feat(otel): integrate OpenTelemetry for Langfuse monitoring and add configuration settings

src/utils/otel.py

@@ -0,0 +1,41 @@
+import os
+import base64
+from src.config.settings import settings
+
+from opentelemetry import trace
+from opentelemetry.sdk.resources import Resource
+from opentelemetry.sdk.trace import TracerProvider
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
+
+_otlp_initialized = False
+
+
+def init_otel():
+    global _otlp_initialized
+    if _otlp_initialized:
+        return
+    if not (
+        settings.LANGFUSE_PUBLIC_KEY
+        and settings.LANGFUSE_SECRET_KEY
+        and settings.OTEL_EXPORTER_OTLP_ENDPOINT
+    ):
+        return
+
+    langfuse_auth = base64.b64encode(
+        f"{settings.LANGFUSE_PUBLIC_KEY}:{settings.LANGFUSE_SECRET_KEY}".encode()
+    ).decode()
+    os.environ["OTEL_EXPORTER_OTLP_ENDPOINT"] = settings.OTEL_EXPORTER_OTLP_ENDPOINT
+    os.environ["OTEL_EXPORTER_OTLP_HEADERS"] = f"Authorization=Basic {langfuse_auth}"
+
+    provider = TracerProvider(
+        resource=Resource.create({"service.name": "evo_ai_agent"})
+    )
+    exporter = OTLPSpanExporter()
+    provider.add_span_processor(BatchSpanProcessor(exporter))
+    trace.set_tracer_provider(provider)
+    _otlp_initialized = True
+
+
+def get_tracer(name: str = "evo_ai_agent"):
+    return trace.get_tracer(name)