fix(dashscope): preserve message outputs without tool calls

[sipercai]

Description

This PR fixes DashScope message-format response parsing when the assistant returns
plain text without a tool_calls field.

DashScope response entities are dict-like objects whose attribute access can raise
KeyError for missing optional fields. The previous parser used
getattr(message, "tool_calls", None), so a result_format="message" text response
without tool_calls could discard the already parsed message.content and omit
gen_ai.output.messages.

This change adds safe field access for DashScope response objects, preserves
assistant text output in this response shape, adds a VCR regression cassette, and
updates the DashScope changelog.

Fixes # (N/A)

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

• tox -e precommit
• OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=SPAN_ONLY tox -c tox-loongsuite.ini -e py311-test-loongsuite-instrumentation-dashscope-latest -- instrumentation-loongsuite/loongsuite-instrumentation-dashscope/tests/test_generation.py::test_generation_call_message_result_without_tool_calls_content_capture -q

Does This PR Require a Core Repo Change?

• Yes. - Link to PR:
• No.

Checklist:

• Followed the style guidelines of this project
• Changelogs have been updated
• Unit tests have been added
• Documentation has been updated

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

This PR fixes DashScope message-format response parsing so assistant text output is preserved when the response omits the optional tool_calls field.

Changes:

• Add a safe field accessor to avoid KeyError from DashScope’s dict-like response objects and use it throughout generation parsing.
• Add a VCR-based regression test (and cassette) covering result_format="message" responses without tool_calls.
• Update DashScope instrumentation changelog and adjust dependency / tooling configs.

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
instrumentation-loongsuite/loongsuite-instrumentation-dashscope/src/opentelemetry/instrumentation/dashscope/utils/generation.py	Introduces _safe_get and applies it across response parsing to preserve message output without tool_calls.
instrumentation-loongsuite/loongsuite-instrumentation-dashscope/tests/test_generation.py	Adds a regression test validating gen_ai.output.messages is emitted for message-format responses without tool_calls.
instrumentation-loongsuite/loongsuite-instrumentation-dashscope/tests/cassettes/test_generation_call_message_result_without_tool_calls_content_capture.yaml	Adds recorded HTTP interaction for the new regression test.
instrumentation-loongsuite/loongsuite-instrumentation-dashscope/pyproject.toml	Removes version constraints for opentelemetry-util-genai.
instrumentation-loongsuite/loongsuite-instrumentation-dashscope/CHANGELOG.md	Adds an Unreleased “Fixed” entry documenting the parsing fix.
.pre-commit-config.yaml	Excludes _template/ from the rstcheck hook.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[sipercai]

This package depends on the latest opentelemetry-util-genai implementation maintained in the LoongSuite repo, so we intentionally avoid pinning or version-constraining this dependency.