import os
import tempfile
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.figure_factory as ff
import plotly.graph_objects as go
from scipy.cluster.hierarchy import linkage, leaves_list
from typing import Union, Tuple, List


def _save_fig(fig: go.Figure, prefix: str, output_dir: str) -> str:
    """
    Save a Plotly figure as a high-res PNG and return the file path.
    """
    os.makedirs(output_dir, exist_ok=True)
    tmp = tempfile.NamedTemporaryFile(suffix='.png', prefix=prefix, dir=output_dir, delete=False)
    path = tmp.name
    tmp.close()
    fig.write_image(path, scale=3)
    return path


def histogram_tool(
    file_path: str,
    column: str,
    bins: int = 30,
    kde: bool = True,
    output_dir: str = '/tmp'
) -> Union[Tuple[ff.FigureFactory, str], str]:
    """
    Create a histogram with optional KDE overlay for a given numeric column.

    Returns (figure, png_path) or error string.
    """
    # Load
    ext = os.path.splitext(file_path)[1].lower()
    df = pd.read_excel(file_path) if ext in ('.xls','.xlsx') else pd.read_csv(file_path)

    # Validate
    if column not in df.columns:
        return f"❌ Column '{column}' not found."
    series = pd.to_numeric(df[column], errors='coerce').dropna()
    if series.empty:
        return f"❌ No numeric data in '{column}'."

    # Build histogram + KDE
    if kde:
        fig = ff.create_distplot([series], [column], bin_size=(series.max()-series.min())/bins)
    else:
        fig = px.histogram(series, nbins=bins, title=f"Histogram – {column}", template='plotly_dark')
    fig.update_layout(template='plotly_dark')

    # Save
    img_path = _save_fig(fig, f"hist_{column}_", output_dir)
    return fig, img_path


def boxplot_tool(
    file_path: str,
    column: str,
    output_dir: str = '/tmp'
) -> Union[Tuple[px.Figure, str], str]:
    """
    Create a box plot with outliers for a numeric column.

    Returns (figure, png_path) or error string.
    """
    ext = os.path.splitext(file_path)[1].lower()
    df = pd.read_excel(file_path) if ext in ('.xls','.xlsx') else pd.read_csv(file_path)
    if column not in df.columns:
        return f"❌ Column '{column}' not found."
    series = pd.to_numeric(df[column], errors='coerce').dropna()
    if series.empty:
        return f"❌ No numeric data in '{column}'."

    fig = px.box(series, points='outliers', title=f"Boxplot – {column}", template='plotly_dark')
    img_path = _save_fig(fig, f"box_{column}_", output_dir)
    return fig, img_path


def violin_tool(
    file_path: str,
    column: str,
    output_dir: str = '/tmp'
) -> Union[Tuple[px.Figure, str], str]:
    """
    Create a violin plot with inner box for a numeric column.

    Returns (figure, png_path) or error string.
    """
    ext = os.path.splitext(file_path)[1].lower()
    df = pd.read_excel(file_path) if ext in ('.xls','.xlsx') else pd.read_csv(file_path)
    if column not in df.columns:
        return f"❌ Column '{column}' not found."
    series = pd.to_numeric(df[column], errors='coerce').dropna()
    if series.empty:
        return f"❌ No numeric data in '{column}'."

    fig = px.violin(series, box=True, points='all', title=f"Violin – {column}", template='plotly_dark')
    img_path = _save_fig(fig, f"violin_{column}_", output_dir)
    return fig, img_path


def scatter_matrix_tool(
    file_path: str,
    columns: List[str],
    output_dir: str = '/tmp',
    size: int = 5
) -> Union[Tuple[px.Figure, str], str]:
    """
    Create an interactive scatter matrix for selected numeric columns.

    Returns (figure, png_path) or error string.
    """
    ext = os.path.splitext(file_path)[1].lower()
    df = pd.read_excel(file_path) if ext in ('.xls','.xlsx') else pd.read_csv(file_path)
    missing = [c for c in columns if c not in df.columns]
    if missing:
        return f"❌ Missing columns: {', '.join(missing)}"
    df_num = df[columns].apply(pd.to_numeric, errors='coerce').dropna()
    if df_num.empty:
        return "❌ No valid numeric data."

    fig = px.scatter_matrix(df_num, dimensions=columns, title="Scatter Matrix", template='plotly_dark')
    fig.update_traces(diagonal_visible=False, marker={'size': size})
    img_path = _save_fig(fig, "scatter_matrix_", output_dir)
    return fig, img_path


def corr_heatmap_tool(
    file_path: str,
    columns: List[str] = None,
    output_dir: str = '/tmp',
    cluster: bool = True
) -> Union[Tuple[px.Figure, str], str]:
    """
    Create a correlation heatmap, with optional hierarchical clustering of variables.

    Returns (figure, png_path) or error string.
    """
    ext = os.path.splitext(file_path)[1].lower()
    df = pd.read_excel(file_path) if ext in ('.xls','.xlsx') else pd.read_csv(file_path)
    df_num = df.select_dtypes(include='number') if columns is None else df[columns]
    df_num = df_num.apply(pd.to_numeric, errors='coerce').dropna(axis=1, how='all')
    if df_num.shape[1] < 2:
        return "❌ Need at least two numeric columns for correlation."

    corr = df_num.corr()
    if cluster:
        link = linkage(corr, method='average')
        order = leaves_list(link)
        corr = corr.iloc[order, order]

    fig = px.imshow(
        corr,
        color_continuous_scale='RdBu',
        title="Correlation Heatmap",
        labels=dict(color="Correlation"),
        template='plotly_dark'
    )
    img_path = _save_fig(fig, "corr_heatmap_", output_dir)
    return fig, img_path