fix: files messages stack (#436)

Author: cotovanu-cristian

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "uipath-langchain"
-version = "0.4.25"
+version = "0.4.26"
 description = "Python SDK that enables developers to build and deploy LangGraph agents to the UiPath Cloud Platform"
 readme = { file = "README.md", content-type = "text/markdown" }
 requires-python = ">=3.11"

…chain/agent/react/multimodal/__init__.py …h_langchain/agent/multimodal/__init__.pysrc/uipath_langchain/agent/react/multimodal/__init__.py renamed to src/uipath_langchain/agent/multimodal/__init__.py src/uipath_langchain/agent/react/multimodal/__init__.py renamed to src/uipath_langchain/agent/multimodal/__init__.py

@@ -1,6 +1,9 @@
 """Multimodal LLM input handling (images, PDFs, etc.)."""
 
-from .invoke import build_file_content_block, llm_call_with_files
+from .invoke import (
+    build_file_content_block,
+    llm_call_with_files,
+)
 from .types import IMAGE_MIME_TYPES, FileInfo
 from .utils import download_file_base64, is_image, is_pdf, sanitize_filename
 

…ngchain/agent/react/multimodal/invoke.py …ath_langchain/agent/multimodal/invoke.pysrc/uipath_langchain/agent/react/multimodal/invoke.py renamed to src/uipath_langchain/agent/multimodal/invoke.py src/uipath_langchain/agent/react/multimodal/invoke.py renamed to src/uipath_langchain/agent/multimodal/invoke.py

@@ -1,5 +1,6 @@
 """LLM invocation with multimodal file attachments."""
 
+import asyncio
 from typing import Any
 
 from langchain_core.language_models import BaseChatModel
@@ -43,6 +44,24 @@ async def build_file_content_block(
     raise ValueError(f"Unsupported mime_type={file_info.mime_type}")
 
 
+async def build_file_content_blocks(files: list[FileInfo]) -> list[DataContentBlock]:
+    """Build content blocks from file attachments.
+
+    Args:
+        files: List of file information to convert to content blocks
+
+    Returns:
+        List of DataContentBlock instances for the files
+    """
+    if not files:
+        return []
+
+    file_content_blocks: list[DataContentBlock] = await asyncio.gather(
+        *[build_file_content_block(file) for file in files]
+    )
+    return file_content_blocks
+
+
 async def llm_call_with_files(
     messages: list[AnyMessage],
     files: list[FileInfo],

…angchain/agent/react/multimodal/types.py …path_langchain/agent/multimodal/types.pysrc/uipath_langchain/agent/react/multimodal/types.py renamed to src/uipath_langchain/agent/multimodal/types.py src/uipath_langchain/agent/react/multimodal/types.py renamed to src/uipath_langchain/agent/multimodal/types.py

@@ -14,7 +14,7 @@
 from uipath_langchain.agent.tools.structured_tool_with_argument_properties import (
     StructuredToolWithArgumentProperties,
 )
-from uipath_langchain.llm import get_payload_handler
+from uipath_langchain.chat.handlers import get_payload_handler
 
 from ..exceptions import AgentTerminationException
 from .constants import (

…angchain/agent/react/multimodal/utils.py …path_langchain/agent/multimodal/utils.pysrc/uipath_langchain/agent/react/multimodal/utils.py renamed to src/uipath_langchain/agent/multimodal/utils.py src/uipath_langchain/agent/react/multimodal/utils.py renamed to src/uipath_langchain/agent/multimodal/utils.py

@@ -1,8 +1,16 @@
+import asyncio
 import uuid
-from typing import Any
+from typing import Any, cast
 
 from langchain_core.language_models import BaseChatModel
-from langchain_core.messages import AnyMessage, HumanMessage, SystemMessage
+from langchain_core.messages import (
+    AnyMessage,
+    BaseMessage,
+    ContentBlock,
+    DataContentBlock,
+    HumanMessage,
+    SystemMessage,
+)
 from langchain_core.messages.tool import ToolCall
 from langchain_core.tools import BaseTool, StructuredTool
 from uipath.agent.models.agent import (
@@ -11,18 +19,20 @@
 from uipath.eval.mocks import mockable
 from uipath.platform import UiPath
 
+from uipath_langchain.agent.multimodal import FileInfo, build_file_content_block
 from uipath_langchain.agent.react.jsonschema_pydantic_converter import create_model
-from uipath_langchain.agent.react.multimodal import FileInfo, llm_call_with_files
 from uipath_langchain.agent.react.types import AgentGraphState
 from uipath_langchain.agent.tools.static_args import handle_static_args
 from uipath_langchain.agent.tools.structured_tool_with_argument_properties import (
     StructuredToolWithArgumentProperties,
 )
-from uipath_langchain.agent.tools.tool_node import (
-    ToolWrapperReturnType,
-)
+from uipath_langchain.agent.tools.tool_node import ToolWrapperReturnType
 from uipath_langchain.agent.tools.utils import sanitize_tool_name
 from uipath_langchain.agent.wrappers import get_job_attachment_wrapper
+from uipath_langchain.chat.helpers import (
+    append_content_blocks_to_message,
+    extract_text_content,
+)
 
