"""
Ashley Wu
Generates insights into the data given
"""
import pandas as pd
import gradio as gr
import utils

def calculate_stats(df):
    """
    Create a dataframe with each column's mean, median, max, min, quartiles for numeric columns and unique values, value counts, and mode for categorical columns as well as missing values 
    
    Args:
        dataframe: The pandas DataFrame to analyze
    
    Returns:
        A DataFrame with column statistics
    """
    df = utils.detect_categories(df)
    df = utils.detect_dates(df)
    
    if df is None or len(df) == 0:
        return None, None

    numeric_data = pd.DataFrame()
    categorical_data = pd.DataFrame() 

    for column in df.columns: 
        dtype = df[column].dtype
        if dtype.kind == 'i' or dtype.kind == 'f' or dtype.kind == 'u' or dtype.kind == 'c':
            numeric_data[column] = df[column]
        elif dtype.name == 'category':
            categorical_data[column] = df[column]
        elif dtype.kind == 'M' or dtype.kind == 'm':
            continue
        else: 
            gr.Warning(f"dType of {column} could not be recognised!", duration=5)
    
    numeric_data = calcualte_numeric_data(numeric_data)
    categorical_data = calculate_categorical_data(categorical_data)
    
    return numeric_data, categorical_data

def calculate_categorical_data(df): 
    """
    Calculates data for categorical data exploration
    
    Args: 
        df: dataframe
    Returns 
        table: dataframe
    """
    table = pd.DataFrame({
        'Column': pd.Series(dtype='string'),
        'Mode': pd.Series(dtype='string'),
        'Missing Values': pd.Series(dtype='int16'),
        'Value Counts': pd.Series(dtype='string'),
    })

    for column in df.columns: 
        mode = df[column].mode()[0]
        count = str(df[column].value_counts().to_dict())
        missing_values = df[column].isnull().sum()
        table.loc[len(table)] = [column, mode, missing_values, count]

    return table


def calcualte_numeric_data(df):
    """
    Calcualtes statistics for numerical data
    
    Args: 
        df: dataframe 
    Returns: 
        table: df with column statistics
    """
    table = pd.DataFrame({
        'Column': pd.Series(dtype='string'),
        'Mean':  pd.Series(dtype='float16'),
        'Median':  pd.Series(dtype='float16'),
        'Max':  pd.Series(dtype='float16'),
        'Min':  pd.Series(dtype='float16'),
        'Bottom Quartile':  pd.Series(dtype='float16'),
        'Top Quartile':  pd.Series(dtype='float16'),
        'Missing Values': pd.Series(dtype='int16'),
    })
    
    for column in df.columns:
        mean = df[column].mean()
        median = df[column].median()
        max_val = df[column].max()
        min_val = df[column].min()
        q25 = df[column].quantile(.25)
        q75 = df[column].quantile(.75)
        missing_values = df[column].isnull().sum()
        table.loc[len(table)] = [column, mean, median, max_val, min_val, q25, q75, missing_values]

    return table

def correlation_matrix(df): 
    """
    Create a correleation matrix for the given dataframe

    Args: 
        df: dataframe 
    Returns: 
        correlation_matrix: numpy array 
    """
    for col in df.columns:
        if df[col].dtype.name == 'object':
            df = df.drop(col, axis=1)
    return df.corr()


def max_min_df(df):
    """
    Find the max and min of a data frame
    
    Arg: 
        df: dataframe 
    Returns: 
        max_min_df: dataframe
    """
    for col in df.columns:
        if df[col].dtype.name == 'object':
            df = df.drop(col, axis=1)

    df_max = df.max()
    df_min = df.min()

    result = pd.DataFrame({
        'Column': df_max.index,
        'Max': df_max.values,
        'Min': df_min.values
    })
    return result

def std(df):
    """
    Finds the standard deviation of the dataframe 
    
    Args:
        df: dataframe 
    Returns: 
        std: dataframe 
    """
    for col in df.columns:
        if df[col].dtype.name == 'object':
            df = df.drop(col, axis=1)

    std_series = df.std()
    result = std_series.to_frame(name='Standard Deviation').reset_index()
    result.columns = ['Column', 'Standard Deviation']
    return result

def var(df):
    """
    Finds the variance of values in a dataframe 
    
    Args: 
        df: dataframe 
    Returns: 
        var: df
    """
    for col in df.columns:
        if df[col].dtype.name == 'object':
            df = df.drop(col, axis=1)
    var_series = df.var()
    result = var_series.to_frame(name='Variance').reset_index()
    result.columns = ['Column', 'Variance']
    return result