Update app.py

app.py

@@ -1,163 +1,89 @@
-# Import necessary libraries
 from kokoro import KPipeline
-
 import soundfile as sf
 import torch
-
-import soundfile as sf
 import os
-from moviepy.editor import VideoFileClip, AudioFileClip, ImageClip
+from moviepy.editor import VideoFileClip, AudioFileClip, ImageClip, concatenate_videoclips, CompositeVideoClip, TextClip
+import moviepy.video.fx.all as vfx
+import moviepy.config as mpy_config
 from PIL import Image
 import tempfile
 import random
 import cv2
 import math
-import os, requests, io, time, re, random
-from moviepy.editor import (
-    VideoFileClip, concatenate_videoclips, AudioFileClip, ImageClip,
-    CompositeVideoClip, TextClip
-)
-import moviepy.video.fx.all as vfx
-import moviepy.config as mpy_config
+import requests
+import re
+import time
 from pydub import AudioSegment
-from pydub.generators import Sine
-
-from PIL import Image, ImageDraw, ImageFont
 import numpy as np
 from bs4 import BeautifulSoup
-import base64
 from urllib.parse import quote
-import pysrt
-from gtts import gTTS
-import gradio as gr  # Import Gradio
+import gradio as gr
+import shutil
 
-# Initialize Kokoro TTS pipeline (using American English)
-pipeline = KPipeline(lang_code='a')  # Use voice 'af_heart' for American English
-# Ensure ImageMagick binary is set
+# Initialize Kokoro TTS pipeline (American English)
+pipeline = KPipeline(lang_code='a')
 mpy_config.change_settings({"IMAGEMAGICK_BINARY": "/usr/bin/convert"})
 
-# ---------------- Global Configuration ---------------- #
+# Global Configuration
 PEXELS_API_KEY = 'BhJqbcdm9Vi90KqzXKAhnEHGsuFNv4irXuOjWtT761U49lRzo03qBGna'
 OPENROUTER_API_KEY = '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'
 OPENROUTER_MODEL = "mistralai/mistral-small-3.1-24b-instruct:free"
 OUTPUT_VIDEO_FILENAME = "final_video.mp4"
 USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
-
-# ---------------- Helper Functions ---------------- #
-# (Your existing helper functions remain unchanged: generate_script, parse_script,
-# search_pexels_videos, search_pexels_images, search_google_images, download_image,
-# download_video, generate_media, generate_tts, apply_kenburns_effect,
-# resize_to_fill, find_mp3_files, add_background_music, create_clip,
-# fix_imagemagick_policy)
-
-# Define these globally as they were in your original code but will be set per run
 TARGET_RESOLUTION = None
 CAPTION_COLOR = None
 TEMP_FOLDER = None
 
