datarobot-genai 0.2.41__tar.gz → 0.2.42__tar.gz

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datarobot-genai
-Version: 0.2.41
+Version: 0.2.42
 Summary: Generic helpers for GenAI
 Project-URL: Homepage, https://github.com/datarobot-oss/datarobot-genai
 Author: DataRobot, Inc.

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "datarobot-genai"
-version = "0.2.41"
+version = "0.2.42"
 description = "Generic helpers for GenAI"
 readme = "README.md"
 requires-python = ">=3.10, <3.13"
@@ -187,6 +187,9 @@ fixable = [
     "I001",   # isort import order within functions
     "PLC0415",# pylint import-outside-toplevel for controlled local imports
 ]
+"src/datarobot_genai/**/*.py" = [
+    "PLC0415",# pylint import-outside-toplevel: allow lazy imports for memory optimization
+]
 
 [tool.ruff.lint.isort]
 force-single-line = true

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/core/agents/base.py

@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from __future__ import annotations
+
 import abc
 import json
 import os
@@ -19,6 +21,7 @@ from collections.abc import AsyncGenerator
 from collections.abc import Mapping
 from typing import Any
 from typing import Generic
+from typing import TypeAlias
 from typing import TypedDict
 from typing import TypeVar
 from typing import cast
@@ -98,7 +101,7 @@ class BaseAgent(Generic[TTool], abc.ABC):
         return get_api_base(self.api_base, deployment_id)
 
     @abc.abstractmethod
-    async def invoke(self, completion_create_params: CompletionCreateParams) -> "InvokeReturn":
+    async def invoke(self, completion_create_params: CompletionCreateParams) -> InvokeReturn:
         raise NotImplementedError("Not implemented")
 
     @classmethod
@@ -167,7 +170,7 @@ class UsageMetrics(TypedDict):
 
 
 # Canonical return type for DRUM-compatible invoke implementations
-InvokeReturn = (
+InvokeReturn: TypeAlias = (
     AsyncGenerator[tuple[str | Event, MultiTurnSample | None, UsageMetrics], None]
     | tuple[str, MultiTurnSample | None, UsageMetrics]
 )

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/core/chat/responses.py

@@ -14,6 +14,8 @@
 
 """OpenAI-compatible response helpers for chat interactions."""
 
+from __future__ import annotations
+
 import asyncio
 import queue
 import time
@@ -24,6 +26,7 @@ from collections.abc import AsyncGenerator
 from collections.abc import AsyncIterator
 from collections.abc import Iterator
 from concurrent.futures import ThreadPoolExecutor
+from typing import TYPE_CHECKING
 from typing import Any
 from typing import TypeVar
 
@@ -38,10 +41,12 @@ from openai.types.chat import ChatCompletionMessage
 from openai.types.chat.chat_completion import Choice
 from openai.types.chat.chat_completion_chunk import Choice as ChunkChoice
 from openai.types.chat.chat_completion_chunk import ChoiceDelta
-from ragas import MultiTurnSample
 
 from datarobot_genai.core.agents import default_usage_metrics
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+
 
 class CustomModelChatResponse(ChatCompletion):
     pipeline_interactions: str | None = None

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/crewai/agent.py

@@ -11,15 +11,20 @@
 # 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 __future__ import annotations
+
+from typing import TYPE_CHECKING
 
 from crewai import LLM
-from ragas import MultiTurnSample
-from ragas.messages import AIMessage
-from ragas.messages import HumanMessage
-from ragas.messages import ToolMessage
 
 from datarobot_genai.core.utils.urls import get_api_base
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+    from ragas.messages import AIMessage
+    from ragas.messages import HumanMessage
+    from ragas.messages import ToolMessage
+
 
 def build_llm(
     *,
@@ -39,4 +44,7 @@ def create_pipeline_interactions_from_messages(
 ) -> MultiTurnSample | None:
     if not messages:
         return None
+    # Lazy import to reduce memory overhead when ragas is not used
+    from ragas import MultiTurnSample
+
     return MultiTurnSample(user_input=messages)

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/crewai/base.py

@@ -26,13 +26,13 @@ from __future__ import annotations
 import abc
 import asyncio
 from collections.abc import AsyncGenerator
+from typing import TYPE_CHECKING
 from typing import Any
 
 from crewai import Crew
 from crewai.events.event_bus import CrewAIEventsBus
 from crewai.tools import BaseTool
 from openai.types.chat import CompletionCreateParams
-from ragas import MultiTurnSample
 
 from datarobot_genai.core.agents.base import BaseAgent
 from datarobot_genai.core.agents.base import InvokeReturn
@@ -44,6 +44,9 @@ from datarobot_genai.core.agents.base import is_streaming
 from .agent import create_pipeline_interactions_from_messages
 from .mcp import mcp_tools_context
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+
 
 class CrewAIAgent(BaseAgent[BaseTool], abc.ABC):
     """Abstract base agent for CrewAI workflows.

