app.py

import os
import re
import shutil
import tempfile
import yaml
from pathlib import Path
from typing import Generator

from smolagents import CodeAgent, LiteLLMModel, MultiStepAgent, PlanningStep, Tool
from smolagents.agent_types import AgentAudio, AgentImage, AgentText
from smolagents.memory import ActionStep, FinalAnswerStep, MemoryStep
from smolagents.models import ChatMessageStreamDelta
from smolagents.utils import _is_package_available
from tools import search, scraper, FileReader, FileWriter, CommitChanges, get_repository_structure # Assuming tools.py is in the same directory

def get_step_footnote_content(step_log: MemoryStep, step_name: str) -> str:
    """Get a footnote string for a step log with duration and token information"""
    step_footnote = f"**{step_name}**"
    if hasattr(step_log, "input_token_count") and hasattr(step_log, "output_token_count"):
        token_str = f" | Input tokens: {step_log.input_token_count:,} | Output tokens: {step_log.output_token_count:,}"
        step_footnote += token_str
    if hasattr(step_log, "duration"):
        step_duration = f" | Duration: {round(float(step_log.duration), 2)}" if step_log.duration else None
        step_footnote += step_duration
    step_footnote_content = f"""<span style="color: #bbbbc2; font-size: 12px;">{step_footnote}</span> """
    return step_footnote_content


def _clean_model_output(model_output: str) -> str:
    """

    Clean up model output by removing trailing tags and extra backticks.



    Args:

        model_output (`str`): Raw model output.



    Returns:

        `str`: Cleaned model output.

    """
    if not model_output:
        return ""
    model_output = model_output.strip()
    # Remove any trailing <end_code> and extra backticks, handling multiple possible formats
    model_output = re.sub(r"```\s*<end_code>", "```", model_output)  # handles ```<end_code>
    model_output = re.sub(r"<end_code>\s*```", "```", model_output)  # handles <end_code>```
    model_output = re.sub(r"```\s*\n\s*<end_code>", "```", model_output)  # handles ```\n<end_code>
    return model_output.strip()


def _format_code_content(content: str) -> str:
    """

    Format code content as Python code block if it's not already formatted.



    Args:

        content (`str`): Code content to format.



    Returns:

        `str`: Code content formatted as a Python code block.

    """
    content = content.strip()
    # Remove existing code blocks and end_code tags
    content = re.sub(r"```.*?\n", "", content)
    content = re.sub(r"\s*<end_code>\s*", "", content)
    content = content.strip()
    # Add Python code block formatting if not already present
    if not content.startswith("```python"):
        content = f"```python\n{content}\n```"
    return content


def _process_action_step(step_log: ActionStep, skip_model_outputs: bool = False) -> Generator:
    """

    Process an [`ActionStep`] and yield appropriate Gradio ChatMessage objects.



    Args:

        step_log ([`ActionStep`]): ActionStep to process.

        skip_model_outputs (`bool`): Whether to skip model outputs.



    Yields:

        `gradio.ChatMessage`: Gradio ChatMessages representing the action step.

    """
    import gradio as gr

    # Output the step number
    step_number = f"Step {step_log.step_number}" if step_log.step_number is not None else "Step"
    if not skip_model_outputs:
        yield gr.ChatMessage(role="assistant", content=f"**{step_number}**", metadata={"status": "done"})

    # First yield the thought/reasoning from the LLM
    if not skip_model_outputs and getattr(step_log, "model_output", ""):
        model_output = _clean_model_output(step_log.model_output)
        yield gr.ChatMessage(role="assistant", content=model_output, metadata={"status": "done"})

    # For tool calls, create a parent message
    if getattr(step_log, "tool_calls", []):
        first_tool_call = step_log.tool_calls[0]
        used_code = first_tool_call.name == "python_interpreter"

        # Process arguments based on type
        args = first_tool_call.arguments
        if isinstance(args, dict):
            content = str(args.get("answer", str(args)))
        else:
            content = str(args).strip()

        # Format code content if needed
        if used_code:
            content = _format_code_content(content)

        # Create the tool call message
        parent_message_tool = gr.ChatMessage(
            role="assistant",
            content=content,
            metadata={
                "title": f"🛠️ Used tool {first_tool_call.name}",
                "status": "done",
            },
        )
        yield parent_message_tool

