feat(llm): Add automatic provider inference for LangChain LLMs

Add automatic provider name detection from LLM module paths to eliminate
manual provider specification. The implementation extracts provider
names from LangChain package naming conventions (e.g.,langchain_openai →
openai) and handles edge cases including community packages, wrapped
classes, and multiple inheritance through MRO traversal.

Author: Pouyanpi

nemoguardrails/actions/llm/utils.py

@@ -46,6 +46,70 @@ def __init__(self, inner_exception: Any):
         self.inner_exception = inner_exception
 
 
+def _infer_provider_from_module(llm: BaseLanguageModel) -> Optional[str]:
+    """Infer provider name from the LLM's module path.
+
+    This function extracts the provider name from LangChain package naming conventions:
+    - langchain_openai -> openai
+    - langchain_anthropic -> anthropic
+    - langchain_google_genai -> google_genai
+    - langchain_nvidia_ai_endpoints -> nvidia_ai_endpoints
+    - langchain_community.chat_models.ollama -> ollama
+
+    For patched/wrapped classes, checks base classes as well.
+
+    Args:
+        llm: The LLM instance
+
+    Returns:
+        The inferred provider name, or None if it cannot be determined
+    """
+    module = type(llm).__module__
+
+    if module.startswith("langchain_"):
+        package = module.split(".")[0]
+        provider = package.replace("langchain_", "")
+
+        if provider == "community":
+            parts = module.split(".")
+            if len(parts) >= 3:
+                provider = parts[-1]
+                return provider
+        else:
+            return provider
+
+    for base_class in type(llm).__mro__[1:]:
+        base_module = base_class.__module__
+        if base_module.startswith("langchain_"):
+            package = base_module.split(".")[0]
+            provider = package.replace("langchain_", "")
+
+            if provider == "community":
+                parts = base_module.split(".")
+                if len(parts) >= 3:
+                    provider = parts[-1]
+                    return provider
+            else:
+                return provider
+
+    return None
+
+
+def get_llm_provider(llm: BaseLanguageModel) -> Optional[str]:
+    """Get the provider name for an LLM instance by inferring from module path.
+
+    This function extracts the provider name from LangChain package naming conventions.
+    See _infer_provider_from_module for details on the inference logic.
+
+    Args:
+        llm: The LLM instance
+
+    Returns:
+        The provider name if it can be inferred, None otherwise
+    """
+    return _infer_provider_from_module(llm)
+
+
 def _infer_model_name(llm: BaseLanguageModel):
     """Helper to infer the model name based from an LLM instance.
 
@@ -126,7 +190,7 @@ def _setup_llm_call_info(
         llm_call_info_var.set(llm_call_info)
 
     llm_call_info.llm_model_name = model_name or _infer_model_name(llm)
-    llm_call_info.llm_provider_name = model_provider
+    llm_call_info.llm_provider_name = model_provider or _infer_provider_from_module(llm)
 
 
 def _prepare_callbacks(

tests/test_actions_llm_utils.py

@@ -0,0 +1,125 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from nemoguardrails.actions.llm.utils import _infer_provider_from_module
+
+
+class MockOpenAILLM:
+    __module__ = "langchain_openai.chat_models"
+
+
+class MockAnthropicLLM:
+    __module__ = "langchain_anthropic.chat_models"
+
+
+class MockNVIDIALLM:
+    __module__ = "langchain_nvidia_ai_endpoints.chat_models"
+
+
+class MockCommunityOllama:
+    __module__ = "langchain_community.chat_models.ollama"
+
+
+class MockUnknownLLM:
+    __module__ = "some_custom_package.models"
+
+
+class MockNVIDIAOriginal:
+    __module__ = "langchain_nvidia_ai_endpoints.chat_models"
+
+
+class MockPatchedNVIDIA(MockNVIDIAOriginal):
+    __module__ = "nemoguardrails.llm.providers._langchain_nvidia_ai_endpoints_patch"
+
+
+def test_infer_provider_openai():
+    llm = MockOpenAILLM()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "openai"
+
+
+def test_infer_provider_anthropic():
+    llm = MockAnthropicLLM()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "anthropic"
+
+
+def test_infer_provider_nvidia_ai_endpoints():
+    llm = MockNVIDIALLM()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "nvidia_ai_endpoints"
+
+
+def test_infer_provider_community_ollama():
+    llm = MockCommunityOllama()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "ollama"
+
+
+def test_infer_provider_unknown():
+    llm = MockUnknownLLM()
+    provider = _infer_provider_from_module(llm)
+    assert provider is None
+
+
+def test_infer_provider_from_patched_class():
+    llm = MockPatchedNVIDIA()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "nvidia_ai_endpoints"
+
+
+def test_infer_provider_checks_base_classes():
+    class BaseOpenAI:
+        __module__ = "langchain_openai.chat_models"
+
+    class CustomWrapper(BaseOpenAI):
+        __module__ = "my_custom_wrapper.llms"
+
+    llm = CustomWrapper()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "openai"
+
+
+def test_infer_provider_multiple_inheritance():
+    class BaseNVIDIA:
+        __module__ = "langchain_nvidia_ai_endpoints.chat_models"
+
+    class Mixin:
+        __module__ = "some_mixin.utils"
+
+    class MultipleInheritance(Mixin, BaseNVIDIA):
+        __module__ = "custom_package.models"
+
+    llm = MultipleInheritance()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "nvidia_ai_endpoints"
+
+
+def test_infer_provider_deeply_nested_inheritance():
+    class Original:
+        __module__ = "langchain_anthropic.chat_models"
+
+    class Wrapper1(Original):
+        __module__ = "wrapper1.models"
+
+    class Wrapper2(Wrapper1):
+        __module__ = "wrapper2.models"
+
+    class Wrapper3(Wrapper2):
+        __module__ = "wrapper3.models"
+
+    llm = Wrapper3()
+    provider = _infer_provider_from_module(llm)
+    assert provider == "anthropic"