+# Helper Functions (unchanged from original, included for completeness)
 def generate_script(user_input):
-    """Generate documentary script with proper OpenRouter handling."""
     headers = {
         'Authorization': f'Bearer {OPENROUTER_API_KEY}',
         'HTTP-Referer': 'https://your-domain.com',
         'X-Title': 'AI Documentary Maker'
     }
-
     prompt = f"""Short Documentary Script GeneratorInstructions:
-
 If I say "use this," just output the script exactly as I gave it.
 If I only give topics, generate a script based on them.
 If I provide a full script, rewrite it without any changes. Make everything short simple and humarous funny and act as serious but humarous. And don't say anything off topic. Also alway say a funny statement to subscribe based on the video topic at the end. Use normal conversational text like a normal person talking and avoid AI phase make the statements humanize and normal conversational
 And always talk negative views don't really wants any positive statement and the full script will be one video about one topic so keep in mind
 Formatting Rules:
-
-
 Title in Square Brackets:
-
-
 Each section starts with a one-word title inside [ ] (max two words if necessary).
 This title will be used as a search term for Pexels footage.
-
-
-
 Casual & Funny Narration:
-
-
 Each section has 5-10 words of narration.
 Keep it natural, funny, and unpredictable (not robotic, poetic, or rhythmic).
-
-
-
 No Special Formatting:
-
-
 No bold, italics, or special characters. You are a assistant AI your task is to create script. You aren't a chatbot. So, don't write extra text
-
-
-
 Generalized Search Terms:
-
-
 If a term is too specific, make it more general for Pexels search.
-
-
-
 Scene-Specific Writing:
-
-
 Each section describes only what should be shown in the video.
-
-
-
 Output Only the Script, and also make it funny and humarous and helirous and also add to subscribe with a funny statement like subscribe now or .....
-
-
 No extra text, just the script.
-
-
-
 Example Output:
 [North Korea]
-
 Top 5 unknown facts about North Korea.
-
 [Invisibility]
-
 North Korea’s internet speed is so fast… it doesn’t exist.
-
 [Leadership]
-
 Kim Jong-un once won an election with 100% votes… against himself.
-
 [Magic]
-
 North Korea discovered time travel. That’s why their news is always from the past.
-
 [Warning]
-
 Subscribe now, or Kim Jong-un will send you a free one-way ticket… to North Korea.
-
 [Freedom]
-
 North Korean citizens can do anything… as long as it's government-approved.
 Now here is the Topic/scrip: {user_input}
 """
-
     data = {
         'model': OPENROUTER_MODEL,
         'messages': [{'role': 'user', 'content': prompt}],
         'temperature': 0.4,
         'max_tokens': 5000
     }
-
     try:
-        response = requests.post(
-            'https://openrouter.ai/api/v1/chat/completions',
-            headers=headers,
-            json=data,
-            timeout=30
-        )
-
+        response = requests.post('https://openrouter.ai/api/v1/chat/completions', headers=headers, json=data, timeout=30)
         if response.status_code == 200:
             response_data = response.json()
             if 'choices' in response_data and len(response_data['choices']) > 0:
@@ -168,22 +94,14 @@ Now here is the Topic/scrip: {user_input}
         else:
             print(f"API Error {response.status_code}: {response.text}")
             return None
-
     except Exception as e:
         print(f"Request failed: {str(e)}")
         return None
 
 def parse_script(script_text):
-    """
-    Parse the generated script into a list of elements.
-    For each section, create two elements:
-      - A 'media' element using the section title as the visual prompt.
-      - A 'tts' element with the narration text, voice info, and computed duration.
-    """
     sections = {}
     current_title = None
     current_text = ""
-
     try:
         for line in script_text.splitlines():
             line = line.strip()
@@ -197,63 +115,50 @@ def parse_script(script_text):
                     current_text = line[bracket_end+1:].strip()
             elif current_title:
                 current_text += line + " "
-
         if current_title:
             sections[current_title] = current_text.strip()
-
         elements = []
         for title, narration in sections.items():
             if not title or not narration:
                 continue
-
             media_element = {"type": "media", "prompt": title, "effects": "fade-in"}
             words = narration.split()
             duration = max(3, len(words) * 0.5)
             tts_element = {"type": "tts", "text": narration, "voice": "en", "duration": duration}
             elements.append(media_element)
             elements.append(tts_element)
-
         return elements
     except Exception as e:
         print(f"Error parsing script: {e}")
         return []
 
