stats_dashboard.py

#stats_dashboard.py
from config import temp_file_path  # Import the global variable
def update_dashboard():
        total_ppts, total_slides, chart_data, latest_html = get_dashboard_stats()
        return (
            gr.update(visible=True),
            gr.update(value=f"<div><h3>Total PPTs: {total_ppts}</h3></div>"),
            gr.update(value=f"<div><h3>Total Slides: {total_slides}</h3></div>"),
            gr.update(value=chart_data),
            gr.update(value=latest_html)
        )
import pandas as pd
import gradio as gr
import os
def get_dashboard_stats():
    # Load metadata CSV
    global temp_file_path
    print('Reading CSV...')
    #metadata_file_name= "Master_metadata.csv"
    # df = pd.read_csv(metadata_file_name)
    #temp_file_path = os.path.join("/tmp", metadata_file_name)
    df = pd.read_csv(temp_file_path)
    
    # Ensure upload_date column is in datetime format
    df["Upload_date"] = pd.to_datetime(df["Upload_date"], errors="coerce")
    print(df)
    # Total unique PPTs and slides
    total_ppts = df["PPT_Unique_ID"].nunique()
    total_slides = len(df)

    # Monthly PPT uploads
    df["month_year"] = df["Upload_date"].dt.to_period("M").astype(str)
    monthly_stats = df.groupby("month_year")["PPT_Unique_ID"].nunique().reset_index()
    monthly_stats.columns = ["Month", "PPT Uploads"]

    # Gradio BarPlot requires a DataFrame
    chart_data = monthly_stats

    # Latest 5 PPTs by upload date
    latest_df = df.drop_duplicates(subset="PPT_Unique_ID").sort_values("Upload_date", ascending=False)
    latest_5 = latest_df[["Suitable_Title", "Slide_Category","Upload_date"]].head(5)
    # Create HTML for the latest PPTs list
    # Create HTML for the latest PPTs list with heading
    latest_html = "<h4 style='margin-bottom: 8px;'>📌 Top 5 Latest Uploaded PPTs</h4><ul style='line-height:1.6em;'>"
    for _, row in latest_5.iterrows():
        title = row["Suitable_Title"]
        category = row["Slide_Category"]
        date_str = row["upload_date"].strftime("%Y-%m-%d") if pd.notnull(row["upload_date"]) else "Unknown Date"
        latest_html += f"<li><b>{title}</b> <br><i>{category}</i> — <span style='color:gray;'>{date_str}</span></li>"
    latest_html += "</ul>"
    return total_ppts, total_slides, chart_data, latest_html