fix: Fix Langchain

[Dwij1704]

Closes #2424 #2429 #2474

• Updated langchain dependency to version 1.0.0 or higher.
• Introduced langchain-classic for legacy support.
• Enhanced README with compatibility information and installation instructions for both LangChain 1.x and LangChain Classic.
• Updated example code to reflect new agent framework usage.
• Added _is_json_parseable helper with optimizes approach
• Handled RemoveMessage from langchain_core
• Handled missing type in _get_attributes_from_message_content to prevent early return

---

Note

Adds LangChain 1.x and Classic compatibility, improves _convert_io to detect stringified JSON with MIME tagging, updates examples/docs, and adjusts deps/tests accordingly.

• Instrumentation (core):
  • Enhance _convert_io to detect stringified JSON and emit application/json MIME type; add _is_json_parseable.
  • Make message role mapping more robust (support RemoveMessage, optional roles) and tolerate missing content.type.
• Docs & Examples:
  • Update README.md with compatibility, install paths for LangChain 1.x and Classic, and new agent-based quickstart.
  • Add examples: examples/langchain_v1_agent.py, examples/langchain_v1_with_middleware.py; update langgraph_agent_supervisor.py to use langchain_classic.
• Dependencies/Config:
  • Bump requirements to langchain>=1.0.0 and add langchain-classic>=1.0.0; update pyproject.toml optional test deps accordingly; refresh examples/requirements.txt.
• Tests:
  • Expand _convert_io test coverage, including JSON-string detection, edge cases, performance, and advanced types.

^{Written by Cursor Bugbot for commit a6c142d. This will update automatically on new commits. Configure here.}

[github-actions]

CLA Assistant Lite bot All contributors have signed the CLA ✍️ ✅

[Dwij1704]

I have read the CLA Document and I hereby sign the CLA

[cursor]

Bug: RemoveMessage triggers unnecessary warning log on every occurrence

The RemoveMessage handling in _map_class_name_to_role returns None to indicate no role is needed. However, in _extract_message_role, when role is None after all strategies are tried, a warning is logged. This means every RemoveMessage processed will trigger a "Unable to determine message role" warning, even though returning None for RemoveMessage is intentional and expected behavior. In LangGraph applications that use message removal frequently, this could cause significant log noise in production.

Additional Locations (1)

• python/instrumentation/openinference-instrumentation-langchain/src/openinference/instrumentation/langchain/_tracer.py#L750-L759