chore(langchain/openai): enable service naming [backport 1.17] (#6444)

Backport 7d675fd from #6407 to 1.17.

This change updates the langchain and openai contribs to support
schematized service names.

== Special Note ==
These changes are gated behind a feature flag and are not expected to be
stable until a release note is generated. Existing customers should be
unaffected.

== Service Naming Summary ==
Datadog is updating its method of determining service names!

Service naming introduces a few concepts:

* Versioned service name "schemas":
  * v0: The existing mechanism for determining service names
* v1: Replacing the 'default' contrib-specific service name with the env
var $DD_SERVICE

These changes are being in conjunction with changes to peer.service
updates to bring a better service-naming experience to customers. In
particular, these changes will help:

* Remove the creation of fake services, such as grpc.client *
Standardize service names across and within tracers * Remove the need to
create as many fake services to take advantage of Datadog features, such
as metrics * Make more sense to humans

== Operation Naming Summary ==
The service name changes mean that generated spans can no longer rely on
`span.service` to differentiate
between processes. A subclass of spans now have a new problem -
differentiating between inbound and outbound requests (for the purposes
of metrics). For instance, spans with names `http.request` used to be
differentiated by the service they belonged to (incoming requests under
`span.service=$DD_SERVICE` and outgoing under
`span.service=$DD_CONTRIB_SERVICE_NAME`). These are now no longer
differentiated.

To remedy these issues, operation names are also being changed as part
of the service naming effort. In the above example, `http.request` turns
into `http.server.request` (incoming) and `http.client.request`
(outgoing).

## Checklist

- [x] Change(s) are motivated and described in the PR description.
- [x] Testing strategy is described if automated tests are not included
in the PR.
- [x] Risk is outlined (performance impact, potential for breakage,
maintainability, etc).
- [x] Change is maintainable (easy to change, telemetry, documentation).
- [x] [Library release note
guidelines](https://ddtrace.readthedocs.io/en/stable/releasenotes.html)
are followed. If no release note is required, add label
`changelog/no-changelog`.
- [x] Documentation is included (in-code, generated user docs, [public
corp docs](https://github.com/DataDog/documentation/)).
- [x] Backport labels are set (if
[applicable](https://ddtrace.readthedocs.io/en/latest/contributing.html#backporting))

## Reviewer Checklist

- [x] Title is accurate.
- [x] No unnecessary changes are introduced.
- [x] Description motivates each change.
- [x] Avoids breaking
[API](https://ddtrace.readthedocs.io/en/stable/versioning.html#interfaces)
changes unless absolutely necessary.
- [x] Testing strategy adequately addresses listed risk(s).
- [x] Change is maintainable (easy to change, telemetry, documentation).
- [x] Release note makes sense to a user of the library.
- [x] Reviewer has explicitly acknowledged and discussed the performance
implications of this PR as reported in the benchmarks PR comment.
- [x] Backport labels are set in a manner that is consistent with the
[release branch maintenance
policy](https://ddtrace.readthedocs.io/en/latest/contributing.html#backporting)

Co-authored-by: Teague Bick <[email protected]>

Author: github-actions[bot]

ddtrace/contrib/openai/patch.py

@@ -8,6 +8,7 @@
 from ddtrace.internal.agent import get_stats_url
 from ddtrace.internal.constants import COMPONENT
 from ddtrace.internal.logger import get_logger
+from ddtrace.internal.schema import schematize_service_name
 from ddtrace.internal.utils.formats import asbool
 from ddtrace.internal.wrapping import wrap
 
@@ -333,7 +334,8 @@ def _patched_make_session(func, args, kwargs):
     This should technically be a ``peer.service`` but this concept doesn't exist yet.
     """
     session = func(*args, **kwargs)
-    Pin.override(session, service="openai")
+    service = schematize_service_name("openai")
+    Pin.override(session, service=service)
     return session
 
 

tests/contrib/langchain/test_langchain.py

@@ -1285,6 +1285,46 @@ def test_openai_integration(langchain, request_vcr, ddtrace_run_python_code_in_s
     assert err == b""
 
