app.py

import streamlit as st
import numpy as np
import pandas as pd
from smolagents import CodeAgent, tool
from typing import Union, List, Dict, Optional
import matplotlib.pyplot as plt
import seaborn as sns
import os
from groq import Groq
from dataclasses import dataclass
import tempfile
import base64
import io
import json
from streamlit_ace import st_ace
from contextlib import contextmanager


class GroqLLM:
    """Compatible LLM interface for smolagents CodeAgent"""

    def __init__(self, model_name="llama-3.1-8B-Instant"):
        self.client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
        self.model_name = model_name

    def __call__(self, prompt: Union[str, dict, List[Dict]]) -> str:
        """Make the class callable as required by smolagents"""
        try:
            # Handle different prompt formats
            if isinstance(prompt, (dict, list)):
                prompt_str = str(prompt)
            else:
                prompt_str = str(prompt)

            # Create a properly formatted message
            completion = self.client.chat.completions.create(
                model=self.model_name,
                messages=[{"role": "user", "content": prompt_str}],
                temperature=0.7,
                max_tokens=1024,
                stream=True,  # Enable streaming
            )

            full_response = ""
            for chunk in completion:
                if chunk.choices[0].delta.content is not None:
                    full_response += chunk.choices[0].delta.content
            return full_response
        except Exception as e:
            error_msg = f"Error generating response: {str(e)}"
            print(error_msg)
            return error_msg


class DataAnalysisAgent(CodeAgent):
    """Extended CodeAgent with dataset awareness"""

    def __init__(self, dataset: pd.DataFrame, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._dataset = dataset

    @property
    def dataset(self) -> pd.DataFrame:
        """Access the stored dataset"""
        return self._dataset

    def run(self, prompt: str, **kwargs) -> str:
        """Override run method to include dataset context"""
        dataset_info = f"""
        Dataset Shape: {self.dataset.shape}
        Columns: {', '.join(self.dataset.columns)}
        Data Types: {self.dataset.dtypes.to_dict()}
        """
        enhanced_prompt = f"""
        Analyze the following dataset:
        {dataset_info}

        Task: {prompt}

        Use the provided tools to analyze this specific dataset and return detailed results.
        """
        return super().run(enhanced_prompt, data=self.dataset, **kwargs)  # Pass data as argument


@tool
def analyze_basic_stats(data: pd.DataFrame) -> str:
    """Calculate basic statistical measures for numerical columns in the dataset.

    This function computes fundamental statistical metrics including mean, median,
    standard deviation, skewness, and counts of missing values for all numerical
    columns in the provided DataFrame.

    Args:
        data: A pandas DataFrame containing the dataset to analyze. The DataFrame
            should contain at least one numerical column for meaningful analysis.

    Returns:
        str: A string containing formatted basic statistics for each numerical column,
            including mean, median, standard deviation, skewness, and missing value counts.
    """
    stats = {}
    numeric_cols = data.select_dtypes(include=[np.number]).columns

    for col in numeric_cols:
        stats[col] = {
            "mean": float(data[col].mean()),
            "median": float(data[col].median()),
            "std": float(data[col].std()),
            "skew": float(data[col].skew()),
            "missing": int(data[col].isnull().sum()),
        }

    return str(stats)


@tool
def generate_correlation_matrix(data: pd.DataFrame) -> str:
    """Generate a visual correlation matrix for numerical columns in the dataset.

    This function creates a heatmap visualization showing the correlations between
    all numerical columns in the dataset. The correlation values are displayed
    using a color-coded matrix for easy interpretation.

    Args:
        data: A pandas DataFrame containing the dataset to analyze. The DataFrame
            should contain at least two numerical columns for correlation analysis.

    Returns:
        str: A base64 encoded string representing the correlation matrix plot image,
            which can be displayed in a web interface or saved as an image file.
    """
    numeric_data = data.select_dtypes(include=[np.number])

    plt.figure(figsize=(10, 8))
    sns.heatmap(numeric_data.corr(), annot=True, cmap="coolwarm")
    plt.title("Correlation Matrix")

    buf = io.BytesIO()
    plt.savefig(buf, format="png")
    plt.close()
    return base64.b64encode(buf.getvalue()).decode()


@tool
def analyze_categorical_columns(data: pd.DataFrame) -> str:
    """Analyze categorical columns in the dataset for distribution and frequencies.