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/crewai/events.py

@@ -16,12 +16,13 @@ from __future__ import annotations
 import importlib
 import json
 import logging
+from typing import TYPE_CHECKING
 from typing import Any
 
-from ragas.messages import AIMessage
-from ragas.messages import HumanMessage
-from ragas.messages import ToolCall
-from ragas.messages import ToolMessage
+if TYPE_CHECKING:
+    from ragas.messages import AIMessage
+    from ragas.messages import HumanMessage
+    from ragas.messages import ToolMessage
 
 # Resolve crewai symbols at runtime to avoid mypy issues with untyped packages
 try:
@@ -65,6 +66,12 @@ class CrewAIEventListener:
         self.messages: list[HumanMessage | AIMessage | ToolMessage] = []
 
     def setup_listeners(self, crewai_event_bus: Any) -> None:
+        # Lazy import to reduce memory overhead when ragas is not used
+        from ragas.messages import AIMessage
+        from ragas.messages import HumanMessage
+        from ragas.messages import ToolCall
+        from ragas.messages import ToolMessage
+
         @crewai_event_bus.on(CrewKickoffStartedEvent)
         def on_crew_execution_started(_: Any, event: Any) -> None:
             self.messages.append(

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/drmcp/tools/clients/microsoft_graph.py

@@ -586,6 +586,102 @@ class MicrosoftGraphClient:
                 f"Microsoft Graph API error {response.status_code}: {response.text}"
             )
 
+    async def update_item_metadata(
+        self,
+        item_id: str,
+        fields_to_update: dict[str, Any],
+        site_id: str | None = None,
+        list_id: str | None = None,
+        drive_id: str | None = None,
+    ) -> dict[str, Any]:
+        """Update metadata on a SharePoint list item or OneDrive/SharePoint drive item.
+
+        For SharePoint list items: Updates custom column values via the fields endpoint.
+        For drive items: Updates properties like name and description.
+
+        Args:
+            item_id: The ID of the item to update.
+            fields_to_update: Key-value pairs of metadata fields to modify.
+            site_id: For SharePoint list items - the site ID.
+            list_id: For SharePoint list items - the list ID.
+            drive_id: For OneDrive/drive items - the drive ID.
+
+        Returns
+        -------
+            The API response containing updated item data.
+
+        Raises
+        ------
+            MicrosoftGraphError: If validation fails or the API request fails.
+        """
+        if not item_id or not item_id.strip():
+            raise MicrosoftGraphError("item_id cannot be empty")
+        if not fields_to_update:
+            raise MicrosoftGraphError("fields_to_update cannot be empty")
+
+        # Determine the endpoint based on provided parameters
+        has_sharepoint_context = site_id is not None and list_id is not None
+        has_drive_context = drive_id is not None
+
+        if has_sharepoint_context and has_drive_context:
+            raise MicrosoftGraphError(
+                "Cannot specify both SharePoint (site_id + list_id) and OneDrive "
+                "(document_library_id) context. Choose one."
+            )
+
+        if not has_sharepoint_context and not has_drive_context:
+            raise MicrosoftGraphError(
+                "Must specify either SharePoint context (site_id + list_id) or "
+                "OneDrive context (document_library_id)."
+            )
+
+        try:
+            if has_sharepoint_context:
+                # PATCH /sites/{site-id}/lists/{list-id}/items/{item-id}/fields
+                url = f"{GRAPH_API_BASE}/sites/{site_id}/lists/{list_id}/items/{item_id}/fields"
+                response = await self._client.patch(url, json=fields_to_update)
+            else:
+                # Drive item: PATCH /drives/{drive-id}/items/{item-id}
+                url = f"{GRAPH_API_BASE}/drives/{drive_id}/items/{item_id}"
+                response = await self._client.patch(url, json=fields_to_update)
+
+            response.raise_for_status()
+            return response.json()
+        except httpx.HTTPStatusError as e:
+            raise self._handle_update_metadata_error(e, item_id) from e
+
+    def _handle_update_metadata_error(
+        self,
+        error: httpx.HTTPStatusError,
+        item_id: str,
+    ) -> MicrosoftGraphError:
+        """Handle HTTP errors for metadata updates and return appropriate MicrosoftGraphError."""
+        status_code = error.response.status_code
+        error_msg = f"Failed to update metadata: HTTP {status_code}"
+
+        if status_code == 400:
+            try:
+                error_data = error.response.json()
+                api_message = error_data.get("error", {}).get("message", "Invalid request")
+                error_msg = f"Bad request updating metadata: {api_message}"
+            except Exception:
+                error_msg = "Bad request: invalid field names or values."
+        elif status_code == 401:
+            error_msg = "Authentication failed. Access token may be expired or invalid."
+        elif status_code == 403:
+            error_msg = (
+                f"Permission denied: you don't have permission to update item '{item_id}'. "
+                "Requires Sites.ReadWrite.All or Files.ReadWrite.All permission."
+            )
+        elif status_code == 404:
+            error_msg = f"Item '{item_id}' not found."
+        elif status_code == 409:
+            error_msg = f"Conflict: item '{item_id}' may have been modified concurrently."
+        elif status_code == 429:
+            error_msg = "Rate limit exceeded. Please try again later."
+
+        return MicrosoftGraphError(error_msg)
+
     async def __aenter__(self) -> "MicrosoftGraphClient":
         """Async context manager entry."""
         return self

