app.py

import pandas as pd
import ydata_profiling
import gradio as gr
from pydantic_settings import BaseSettings
from tempfile import NamedTemporaryFile
import sweetviz as sv
import sketch
#from datatile.summary.df import DataFrameSummary
from autoviz.AutoViz_Class import AutoViz_Class
from traceml.summary.df import DataFrameSummary
import os
df1=gr.State()

def variable_table(df):
    """
    Analyzes a DataFrame and categorizes variables with colorful HTML formatting.
    
    Args:
      df (pandas.DataFrame): The DataFrame to analyze.
    
    Returns:
      str: HTML code representing the analysis results with colorful highlights.
    """
    # Analyze variable types
    numerical_vars=[]
    text_vars=[]
    categorical_vars=[]
    # Identify categorical, numerical, and text columns
    # Iterate over columns
    for col in df.columns:
        print(col )
        unique_values = df[col].nunique()
        total_values = df[col].count()
        dtype_kind = df[col].dtype.kind
        print( 'unique value-',unique_values,' total_values ',total_values,dtype_kind)
        # Check criteria for numerical, text, and categorical variables
        if (dtype_kind == 'f' or dtype_kind == 'i') and unique_values >= total_values / 10:
            numerical_vars.append(col)
            print('Numerical')
        elif (dtype_kind == 'O' or dtype_kind == 'b') and unique_values > total_values / 10:
            text_vars.append(col)
            print('Text')
        else:
            categorical_vars.append(col)
            print('categorical')
    print(categorical_vars,numerical_vars,text_vars)

    # Build HTML table with styles
    table_style = 'border: 1px solid #ddd; border-collapse: collapse; text-align: left; font-size: 14px;'
    header_style = 'background-color: #f2f2f2; padding: 5px 10px;'
    data_style = 'padding: 5px 10px; border-bottom: 1px solid #ddd;'
    category_color = '#90ee90'  # Light green for categorical
    numerical_color = '#add8e6'  # Light blue for numerical
    text_color = '#ffd9b3'  # Light yellow for text
    
    html = f"<table style='{table_style}'>"
    html += f"<tr><th style='{header_style}'>Variable Type</th><th style='{header_style}'>Columns</th></tr>"
    
    # Add rows for each variable type with coloring
    if categorical_vars:
        html += f"<tr style='background-color: {category_color};'><td>Categorical</td><td style='{data_style}'>{', '.join(categorical_vars)}</td></tr>"
    if numerical_vars:
        html += f"<tr style='background-color: {numerical_color};'><td>Numerical</td><td style='{data_style}'>{', '.join(numerical_vars)}</td></tr>"
    if text_vars:
        html += f"<tr style='background-color: {text_color};'><td>Text</td><td style='{data_style}'>{', '.join(text_vars)}</td></tr>"
    
    # Handle cases where no variables are found
    if not (categorical_vars or numerical_vars or text_vars):
        html += "<tr><td>No variables found!</td></tr>"
    
    html += "</table>"
    return html

def generate_report(file, type):
    df = pd.read_csv(file) if file.name.endswith(".csv") else pd.read_excel(file)
    

    pandas_html_report =ydata_profiling.ProfileReport(df).to_html() 
    temp_file1 = NamedTemporaryFile(delete=False, suffix=".html")
    temp_file1.write(pandas_html_report.encode('utf-8'))
    temp_file1.close()
    print('Pandas completed')
    
    # dataprep_report = create_report(df)
    # temp_file2 = NamedTemporaryFile(delete=False, suffix=".html")
    # temp_file2.write(dataprep_report.encode('utf-8'))
    # temp_file2.close()

    def convert_mixed_to_str(df):
  
      mixed_cols = df.select_dtypes(include=['object']).columns  # Detect object dtype (usually mixed)
      for col in mixed_cols:
        df[col] = df[col].astype(str)  # Convert mixed columns to strings
      return df

    df = convert_mixed_to_str(df.copy())
    df.columns = df.columns.str.upper()
    df1.value=df
    sweetviz_report = sv.analyze(df)
    #sweetviz_report=sweetviz_report.show_html()
    #print(type(sweetviz_report))
    report=sweetviz_report.show_html(  filepath='SWEETVIZ_REPORT.html', 
            open_browser=False, 
            layout='widescreen', 
            scale=None)
    HTMLFileToBeOpened = open('SWEETVIZ_REPORT.html', "r") 
  
    # Reading the file and storing in a variable 
    contents = HTMLFileToBeOpened.read() 
    print('Pandas completed')
    
  
    temp_file3 = NamedTemporaryFile(delete=False, suffix=".html")
    temp_file3.write(contents.encode('utf-8'))
    temp_file3.close()
    dfviz=df
    try:
        AV = AutoViz_Class()
        dft = AV.AutoViz(filename="", sep=',', dfte=df, header=0, verbose=1,
                        lowess=False,chart_format='html',max_rows_analyzed=150000,
                         max_cols_analyzed=30,save_plot_dir=None)
        reports = []
        print(os.listdir("./"))
        output_dir_html="./AutoViz_Plots/AutoViz"
        for filename in os.listdir(output_dir_html):
            if filename.endswith(".html"):
                title = os.path.splitext(filename)[0]  # Extract title from filename (without extension)
                reports.append(title)
                reports.append(output_dir_html+'/'+filename)
                print(reports)
                
        dft = AV.AutoViz(filename='', sep=',', header=0, verbose=2,dfte=df,
                      lowess=False,chart_format='png',max_rows_analyzed=150000,
                      max_cols_analyzed=30,save_plot_dir=None)
        
        imgs = []
        output_dir_img="./AutoViz_Plots/AutoViz"
        for filename in os.listdir(output_dir_img):
            if filename.endswith(".png"):
                title = os.path.splitext(filename)[0]  # Extract title from filename (without extension)
                imgs.append(title)
                imgs.append(output_dir_img+'/'+filename)
                print(imgs)
                
    except Exception as e:
        dfviz=df
        print(e)
        print('Exception during Autoviz')

