app.py

import os
import sys
import subprocess
import base64
import json
from io import StringIO
from typing import Dict, List

import streamlit as st
import torch
from transformers import pipeline, AutoModelForSeq2SeqLM, AutoTokenizer, HfApi
from pylint import lint
import black

from setuptools import setup, Extension
from Cython.Build import cythonize

extensions = [
    Extension("app", ["app.pyx"]),
]

setup(
    ext_modules=cythonize(extensions),
)

# Add your Hugging Face API token here
hf_token = st.secrets["huggingface"]

# Constants
PROJECT_ROOT = "./projects"
AGENT_DIRECTORY = "./agents"
AVAILABLE_CODE_GENERATIVE_MODELS = ["codegen", "gpt-neo", "codeparrot"]

# Global state management
if "chat_history" not in st.session_state:
    st.session_state.chat_history = []
if "terminal_history" not in st.session_state:
    st.session_state.terminal_history = []
if "workspace_projects" not in st.session_state:
    st.session_state.workspace_projects = {}
if "available_agents" not in st.session_state:
    st.session_state.available_agents = []

# Load pre-trained models
@st.cache(allow_output_mutation=True)
def load_models():
    rag_retriever = pipeline("question-answering", model="facebook/rag-token-nq")
    chat_model = AutoModelForSeq2SeqLM.from_pretrained("microsoft/DialoGPT-medium")
    tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-medium")
    return rag_retriever, chat_model, tokenizer

rag_retriever, chat_model, tokenizer = load_models()

def process_input(user_input: str) -> str:
    # Input pipeline: Tokenize and preprocess user input
    input_ids = tokenizer(user_input, return_tensors="pt").input_ids
    attention_mask = tokenizer(user_input, return_tensors="pt").attention_mask

    # RAG model: Generate response
    with torch.no_grad():
        rag_output = rag_retriever(question=user_input, context=user_input)
        rag_answer = rag_output['answer']

    # Chat model: Refine response
    chat_input = tokenizer(rag_answer, return_tensors="pt")
    with torch.no_grad():
        chat_output = chat_model.generate(**chat_input)
        refined_response = tokenizer.decode(chat_output[0], skip_special_tokens=True)

    return refined_response

class AIAgent:
    def __init__(self, name: str, description: str, skills: List[str], hf_api=None):
        self.name = name
        self.description = description
        self.skills = skills
        self._hf_api = hf_api
        self._hf_token = hf_token

    @property
    def hf_api(self):
        if not self._hf_api and self.has_valid_hf_token():
            self._hf_api = HfApi(token=self._hf_token)
        return self._hf_api

    def has_valid_hf_token(self):
        return bool(self._hf_token)

    def create_agent_prompt(self):
        return f"Name: {self.name}\nDescription: {self.description}\nSkills:\n" + "\n".join(self.skills)

    async def autonomous_build(self, chat_history: List[str], workspace_projects: Dict[str, str], project_name: str, selected_model: str):
        summary = "Chat History:\n" + "\n".join(chat_history)
        summary += "\n\nWorkspace Projects:\n" + "\n".join(workspace_projects.items())

        next_step = "Based on the current state, the next logical step is to implement the main application logic."

        project_path = os.path.join(PROJECT_ROOT, project_name)
        if not os.path.exists(project_path):
            os.makedirs(project_path)

        requirements_file = os.path.join(project_path, "requirements.txt")
        if not os.path.exists(requirements_file):
            with open(requirements_file, "w") as f:
                f.write("# Add your project's dependencies here\n")

        app_file = os.path.join(project_path, "app.py")
        if not os.path.exists(app_file):
            with open(app_file, "w") as f:
                f.write("# Your project's main application logic goes here\n")

        if "create a gui" in summary.lower():
            gui_code = generate_code(
                "Create a simple GUI for this application", selected_model)
            with open(app_file, "a") as f:
                f.write(gui_code)

        build_command = "pip install -r requirements.txt && python app.py"
        try:
            result = subprocess.run(
                build_command, shell=True, capture_output=True, text=True, cwd=project_path)
            st.write(f"Build Output:\n{result.stdout}")
            if result.stderr:
                st.error(f"Build Errors:\n{result.stderr}")
        except Exception as e:
            st.error(f"Build Error: {e}")