    This function examines categorical columns to identify unique values, top categories,
    and missing value counts, providing insights into the categorical data distribution.

    Args:
        data: A pandas DataFrame containing the dataset to analyze. The DataFrame
            should contain at least one categorical column for meaningful analysis.

    Returns:
        str: A string containing formatted analysis results for each categorical column,
            including unique value counts, top categories, and missing value counts.
    """
    categorical_cols = data.select_dtypes(include=["object", "category"]).columns
    analysis = {}

    for col in categorical_cols:
        analysis[col] = {
            "unique_values": int(data[col].nunique()),
            "top_categories": data[col].value_counts().head(5).to_dict(),
            "missing": int(data[col].isnull().sum()),
        }

    return str(analysis)


@tool
def suggest_features(data: pd.DataFrame) -> str:
    """Suggest potential feature engineering steps based on data characteristics.

    This function analyzes the dataset's structure and statistical properties to
    recommend possible feature engineering steps that could improve model performance.

    Args:
        data: A pandas DataFrame containing the dataset to analyze. The DataFrame
            can contain both numerical and categorical columns.

    Returns:
        str: A string containing suggestions for feature engineering based on
            the characteristics of the input data.
    """
    suggestions = []
    numeric_cols = data.select_dtypes(include=[np.number]).columns
    categorical_cols = data.select_dtypes(include=["object", "category"]).columns

    if len(numeric_cols) >= 2:
        suggestions.append("Consider creating interaction terms between numerical features")

    if len(categorical_cols) > 0:
        suggestions.append("Consider one-hot encoding for categorical variables")

    for col in numeric_cols:
        if data[col].skew() > 1 or data[col].skew() < -1:
            suggestions.append(f"Consider log transformation for {col} due to skewness")

    return "\n".join(suggestions)


@tool
def describe_data(data: pd.DataFrame) -> str:
    """Generates a comprehensive descriptive statistics report for the entire DataFrame.

    Args:
        data: A pandas DataFrame containing the dataset to analyze.

    Returns:
        str: String representation of the descriptive statistics
    """

    return data.describe(include="all").to_string()


@tool
def execute_code(code_string: str, data: pd.DataFrame) -> str:
    """Executes python code and returns results as a string.

    Args:
        code_string (str): Python code to execute.
        data (pd.DataFrame): The dataframe to use in the code
    Returns:
        str: The result of executing the code or an error message
    """
    try:
        # This dictionary will be available to the code
        local_vars = {"data": data, "pd": pd, "np": np, "plt": plt, "sns": sns}

        # Execute the code with the passed variables
        exec(code_string, local_vars)

        if "result" in local_vars:
            if isinstance(local_vars["result"], (pd.DataFrame, pd.Series)):
                return local_vars["result"].to_string()
            elif isinstance(local_vars["result"], plt.Figure):
                buf = io.BytesIO()
                local_vars["result"].savefig(buf, format="png")
                plt.close(local_vars["result"])
                return f"data:image/png;base64,{base64.b64encode(buf.getvalue()).decode()}"
            else:
                return str(local_vars["result"])
        else:
            return "Code executed successfully, but no variable called 'result' was assigned."

    except Exception as e:
        return f"Error executing code: {str(e)}"


@st.cache_data
def load_data(uploaded_file):
    """Loads data from an uploaded file with caching."""
    try:
        if uploaded_file.name.endswith(".csv"):
            return pd.read_csv(uploaded_file)
        elif uploaded_file.name.endswith((".xls", ".xlsx")):
            return pd.read_excel(uploaded_file)
        elif uploaded_file.name.endswith(".json"):
            return pd.read_json(uploaded_file)
        else:
            raise ValueError(
                "Unsupported file format. Please upload a CSV, Excel, or JSON file."
            )
    except Exception as e:
        st.error(f"Error loading data: {e}")
        return None


def main():
    st.title("Data Analysis Assistant")
    st.write("Upload your dataset and get automated analysis with natural language interaction.")

