dhti-elixir-base 1.5.0__tar.gz → 1.5.2__tar.gz

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/CHANGELOG.md

@@ -2,14 +2,13 @@
 
 ## [Unreleased](https://github.com/dermatologist/dhti-elixir-base/tree/HEAD)
 
-[Full Changelog](https://github.com/dermatologist/dhti-elixir-base/compare/v1.4.1...HEAD)
-
-**Implemented enhancements:**
-
-- Add CrewAI wrappers for LLMs, Tools and Agents [\#77](https://github.com/dermatologist/dhti-elixir-base/issues/77)
+[Full Changelog](https://github.com/dermatologist/dhti-elixir-base/compare/v1.5.1...HEAD)
 
 **Merged pull requests:**
 
+- Refactor code structure for improved readability and maintainability [\#106](https://github.com/dermatologist/dhti-elixir-base/pull/106) ([dermatologist](https://github.com/dermatologist))
+- \[Automated\] Dependencies upgrade [\#84](https://github.com/dermatologist/dhti-elixir-base/pull/84) ([github-actions[bot]](https://github.com/apps/github-actions))
+- Feature/fix return type 1 [\#79](https://github.com/dermatologist/dhti-elixir-base/pull/79) ([dermatologist](https://github.com/dermatologist))
 - Add CrewAI wrappers for LLMs, Tools, and Agents [\#78](https://github.com/dermatologist/dhti-elixir-base/pull/78) ([Copilot](https://github.com/apps/copilot-swe-agent))
 - \[Automated\] Dependencies upgrade [\#76](https://github.com/dermatologist/dhti-elixir-base/pull/76) ([github-actions[bot]](https://github.com/apps/github-actions))
 - fix\(routes\): Improve response handling in invoke\_chain function [\#73](https://github.com/dermatologist/dhti-elixir-base/pull/73) ([dermatologist](https://github.com/dermatologist))
@@ -40,6 +39,18 @@
 - Feature/uv build 1 [\#40](https://github.com/dermatologist/dhti-elixir-base/pull/40) ([dermatologist](https://github.com/dermatologist))
 - Feature/mcp 1 [\#39](https://github.com/dermatologist/dhti-elixir-base/pull/39) ([dermatologist](https://github.com/dermatologist))
 
+## [v1.5.1](https://github.com/dermatologist/dhti-elixir-base/tree/v1.5.1) (2026-02-10)
+
+[Full Changelog](https://github.com/dermatologist/dhti-elixir-base/compare/v1.5.0...v1.5.1)
+
+## [v1.5.0](https://github.com/dermatologist/dhti-elixir-base/tree/v1.5.0) (2026-02-09)
+
+[Full Changelog](https://github.com/dermatologist/dhti-elixir-base/compare/v1.4.1...v1.5.0)
+
+**Implemented enhancements:**
+
+- Add CrewAI wrappers for LLMs, Tools and Agents [\#77](https://github.com/dermatologist/dhti-elixir-base/issues/77)
+
 ## [v1.4.1](https://github.com/dermatologist/dhti-elixir-base/tree/v1.4.1) (2026-01-18)
 
 [Full Changelog](https://github.com/dermatologist/dhti-elixir-base/compare/v1.4.0...v1.4.1)

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dhti-elixir-base
-Version: 1.5.0
+Version: 1.5.2
 Summary: DHTI Elixir Base
 Project-URL: Documentation, https://github.com/dermatologist/dhti
 Project-URL: Homepage, https://nuchange.ca
@@ -32,7 +32,7 @@ Requires-Dist: kink
 Requires-Dist: langchain-community
 Requires-Dist: langchain-core
 Requires-Dist: langchain-mcp-adapters
-Requires-Dist: langchain<1.1.0,>=1.0.0
+Requires-Dist: langchain>=1.0.0
 Requires-Dist: langgraph
 Requires-Dist: langserve
 Requires-Dist: mcp

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/pyproject.toml

@@ -18,7 +18,7 @@ classifiers = [
 ]
 dependencies = [
   "langchain-core",
-  "langchain>=1.0.0,<1.1.0",
+  "langchain>=1.0.0",
   "kink",
   "python-dotenv",
   "httpx",

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/src/dhti_elixir_base/crewai/agent_wrapper.py

@@ -100,23 +100,24 @@ class CrewAIAgentWrapper(CrewAIAgent):
             try:
                 from crewai.tools import BaseTool as CrewAIBaseTool
             except ImportError:
-                from crewai_tools import BaseTool as CrewAIBaseTool
+                from crewai.tools.base_tool import BaseTool as CrewAIBaseTool
 
             from .langchain_tool_wrapper import CrewAILangChainToolWrapper
 
             validated_tools = []
-            for tool in tools if isinstance(tools, list) else [tools]:
-                # If it's already a BaseTool, use it as is
-                if isinstance(tool, CrewAIBaseTool):
-                    validated_tools.append(tool)
-                # Try to wrap as a LangChain tool
-                else:
-                    try:
+            tool_list = tools if isinstance(tools, list) else [tools]
+            for tool in tool_list:
+                try:
+                    # If it's already a BaseTool, use it as is
+                    if isinstance(tool, CrewAIBaseTool):
+                        validated_tools.append(tool)
+                    # Try to wrap as a LangChain tool
+                    else:
                         wrapped = CrewAILangChainToolWrapper(langchain_tool=tool)
                         validated_tools.append(wrapped)
-                    except Exception:
-                        # Skip tools that can't be wrapped
-                        pass
+                except Exception as e:
+                    # Log but skip tools that can't be wrapped
+                    pass
             tools = validated_tools if validated_tools else None
 