        return summary, next_step

    def deploy_built_space_to_hf(self):
        if not self.has_valid_hf_token():
            st.error("Invalid Hugging Face token. Please check your configuration.")
            return

        try:
            files = get_built_space_files()
            create_space_on_hugging_face(self.hf_api, self.name, self.description, True, files)
            st.success(f"Successfully deployed {self.name} to Hugging Face Spaces!")
        except Exception as e:
            st.error(f"Error deploying to Hugging Face Spaces: {str(e)}")

def get_built_space_files() -> Dict[str, str]:
    return {
        "app.py": "# Your Streamlit app code here",
        "requirements.txt": "streamlit\ntransformers"
    }

def save_agent_to_file(agent: AIAgent):
    if not os.path.exists(AGENT_DIRECTORY):
        os.makedirs(AGENT_DIRECTORY)
    file_path = os.path.join(AGENT_DIRECTORY, f"{agent.name}.txt")
    with open(file_path, "w") as file:
        file.write(agent.create_agent_prompt())
    st.session_state.available_agents.append(agent.name)

def load_agent_prompt(agent_name: str) -> str:
    file_path = os.path.join(AGENT_DIRECTORY, f"{agent_name}.txt")
    if os.path.exists(file_path):
        with open(file_path, "r") as file:
            agent_prompt = file.read()
        return agent_prompt
    else:
        return None

def create_agent_from_text(name: str, text: str) -> str:
    skills = text.split("\n")
    agent = AIAgent(name, "AI agent created from text input.", skills)
    save_agent_to_file(agent)
    return agent.create_agent_prompt()

def chat_interface_with_agent(input_text: str, agent_name: str) -> str:
    agent_prompt = load_agent_prompt(agent_name)
    if agent_prompt is None:
        return f"Agent {agent_name} not found."

    model_name = "microsoft/DialoGPT-medium"
    try:
        generator = pipeline("text-generation", model=model_name)
        generated_response = generator(
            f"{agent_prompt}\n\nUser: {input_text}\nAgent:", max_length=100, do_sample=True, top_k=50)[0]["generated_text"]
        return generated_response
    except Exception as e:
        return f"Error loading model: {e}"

def terminal_interface(command: str, project_name: str = None) -> str:
    if project_name:
        project_path = os.path.join(PROJECT_ROOT, project_name)
        if not os.path.exists(project_path):
            return f"Project {project_name} does not exist."
        result = subprocess.run(
            command, shell=True, capture_output=True, text=True, cwd=project_path)
    else:
        result = subprocess.run(command, shell=True, capture_output=True, text=True)
    return result.stdout

def code_editor_interface(code: str) -> str:
    try:
        formatted_code = black.format_str(code, mode=black.FileMode())
    except black.NothingChanged:
        formatted_code = code

    result = StringIO()
    sys.stdout = result
    sys.stderr = result

    lint.Run(['--rcfile=/dev/null', '-'], exit=False)
    lint_message = result.getvalue()

    sys.stdout = sys.__stdout__
    sys.stderr = sys.__stderr__

    return formatted_code, lint_message

def summarize_text(text: str) -> str:
    summarizer = pipeline("summarization")
    summary = summarizer(text, max_length=130, min_length=30, do_sample=False)
    return summary[0]['summary_text']

def sentiment_analysis(text: str) -> str:
    analyzer = pipeline("sentiment-analysis")
    result = analyzer(text)
    return result[0]['label']

def translate_code(code: str, source_language: str, target_language: str) -> str:
    translator = pipeline("translation", model="Helsinki-NLP/opus-mt-en-ROMANCE")
    translated_code = translator(code, max_length=512)[0]['translation_text']
    return translated_code

def generate_code(code_idea: str, model_name: str) -> str:
    try:
        generator = pipeline('text-generation', model=model_name)
        generated_code = generator(code_idea, max_length=1000, num_return_sequences=1)[0]['generated_text']
        return generated_code
    except Exception as e:
        return f"Error generating code: {e}"

def chat_interface(input_text: str) -> str:
    chatbot = pipeline("text-generation", model="microsoft/DialoGPT-medium")
    response = chatbot(input_text, max_length=50, num_return_sequences=1)[0]['generated_text']
    return response

def workspace_interface(project_name: str) -> str:
    project_path = os.path.join(PROJECT_ROOT, project_name)
    if not os.path.exists(project_path):
        os.makedirs(project_path)
        st.session_state.workspace_projects[project_name] = {'files': []}
        return f"Project '{project_name}' created successfully."
    else:
        return f"Project '{project_name}' already exists."

