feat(gooddata-eval): name gooddata-eval in Langfuse requests and spans

Langfuse groups API traffic by User-Agent, so every gooddata-eval call
showed up as python-httpx/0.28.1, indistinguishable from other callers.
Its exported root spans carried only service.name, so a trace did not
say which build or CI run produced it.

Every client from make_http_client now sends
gooddata-eval/<version> (<ci|local>; run=<GITHUB_RUN_ID>; sha=<12
chars of GITHUB_SHA>), with "local" for any value unset or unsafe in a
header. The OTLP resource adds service.version and, when set,
github.run_id, github.workflow, github.ref_name and github.event_name.

jira: trivial
risk: low

Author: tychtjan

packages/gooddata-eval/src/gooddata_eval/core/langfuse/_env.py

@@ -5,10 +5,25 @@
 
 import base64
 import os
+import re
 
 import httpx
 
+from gooddata_eval._version import __version__
+
 _DEFAULT_BASE_URL = "https://us.cloud.langfuse.com"
+SERVICE_NAME = "gooddata-eval"
+
+_LOCAL = "local"
+_HEADER_SAFE = re.compile(r"[A-Za-z0-9._-]+")
+_SHORT_SHA_LENGTH = 12
+# Resource attribute -> the GitHub Actions variable it is read from.
+_GITHUB_RESOURCE = {
+    "github.run_id": "GITHUB_RUN_ID",
+    "github.workflow": "GITHUB_WORKFLOW",
+    "github.ref_name": "GITHUB_REF_NAME",
+    "github.event_name": "GITHUB_EVENT_NAME",
+}
 
 
 def resolve_base_url() -> str:
@@ -30,10 +45,34 @@ def basic_auth_header() -> str:
     return f"Basic {creds}"
 
 
+def _header_value(name: str) -> str:
+    """The variable's value, or "local" when it is unset, blank or unsafe in a header."""
+    value = os.environ.get(name, "").strip()
+    return value if _HEADER_SAFE.fullmatch(value) else _LOCAL
+
+
+def user_agent() -> str:
+    """``gooddata-eval/<version> (<ci|local>; run=<GITHUB_RUN_ID>; sha=<short GITHUB_SHA>)``."""
+    env = "ci" if os.environ.get("GITHUB_ACTIONS") == "true" else _LOCAL
+    sha = _header_value("GITHUB_SHA")
+    short_sha = sha[:_SHORT_SHA_LENGTH] if sha != _LOCAL else _LOCAL
+    return f"{SERVICE_NAME}/{__version__} ({env}; run={_header_value('GITHUB_RUN_ID')}; sha={short_sha})"
+
+
+def resource_attributes() -> dict[str, str]:
+    """OTLP resource attributes of gooddata-eval's spans, with the GitHub run context when set."""
+    attributes = {"service.name": SERVICE_NAME, "service.version": __version__}
+    for key, variable in _GITHUB_RESOURCE.items():
+        value = os.environ.get(variable, "").strip()
+        if value:
+            attributes[key] = value
+    return attributes
+
+
 def make_http_client(*, timeout: float, transport: httpx.BaseTransport | None = None) -> httpx.Client:
     return httpx.Client(
         base_url=resolve_base_url(),
-        headers={"Authorization": basic_auth_header()},
+        headers={"Authorization": basic_auth_header(), "User-Agent": user_agent()},
         timeout=timeout,
         transport=transport,
     )

packages/gooddata-eval/src/gooddata_eval/core/langfuse/otlp.py

@@ -11,6 +11,7 @@
 from typing import TYPE_CHECKING, Any
 
 from gooddata_eval._version import __version__
+from gooddata_eval.core.langfuse import _env
 
 if TYPE_CHECKING:
     import httpx
@@ -42,7 +43,6 @@
 ATTR_EXPERIMENT_ITEM_EXPECTED_OUTPUT = "langfuse.experiment.item.expected_output"
 ATTR_EXPERIMENT_ITEM_METADATA_PREFIX = "langfuse.experiment.item.metadata"
 
-_SERVICE_NAME = "gooddata-eval"
 _EPOCH = datetime(1970, 1, 1, tzinfo=timezone.utc)
 