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/drmcp/tools/microsoft_graph/tools.py

@@ -16,6 +16,7 @@
 
 import logging
 from typing import Annotated
+from typing import Any
 from typing import Literal
 
 from fastmcp.exceptions import ToolError
@@ -23,6 +24,7 @@ from fastmcp.tools.tool import ToolResult
 
 from datarobot_genai.drmcp.core.mcp_instance import dr_mcp_tool
 from datarobot_genai.drmcp.tools.clients.microsoft_graph import MicrosoftGraphClient
+from datarobot_genai.drmcp.tools.clients.microsoft_graph import MicrosoftGraphError
 from datarobot_genai.drmcp.tools.clients.microsoft_graph import get_microsoft_graph_access_token
 from datarobot_genai.drmcp.tools.clients.microsoft_graph import validate_site_url
 
@@ -261,7 +263,8 @@ async def microsoft_graph_share_item(
         "create",
         "file",
         "write",
-    }
+    },
+    enabled=False,
 )
 async def microsoft_create_file(
     *,
@@ -328,3 +331,145 @@ async def microsoft_create_file(
             "parentFolderId": created_file.parent_folder_id,
         },
     )
+
+
+@dr_mcp_tool(
+    tags={
+        "microsoft",
+        "graph api",
+        "sharepoint",
+        "onedrive",
+        "metadata",
+        "update",
+        "fields",
+        "compliance",
+    },
+    enabled=False,
+)
+async def microsoft_update_metadata(
+    *,
+    item_id: Annotated[str, "The ID of the file or list item to update."],
+    fields_to_update: Annotated[
+        dict[str, Any],
+        "Key-value pairs of metadata fields to modify. "
+        "For SharePoint list items: any custom column values. "
+        "For drive items: 'name' and/or 'description'.",
+    ],
+    site_id: Annotated[
+        str | None,
+        "The site ID (required for SharePoint list items, along with list_id).",
+    ] = None,
+    list_id: Annotated[
+        str | None,
+        "The list ID (required for SharePoint list items, along with site_id).",
+    ] = None,
+    document_library_id: Annotated[
+        str | None,
+        "The drive ID (required for OneDrive/drive item updates). "
+        "Cannot be used together with site_id and list_id.",
+    ] = None,
+) -> ToolResult | ToolError:
+    """
+    Update metadata on a SharePoint list item or OneDrive/SharePoint drive item.
+
+    **SharePoint List Items:** Provide site_id and list_id to update custom
+    column values on a list item. All custom columns can be updated.
+
+    **OneDrive/Drive Items:** Provide document_library_id to update drive item
+    properties. Only 'name' and 'description' fields can be updated.
+
+    **Context Requirements:**
+    - For SharePoint list items: Both site_id AND list_id are required
+    - For OneDrive/drive items: document_library_id is required
+    - Cannot specify both contexts simultaneously
+
+    **Examples:**
+    - SharePoint list item: Update a 'Status' column to 'Approved'
+    - Drive item: Rename a file or update its description
+
+    **Permissions:**
+    - Requires Sites.ReadWrite.All or Files.ReadWrite.All permission
+    """
+    if not item_id or not item_id.strip():
+        raise ToolError("Error: item_id is required.")
+    if not fields_to_update:
+        raise ToolError("Error: fields_to_update is required and cannot be empty.")
+
+    # Validate context parameters
+    has_sharepoint_context = site_id is not None and list_id is not None
+    has_partial_sharepoint_context = (site_id is not None) != (list_id is not None)
+    has_drive_context = document_library_id is not None
+
+    if has_partial_sharepoint_context:
+        raise ToolError(
+            "Error: For SharePoint list items, both site_id and list_id must be provided."
+        )
+
+    if has_sharepoint_context and has_drive_context:
+        raise ToolError(
+            "Error: Cannot specify both SharePoint (site_id + list_id) and OneDrive "
+            "(document_library_id) context. Choose one."
+        )
+
+    if not has_sharepoint_context and not has_drive_context:
+        raise ToolError(
+            "Error: Must specify either SharePoint context (site_id + list_id) or "