def add_code_to_workspace(project_name: str, code: str, file_name: str) -> str:
    project_path = os.path.join(PROJECT_ROOT, project_name)
    if not os.path.exists(project_path):
        return f"Project '{project_name}' does not exist."

    file_path = os.path.join(project_path, file_name)
    with open(file_path, "w") as file:
        file.write(code)
    st.session_state.workspace_projects[project_name]['files'].append(file_name)
    return f"Code added to '{file_name}' in project '{project_name}'."

def create_space_on_hugging_face(api, name, description, public, files, entrypoint="app.py"):
    try:
        repo = api.create_repo(name, exist_ok=True, private=not public)
        for filename, content in files.items():
            api.upload_file(
                path_or_fileobj=content.encode(),
                path_in_repo=filename,
                repo_id=repo.repo_id,
                repo_type="space",
            )
        return repo
    except Exception as e:
        st.error(f"Error creating Hugging Face Space: {str(e)}")
        return None

# Streamlit App
st.title("AI Agent Creator")

# Sidebar navigation
st.sidebar.title("Navigation")
app_mode = st.sidebar.selectbox(
    "Choose the app mode", ["AI Agent Creator", "Tool Box", "Workspace Chat App"])

if app_mode == "AI Agent Creator":
    st.header("Create an AI Agent from Text")

    st.subheader("From Text")
    agent_name = st.text_input("Enter agent name:")
    text_input = st.text_area("Enter skills (one per line):")
    if st.button("Create Agent"):
        agent_prompt = create_agent_from_text(agent_name, text_input)
        st.success(f"Agent '{agent_name}' created and saved successfully.")
        st.session_state.available_agents.append(agent_name)

elif app_mode == "Tool Box":
    st.header("AI-Powered Tools")

    st.subheader("Chat with CodeCraft")
    chat_input = st.text_area("Enter your message:")
    if st.button("Send"):
        chat_response = chat_interface(chat_input)
        st.session_state.chat_history.append((chat_input, chat_response))
        st.write(f"CodeCraft: {chat_response}")

    st.subheader("Terminal")
    terminal_input = st.text_input("Enter a command:")
    if st.button("Run"):
        terminal_output = terminal_interface(terminal_input)
        st.session_state.terminal_history.append(
            (terminal_input, terminal_output))
        st.code(terminal_output, language="bash")

    st.subheader("Code Editor")
    code_editor = st.text_area("Write your code:", height=300)
    if st.button("Format & Lint"):
        formatted_code, lint_message = code_editor_interface(code_editor)
        st.code(formatted_code, language="python")
        st.info(lint_message)

    st.subheader("Summarize Text")
    text_to_summarize = st.text_area("Enter text to summarize:")
    if st.button("Summarize"):
        summary = summarize_text(text_to_summarize)
        st.write(f"Summary: {summary}")

    st.subheader("Sentiment Analysis")
    sentiment_text = st.text_area("Enter text for sentiment analysis:")
    if st.button("Analyze Sentiment"):
        sentiment = sentiment_analysis(sentiment_text)
        st.write(f"Sentiment: {sentiment}")

    st.subheader("Translate Code")
    code_to_translate = st.text_area("Enter code to translate:")
    source_language = st.text_input("Enter source language (e.g., 'Python'):")
    target_language = st.text_input(
        "Enter target language (e.g., 'JavaScript'):")
    if st.button("Translate Code"):
        translated_code = translate_code(
            code_to_translate, source_language, target_language)
        st.code(translated_code, language=target_language.lower())

    st.subheader("Code Generation")
    code_idea = st.text_input("Enter your code idea:")
    if st.button("Generate Code"):
        generated_code = generate_code(code_idea, "gpt2")
        st.code(generated_code, language="python")

elif app_mode == "Workspace Chat App":
    st.header("Workspace Chat App")

    st.subheader("Create a New Project")
    project_name = st.text_input("Enter project name:")
    if st.button("Create Project"):
        workspace_status = workspace_interface(project_name)
        st.success(workspace_status)

    st.subheader("Add Code to Workspace")
    code_to_add = st.text_area("Enter code to add to workspace:")