 def search_pexels_videos(query, pexels_api_key):
-    """Search for a video on Pexels by query and return a random HD video."""
     headers = {'Authorization': pexels_api_key}
     base_url = "https://api.pexels.com/videos/search"
     num_pages = 3
     videos_per_page = 15
-
     max_retries = 3
     retry_delay = 1
-
     search_query = query
     all_videos = []
-
     for page in range(1, num_pages + 1):
         for attempt in range(max_retries):
             try:
                 params = {"query": search_query, "per_page": videos_per_page, "page": page}
                 response = requests.get(base_url, headers=headers, params=params, timeout=10)
-
                 if response.status_code == 200:
                     data = response.json()
                     videos = data.get("videos", [])
-
                     if not videos:
                         print(f"No videos found on page {page}.")
                         break
-
                     for video in videos:
                         video_files = video.get("video_files", [])
                         for file in video_files:
                             if file.get("quality") == "hd":
                                 all_videos.append(file.get("link"))
                                 break
-
                     break
-
                 elif response.status_code == 429:
                     print(f"Rate limit hit (attempt {attempt+1}/{max_retries}). Retrying in {retry_delay} seconds...")
                     time.sleep(retry_delay)
@@ -264,18 +169,12 @@ def search_pexels_videos(query, pexels_api_key):
                         print(f"Retrying in {retry_delay} seconds...")
                         time.sleep(retry_delay)
                         retry_delay *= 2
-                    else:
-                        break
-
             except requests.exceptions.RequestException as e:
                 print(f"Request exception: {e}")
                 if attempt < max_retries - 1:
                     print(f"Retrying in {retry_delay} seconds...")
                     time.sleep(retry_delay)
                     retry_delay *= 2
-                else:
-                    break
-
     if all_videos:
         random_video = random.choice(all_videos)
         print(f"Selected random video from {len(all_videos)} HD videos")
@@ -285,18 +184,14 @@ def search_pexels_videos(query, pexels_api_key):
         return None
 
 def search_pexels_images(query, pexels_api_key):
-    """Search for an image on Pexels by query."""
     headers = {'Authorization': pexels_api_key}
     url = "https://api.pexels.com/v1/search"
     params = {"query": query, "per_page": 5, "orientation": "landscape"}
-
     max_retries = 3
     retry_delay = 1
-
     for attempt in range(max_retries):
         try:
             response = requests.get(url, headers=headers, params=params, timeout=10)
-
             if response.status_code == 200:
                 data = response.json()
                 photos = data.get("photos", [])
@@ -307,7 +202,6 @@ def search_pexels_images(query, pexels_api_key):
                 else:
                     print(f"No images found for query: {query}")
                     return None
-
             elif response.status_code == 429:
                 print(f"Rate limit hit (attempt {attempt+1}/{max_retries}). Retrying in {retry_delay} seconds...")
                 time.sleep(retry_delay)
@@ -318,32 +212,27 @@ def search_pexels_images(query, pexels_api_key):
                     print(f"Retrying in {retry_delay} seconds...")
                     time.sleep(retry_delay)
                     retry_delay *= 2
-
         except requests.exceptions.RequestException as e:
             print(f"Request exception: {e}")
             if attempt < max_retries - 1:
                 print(f"Retrying in {retry_delay} seconds...")
                 time.sleep(retry_delay)
                 retry_delay *= 2
-
     print(f"No Pexels images found for query: {query} after all attempts")
     return None
 
 def search_google_images(query):
-    """Search for images on Google Images (for news-related queries)"""
     try:
         search_url = f"https://www.google.com/search?q={quote(query)}&tbm=isch"
         headers = {"User-Agent": USER_AGENT}
         response = requests.get(search_url, headers=headers, timeout=10)
         soup = BeautifulSoup(response.text, "html.parser")
-
         img_tags = soup.find_all("img")
         image_urls = []
         for img in img_tags:
             src = img.get("src", "")
             if src.startswith("http") and "gstatic" not in src:
                 image_urls.append(src)
-
         if image_urls:
             return random.choice(image_urls[:5]) if len(image_urls) >= 5 else image_urls[0]
         else:
@@ -354,19 +243,15 @@ def search_google_images(query):
         return None
 
 def download_image(image_url, filename):
-    """Download an image from a URL to a local file with enhanced error handling."""
     try:
         headers = {"User-Agent": USER_AGENT}
         print(f"Downloading image from: {image_url} to {filename}")
         response = requests.get(image_url, headers=headers, stream=True, timeout=15)
         response.raise_for_status()
-
         with open(filename, 'wb') as f:
             for chunk in response.iter_content(chunk_size=8192):
                 f.write(chunk)
-
         print(f"Image downloaded successfully to: {filename}")
-
         try:
             img = Image.open(filename)
             img.verify()
@@ -381,7 +266,6 @@ def download_image(image_url, filename):
             if os.path.exists(filename):
                 os.remove(filename)
             return None
-
     except requests.exceptions.RequestException as e_download:
         print(f"Image download error: {e_download}")
         if os.path.exists(filename):
@@ -394,7 +278,6 @@ def download_image(image_url, filename):
         return None
 
