generate.py

import os
import re
import io
import time
import yaml
import pandas as pd
from pptx import Presentation
import comtypes.client

def load_config(path="config.yaml"):
    with open(path, "r") as f:
        return yaml.safe_load(f)

def load_data(file_path):
    if file_path.endswith(".csv"):
        return pd.read_csv(file_path, dtype=str).fillna("")
    return pd.read_excel(file_path, dtype=str).fillna("")

def sanitize_filename(text):
    return re.sub(r'[\\/*?:"<>|]', "", str(text)).replace(" ", "_")

def replace_text(shape, replacements):
    if not shape.has_text_frame:
        return

    for paragraph in shape.text_frame.paragraphs:
        full_text = "".join(run.text for run in paragraph.runs)

        new_text = full_text
        for key, value in replacements.items():
            new_text = new_text.replace(key, str(value))

        if new_text == full_text:
            continue

        current_index = 0
        for run in paragraph.runs:
            run_length = len(run.text)
            run.text = new_text[current_index:current_index + run_length]
            current_index += run_length

        if current_index < len(new_text):
            paragraph.runs[-1].text += new_text[current_index:]

def convert_to_pdf(powerpoint, ppt_path, pdf_path):
    # ppt_path = os.path.abspath(ppt_path)
    # pdf_path = os.path.abspath(pdf_path)
    presentation = powerpoint.Presentations.Open(ppt_path, WithWindow=False)
    presentation.SaveAs(pdf_path, 32)
    presentation.Close()
    os.remove(ppt_path)


def generate_certificates(config, df):
    start_time = time.time()

    cert_type = config["certificate"]["type"]
    template_path = config["certificate"]["templates"][cert_type]
    output_folder = config["output"]["folder"]

    name_column = config["input"]["name_column"]
    uid_column = config["input"]["uid_column"]
    ref_column = config["input"]["ref_column"]
    dept_column = config["input"]["dept_column"]
    role_column = config["input"].get("role_column")

    event_name = config["event"]["name"]
    event_date = config["event"]["date"]

    os.makedirs(output_folder, exist_ok=True)

    with open(template_path, "rb") as f:
        template_bytes = f.read()

    ppt_files = []
    success, fail = 0, 0

    print("🚀 Generating PPT files...")

    for row in df.itertuples(index=False):
        
        try: 
            uid = str(getattr(row, uid_column, "")).strip()
            if not uid:
                # print("⚠️ Skipping row with empty UID")
                continue
            
            name = getattr(row, name_column, "")
            ref = getattr(row, ref_column, "")
            dept = getattr(row, dept_column, "")
            role = getattr(row, role_column, "") if role_column else ""

            prs = Presentation(io.BytesIO(template_bytes))

            replacements = {
                "{{name}}": name,
                "{{uid}}": uid,
                "{{ref}}": ref,
                "{{dept}}": dept,
                "{{event_name}}": event_name,
                "{{event_date}}": event_date
            }

            if role_column:
                replacements["{{role}}"] = role

            for slide in prs.slides:
                for shape in slide.shapes:
                    replace_text(shape, replacements)
                
            safe_event = sanitize_filename(event_name)
            safe_ref = sanitize_filename(ref)
            safe_uid = sanitize_filename(uid)

            filename = f"{safe_event}_{safe_ref}_{safe_uid}"

            ppt_path = os.path.abspath(os.path.join(output_folder, filename + ".pptx"))

            prs.save(ppt_path)
            ppt_files.append(ppt_path)

            success+=1
            
        except Exception as e:
            print(f"❌ {uid} - {e}")
            fail += 1

    print(f"PPT Generated → {success}")
    print("🚀 Converting to PDF...")

    powerpoint = comtypes.client.CreateObject("Powerpoint.Application")
    powerpoint.Visible = 1
    powerpoint.WindowState = 2

    try:
        for ppt_path in ppt_files:
            try:
                pdf_path = ppt_path.replace(".pptx", ".pdf")
                convert_to_pdf(powerpoint, ppt_path, pdf_path)
                print(f"✅ {os.path.basename(pdf_path)}")
            except Exception as e:
                print(f"❌ Conversion failed: {ppt_path} - {e}")
                fail += 1
    finally:
        powerpoint.Quit()

    end_time = time.time()

    print(f"✅ Success: {success} | ❌ Failure: {fail} | Time: {round(end_time - start_time, 2)} sec")

def main():
    config = load_config()
    df = load_data(config["input"]["file"])
    generate_certificates(config, df)

if __name__=="__main__":
    main()