 
+@pytest.mark.skipif(sys.version_info < (3, 10, 0), reason="Requires unnecessary cassette file for Python 3.9")
+@pytest.mark.snapshot(ignores=["metrics.langchain.tokens.total_cost"])
+@pytest.mark.parametrize("schema_version", [None, "v0", "v1"])
+@pytest.mark.parametrize("service_name", [None, "mysvc"])
+def test_openai_service_name(
+    langchain, request_vcr, ddtrace_run_python_code_in_subprocess, schema_version, service_name
+):
+    env = os.environ.copy()
+    pypath = [os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(__file__))))]
+    if "PYTHONPATH" in env:
+        pypath.append(env["PYTHONPATH"])
+    env.update(
+        {
+            "PYTHONPATH": ":".join(pypath),
+            # Disable metrics because the test agent doesn't support metrics
+            "DD_LANGCHAIN_METRICS_ENABLED": "false",
+            "DD_OPENAI_METRICS_ENABLED": "false",
+            "OPENAI_API_KEY": "<not-a-real-key>",
+        }
+    )
+    if service_name:
+        env["DD_SERVICE"] = service_name
+    if schema_version:
+        env["DD_TRACE_SPAN_ATTRIBUTE_SCHEMA"] = schema_version
+    out, err, status, pid = ddtrace_run_python_code_in_subprocess(
+        """
+from langchain.llms import OpenAI
+import ddtrace
+from tests.contrib.langchain.test_langchain import get_request_vcr
+llm = OpenAI()
+with get_request_vcr().use_cassette("openai_completion_sync.yaml"):
+    llm("Can you explain what Descartes meant by 'I think, therefore I am'?")
+""",
+        env=env,
+    )
+    assert status == 0, err
+    assert out == b""
+    assert err == b""
+
+
 @pytest.mark.parametrize(
     "ddtrace_config_langchain",
     [

tests/contrib/openai/test_openai.py

@@ -2092,3 +2092,40 @@ def test_est_tokens():
         )
         == 97
     )  # oracle: 92
+
+
+@pytest.mark.snapshot(ignores=["meta.http.useragent"], async_mode=False)
+@pytest.mark.parametrize("schema_version", [None, "v0", "v1"])
+@pytest.mark.parametrize("service_name", [None, "mysvc"])
+def test_integration_service_name(openai_api_key, ddtrace_run_python_code_in_subprocess, schema_version, service_name):
+    env = os.environ.copy()
+    pypath = [os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(__file__))))]
+    if "PYTHONPATH" in env:
+        pypath.append(env["PYTHONPATH"])
+    env.update(
+        {
+            "OPENAI_API_KEY": openai_api_key,
+            "PYTHONPATH": ":".join(pypath),
+            # Disable metrics because the test agent doesn't support metrics
+            "DD_OPENAI_METRICS_ENABLED": "false",
+        }
+    )
+    if schema_version:
+        env["DD_TRACE_SPAN_ATTRIBUTE_SCHEMA"] = schema_version
+    if service_name:
+        env["DD_SERVICE"] = service_name
+    out, err, status, pid = ddtrace_run_python_code_in_subprocess(
+        """
+import openai
+import ddtrace
+from tests.contrib.openai.test_openai import FilterOrg, get_openai_vcr
+pin = ddtrace.Pin.get_from(openai)
+pin.tracer.configure(settings={"FILTERS": [FilterOrg()]})
+with get_openai_vcr().use_cassette("completion_2.yaml"):
+    resp = openai.Completion.create(model="ada", prompt="hello world")
+""",
+        env=env,
+    )
+    assert status == 0, err
+    assert out == b""
+    assert err == b""

tests/snapshots/tests.contrib.langchain.test_langchain.test_openai_service_name[None-None].json

