refactor: streamline project and logstream resolution in GalileoOTLPE… (#493)

Author: Focadecombate

src/galileo/otel.py

@@ -1,9 +1,7 @@
 import contextvars
 import json
 import logging
-import os
 import typing
-import uuid
 from _contextvars import ContextVar
 from collections.abc import Generator
 from contextlib import contextmanager
@@ -14,6 +12,7 @@
 
 from galileo.config import GalileoPythonConfig
 from galileo.decorator import _experiment_id_context, _log_stream_context, _project_context, _session_id_context
+from galileo.utils.env_helpers import _get_log_stream_or_default, _get_project_or_default
 from galileo.utils.retrievers import document_adapter
 from galileo_core.schemas.logging.span import RetrieverSpan, ToolSpan, WorkflowSpan
 from galileo_core.schemas.logging.span import Span as GalileoSpan
@@ -128,29 +127,12 @@ def __init__(self, project: Optional[str] = None, logstream: Optional[str] = Non
         if not api_key:
             raise ValueError("API key is required.")
 
-        if project is not None:
-            _project_context.set(project)
-        if logstream is not None:
-            _log_stream_context.set(logstream)
-
-        # Resolve project and logstream from parameters, context, or environment variables
-        # Check context first, then environment, then generate/use defaults
-        ctx_project = _project_context.get(None)
-        ctx_logstream = _log_stream_context.get(None)
-
-        if ctx_project is not None:
-            self.project = ctx_project
-        elif "GALILEO_PROJECT" in os.environ:
-            self.project = os.environ["GALILEO_PROJECT"]
-        else:
-            self.project = f"project_{uuid.uuid4()}"
-
-        if ctx_logstream is not None:
-            self.logstream = ctx_logstream
-        elif "GALILEO_LOG_STREAM" in os.environ:
-            self.logstream = os.environ["GALILEO_LOG_STREAM"]
-        else:
-            self.logstream = "default"
+        # Resolve project and logstream: param first, then context var, then env var with default fallback
+        ctx_project = project if project is not None else _project_context.get(None)
+        ctx_logstream = logstream if logstream is not None else _log_stream_context.get(None)
+
+        self.project = _get_project_or_default(ctx_project)
+        self.logstream = _get_log_stream_or_default(ctx_logstream)
 
         exporter_headers = {"Galileo-API-Key": api_key, "project": self.project, "logstream": self.logstream}
 
@@ -240,13 +222,13 @@ def __init__(
                 "OpenTelemetry packages are not installed. "
                 "Install optional OpenTelemetry dependencies with: pip install galileo[otel]"
             )
-        if project is not None:
-            _project_context.set(project)
-        if logstream is not None:
-            _log_stream_context.set(logstream)
 
-        self._project = _project_context.get()
-        self._logstream = _log_stream_context.get()
+        # Resolve project and logstream: param first, then context var, then env var with default fallback
+        ctx_project = project if project is not None else _project_context.get(None)
+        ctx_logstream = logstream if logstream is not None else _log_stream_context.get(None)
+
+        self._project = _get_project_or_default(ctx_project)
+        self._logstream = _get_log_stream_or_default(ctx_logstream)
 
         # Create the exporter using the config-based approach
         self._exporter = GalileoOTLPExporter(**kwargs)
@@ -259,8 +241,9 @@ def __init__(
     def on_start(self, span: Span, parent_context: Optional[context.Context] = None) -> None:
         """Handle span start events by delegating to the underlying processor."""
         # Set Galileo context attributes on the span
-        project = _project_context.get(self._project)
-        log_stream = _log_stream_context.get(self._logstream)
+        # Use context var if set and not None, otherwise fall back to instance defaults
+        project = _project_context.get(None) or self._project
+        log_stream = _log_stream_context.get(None) or self._logstream
         experiment_id = _experiment_id_context.get(None)
         session_id = _session_id_context.get(None)
 

tests/test_otel.py

@@ -1,7 +1,6 @@
 import json
 import os
 import re
-import uuid
 from unittest.mock import Mock, patch
 
 import pytest
@@ -80,22 +79,19 @@ def test_init_and_parameter_priority(self, mock_otlp_init, mock_config, clear_en
     def test_init_with_env_variables(self, mock_otlp_init, mock_config, clear_env_vars):
         """Test initialization using environment variables."""
         with patch.dict(os.environ, {"GALILEO_PROJECT": "env-project", "GALILEO_LOG_STREAM": "env-logstream"}):
-            with patch("galileo.otel.uuid.uuid4") as mock_uuid:
-                exporter = GalileoOTLPExporter()
-                mock_uuid.assert_not_called()  # No UUID generation when env var provided
-                assert exporter.project == "env-project"
-                assert exporter.logstream == "env-logstream"
+            exporter = GalileoOTLPExporter()
+            assert exporter.project == "env-project"
+            assert exporter.logstream == "env-logstream"
 
     @pytest.mark.skipif(not OTEL_AVAILABLE, reason="OpenTelemetry not available")
-    @patch("galileo.otel.uuid.uuid4")
     @patch("galileo.otel.OTLPSpanExporter.__init__", return_value=None)
-    def test_init_auto_generate_project(self, mock_otlp_init, mock_uuid, mock_config, clear_env_vars):
-        """Test automatic project generation when no project is provided."""
-        mock_uuid.return_value = uuid.UUID("12345678-1234-5678-9012-123456789012")
+    def test_init_uses_default_project(self, mock_otlp_init, mock_config, clear_env_vars):
+        """Test default project name is used when no project is provided."""
         GalileoOTLPExporter()
 
         call_kwargs = mock_otlp_init.call_args[1]
-        assert call_kwargs["headers"]["project"] == "project_12345678-1234-5678-9012-123456789012"
+        assert call_kwargs["headers"]["project"] == "default"
+        assert call_kwargs["headers"]["logstream"] == "default"
 
     @pytest.mark.skipif(not OTEL_AVAILABLE, reason="OpenTelemetry not available")
     @patch("galileo.otel.OTLPSpanExporter.__init__", return_value=None)
@@ -391,7 +387,8 @@ def test_processor_on_start_sets_span_attributes(self, mock_processor_deps, rese
         assert ("galileo.project.name", "test-project") in actual_calls
         assert ("galileo.session.id", "test-session") in actual_calls
 
-        # Test with only project set (None values should be skipped)
+        # Test that processor always sets project and logstream (using env var fallbacks)
+        # When context is None, it falls back to env vars (set in conftest.py)
         _log_stream_context.set(None)
         _experiment_id_context.set(None)
         _session_id_context.set(None)
@@ -400,8 +397,12 @@ def test_processor_on_start_sets_span_attributes(self, mock_processor_deps, rese
         mock_span2 = Mock()
         processor2.on_start(mock_span2, None)
 
-        assert mock_span2.set_attribute.call_count == 1
-        mock_span2.set_attribute.assert_called_with("galileo.project.name", "test-project")
+        # Project and logstream are always set (env var fallbacks when context is None)
+        assert mock_span2.set_attribute.call_count == 2
+        actual_calls = {(args[0], args[1]) for args, _ in mock_span2.set_attribute.call_args_list}
+        assert ("galileo.project.name", "test-project") in actual_calls
+        # Falls back to GALILEO_LOG_STREAM env var set in conftest.py
+        assert ("galileo.logstream.name", "test-log-stream") in actual_calls
 
     @pytest.mark.skipif(not OTEL_AVAILABLE, reason="OpenTelemetry not available")
     @patch("galileo.otel.OTLPSpanExporter.export")