script_generator.py

import os
import gradio as gr
import wikipediaapi
from groq import Groq
import torch
from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
from deep_translator import GoogleTranslator
import yake
from datetime import datetime
import shutil
import glob

# ✅ Set API Key
# os.environ["GROQ_API_KEY"] = "your_api_key_here"
# client = Groq(api_key=os.environ["GROQ_API_KEY"])

# ✅ Set API Key
os.environ["GROQ_API_KEY"] = "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"    # Replace with your API key
client = Groq(api_key=os.environ["GROQ_API_KEY"])

# # ✅ Load M2M-100 Model
# model_name = "facebook/m2m100_418M"
# tokenizer = M2M100Tokenizer.from_pretrained(model_name)
# model = M2M100ForConditionalGeneration.from_pretrained(model_name)

def fetch_wikipedia_summary(topic):
    wiki_wiki = wikipediaapi.Wikipedia(
        user_agent="EducationalScriptApp/1.0",
        language="en"
    )
    page = wiki_wiki.page(topic)
    return page.summary if page.exists() else "No Wikipedia summary available."

def generate_script(topic, duration):
    try:
        factual_content = fetch_wikipedia_summary(topic)
        words_per_minute = 130
        target_words = duration * words_per_minute

        response = client.chat.completions.create(
            messages=[{"role": "user", "content": f"Format the following factual content into a well-structured educational script in English with approximately {target_words} words: \n{factual_content}"}],
            model="llama-3.3-70b-versatile"
        )
        return response.choices[0].message.content
    except Exception as e:
        return f"❌ Error in script generation: {str(e)}"


# ✅ Function to Extract Keywords Using YAKE
def extract_keywords(script):
    try:
        kw_extractor = yake.KeywordExtractor(
            lan="en",  # Language
            n=3,  # Max number of words in a keyword phrase (trigrams)
            dedupLim=0.9,  # Reduce redundant phrases
            # top=10  # Extract top 10 keywords
        )

        keywords = kw_extractor.extract_keywords(script)
        return ", ".join([kw[0] for kw in keywords])  # ✅ Extract only the keyword text
    except Exception as e:
        return f"❌ Error extracting keywords: {str(e)}"


def save_keywords_file(keywords, topic):
    today = datetime.today().strftime('%Y_%b_%d')
    filename = f"Keywords/{topic}_Keyword_{today}.txt"
    os.makedirs(os.path.dirname(filename), exist_ok=True)
    with open(filename, "w", encoding="utf-8") as f:
        f.write(keywords)
    return filename


# # ✅ Function to Translate English Script to Urdu
# def translate_to_urdu(english_script):
#     try:
#         google_translation = GoogleTranslator(source='en', target='ur').translate(english_script)
#         tokenizer.src_lang = "en"
#         max_length = 500
#         input_chunks = [google_translation[i:i+max_length] for i in range(0, len(google_translation), max_length)]
#         refined_chunks = []
#         for chunk in input_chunks:
#             inputs = tokenizer(chunk, return_tensors="pt", truncation=True, max_length=1024).to(model.device)
#             translated_tokens = model.generate(
#                 **inputs,
#                 max_length=1024,
#                 no_repeat_ngram_size=2,
#                 forced_bos_token_id=tokenizer.get_lang_id("ur"),
#                 num_beams=2
#             )
#             refined_chunks.append(tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0])
#         return " ".join(refined_chunks)
#     except Exception as e:
#         return f"❌ Error in translation: {str(e)}"


def translate_to_urdu(english_script):
    try:
        # ✅ Define a max chunk size (Google Translator has a limit)
        max_chunk_size = 4500  # Stay below 5000 to be safe
        chunks = [english_script[i:i + max_chunk_size] for i in range(0, len(english_script), max_chunk_size)]

        translated_chunks = []
        for chunk in chunks:
            translated_chunk = GoogleTranslator(source='en', target='ur').translate(chunk)
            translated_chunks.append(translated_chunk)

        return " ".join(translated_chunks)  # ✅ Join all translated chunks
    except Exception as e:
        return f"❌ Error in translation: {str(e)}"



