utf8 bug fix

[dibahlfi]

This PR introduces two independent fixes in Cosmos Python request/response handling.

Fix 1: Content-Length calculation for request bodies-
The Content-Length header is updated to use the UTF-8 encoded byte length instead of the Unicode character count.
Previously, len(body) was used, which returns character length. For non-ASCII payloads, this can under-report the actual size. For example, the string { "name": "café" } has a character length of 15 but a UTF-8 byte length of 16 because "é" is multi-byte.
This mismatch can lead to incorrect wire-level payload sizing and potential request failures. The fix ensures accurate byte-level accounting across both sync and async request paths.

Fix 2: Optional fallback for malformed UTF-8 in responses-
Adds an opt-in mechanism to handle response bodies that contain invalid UTF-8 sequences.
Default behavior remains unchanged: strict decoding raises UnicodeDecodeError when malformed UTF-8 is encountered.
An environment variable controls the fallback behavior:
COSMOS.CHARSET_DECODER_ERROR_ACTION_ON_MALFORMED_INPUT
REPLACE: invalid bytes are replaced with the Unicode replacement character
IGNORE: invalid bytes are skipped during decoding
This enables workloads to continue when encountering malformed data without changing default strict behavior.

Compatibility and risk-
No change to default decoding behavior (still strict).
Fallback behavior is explicitly opt-in via environment variable.
Content-Length change only corrects byte calculation and does not alter request semantics.

[Copilot]

Pull request overview

This PR improves Cosmos Python SDK reliability around request framing and response decoding by (1) correcting Content-Length calculation for UTF-8 payloads and (2) introducing an opt-in, environment-variable-driven fallback when decoding malformed UTF-8 response bodies.

Changes:

• Compute Content-Length using UTF-8 byte length (len(body.encode("utf-8"))) in both sync and async request paths.
• Add _response_decoding.decode_response_body() to keep strict UTF-8 decoding by default while enabling REPLACE/IGNORE fallback via COSMOS.CHARSET_DECODER_ERROR_ACTION_ON_MALFORMED_INPUT.
• Add unit tests covering decoding behavior (strict + opt-in fallback) and regression tests for Content-Length wiring in sync/async paths.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
sdk/cosmos/azure-cosmos/azure/cosmos/_synchronized_request.py	Uses shared decode helper for response bodies; fixes sync Content-Length UTF-8 byte accounting.
sdk/cosmos/azure-cosmos/azure/cosmos/aio/_asynchronous_request.py	Uses shared decode helper for response bodies; fixes async Content-Length UTF-8 byte accounting.
sdk/cosmos/azure-cosmos/azure/cosmos/_response_decoding.py	New helper module implementing strict decode + opt-in permissive fallback with logging.
sdk/cosmos/azure-cosmos/tests/test_response_decoding.py	New tests validating strict behavior, actionable error hinting, and env-var-driven fallback modes.
sdk/cosmos/azure-cosmos/tests/test_content_length_encoding.py	New regression tests validating UTF-8 byte-length Content-Length (including sync/async wiring).
sdk/cosmos/azure-cosmos/CHANGELOG.md	Documents the Content-Length fix and the opt-in malformed UTF-8 decode fallback.

[dibahlfi]

@sdkReviewAgent-2

[xinlian12]

✅ Review complete (44:43)

Posted 2 inline comment(s).

_{Steps: ✓ context, correctness, cross-sdk, design, history, past-prs, synthesis, test-coverage}

[dibahlfi]

/azp run python - cosmos - tests

[dibahlfi]

@sdkReviewAgent-2

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).