GenAI config labels loss (#2005)

Co-authored-by: Cursor Agent <cursoragent@cursor.com>

Author: jxnl

instructor/providers/gemini/utils.py

@@ -330,11 +330,13 @@ def update_genai_kwargs(
             }
         )
 
-    # Extract thinking_config from user's config object if provided
-    # This ensures thinking_config inside config parameter is not ignored
+    # Extract thinking_config from user's config if provided (dict or object)
+    # This ensures thinking_config inside config parameter is not ignored.
     user_config = new_kwargs.get("config")
     user_thinking_config = None
-    if user_config is not None and hasattr(user_config, "thinking_config"):
+    if isinstance(user_config, dict):
+        user_thinking_config = user_config.get("thinking_config")
+    elif user_config is not None and hasattr(user_config, "thinking_config"):
         user_thinking_config = user_config.thinking_config
 
     # Handle thinking_config parameter - prioritize kwarg over config.thinking_config
@@ -345,19 +347,25 @@ def update_genai_kwargs(
     if thinking_config is not None:
         base_config["thinking_config"] = thinking_config
 
-    # Extract other relevant fields from user's config object
-    # This ensures fields like automatic_function_calling are not ignored
+    # Extract other relevant fields from user's config (dict or object).
+    # This ensures fields like automatic_function_calling / labels / cached_content
+    # are not ignored when config is passed as a dict.
     if user_config is not None:
         config_fields_to_merge = [
             "automatic_function_calling",
             "labels",
             "cached_content",
         ]
         for field in config_fields_to_merge:
-            if hasattr(user_config, field):
+            if isinstance(user_config, dict):
+                field_value = user_config.get(field)
+            elif hasattr(user_config, field):
                 field_value = getattr(user_config, field)
-                if field_value is not None and field not in base_config:
-                    base_config[field] = field_value
+            else:
+                field_value = None
+
+            if field_value is not None and field not in base_config:
+                base_config[field] = field_value
 
     return base_config
 
@@ -882,12 +890,16 @@ def handle_genai_structured_outputs(
     if new_kwargs.get("stream", False) and not issubclass(response_model, PartialBase):
         response_model = Partial[response_model]
 
-    # Extract thinking_config from user-provided config object if present
-    # This fixes issue #1966 where thinking_config inside config was ignored
+    # Extract thinking_config and cached_content from user-provided config (dict or object).
+    # This fixes issue #1966 (thinking_config ignored) and ensures cached_content
+    # is detected even when config is provided as a dict.
     user_config = new_kwargs.get("config")
     user_thinking_config = None
     user_cached_content = None
-    if user_config is not None:
+    if isinstance(user_config, dict):
+        user_thinking_config = user_config.get("thinking_config")
+        user_cached_content = user_config.get("cached_content")
+    elif user_config is not None:
         if hasattr(user_config, "thinking_config"):
             user_thinking_config = user_config.thinking_config
         if hasattr(user_config, "cached_content"):
@@ -965,12 +977,16 @@ def handle_genai_tools(
     if new_kwargs.get("stream", False) and not issubclass(response_model, PartialBase):
         response_model = Partial[response_model]
 
-    # Extract thinking_config and cached_content from user-provided config object if present
-    # This fixes issue #1966 where thinking_config inside config was ignored
+    # Extract thinking_config and cached_content from user-provided config (dict or object).
+    # This fixes issue #1966 (thinking_config ignored) and ensures cached_content
+    # is detected even when config is provided as a dict.
     user_config = new_kwargs.get("config")
     user_thinking_config = None
     user_cached_content = None
-    if user_config is not None:
+    if isinstance(user_config, dict):
+        user_thinking_config = user_config.get("thinking_config")
+        user_cached_content = user_config.get("cached_content")
+    elif user_config is not None:
         if hasattr(user_config, "thinking_config"):
             user_thinking_config = user_config.thinking_config
         if hasattr(user_config, "cached_content"):