 
@@ -112,7 +112,7 @@ class Span:
 
 
 def encode_export_request(
-    spans: list[Span], *, scope_name: str = _SERVICE_NAME, scope_version: str = __version__
+    spans: list[Span], *, scope_name: str = _env.SERVICE_NAME, scope_version: str = __version__
 ) -> dict[str, Any]:
     """Build the OTLP/JSON export request body for `POST /api/public/otel/v1/traces`."""
     otlp_spans = [
@@ -131,7 +131,7 @@ def encode_export_request(
     return {
         "resourceSpans": [
             {
-                "resource": {"attributes": [{"key": "service.name", "value": {"stringValue": _SERVICE_NAME}}]},
+                "resource": {"attributes": [otlp_attribute(k, v) for k, v in _env.resource_attributes().items()]},
                 "scopeSpans": [{"scope": {"name": scope_name, "version": scope_version}, "spans": otlp_spans}],
             }
         ]

packages/gooddata-eval/tests/test_langfuse_env.py

@@ -3,8 +3,38 @@
 
 import base64
 
+import httpx
 import pytest
-from gooddata_eval.core.langfuse._env import basic_auth_header, credentials_present, make_http_client, resolve_base_url
+from gooddata_eval._version import __version__
+from gooddata_eval.core.langfuse._env import (
+    basic_auth_header,
+    credentials_present,
+    make_http_client,
+    resolve_base_url,
+    resource_attributes,
+    user_agent,
+)
+
+_GITHUB_ENV = {
+    "GITHUB_ACTIONS": "true",
+    "GITHUB_RUN_ID": "18273645",
+    "GITHUB_SHA": "0123456789abcdef0123456789abcdef01234567",
+    "GITHUB_WORKFLOW": "AI agent tests (staging)",
+    "GITHUB_REF_NAME": "master",
+    "GITHUB_EVENT_NAME": "schedule",
+}
+
+
+@pytest.fixture
+def outside_ci(monkeypatch):
+    for name in _GITHUB_ENV:
+        monkeypatch.delenv(name, raising=False)
+
+
+@pytest.fixture
+def in_ci(monkeypatch):
+    for name, value in _GITHUB_ENV.items():
+        monkeypatch.setenv(name, value)
 
 
 def test_resolve_base_url_prefers_base_url_over_host(monkeypatch):
@@ -68,3 +98,47 @@ def test_make_http_client_uses_resolved_base_url_and_auth(monkeypatch):
         assert client.headers["Authorization"].startswith("Basic ")
     finally:
         client.close()
+
+
+def test_a_ci_run_names_the_package_its_run_and_short_sha(in_ci):
+    assert user_agent() == f"gooddata-eval/{__version__} (ci; run=18273645; sha=0123456789ab)"
+
+
+def test_outside_ci_the_user_agent_falls_back_to_local(outside_ci):
+    assert user_agent() == f"gooddata-eval/{__version__} (local; run=local; sha=local)"
+
+
+@pytest.mark.parametrize("value", ["", "18273645\r\nX-Injected: 1", "run id"])
+def test_a_blank_or_unsafe_run_id_falls_back_to_local(outside_ci, monkeypatch, value):
+    monkeypatch.setenv("GITHUB_RUN_ID", value)
+    assert "run=local;" in user_agent()
+
+
+def test_every_request_of_the_http_client_carries_the_user_agent(in_ci, monkeypatch):
+    monkeypatch.setenv("LANGFUSE_PUBLIC_KEY", "pk-test")
+    monkeypatch.setenv("LANGFUSE_SECRET_KEY", "sk-test")
+    sent: list[httpx.Request] = []
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        sent.append(request)
+        return httpx.Response(200, json={})
+
+    with make_http_client(timeout=5, transport=httpx.MockTransport(handler)) as client:
+        client.get("/api/public/v2/observations")
+
+    assert sent[0].headers["User-Agent"] == user_agent()
+
+
+def test_a_ci_run_puts_its_github_context_on_the_resource(in_ci):
+    assert resource_attributes() == {
+        "service.name": "gooddata-eval",
+        "service.version": __version__,
+        "github.run_id": "18273645",
+        "github.workflow": "AI agent tests (staging)",
+        "github.ref_name": "master",
+        "github.event_name": "schedule",
+    }
+
+
+def test_outside_ci_the_resource_names_only_the_service(outside_ci):
+    assert resource_attributes() == {"service.name": "gooddata-eval", "service.version": __version__}

packages/gooddata-eval/tests/test_langfuse_otlp.py

@@ -7,6 +7,7 @@
 
 import httpx
 import pytest
+from gooddata_eval.core.langfuse._env import resource_attributes
 from gooddata_eval.core.langfuse.otlp import (
     Span,
     encode_export_request,
@@ -126,7 +127,7 @@ def test_encode_export_request_shape():
     resource_spans = request["resourceSpans"]
     assert len(resource_spans) == 1
     resource = resource_spans[0]["resource"]
-    assert resource["attributes"] == [{"key": "service.name", "value": {"stringValue": "gooddata-eval"}}]
+    assert {a["key"]: a["value"]["stringValue"] for a in resource["attributes"]} == resource_attributes()
 
     scope_spans = resource_spans[0]["scopeSpans"]
     assert len(scope_spans) == 1