         # Initialize CrewAI Agent

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/src/dhti_elixir_base/crewai/chain_tool_wrapper.py

@@ -21,7 +21,7 @@ from pydantic import PrivateAttr
 try:
     from crewai.tools import BaseTool as CrewAIBaseTool
 except ImportError:
-    from crewai_tools import BaseTool as CrewAIBaseTool
+    from crewai.tools.base_tool import BaseTool as CrewAIBaseTool
 
 from ..chain import BaseChain
 
@@ -114,30 +114,39 @@ class CrewAIChainToolWrapper(CrewAIBaseTool):
 
         Returns:
             str: The result of the chain execution
+
+        Raises:
+            ValueError: If neither positional nor keyword arguments are provided
+            RuntimeError: If chain execution fails
         """
-        # If kwargs are provided, use them directly
-        if kwargs:
-            result = self._dhti_chain.invoke(**kwargs)
-        # If a single positional arg is provided, treat it as the input
-        elif args:
-            result = self._dhti_chain.invoke(input=args[0])
-        else:
-            raise ValueError(
-                "Either provide input as a keyword argument or as a positional argument"
-            )
-
-        # Convert result to string
-        if isinstance(result, dict):
-            # If result is a dict, try to extract the most relevant value
-            if "cards" in result:
-                # Handle CDS Hook response format
-                return str(result.get("cards", []))
-            elif "output" in result:
-                return str(result["output"])
+        try:
+            # If kwargs are provided, use them directly
+            if kwargs:
+                result = self._dhti_chain.invoke(**kwargs)
+            # If a single positional arg is provided, treat it as the input
+            elif args:
+                result = self._dhti_chain.invoke(input=args[0])
             else:
-                return str(result)
-
-        return str(result)
+                raise ValueError(
+                    "Either provide input as a keyword argument or as a positional argument"
+                )
+
+            # Convert result to string
+            if isinstance(result, dict):
+                # If result is a dict, try to extract the most relevant value
+                if "cards" in result:
+                    # Handle CDS Hook response format
+                    return str(result.get("cards", []))
+                elif "output" in result:
+                    return str(result["output"])
+                else:
+                    return str(result)
+
+            return str(result)
+        except ValueError:
+            raise
+        except Exception as e:
+            raise RuntimeError(f"Chain execution failed: {e}") from e
 
     def __str__(self) -> str:
         """Return string representation of the wrapper."""

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/src/dhti_elixir_base/crewai/langchain_tool_wrapper.py

@@ -21,7 +21,7 @@ from pydantic import PrivateAttr
 try:
     from crewai.tools import BaseTool as CrewAIBaseTool
 except ImportError:
-    from crewai_tools import BaseTool as CrewAIBaseTool
+    from crewai.tools.base_tool import BaseTool as CrewAIBaseTool
 
 from langchain_core.tools import BaseTool as LangChainBaseTool
 
@@ -104,11 +104,11 @@ class CrewAILangChainToolWrapper(CrewAIBaseTool):
 
         Returns:
             str: The result of the tool execution
+
+        Raises:
+            AttributeError: If no valid invocation method is found on the tool
         """
-        # Try different invocation methods based on the tool type
         try:
-            result = None
-
             # Try using the run method (common in LangChain tools)
             if hasattr(self._langchain_tool, "run"):
                 try:
@@ -117,18 +117,18 @@ class CrewAILangChainToolWrapper(CrewAIBaseTool):
                     elif kwargs:
                         result = self._langchain_tool.run(**kwargs)
                     else:
-                        result = self._langchain_tool.run()
+                        result = self._langchain_tool.run(tool_input={})
                     return str(result)
                 except (AttributeError, TypeError):
                     # run method doesn't exist or failed, try next option
                     pass
 
-            # If not run, try using the invoke method (newer LangChain tools)
+            # Try using the invoke method (newer LangChain tools)
             if hasattr(self._langchain_tool, "invoke"):
                 try:
                     if args and not kwargs:
                         result = self._langchain_tool.invoke(
-                            args[0] if len(args) == 1 else args
+                            args[0] if len(args) == 1 else args # type: ignore
                         )
                     elif kwargs:
                         result = self._langchain_tool.invoke(kwargs)
@@ -139,7 +139,7 @@ class CrewAILangChainToolWrapper(CrewAIBaseTool):
                     # invoke method doesn't exist or failed, try next option
                     pass
 