@@ -0,0 +1,112 @@
+[[
+  {
+    "name": "langchain.request",
+    "service": "",
+    "resource": "langchain.llms.openai.OpenAI",
+    "trace_id": 0,
+    "span_id": 1,
+    "parent_id": 0,
+    "type": "",
+    "error": 0,
+    "meta": {
+      "_dd.p.dm": "-0",
+      "langchain.request.api_key": "...key>",
+      "langchain.request.model": "text-davinci-003",
+      "langchain.request.openai.parameters.frequency_penalty": "0",
+      "langchain.request.openai.parameters.max_tokens": "256",
+      "langchain.request.openai.parameters.model_name": "text-davinci-003",
+      "langchain.request.openai.parameters.n": "1",
+      "langchain.request.openai.parameters.presence_penalty": "0",
+      "langchain.request.openai.parameters.request_timeout": "None",
+      "langchain.request.openai.parameters.temperature": "0.7",
+      "langchain.request.openai.parameters.top_p": "1",
+      "langchain.request.prompts.0": "Can you explain what Descartes meant by 'I think, therefore I am'?",
+      "langchain.request.provider": "openai",
+      "langchain.request.type": "llm",
+      "langchain.response.completions.0.finish_reason": "stop",
+      "langchain.response.completions.0.logprobs": "None",
+      "langchain.response.completions.0.text": "\\n\\nDescartes' famous phrase \"I think, therefore I am\" is a fundamental statement of his philosophical approach, known as Cartes...",
+      "language": "python",
+      "runtime-id": "d24e6145724d4113a14c083bbe74f8bf"
+    },
+    "metrics": {
+      "_dd.agent_psr": 1.0,
+      "_dd.measured": 1,
+      "_dd.top_level": 1,
+      "_dd.tracer_kr": 1.0,
+      "_sampling_priority_v1": 1,
+      "langchain.tokens.completion_tokens": 95,
+      "langchain.tokens.prompt_tokens": 17,
+      "langchain.tokens.total_cost": 0.0022400000000000002,
+      "langchain.tokens.total_tokens": 112,
+      "process_id": 1184
+    },
+    "duration": 9396250,
+    "start": 1689781623043723458
+  },
+     {
+       "name": "openai.request",
+       "service": "",
+       "resource": "createCompletion",
+       "trace_id": 0,
+       "span_id": 2,
+       "parent_id": 1,
+       "type": "",
+       "error": 0,
+       "meta": {
+         "component": "openai",
+         "openai.api_base": "https://api.openai.com/v1",
+         "openai.api_type": "open_ai",
+         "openai.organization.name": "datadog-4",
+         "openai.request.endpoint": "/v1/completions",
+         "openai.request.frequency_penalty": "0",
+         "openai.request.max_tokens": "256",
+         "openai.request.method": "POST",
+         "openai.request.model": "text-davinci-003",
+         "openai.request.n": "1",
+         "openai.request.presence_penalty": "0",
+         "openai.request.prompt.0": "Can you explain what Descartes meant by 'I think, therefore I am'?",
+         "openai.request.temperature": "0.7",
+         "openai.request.top_p": "1",
+         "openai.response.choices.0.finish_reason": "stop",
+         "openai.response.choices.0.logprobs": "returned",
+         "openai.response.choices.0.text": "\\n\\nDescartes' famous phrase \"I think, therefore I am\" is a fundamental statement of his philosophical approach, known as Cartes...",
+         "openai.response.id": "cmpl-7TZs9SoIjYLVNCYmAqIz9ihMX1grP",
+         "openai.response.model": "text-davinci-003",
+         "openai.user.api_key": "sk-...key>"
+       },
+       "metrics": {
+         "_dd.measured": 1,
+         "openai.organization.ratelimit.requests.remaining": 2999,
+         "openai.organization.ratelimit.tokens.remaining": 249744,
+         "openai.response.choices_count": 1,
+         "openai.response.created": 1687283761
+       },
+       "duration": 8934709,
+       "start": 1689781623044016166
+     },
+        {
+          "name": "requests.request",
+          "service": "openai",
+          "resource": "requests.request",
+          "trace_id": 0,
+          "span_id": 3,
+          "parent_id": 2,
+          "type": "http",
+          "error": 0,
+          "meta": {
+            "component": "requests",
+            "http.method": "POST",
+            "http.status_code": "200",
+            "http.url": "https://api.openai.com/v1/completions",
+            "http.useragent": "OpenAI/v1 PythonBindings/0.27.8",
+            "out.host": "api.openai.com",
+            "span.kind": "client"
+          },
+          "metrics": {
+            "_dd.measured": 1,
+            "_dd.top_level": 1
+          },
+          "duration": 2049792,
+          "start": 1689781623047737333
+        }]]

tests/snapshots/tests.contrib.langchain.test_langchain.test_openai_service_name[None-v0].json