 ANALYZE_FILES_SYSTEM_MESSAGE = (
     "Process the provided files to complete the given task. "
@@ -50,8 +60,8 @@ async def tool_fn(**kwargs: Any):
         if "attachments" not in kwargs:
             raise ValueError("Argument 'attachments' is not available")
 
-        analysisTask = kwargs["analysisTask"]
-        if not analysisTask:
+        analysis_task = kwargs["analysisTask"]
+        if not analysis_task:
             raise ValueError("Argument 'analysisTask' is not available")
 
         attachments = kwargs["attachments"]
@@ -60,12 +70,17 @@ async def tool_fn(**kwargs: Any):
         if not files:
             return {"analysisResult": "No attachments provided to analyze."}
 
+        human_message = HumanMessage(content=analysis_task)
+        human_message_with_files = await add_files_to_message(human_message, files)
+
         messages: list[AnyMessage] = [
             SystemMessage(content=ANALYZE_FILES_SYSTEM_MESSAGE),
-            HumanMessage(content=analysisTask),
+            cast(AnyMessage, human_message_with_files),
         ]
-        result = await llm_call_with_files(messages, files, llm)
-        return result
+        result = await llm.ainvoke(messages)
+
+        analysis_result = extract_text_content(result)
+        return analysis_result
 
     job_attachment_wrapper = get_job_attachment_wrapper(output_type=output_model)
 
@@ -125,3 +140,27 @@ async def _resolve_job_attachment_arguments(
         file_infos.append(file_info)
 
     return file_infos
+
+
+async def add_files_to_message(
+    message: BaseMessage,
+    files: list[FileInfo],
+) -> BaseMessage:
+    """Add file attachments to a message.
+
+    Args:
+        message: The message to add files to (any BaseMessage subclass)
+        files: List of file attachments to add
+
+    Returns:
+        New message of the same type with file content blocks appended
+    """
+    if not files:
+        return message
+
+    file_content_blocks: list[DataContentBlock] = await asyncio.gather(
+        *[build_file_content_block(file) for file in files]
+    )
+    return append_content_blocks_to_message(
+        message, cast(list[ContentBlock], file_content_blocks)
+    )

src/uipath_langchain/agent/react/llm_node.py

@@ -30,7 +30,7 @@ def _create_openai_llm(
     agenthub_config: str,
     byo_connection_id: str | None = None,
 ) -> BaseChatModel:
-    """Create UiPathChatOpenAI for OpenAI models via passthrough."""
+    """Create UiPathChatOpenAI for OpenAI models via LLMGateway."""
     from uipath_langchain.chat.openai import UiPathChatOpenAI
 
     azure_open_ai_latest_api_version = "2025-04-01-preview"
@@ -70,7 +70,7 @@ def _create_bedrock_llm(
     agenthub_config: str,
     byo_connection_id: str | None = None,
 ) -> BaseChatModel:
-    """Create UiPathChatBedrockConverse for Claude models via passthrough."""
+    """Create UiPathChatBedrockConverse for Claude models via LLMGateway."""
     from uipath_langchain.chat.bedrock import (
         UiPathChatBedrock,
         UiPathChatBedrockConverse,
@@ -107,7 +107,7 @@ def _create_vertex_llm(
     agenthub_config: str,
     byo_connection_id: str | None = None,
 ) -> BaseChatModel:
-    """Create UiPathChatVertex for Gemini models via passthrough."""
+    """Create UiPathChatVertex for Gemini models via LLMGateway."""
     from uipath_langchain.chat.vertex import UiPathChatVertex
 
     match api_flavor:
@@ -185,7 +185,7 @@ def get_chat_model(
     agenthub_config: str,
     byo_connection_id: str | None = None,
 ) -> BaseChatModel:
-    """Create and configure LLM instance using passthrough API.
+    """Create and configure LLM instance using LLMGateway API.
 
     Fetches available models from the discovery API and selects the appropriate
     LLM class based on the apiFlavor field from the matching model configuration.

src/uipath_langchain/agent/tools/internal_tools/analyze_files_tool.py

@@ -3,6 +3,7 @@
 from .base import ModelPayloadHandler
 from .bedrock_converse import BedrockConversePayloadHandler
 from .bedrock_invoke import BedrockInvokePayloadHandler
+from .handler_factory import get_payload_handler
 from .openai_completions import OpenAICompletionsPayloadHandler
 from .openai_responses import OpenAIResponsesPayloadHandler
 from .vertex_gemini import VertexGeminiPayloadHandler
@@ -14,4 +15,5 @@
     "OpenAICompletionsPayloadHandler",
     "OpenAIResponsesPayloadHandler",
     "VertexGeminiPayloadHandler",
+    "get_payload_handler",
 ]