    # Display execution logs if they exist
    if getattr(step_log, "observations", "") and step_log.observations.strip():
        log_content = step_log.observations.strip()
        if log_content:
            log_content = re.sub(r"^Execution logs:\s*", "", log_content)
            yield gr.ChatMessage(
                role="assistant",
                content=f"```bash\n{log_content}\n",
                metadata={"title": "📝 Execution Logs", "status": "done"},
            )

    # Display any images in observations
    if getattr(step_log, "observations_images", []):
        for image in step_log.observations_images:
            path_image = AgentImage(image).to_string()
            yield gr.ChatMessage(
                role="assistant",
                content={"path": path_image, "mime_type": f"image/{path_image.split('.')[-1]}"},
                metadata={"title": "🖼️ Output Image", "status": "done"},
            )

    # Handle errors
    if getattr(step_log, "error", None):
        yield gr.ChatMessage(
            role="assistant", content=str(step_log.error), metadata={"title": "💥 Error", "status": "done"}
        )

    # Add step footnote and separator
    yield gr.ChatMessage(
        role="assistant", content=get_step_footnote_content(step_log, step_number), metadata={"status": "done"}
    )
    yield gr.ChatMessage(role="assistant", content="-----", metadata={"status": "done"})


def _process_planning_step(step_log: PlanningStep, skip_model_outputs: bool = False) -> Generator:
    """

    Process a [`PlanningStep`] and yield appropriate gradio.ChatMessage objects.



    Args:

        step_log ([`PlanningStep`]): PlanningStep to process.



    Yields:

        `gradio.ChatMessage`: Gradio ChatMessages representing the planning step.

    """
    import gradio as gr

    if not skip_model_outputs:
        yield gr.ChatMessage(role="assistant", content="**Planning step**", metadata={"status": "done"})
        yield gr.ChatMessage(role="assistant", content=step_log.plan, metadata={"status": "done"})
    yield gr.ChatMessage(
        role="assistant", content=get_step_footnote_content(step_log, "Planning step"), metadata={"status": "done"}
    )
    yield gr.ChatMessage(role="assistant", content="-----", metadata={"status": "done"})


def _process_final_answer_step(step_log: FinalAnswerStep) -> Generator:
    """

    Process a [`FinalAnswerStep`] and yield appropriate gradio.ChatMessage objects.



    Args:

        step_log ([`FinalAnswerStep`]): FinalAnswerStep to process.



    Yields:

        `gradio.ChatMessage`: Gradio ChatMessages representing the final answer.

    """
    import gradio as gr

    final_answer = step_log.final_answer
    if isinstance(final_answer, AgentText):
        yield gr.ChatMessage(
            role="assistant",
            content=f"**Final answer:**\n{final_answer.to_string()}\n",
            metadata={"status": "done"},
        )
    elif isinstance(final_answer, AgentImage):
        yield gr.ChatMessage(
            role="assistant",
            content={"path": final_answer.to_string(), "mime_type": "image/png"},
            metadata={"status": "done"},
        )
    elif isinstance(final_answer, AgentAudio):
        yield gr.ChatMessage(
            role="assistant",
            content={"path": final_answer.to_string(), "mime_type": "audio/wav"},
            metadata={"status": "done"},
        )
    else:
        yield gr.ChatMessage(
            role="assistant", content=f"**Final answer:** {str(final_answer)}", metadata={"status": "done"}
        )


def pull_messages_from_step(step_log: MemoryStep, skip_model_outputs: bool = False):
    """Extract ChatMessage objects from agent steps with proper nesting.



    Args:

        step_log: The step log to display as gr.ChatMessage objects.

        skip_model_outputs: If True, skip the model outputs when creating the gr.ChatMessage objects:

            This is used for instance when streaming model outputs have already been displayed.

