feat(langgraph): submit spans from langgraph to APM and LLMObs

[sabrenner]

This PR adds a simple integration and corresponding LLMObs plugin for the LangGraph framework.

Enablement

For both APM and LLMObs, this integration is disabled by default, and only enabled with DD_TRACE_LANGGRAPH_ENABLED=true or patch(langgraph=True), along with _DD_TRACE_LANGGRAPH_ENABLED=true, due to its experimental nature.

APM Tracing

The APM integration adds basic tracing for LangGraph Pregel graph invocations, as well as the execution of each node/step in the graph (traced as RunnableSeq).

LLM Observability Plugin

The LLMObs plugin submits the traces to LLM Observability.

Graph Invocation/Streaming Output Tagging

Because we patch the (a)stream method for both streams and (a)invoke, since they rely on (a)stream under the hood, we are bound to the yields from the stream. The object shape and value depend on the stream_mode passed in to the stream function. For invoke, it defaults to "values", which is perfect for our use case. However, if one were to call stream directly, and either pass in nothing or something like stream_mode=["updates", "debug"], the output we would have for the graph span would appear as:

{
  "type": "task_result",
  "timestamp": "2025-01-03T14:45:26.087608+00:00",
  "step": "3",
  "payload": {
    "id": "399cfe0a-aefd-f67c-6909-fcea80ef508b",
    "name": "e",
    "error": "None",
    "result": [
      "('a_list', ['e'])"
    ],
    "interrupts": []
  }
}

This is a known limitation for now, and a more thought-out solution for either adding on "values" to stream_mode and omitting their yields if it wasn't already present, or having some kind of clever joining, can be added in a later PR or by feature request.

Checklist

• PR author has checked that all the criteria below are met
• The PR description includes an overview of the change
• The PR description articulates the motivation for the change
• The change includes tests OR the PR description describes a testing strategy
• The PR description notes risks associated with the change, if any
• Newly-added code is easy to change
• The change follows the library release note guidelines
• The change includes or references documentation updates if necessary
• Backport labels are set (if applicable)

Reviewer Checklist

• Reviewer has checked that all the criteria below are met
• Title is accurate
• All changes are related to the pull request's stated goal
• Avoids breaking API changes
• Testing strategy adequately addresses listed risks
• Newly-added code is easy to change
• Release note makes sense to a user of the library
• If necessary, author has acknowledged and discussed the performance implications of this PR as reported in the benchmarks PR comment
• Backport labels are set in a manner that is consistent with the release branch maintenance policy

[github-actions]

CODEOWNERS have been resolved as:

.riot/requirements/115283d.txt                                          @DataDog/apm-python
.riot/requirements/18551a1.txt                                          @DataDog/apm-python
.riot/requirements/2bd0151.txt                                          @DataDog/apm-python
.riot/requirements/2e4fbd3.txt                                          @DataDog/apm-python
.riot/requirements/7dd4bcd.txt                                          @DataDog/apm-python
ddtrace/contrib/internal/langgraph/patch.py                             @DataDog/ml-observability
ddtrace/contrib/langgraph/__init__.py                                   @DataDog/ml-observability
ddtrace/llmobs/_integrations/langgraph.py                               @DataDog/ml-observability
tests/contrib/langgraph/__init__.py                                     @DataDog/ml-observability
tests/contrib/langgraph/conftest.py                                     @DataDog/ml-observability
tests/contrib/langgraph/test_langgraph.py                               @DataDog/ml-observability
tests/contrib/langgraph/test_langgraph_llmobs.py                        @DataDog/ml-observability
tests/contrib/langgraph/test_langgraph_patch.py                         @DataDog/ml-observability
.github/CODEOWNERS                                                      @DataDog/python-guild @DataDog/apm-core-python
ddtrace/_monkey.py                                                      @DataDog/apm-core-python
ddtrace/llmobs/_constants.py                                            @DataDog/ml-observability
ddtrace/llmobs/_integrations/base.py                                    @DataDog/ml-observability
ddtrace/llmobs/_integrations/langchain.py                               @DataDog/ml-observability
ddtrace/llmobs/_integrations/utils.py                                   @DataDog/ml-observability
ddtrace/llmobs/_llmobs.py                                               @DataDog/ml-observability
ddtrace/llmobs/_utils.py                                                @DataDog/ml-observability
riotfile.py                                                             @DataDog/apm-python
tests/llmobs/suitespec.yml                                              @DataDog/ml-observability

[pr-commenter]

Benchmarks

Benchmark execution time: 2025-01-17 16:43:15

Comparing candidate commit 1a2dab7 in PR branch sabrenner/langgraph with baseline commit f67a358 in branch main.

Found 0 performance improvements and 4 performance regressions! Performance is the same for 390 metrics, 2 unstable metrics.

scenario:iast_aspects-lower_aspect

• 🟥 execution_time [+253.973ns; +297.848ns] or [+9.423%; +11.050%]

scenario:iast_aspects-ospathdirname_aspect

• 🟥 execution_time [+321.279ns; +390.856ns] or [+8.648%; +10.521%]

scenario:iast_aspects-ospathsplit_aspect

• 🟥 execution_time [+351.346ns; +414.758ns] or [+9.132%; +10.780%]

scenario:iast_aspects-translate_aspect

• 🟥 execution_time [+274.107ns; +325.343ns] or [+7.194%; +8.539%]

[datadog-dd-trace-py-rkomorn]

Datadog Report

Branch report: sabrenner/langgraph
Commit report: 1a2dab7
Test service: dd-trace-py

✅ 0 Failed, 130 Passed, 1468 Skipped, 4m 12.31s Total duration (35m 22.88s time saved)

[erikayasuda]

LGTM for guild/core files 👍

[Yun-Kim]

Final review for me, everything looks good except for a couple small comments!