tests/test_genai_config_merging.py

@@ -397,3 +397,124 @@ class TestModel(BaseModel):
     assert result_config.tools is not None
     assert result_config.tool_config is not None
     assert result_config.system_instruction is not None
+
+
+def test_update_genai_kwargs_config_dict_labels():
+    """Test that labels is merged when config is provided as a dict (issue #1759)."""
+    kwargs = {"config": {"labels": {"env": "prod", "team": "ml"}}}
+    base_config: dict[str, object] = {}
+
+    result = update_genai_kwargs(kwargs, base_config)
+
+    assert result["labels"] == {"env": "prod", "team": "ml"}
+
+
+def test_update_genai_kwargs_config_dict_cached_content():
+    """Test that cached_content is merged when config is provided as a dict."""
+    kwargs = {"config": {"cached_content": "caches/dict123"}}
+    base_config: dict[str, object] = {}
+
+    result = update_genai_kwargs(kwargs, base_config)
+
+    assert result["cached_content"] == "caches/dict123"
+
+
+def test_update_genai_kwargs_config_dict_thinking_config():
+    """Test that thinking_config is merged when config is provided as a dict."""
+    thinking_config = {"thinking_budget": 1234}
+    kwargs = {"config": {"thinking_config": thinking_config}}
+    base_config: dict[str, object] = {}
+
+    result = update_genai_kwargs(kwargs, base_config)
+
+    assert result["thinking_config"] == thinking_config
+
+
+def test_handle_genai_structured_outputs_preserves_labels_from_config_dict():
+    """Test that labels are preserved when config is provided as a dict (issue #1759)."""
+    from pydantic import BaseModel
+
+    from instructor.providers.gemini.utils import handle_genai_structured_outputs
+
+    class TestModel(BaseModel):
+        name: str
+
+    new_kwargs = {
+        "messages": [{"role": "user", "content": "Hello"}],
+        "config": {"labels": {"tenant": "acme", "cost-center": "123"}},
+    }
+
+    _, result_kwargs = handle_genai_structured_outputs(TestModel, new_kwargs)
+
+    result_config = result_kwargs["config"]
+    assert result_config.labels == {"tenant": "acme", "cost-center": "123"}
+
+
+def test_handle_genai_tools_preserves_labels_from_config_dict():
+    """Test that labels are preserved in tools mode when config is a dict (issue #1759)."""
+    from pydantic import BaseModel
+
+    from instructor.providers.gemini.utils import handle_genai_tools
+
+    class TestModel(BaseModel):
+        name: str
+
+    new_kwargs = {
+        "messages": [{"role": "user", "content": "Hello"}],
+        "config": {"labels": {"tenant": "acme", "cost-center": "123"}},
+    }
+
+    _, result_kwargs = handle_genai_tools(TestModel, new_kwargs)
+
+    result_config = result_kwargs["config"]
+    assert result_config.labels == {"tenant": "acme", "cost-center": "123"}
+
+
+def test_handle_genai_structured_outputs_skips_system_instruction_with_cached_content_dict():
+    """Test cached_content dict config disables system_instruction in structured outputs."""
+    from pydantic import BaseModel
+
+    from instructor.providers.gemini.utils import handle_genai_structured_outputs
+
+    class TestModel(BaseModel):
+        name: str
+
+    new_kwargs = {
+        "messages": [
+            {"role": "system", "content": "You are a helpful assistant."},
+            {"role": "user", "content": "Hello"},
+        ],
+        "config": {"cached_content": "caches/dict-cache-1"},
+    }
+
+    _, result_kwargs = handle_genai_structured_outputs(TestModel, new_kwargs)
+
+    result_config = result_kwargs["config"]
+    assert result_config.cached_content == "caches/dict-cache-1"
+    assert result_config.system_instruction is None
+
+
+def test_handle_genai_tools_skips_tools_and_system_instruction_with_cached_content_dict():
+    """Test cached_content dict config disables tools/tool_config/system_instruction in tools mode."""
+    from pydantic import BaseModel
+
+    from instructor.providers.gemini.utils import handle_genai_tools
+
+    class TestModel(BaseModel):
+        name: str
+
+    new_kwargs = {
+        "messages": [
+            {"role": "system", "content": "You are a helpful assistant."},
+            {"role": "user", "content": "Hello"},
+        ],
+        "config": {"cached_content": "caches/dict-cache-2"},
+    }
+
+    _, result_kwargs = handle_genai_tools(TestModel, new_kwargs)
+
+    result_config = result_kwargs["config"]
+    assert result_config.cached_content == "caches/dict-cache-2"
+    assert result_config.system_instruction is None
+    assert result_config.tools is None
+    assert result_config.tool_config is None