    # Initialize session state
    if "data" not in st.session_state:
        st.session_state["data"] = None
    if "agent" not in st.session_state:
        st.session_state["agent"] = None
    if "custom_code" not in st.session_state:
        st.session_state["custom_code"] = ""

    uploaded_file = st.file_uploader("Choose a CSV, Excel, or JSON file", type=["csv", "xlsx", "xls", "json"])

    if uploaded_file:
        with st.spinner("Loading and processing your data..."):
            data = load_data(uploaded_file)
            if data is not None:
                st.session_state["data"] = data

                st.session_state["agent"] = DataAnalysisAgent(
                    dataset=data,
                    tools=[
                        analyze_basic_stats,
                        generate_correlation_matrix,
                        analyze_categorical_columns,
                        suggest_features,
                        describe_data,
                        execute_code,
                    ],
                    model=GroqLLM(),
                    additional_authorized_imports=["pandas", "numpy", "matplotlib", "seaborn"],
                )
                st.success(
                    f"Successfully loaded dataset with {data.shape[0]} rows and {data.shape[1]} columns"
                )
                st.subheader("Data Preview")
                st.dataframe(data.head())

    if st.session_state["data"] is not None:
        analysis_type = st.selectbox(
            "Choose analysis type",
            [
                "Basic Statistics",
                "Correlation Analysis",
                "Categorical Analysis",
                "Feature Engineering",
                "Data Description",
                "Custom Code",
                "Custom Question",
            ],
        )

        if analysis_type == "Basic Statistics":
            with st.spinner("Analyzing basic statistics..."):
                result = st.session_state["agent"].run(
                    "Use the analyze_basic_stats tool to analyze this dataset and "
                    "provide insights about the numerical distributions."
                )
                st.write(result)

        elif analysis_type == "Correlation Analysis":
            with st.spinner("Generating correlation matrix..."):
                result = st.session_state["agent"].run(
                    "Use the generate_correlation_matrix tool to analyze correlations "
                    "and explain any strong relationships found."
                )
                if isinstance(result, str) and result.startswith("data:image") or "," in result:
                    st.image(f"data:image/png;base64,{result.split(',')[-1]}")
                else:
                    st.write(result)

        elif analysis_type == "Categorical Analysis":
            with st.spinner("Analyzing categorical columns..."):
                result = st.session_state["agent"].run(
                    "Use the analyze_categorical_columns tool to examine the "
                    "categorical variables and explain the distributions."
                )
                st.write(result)

        elif analysis_type == "Feature Engineering":
            with st.spinner("Generating feature suggestions..."):
                result = st.session_state["agent"].run(
                    "Use the suggest_features tool to recommend potential "
                    "feature engineering steps for this dataset."
                )
                st.write(result)

        elif analysis_type == "Data Description":
            with st.spinner("Generating data description"):
                result = st.session_state["agent"].run(
                    "Use the describe_data tool to generate a comprehensive description "
                    "of the data."
                )
                st.write(result)

        elif analysis_type == "Custom Code":
            st.session_state["custom_code"] = st_ace(
                placeholder="Enter your Python code here...",
                language="python",
                theme="github",
                key="code_editor",
                value=st.session_state["custom_code"],
            )
            if st.button("Run Code"):
                with st.spinner("Executing custom code..."):
                    result = st.session_state["agent"].run(
                        f"Execute the following code and return any 'result' variable"
                        f"```python\n{st.session_state['custom_code']}\n```"
                    )
                    if isinstance(result, str) and result.startswith("data:image"):
                        st.image(f"{result}")
                    else:
                        st.write(result)

        elif analysis_type == "Custom Question":
            question = st.text_input("What would you like to know about your data?")
            if question:
                with st.spinner("Analyzing..."):
                    result = st.session_state["agent"].run(question, stream=True)  # Pass stream argument here
                    st.write(result)


if __name__ == "__main__":
    main()