@@ -0,0 +1,112 @@
+[[
+  {
+    "name": "langchain.request",
+    "service": "",
+    "resource": "langchain.llms.openai.OpenAI",
+    "trace_id": 0,
+    "span_id": 1,
+    "parent_id": 0,
+    "type": "",
+    "error": 0,
+    "meta": {
+      "_dd.p.dm": "-0",
+      "langchain.request.api_key": "...key>",
+      "langchain.request.model": "text-davinci-003",
+      "langchain.request.openai.parameters.frequency_penalty": "0",
+      "langchain.request.openai.parameters.max_tokens": "256",
+      "langchain.request.openai.parameters.model_name": "text-davinci-003",
+      "langchain.request.openai.parameters.n": "1",
+      "langchain.request.openai.parameters.presence_penalty": "0",
+      "langchain.request.openai.parameters.request_timeout": "None",
+      "langchain.request.openai.parameters.temperature": "0.7",
+      "langchain.request.openai.parameters.top_p": "1",
+      "langchain.request.prompts.0": "Can you explain what Descartes meant by 'I think, therefore I am'?",
+      "langchain.request.provider": "openai",
+      "langchain.request.type": "llm",
+      "langchain.response.completions.0.finish_reason": "stop",
+      "langchain.response.completions.0.logprobs": "None",
+      "langchain.response.completions.0.text": "\\n\\nDescartes' famous phrase \"I think, therefore I am\" is a fundamental statement of his philosophical approach, known as Cartes...",
+      "language": "python",
+      "runtime-id": "ab50b44eb3054abf9bc73864166b5721"
+    },
+    "metrics": {
+      "_dd.agent_psr": 1.0,
+      "_dd.measured": 1,
+      "_dd.top_level": 1,
+      "_dd.tracer_kr": 1.0,
+      "_sampling_priority_v1": 1,
+      "langchain.tokens.completion_tokens": 95,
+      "langchain.tokens.prompt_tokens": 17,
+      "langchain.tokens.total_cost": 0.0022400000000000002,
+      "langchain.tokens.total_tokens": 112,
+      "process_id": 1197
+    },
+    "duration": 9884916,
+    "start": 1689781625795789668
+  },
+     {
+       "name": "openai.request",
+       "service": "",
+       "resource": "createCompletion",
+       "trace_id": 0,
+       "span_id": 2,
+       "parent_id": 1,
+       "type": "",
+       "error": 0,
+       "meta": {
+         "component": "openai",
+         "openai.api_base": "https://api.openai.com/v1",
+         "openai.api_type": "open_ai",
+         "openai.organization.name": "datadog-4",
+         "openai.request.endpoint": "/v1/completions",
+         "openai.request.frequency_penalty": "0",
+         "openai.request.max_tokens": "256",
+         "openai.request.method": "POST",
+         "openai.request.model": "text-davinci-003",
+         "openai.request.n": "1",
+         "openai.request.presence_penalty": "0",
+         "openai.request.prompt.0": "Can you explain what Descartes meant by 'I think, therefore I am'?",
+         "openai.request.temperature": "0.7",
+         "openai.request.top_p": "1",
+         "openai.response.choices.0.finish_reason": "stop",
+         "openai.response.choices.0.logprobs": "returned",
+         "openai.response.choices.0.text": "\\n\\nDescartes' famous phrase \"I think, therefore I am\" is a fundamental statement of his philosophical approach, known as Cartes...",
+         "openai.response.id": "cmpl-7TZs9SoIjYLVNCYmAqIz9ihMX1grP",
+         "openai.response.model": "text-davinci-003",
+         "openai.user.api_key": "sk-...key>"
+       },
+       "metrics": {
+         "_dd.measured": 1,
+         "openai.organization.ratelimit.requests.remaining": 2999,
+         "openai.organization.ratelimit.tokens.remaining": 249744,
+         "openai.response.choices_count": 1,
+         "openai.response.created": 1687283761
+       },
+       "duration": 9303209,
+       "start": 1689781625796142209
+     },
+        {
+          "name": "requests.request",
+          "service": "openai",
+          "resource": "requests.request",
+          "trace_id": 0,
+          "span_id": 3,
+          "parent_id": 2,
+          "type": "http",
+          "error": 0,
+          "meta": {
+            "component": "requests",
+            "http.method": "POST",
+            "http.status_code": "200",
+            "http.url": "https://api.openai.com/v1/completions",
+            "http.useragent": "OpenAI/v1 PythonBindings/0.27.8",
+            "out.host": "api.openai.com",
+            "span.kind": "client"
+          },
+          "metrics": {
+            "_dd.measured": 1,
+            "_dd.top_level": 1
+          },
+          "duration": 2005833,
+          "start": 1689781625799821543
+        }]]