new test cases. Please address those failures.

[v-pratap]

Done. Added // NOLINTNEXTLINE(performance-unnecessary-copy-initialization) where absl::Cord is intentionally copy-constructed in the tests, and verified clang-tidy passes cleanly.

[v-pratap]

Updated the PR description and addressed all review comments. Thanks for the review!