-            # Check if __call__ is explicitly set onthe object (for test mocks)
+            # Check if __call__ is explicitly set on the object (for test mocks)
             if "__call__" in self._langchain_tool.__dict__:
                 __call_method = self._langchain_tool.__dict__["__call__"]
                 if args and not kwargs:
@@ -165,11 +165,14 @@ class CrewAILangChainToolWrapper(CrewAIBaseTool):
                     pass
 
             # If we got here, no valid method was found
+            tool_type = type(self._langchain_tool).__name__
             raise AttributeError(
-                f"LangChain tool {type(self._langchain_tool)} does not have "
-                "run, invoke, or __call__ methods"
+                f"LangChain tool '{tool_type}' does not have any of the following: "
+                "run(), invoke(), or __call__() methods"
             )
 
+        except AttributeError:
+            raise
         except Exception as e:
             return f"Error executing LangChain tool: {e!s}"
 

{dhti_elixir_base-1.5.0 → dhti_elixir_base-1.5.2}/src/dhti_elixir_base/crewai/llm_wrapper.py

@@ -17,6 +17,7 @@ limitations under the License.
 from typing import Any
 
 from crewai.llms.base_llm import BaseLLM as CrewAIBaseLLM
+from crewai.types.usage_metrics import UsageMetrics
 
 from ..chatllm import BaseChatLLM
 from ..llm import BaseLLM
@@ -72,40 +73,51 @@ class CrewAILLMWrapper(CrewAIBaseLLM):
 
         Returns:
             str: The generated response from the LLM
-        """
-        # Convert messages to the format expected by DHTI LLMs
-        if isinstance(self._dhti_llm, BaseChatLLM):
-            # For BaseChatLLM, convert to LangChain messages
-            from langchain_core.messages import AIMessage, HumanMessage, SystemMessage
-
-            lc_messages = []
-            for msg in messages:
-                role = msg.get("role", "user")
-                content = msg.get("content", "")
-
-                if role == "system":
-                    lc_messages.append(SystemMessage(content=content))
-                elif role == "assistant":
-                    lc_messages.append(AIMessage(content=content))
-                else:  # user or any other role
-                    lc_messages.append(HumanMessage(content=content))
-
-            result = self._dhti_llm.invoke(lc_messages)
-            return result.content if hasattr(result, "content") else str(result)
-        else:
-            # For BaseLLM, combine messages into a single prompt
-            prompt = "\n".join(
-                [
-                    f"{msg.get('role', 'user')}: {msg.get('content', '')}"
-                    for msg in messages
-                ]
-            )
-            return self._dhti_llm.invoke(prompt)
 
-    @property
-    def is_litellm(self) -> bool:
-        """Return whether this LLM is a LiteLLM provider."""
-        return False
+        Raises:
+            ValueError: If messages list is empty or malformed
+            RuntimeError: If LLM invocation fails
+        """
+        if not messages:
+            raise ValueError("Messages list cannot be empty")
+
+        try:
+            # Convert messages to the format expected by DHTI LLMs
+            if isinstance(self._dhti_llm, BaseChatLLM):
+                # For BaseChatLLM, convert to LangChain messages
+                from langchain_core.messages import (
+                    AIMessage,
+                    HumanMessage,
+                    SystemMessage,
+                )
+
+                lc_messages = []
+                for msg in messages:
+                    role = msg.get("role", "user")
+                    content = msg.get("content", "")
+
+                    if role == "system":
+                        lc_messages.append(SystemMessage(content=content))
+                    elif role == "assistant":
+                        lc_messages.append(AIMessage(content=content))
+                    else:  # user or any other role
+                        lc_messages.append(HumanMessage(content=content))
+
+                result = self._dhti_llm.invoke(lc_messages)
+                return result.content if hasattr(result, "content") else str(result)  # type: ignore
+            else:
+                # For BaseLLM, combine messages into a single prompt
+                prompt = "\n".join(
+                    [
+                        f"{msg.get('role', 'user')}: {msg.get('content', '')}"
+                        for msg in messages
+                    ]
+                )
+                return self._dhti_llm.invoke(prompt)
+        except Exception as e:
+            raise RuntimeError(f"Failed to invoke LLM: {e}") from e
+
+    is_litellm: bool = False
 
     @property
     def model(self) -> str:
@@ -117,19 +129,24 @@ class CrewAILLMWrapper(CrewAIBaseLLM):
         """Return the provider name."""
         return "dhti-elixir"
 
+    @provider.setter
+    def provider(self, value: str) -> None:
+        # Read-only property; setter is a no-op to satisfy base class signature
+        pass
+
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
         # Default context window size
         # This can be overridden in subclasses for specific models
-        return 4096
+        return 4096 * 2
 
-    def get_token_usage_summary(self) -> dict[str, Any]:
+    def get_token_usage_summary(self) -> UsageMetrics:
         """Get token usage summary."""
-        return {
-            "prompt_tokens": 0,
-            "completion_tokens": 0,
-            "total_tokens": 0,
-        }
+        return UsageMetrics(
+            prompt_tokens=0,
+            completion_tokens=0,
+            total_tokens=0,
+        )
 
     def supports_stop_words(self) -> bool:
         """Check if the model supports stop words."""