Add otel langchain instrumentation

[Radhika Gupta (rads-1996)]

No description provided.

[Copilot]

Pull request overview

This PR introduces a new opentelemetry-instrumentation-langchain package that instruments LangChain chat model executions (notably ChatOpenAI / ChatBedrock) using OpenTelemetry GenAI semantic conventions, along with a test suite that validates spans, metrics, and logs via VCR cassettes.

Changes:

• Added a LangChain instrumentor that injects an OpenTelemetry callback handler into LangChain’s callback manager.
• Implemented an invocation-state manager and callback handler logic to start/stop/fail LLM invocations and populate GenAI attributes.
• Added pytest coverage (including VCR cassettes) to assert emitted spans/metrics/logs for success and error scenarios.

Reviewed changes

Copilot reviewed 13 out of 14 changed files in this pull request and generated 19 comments.

Show a summary per file

File	Description
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/init.py	Defines LangChainInstrumentor and wrap/unwrap wiring for callback manager injection
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/callback_handler.py	Implements LangChain callback handler to build GenAI invocation objects and stop/fail them
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/invocation_manager.py	Stores invocation state keyed by run_id with basic parent/child tracking
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/span_manager.py	Adds a span manager abstraction for run_id-based span lifecycle management
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/package.py	Declares instrumentation dependency metadata
src/_langchain/opentelemetry-instrumentation-langchain/src/opentelemetry/instrumentation/langchain/version.py	Adds package version definition
src/_langchain/opentelemetry-instrumentation-langchain/tests/conftest.py	Test fixtures for OTel SDK in-memory exporters + VCR configuration/scrubbing
src/_langchain/opentelemetry-instrumentation-langchain/tests/test_llm_call.py	End-to-end assertions for spans/metrics/logs for ChatOpenAI invoke success/error
src/_langchain/opentelemetry-instrumentation-langchain/tests/test_invocation_manager.py	Unit tests for _InvocationManager behavior
src/_langchain/opentelemetry-instrumentation-langchain/tests/cassettes/test_chat_openai_gpt_3_5_turbo_model_llm_call.yaml	Recorded success response cassette
src/_langchain/opentelemetry-instrumentation-langchain/tests/cassettes/test_chat_openai_gpt_3_5_turbo_model_llm_call_with_error.yaml	Recorded auth error response cassette
src/_langchain/opentelemetry-instrumentation-langchain/pyproject.toml	Hatch-based packaging config for the new instrumentation package
src/_langchain/opentelemetry-instrumentation-langchain/dev_requirements.txt	Dev/test requirements for the new package

---

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

[Ram Thiru (ramthi)]

Do we need to copy the code over or can we take a reference to OTel instrumentation

[Radhika Gupta (rads-1996)]

This instrumentation package in otel is not published yet, therefore we need to copy the code till the upstream package is released and in a good state. Until then we will maintain this internally in the distro.