    """
    if not _is_package_available("gradio"):
        raise ModuleNotFoundError(
            "Please install 'gradio' extra to use the GradioUI: `pip install 'smolagents[gradio]'`"
        )
    if isinstance(step_log, ActionStep):
        yield from _process_action_step(step_log, skip_model_outputs)
    elif isinstance(step_log, PlanningStep):
        yield from _process_planning_step(step_log, skip_model_outputs)
    elif isinstance(step_log, FinalAnswerStep):
        yield from _process_final_answer_step(step_log)
    else:
        raise ValueError(f"Unsupported step type: {type(step_log)}")


def stream_to_gradio(

    agent,

    task: str,

    task_images: list | None = None,

    reset_agent_memory: bool = False,

    additional_args: dict | None = None,

):
    """Runs an agent with the given task and streams the messages from the agent as gradio ChatMessages."""
    if not _is_package_available("gradio"):
        raise ModuleNotFoundError(
            "Please install 'gradio' extra to use the GradioUI: `pip install 'smolagents[gradio]'`"
        )
    intermediate_text = ""
    for step_log in agent.run(
        task, images=task_images, stream=True, reset=reset_agent_memory, additional_args=additional_args
    ):
        # Track tokens if model provides them
        if getattr(agent.model, "last_input_token_count", None) is not None:
            if isinstance(step_log, (ActionStep, PlanningStep)):
                step_log.input_token_count = agent.model.last_input_token_count
                step_log.output_token_count = agent.model.last_output_token_count

        if isinstance(step_log, MemoryStep):
            intermediate_text = ""
            for message in pull_messages_from_step(
                step_log,
                # If we're streaming model outputs, no need to display them twice
                skip_model_outputs=getattr(agent, "stream_outputs", False),
            ):
                yield message
        elif isinstance(step_log, ChatMessageStreamDelta):
            intermediate_text += step_log.content or ""
            yield intermediate_text


class GradioUI:
    """A one-line interface to launch your agent in Gradio"""

    def __init__(self, agent_name: str = "Agent interface", agent_description: str | None = None):
        if not _is_package_available("gradio"):
            raise ModuleNotFoundError(
                "Please install 'gradio' extra to use the GradioUI: `pip install 'smolagents[gradio]'`"
            )
        self.agent_name = agent_name
        self.agent_description = agent_description

    def _initialize_agents(self, space_id: str, temp_dir: str):
        """Initializes agents and tools for a given space_id and temporary directory."""
        # Initialize model with your API key
        model = LiteLLMModel(model_id="gemini/gemini-2.0-flash")

        # Get repository structure
        repo_structure = get_repository_structure(space_id=space_id)

        # Load prompt templates
        try:
            with open("planning_agent_prompt_templates.yaml", "r") as f:
                planning_agent_prompt_templates = yaml.safe_load(f)

            with open("swe_agent_prompt_templates.yaml", "r") as f:
                swe_agent_prompt_templates = yaml.safe_load(f)
        except FileNotFoundError as e:
            print(f"Error loading prompt templates: {e}")
            print("Please ensure 'planning_agent_prompt_templates.yaml' and 'swe_agent_prompt_templates.yaml' are in the same directory.")
            return None, None # Indicate failure

        # Enhance prompts with repository structure
        planning_agent_prompt_templates["system_prompt"] = planning_agent_prompt_templates["system_prompt"] + "\n\n\n" + repo_structure + "\n\n"
        swe_agent_prompt_templates["system_prompt"] = swe_agent_prompt_templates["system_prompt"] + "\n\n\n" + repo_structure + "\n\n"

        # Initialize tool instances with temp directory
        read_file = FileReader(space_id=space_id, folder_path=temp_dir)
        write_file = FileWriter(space_id=space_id, folder_path=temp_dir)
        commit_changes = CommitChanges(space_id=space_id, folder_path=temp_dir)

        # Initialize SWE Agent with enhanced capabilities and improved description
        swe_agent = CodeAgent(
            model=model,
            prompt_templates=swe_agent_prompt_templates,
            verbosity_level=1,
            tools=[search, scraper, read_file, write_file],
            name="swe_agent",
            description="An expert Software Engineer capable of designing, developing, and debugging code. This agent can read and write files, search the web, and scrape web content to assist in coding tasks. It excels at implementing detailed technical specifications provided by the Planning Agent."
        )