 
   
    while len(reports)<10:
          # Comparing the lengths and appending t and m to output if needed

          t = gr.Markdown(visible=False)
          m = gr.File(visible=False)
          reports.append(t)
          reports.append(m)
          
    while len(imgs)<10:
          # Comparing the lengths and appending t and m to output if needed
    
          t = gr.Markdown(visible=False)
          m = gr.Image(visible=False)
          imgs.append(t)
          imgs.append(m)   
   
    dfs = DataFrameSummary(df)
    sd=dfs.summary()
    sd.index.name = 'PARAMETERS'
    sd1= sd.reset_index(drop=False)
    def highlight_cols(x): 
        df = x.copy() 
        df.loc[:, :] = 'color: purple'
        df[['PARAMETERS']] = 'color: red'
        return df 

    # Applying the style function
    sd1 = sd1.style.apply(highlight_cols, axis = None)
    sd1.columns = sd1.columns.str.upper()


    
    return [temp_file1.name  ,temp_file3.name ,variable_table(df),dfviz,sd1]+reports+imgs

def query_chat(query):
    df=df1.value
    return df.sketch.ask(query)
    
def handle_click(file, sheet_name):
    print('Handle click')  
    results = generate_report(file, sheet_name)
    return results
    
def update_sheets(file):
    sheet_names = pd.ExcelFile(file).sheet_names
    
    print('Sheet updated')
    sheet_dropdown = gr.Dropdown(
            label="Select sheet", choices=[sheet_names], visible=True  # Initially disabled
        )
    return sheet_dropdown
    

with gr.Blocks() as cluster:
    with gr.Column():
        gr.HTML(value="""<h1 style="color: #1999FF; text-shadow: 1px 1px 2px #ddd;">EXCEL ANALYSIS AND INSIGHTS</h1>""")
        with gr.Row():
            file=gr.File(file_types=['.csv', '.xlsx'], label="Upload a CSV or Excel file")
            btn=gr.Button(value="Download Report") 
            
        with gr.Row():
            
            
            gr.HTML(value="""<h2 style="color: #3399FF; text-shadow: 1px 1px 2px #ddd;">PANDAS REPORT</h2>""")
            out1=gr.File(label="Download CSV")
            gr.HTML(value="""<h2 style="color: #3399FF; text-shadow: 1px 1px 2px #ddd;">SWEETVIZ REPORT</h2>""")
            out2=gr.File(label="Download CSV")
        sheet_dropdown = gr.Dropdown(
            label="Select sheet", choices=[], visible=False  # Initially disabled
        )
    
        
        with gr.Column():
            gr.HTML(value="""<h2 style="color: #BA4A00; text-shadow: 1px 1px 2px #ddd;"COLUMNS IN FILE</h2>""")
            #gr.Markdown("***Uploaded File***")
            var=gr.HTML()
        with gr.Column():
            gr.HTML(value="""<h2 style="color: #BA4A00; text-shadow: 1px 1px 2px #ddd;"UPLOADED FILE</h2>""")
            #gr.Markdown("#Uploaded File")
            dataframe1=gr.Dataframe()
        with gr.Column():
            gr.HTML(value="""<h2 style="color: #BA4A00; text-shadow: 1px 1px 2px #ddd;"COLUMNS STATS AND ANALYSIS</h2>""")
            #gr.Markdown("#Columns Analysis")
            dataframe2=gr.Dataframe()

        with gr.Column():
            gr.HTML(value="""<h2 style="color: #BA4A00; text-shadow: 1px 1px 2px #ddd;"ASK QUESTIONS ON THE COLUMNS </h2>""")
            with gr.Row():
                query=gr.Textbox(label="Enter any question on data columns",
                info="Natural language query",
                lines=1,
                value="What is maximum value of first column")
                outlabel=gr.HighlightedText(label='Query Result ')
            query_btn=gr.Button(value='Submit Query')
            query_btn.click(query_chat,inputs=query,outputs=outlabel)
            
            

        #Autoviz insights graphs
  
        gr.HTML(value="""<h3 style="color: #1999FF; text-shadow: 1px 1px 2px #ddd;">Insight Visualisations</h3>""")
        with gr.Column():
          plots2 = []
          for i in range(5):
              t = gr.Markdown()
              with gr.Column(scale=1):
                m=gr.Image()
                plots2.append(t)
                plots2.append(m)
                print(type(plots2[0]))
        gr.HTML(value="""<h3 style="color: #1999FF; text-shadow: 1px 1px 2px #ddd;">Insight graphs</h3>""")
        with gr.Row():           
          plots1 = []
          for i in range(5):
            with gr.Row():
              t = gr.Markdown()
              m=gr.File()
              plots1.append(t)
              plots1.append(m)
              print(type(plots1[0]))
    
    btn.click(handle_click, inputs=[file, sheet_dropdown], outputs=[out1, out2, var,dataframe1, dataframe2]+plots1+plots2)
    file.change(update_sheets, inputs=file, outputs=sheet_dropdown) 
    #btn.click(generate_report,inputs=[file],outputs=[out1,out2,var,dataframe1,dataframe2])
cluster.launch(debug=True)