def save_english_file(content, topic):
    today = datetime.today().strftime('%Y_%b_%d')  # Format: 2025_Feb_21
    filename = f"English_Scripts/{topic}_Eng_{today}.txt"
    os.makedirs(os.path.dirname(filename), exist_ok=True)  # Ensure directory exists
    with open(filename, "w", encoding="utf-8") as f:
        f.write(content)
    return filename


def save_urdu_file(content, topic):
    today = datetime.today().strftime('%Y_%b_%d')
    filename = f"Urdu_Scripts/{topic}_Urdu_{today}.txt"
    os.makedirs(os.path.dirname(filename), exist_ok=True)
    with open(filename, "w", encoding="utf-8") as f:
        f.write(content)
    return filename


def save_final_urdu_file(topic, content):
    date_str = datetime.now().strftime("%Y_%b_%d")
    filename = f"Urdu_Final/{topic}_Urdu_Final_{date_str}.txt"  # ✅ Corrected file path
    os.makedirs(os.path.dirname(filename), exist_ok=True)  # ✅ Ensure the directory exists
    with open(filename, "w", encoding="utf-8") as f:
        f.write(content)
    return filename


def finalize_process():
    return "✅ Script Generation Completed Successfully!"


def clear_old_files():
    # ✅ Define all directories where files are stored
    directories = ["English_Scripts", "Urdu_Scripts", "Urdu_Final", "Keywords"]

    for directory in directories:
        if os.path.exists(directory):  # ✅ Check if directory exists
            files = glob.glob(f"{directory}/*")  # ✅ Get all files in the directory
            for file in files:
                try:
                    os.remove(file)  # ✅ Delete each file
                except Exception as e:
                    print(f"❌ Error deleting {file}: {e}")

    return "", "", "", "", ""  # ✅ Clear all textboxes in UI




# ✅ Gradio UI
with gr.Blocks() as app:
    gr.Markdown("# 🎬 AI-Powered Educational Script Generator")

    topic_input = gr.Textbox(label="Enter Topic")
    duration_input = gr.Slider(minimum=1, maximum=30, step=1, label="Duration (minutes)")


    generate_button = gr.Button("Generate English Script")
    eng_output = gr.Textbox(label="Generated English Script", interactive=False)
    download_english_button = gr.Button("Download English Script")
    download_english_button.click(save_english_file, inputs=[eng_output, topic_input], outputs=[gr.File()])


    # ✅ Keyword Extraction Section
    extract_keywords_btn = gr.Button("🔑 Extract Keywords")
    keyword_output = gr.Textbox(label="🔍 Extracted Keywords", interactive=True)
    download_keywords_btn = gr.Button("⬇️ Download Keywords")
    download_keywords_btn.click(save_keywords_file, inputs=[keyword_output, topic_input], outputs=[gr.File()])

    translate_button = gr.Button("Generate Urdu Script")
    urdu_output = gr.Textbox(label="Translated Urdu Script", interactive=False, rtl=True)
    download_urdu_button = gr.Button("Download Urdu Script")
    download_urdu_button.click(save_urdu_file, inputs=[urdu_output, topic_input], outputs=[gr.File()])

    
    final_edited_urdu_output = gr.Textbox(label="Edited Urdu Script", interactive=True, rtl=True)
    download_final_urdu_button = gr.Button("Download Final Urdu Script")
    download_final_urdu_button.click(save_final_urdu_file, inputs=[topic_input, final_edited_urdu_output], outputs=[gr.File()])


    # ✅ Button Actions
    # generate_button.click(generate_script, inputs=[topic_input, duration_input], outputs=[eng_output])
    generate_button.click(generate_script, inputs=[topic_input, duration_input],  outputs=[eng_output])
    extract_keywords_btn.click(extract_keywords, inputs=[eng_output], outputs=[keyword_output])
    translate_button.click(translate_to_urdu, inputs=[eng_output], outputs=[urdu_output])

    status_output = gr.Textbox(label="Status", interactive=False)
    finalize_button = gr.Button("Finalize Process")
    finalize_button.click(finalize_process, outputs=[status_output])

    generate_button.click(
    lambda topic, duration: (*clear_old_files(), generate_script(topic, duration)),  
    inputs=[topic_input, duration_input],  
    outputs=[keyword_output, urdu_output, final_edited_urdu_output, status_output]    )

    


app.launch()