        # Initialize Planning Agent with improved planning focus and description
        planning_agent = CodeAgent(
            model=model,
            prompt_templates=planning_agent_prompt_templates,
            verbosity_level=1,
            tools=[search, scraper, read_file, write_file, commit_changes],
            managed_agents=[swe_agent],
            name="planning_agent",
            description="A high-level planning agent responsible for breaking down complex user requests into actionable steps and delegating coding tasks to the Software Engineer Agent. It focuses on strategy, task decomposition, and coordinating the overall development process.",
            stream_outputs=True
        )

        return planning_agent, swe_agent

    def _update_space_id_and_agents(self, new_space_id: str, current_state: dict):
        """Handles space_id change, cleans up old temp dir, creates new, and re-initializes agents."""
        import gradio as gr

        old_temp_dir = current_state.get("temp_dir")
        if old_temp_dir and os.path.exists(old_temp_dir):
            print(f"Cleaning up old temporary directory: {old_temp_dir}")
            shutil.rmtree(old_temp_dir)

        if not new_space_id:
            current_state["space_id"] = None
            current_state["temp_dir"] = None
            current_state["agent"] = None
            current_state["file_upload_folder"] = None
            print("Space ID is empty. Agents are not initialized.")
            return new_space_id, None, None, None, gr.Textbox("Please enter a Space ID to initialize agents.", visible=True)

        # Create new temporary directory
        temp_dir = tempfile.mkdtemp(prefix=f"ai_workspace_{new_space_id.replace('/', '_')}_")
        file_upload_folder = Path(temp_dir) / "uploads"
        file_upload_folder.mkdir(parents=True, exist_ok=True)

        # Initialize agents
        planning_agent, _ = self._initialize_agents(new_space_id, temp_dir)

        if planning_agent is None:
             # Handle initialization failure
            shutil.rmtree(temp_dir) # Clean up the newly created temp dir
            current_state["space_id"] = None
            current_state["temp_dir"] = None
            current_state["agent"] = None
            current_state["file_upload_folder"] = None
            return new_space_id, None, None, None, gr.Textbox("Failed to initialize agents. Check console for errors.", visible=True)


        # Update session state
        current_state["space_id"] = new_space_id
        current_state["temp_dir"] = temp_dir
        current_state["agent"] = planning_agent
        current_state["file_upload_folder"] = file_upload_folder

        print(f"Initialized agents for Space ID: {new_space_id} in {temp_dir}")
        return new_space_id, [], [], file_upload_folder, gr.Textbox(f"Agents initialized for Space ID: {new_space_id}", visible=True)


    def interact_with_agent(self, prompt, messages, session_state):
        import gradio as gr

        agent = session_state.get("agent")
        if agent is None:
            messages.append(gr.ChatMessage(role="assistant", content="Please enter a Space ID and initialize the agents first."))
            yield messages
            return

        try:
            messages.append(gr.ChatMessage(role="user", content=prompt, metadata={"status": "done"}))
            yield messages

            for msg in stream_to_gradio(agent, task=prompt, reset_agent_memory=False):
                if isinstance(msg, gr.ChatMessage):
                    messages.append(msg)
                elif isinstance(msg, str):  # Then it's only a completion delta
                    try:
                        if messages[-1].metadata["status"] == "pending":
                            messages[-1].content = msg
                        else:
                            messages.append(
                                gr.ChatMessage(role="assistant", content=msg, metadata={"status": "pending"})
                            )
                    except Exception as e:
                        raise e
                yield messages

            yield messages
        except Exception as e:
            print(f"Error in interaction: {str(e)}")
            messages.append(gr.ChatMessage(role="assistant", content=f"Error: {str(e)}"))
            yield messages

    def upload_file(self, file, file_uploads_log, session_state, allowed_file_types=None):
        """

        Handle file uploads, default allowed types are .pdf, .docx, and .txt

        """
        import gradio as gr

        file_upload_folder = session_state.get("file_upload_folder")
        if file_upload_folder is None:
             return gr.Textbox("Please enter a Space ID and initialize agents before uploading files.", visible=True), file_uploads_log

        if file is None:
            return gr.Textbox(value="No file uploaded", visible=True), file_uploads_log

        if allowed_file_types is None:
            allowed_file_types = [".pdf", ".docx", ".txt"]

        file_ext = os.path.splitext(file.name)[1].lower()
        if file_ext not in allowed_file_types:
            return gr.Textbox("File type disallowed", visible=True), file_uploads_log

        # Sanitize file name
        original_name = os.path.basename(file.name)
        sanitized_name = re.sub(
            r"[^\w\-.]", "_", original_name
        )  # Replace any non-alphanumeric, non-dash, or non-dot characters with underscores