+            "OneDrive context (document_library_id)."
+        )
+
+    access_token = await get_microsoft_graph_access_token()
+    if isinstance(access_token, ToolError):
+        raise access_token
+
+    try:
+        async with MicrosoftGraphClient(access_token=access_token) as client:
+            result = await client.update_item_metadata(
+                item_id=item_id.strip(),
+                fields_to_update=fields_to_update,
+                site_id=site_id,
+                list_id=list_id,
+                drive_id=document_library_id,
+            )
+    except MicrosoftGraphError as e:
+        logger.error(f"Microsoft Graph error updating metadata: {e}")
+        raise ToolError(str(e))
+    except Exception as e:
+        logger.error(f"Unexpected error updating metadata: {e}", exc_info=True)
+        raise ToolError(f"An unexpected error occurred while updating metadata: {str(e)}")
+
+    context_type = "sharepoint_list_item" if has_sharepoint_context else "drive_item"
+
+    # Build structured content similar to microsoft_create_file pattern
+    structured: dict[str, Any] = {
+        "item_id": item_id,
+        "context_type": context_type,
+        "fields_updated": list(fields_to_update.keys()),
+    }
+
+    # Add context-specific IDs for traceability
+    if has_sharepoint_context:
+        structured["site_id"] = site_id
+        structured["list_id"] = list_id
+    else:
+        structured["document_library_id"] = document_library_id
+
+    # Include relevant response data
+    if isinstance(result, dict):
+        # For drive items, include key properties if present
+        if has_drive_context:
+            if "id" in result:
+                structured["id"] = result["id"]
+            if "name" in result:
+                structured["name"] = result["name"]
+            if "webUrl" in result:
+                structured["webUrl"] = result["webUrl"]
+            if "description" in result:
+                structured["description"] = result.get("description")
+        # For list items, the response is the fields object itself
+        else:
+            structured["updated_fields"] = result
+
+    return ToolResult(
+        content=f"Metadata updated successfully for item '{item_id}'.",
+        structured_content=structured,
+    )

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/langgraph/agent.py

@@ -11,9 +11,12 @@
 # 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 __future__ import annotations
+
 import abc
 import logging
 from collections.abc import AsyncGenerator
+from typing import TYPE_CHECKING
 from typing import Any
 from typing import cast
 
@@ -34,8 +37,6 @@ from langgraph.graph import MessagesState
 from langgraph.graph import StateGraph
 from langgraph.types import Command
 from openai.types.chat import CompletionCreateParams
-from ragas import MultiTurnSample
-from ragas.integrations.langgraph import convert_to_ragas_messages
 
 from datarobot_genai.core.agents.base import BaseAgent
 from datarobot_genai.core.agents.base import InvokeReturn
@@ -44,6 +45,9 @@ from datarobot_genai.core.agents.base import extract_user_prompt_content
 from datarobot_genai.core.agents.base import is_streaming
 from datarobot_genai.langgraph.mcp import mcp_tools_context
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+
 logger = logging.getLogger(__name__)
 
 
@@ -337,5 +341,9 @@ class LangGraphAgent(BaseAgent[BaseTool], abc.ABC):
                 if v is not None:
                     messages.extend(v.get("messages", []))
         messages = [m for m in messages if not isinstance(m, ToolMessage)]
+        # Lazy import to reduce memory overhead when ragas is not used
+        from ragas import MultiTurnSample
+        from ragas.integrations.langgraph import convert_to_ragas_messages
+
         ragas_trace = convert_to_ragas_messages(messages)
         return MultiTurnSample(user_input=ragas_trace)

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/llama_index/agent.py

@@ -11,17 +11,17 @@
 # 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 __future__ import annotations
 