        # Save the uploaded file to the specified folder
        file_path = os.path.join(file_upload_folder, os.path.basename(sanitized_name))
        shutil.copy(file.name, file_path)

        return gr.Textbox(f"File uploaded: {file_path}", visible=True), file_uploads_log + [file_path]

    def log_user_message(self, text_input, file_uploads_log):
        import gradio as gr

        return (
            text_input
            + (
                f"\nYou have been provided with these files, which might be helpful or not: {file_uploads_log}"
                if len(file_uploads_log) > 0
                else ""
            ),
            "",
            gr.Button(interactive=False),
        )

    def launch(self, share: bool = True, **kwargs):
        self.create_app().launch(debug=True, share=share, **kwargs)

    def create_app(self):
        import gradio as gr

        with gr.Blocks(theme="ocean", fill_height=True) as demo:
            # Add session state to store session-specific data
            session_state = gr.State({})
            stored_messages = gr.State([])
            file_uploads_log = gr.State([])
            space_id_state = gr.State(None) # State to hold the current space_id
            temp_dir_state = gr.State(None) # State to hold the current temp_dir
            file_upload_folder_state = gr.State(None) # State to hold the current file upload folder

            with gr.Sidebar():
                gr.Markdown(
                    f"# {self.agent_name.replace('_', ' ').capitalize()}"
                    "\n> This web ui allows you to interact with a `smolagents` agent that can use tools and execute steps to complete tasks."
                    + (f"\n\n**Agent description:**\n{self.agent_description}" if self.agent_description else "")
                )

                # Add Space ID input
                space_id_input = gr.Textbox(
                    label="Hugging Face Space ID",
                    placeholder="Enter your Space ID (e.g., username/space-name)",
                    interactive=True
                )
                initialization_status = gr.Textbox(label="Initialization Status", interactive=False, visible=True)

                # Trigger agent initialization when space_id changes
                space_id_input.change(
                    self._update_space_id_and_agents,
                    [space_id_input, session_state],
                    [space_id_state, stored_messages, file_uploads_log, file_upload_folder_state, initialization_status]
                )


                with gr.Group():
                    gr.Markdown("**Your request**", container=True)
                    text_input = gr.Textbox(
                        lines=3,
                        label="Chat Message",
                        container=False,
                        placeholder="Enter your prompt here and press Shift+Enter or press the button",
                    )
                    submit_btn = gr.Button("Submit", variant="primary")

                gr.HTML(
                    "<br><br><h4><center>Powered by <a target='_blank' href='https://github.com/huggingface/smolagents'><b>smolagents</b></a></center></h4>"
                )

            # Main chat interface
            chatbot = gr.Chatbot(
                label="Agent",
                type="messages",
                avatar_images=(
                    None,
                    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/smolagents/mascot_smol.png",
                ),
                resizeable=True,
                scale=1,
            )

            # Set up event handlers
            text_input.submit(
                self.log_user_message,
                [text_input, file_uploads_log],
                [stored_messages, text_input, submit_btn],
            ).then(self.interact_with_agent, [stored_messages, chatbot, session_state], [chatbot]).then(
                lambda: (
                    gr.Textbox(
                        interactive=True, placeholder="Enter your prompt here and press Shift+Enter or the button"
                    ),
                    gr.Button(interactive=True),
                ),
                None,
                [text_input, submit_btn],
            )

            submit_btn.click(
                self.log_user_message,
                [text_input, file_uploads_log],
                [stored_messages, text_input, submit_btn],
            ).then(self.interact_with_agent, [stored_messages, chatbot, session_state], [chatbot]).then(
                lambda: (
                    gr.Textbox(
                        interactive=True, placeholder="Enter your prompt here and press Shift+Enter or the button"
                    ),
                    gr.Button(interactive=True),
                ),
                None,
                [text_input, submit_btn],
            )

        return demo

def run_gradio_agent():
    """Runs the Gradio UI for the agent."""
    # The GradioUI class now handles space_id input and agent initialization internally
    GradioUI(
        agent_name="SmolAgents Gradio Interface",
        agent_description="Interact with a SmolAgents planning agent capable of code generation and execution."
    ).launch()

if __name__ == "__main__":
    # The script now directly launches the Gradio UI
    run_gradio_agent()