+from typing import TYPE_CHECKING
 from typing import cast
 
 from llama_index.core.base.llms.types import LLMMetadata
 from llama_index.core.workflow import Event
 from llama_index.llms.litellm import LiteLLM
-from ragas import MultiTurnSample
-from ragas.integrations.llama_index import convert_to_ragas_messages
-from ragas.messages import AIMessage
-from ragas.messages import HumanMessage
-from ragas.messages import ToolMessage
+
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
 
 
 class DataRobotLiteLLM(LiteLLM):
@@ -44,6 +44,13 @@ def create_pipeline_interactions_from_events(
 ) -> MultiTurnSample | None:
     if not events:
         return None
+    # Lazy import to reduce memory overhead when ragas is not used
+    from ragas import MultiTurnSample
+    from ragas.integrations.llama_index import convert_to_ragas_messages
+    from ragas.messages import AIMessage
+    from ragas.messages import HumanMessage
+    from ragas.messages import ToolMessage
+
     # convert_to_ragas_messages expects a list[Event]
     ragas_trace = convert_to_ragas_messages(list(events))
     ragas_messages = cast(list[HumanMessage | AIMessage | ToolMessage], ragas_trace)

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/llama_index/base.py

@@ -25,11 +25,11 @@ from __future__ import annotations
 import abc
 import inspect
 from collections.abc import AsyncGenerator
+from typing import TYPE_CHECKING
 from typing import Any
 
 from llama_index.core.tools import BaseTool
 from openai.types.chat import CompletionCreateParams
-from ragas import MultiTurnSample
 
 from datarobot_genai.core.agents.base import BaseAgent
 from datarobot_genai.core.agents.base import InvokeReturn
@@ -41,6 +41,9 @@ from datarobot_genai.core.agents.base import is_streaming
 from .agent import create_pipeline_interactions_from_events
 from .mcp import load_mcp_tools
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+
 
 class LlamaIndexAgent(BaseAgent[BaseTool], abc.ABC):
     """Abstract base agent for LlamaIndex workflows."""

{datarobot_genai-0.2.41 → datarobot_genai-0.2.42}/src/datarobot_genai/nat/agent.py

@@ -11,9 +11,12 @@
 # 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 __future__ import annotations
+
 import asyncio
 import logging
 from collections.abc import AsyncGenerator
+from typing import TYPE_CHECKING
 from typing import Any
 
 from nat.builder.context import Context
@@ -23,10 +26,6 @@ from nat.data_models.intermediate_step import IntermediateStep
 from nat.data_models.intermediate_step import IntermediateStepType
 from nat.utils.type_utils import StrPath
 from openai.types.chat import CompletionCreateParams
-from ragas import MultiTurnSample
-from ragas.messages import AIMessage
-from ragas.messages import HumanMessage
-from ragas.messages import ToolMessage
 
 from datarobot_genai.core.agents.base import BaseAgent
 from datarobot_genai.core.agents.base import InvokeReturn
@@ -36,13 +35,22 @@ from datarobot_genai.core.agents.base import is_streaming
 from datarobot_genai.core.mcp.common import MCPConfig
 from datarobot_genai.nat.helpers import load_workflow
 
+if TYPE_CHECKING:
+    from ragas import MultiTurnSample
+    from ragas.messages import AIMessage
+    from ragas.messages import HumanMessage
+
 logger = logging.getLogger(__name__)
 
 
 def convert_to_ragas_messages(
     steps: list[IntermediateStep],
-) -> list[HumanMessage | AIMessage | ToolMessage]:
-    def _to_ragas(step: IntermediateStep) -> HumanMessage | AIMessage | ToolMessage:
+) -> list[HumanMessage | AIMessage]:
+    # Lazy import to reduce memory overhead when ragas is not used
+    from ragas.messages import AIMessage
+    from ragas.messages import HumanMessage
+
+    def _to_ragas(step: IntermediateStep) -> HumanMessage | AIMessage:
         if step.event_type == IntermediateStepType.LLM_START:
             return HumanMessage(content=_parse(step.data.input))
         elif step.event_type == IntermediateStepType.LLM_END:
@@ -78,6 +86,9 @@ def create_pipeline_interactions_from_steps(
 ) -> MultiTurnSample | None:
     if not steps:
         return None
+    # Lazy import to reduce memory overhead when ragas is not used
+    from ragas import MultiTurnSample
+
     ragas_trace = convert_to_ragas_messages(steps)
     return MultiTurnSample(user_input=ragas_trace)