 def download_video(video_url, filename):
-    """Download a video from a URL to a local file."""
     try:
         response = requests.get(video_url, stream=True, timeout=30)
         response.raise_for_status()
@@ -409,14 +292,8 @@ def download_video(video_url, filename):
             os.remove(filename)
         return None
 
-def generate_media(prompt, user_image=None, current_index=0, total_segments=1):
-    """
-    Generate a visual asset by first searching for a video or using a specific search strategy.
-    For news-related queries, use Google Images.
-    Returns a dict: {'path': <file_path>, 'asset_type': 'video' or 'image'}.
-    """
+def generate_media(prompt, current_index=0, total_segments=1):
     safe_prompt = re.sub(r'[^\w\s-]', '', prompt).strip().replace(' ', '_')
-
     if "news" in prompt.lower():
         print(f"News-related query detected: {prompt}. Using Google Images...")
         image_file = os.path.join(TEMP_FOLDER, f"{safe_prompt}_news.jpg")
@@ -428,7 +305,6 @@ def generate_media(prompt, user_image=None, current_index=0, total_segments=1):
                 return {"path": downloaded_image, "asset_type": "image"}
         else:
             print(f"Google Images search failed for prompt: {prompt}")
-
     if random.random() < 0.45:
         video_file = os.path.join(TEMP_FOLDER, f"{safe_prompt}_video.mp4")
         video_url = search_pexels_videos(prompt, PEXELS_API_KEY)
@@ -439,7 +315,6 @@ def generate_media(prompt, user_image=None, current_index=0, total_segments=1):
                 return {"path": downloaded_video, "asset_type": "video"}
         else:
             print(f"Pexels video search failed for prompt: {prompt}")
-
     image_file = os.path.join(TEMP_FOLDER, f"{safe_prompt}.jpg")
     image_url = search_pexels_images(prompt, PEXELS_API_KEY)
     if image_url:
@@ -449,7 +324,6 @@ def generate_media(prompt, user_image=None, current_index=0, total_segments=1):
             return {"path": downloaded_image, "asset_type": "image"}
         else:
             print(f"Pexels image download failed for prompt: {prompt}")
-
     fallback_terms = ["nature", "people", "landscape", "technology", "business"]
     for term in fallback_terms:
         print(f"Trying fallback image search with term: {term}")
@@ -464,30 +338,15 @@ def generate_media(prompt, user_image=None, current_index=0, total_segments=1):
                 print(f"Fallback image download failed for term: {term}")
         else:
             print(f"Fallback image search failed for term: {term}")
-
     print(f"Failed to generate visual asset for prompt: {prompt}")
     return None
 
-def generate_silent_audio(duration, sample_rate=24000):
-    """Generate a silent WAV audio file lasting 'duration' seconds."""
-    num_samples = int(duration * sample_rate)
-    silence = np.zeros(num_samples, dtype=np.float32)
-    silent_path = os.path.join(TEMP_FOLDER, f"silent_{int(time.time())}.wav")
-    sf.write(silent_path, silence, sample_rate)
-    print(f"Silent audio generated: {silent_path}")
-    return silent_path
-
 def generate_tts(text, voice):
-    """
-    Generate TTS audio using Kokoro, falling back to gTTS or silent audio if needed.
-    """
     safe_text = re.sub(r'[^\w\s-]', '', text[:10]).strip().replace(' ', '_')
     file_path = os.path.join(TEMP_FOLDER, f"tts_{safe_text}.wav")
-
     if os.path.exists(file_path):
         print(f"Using cached TTS for text '{text[:10]}...'")
         return file_path
-
     try:
         kokoro_voice = 'af_heart' if voice == 'en' else voice
         generator = pipeline(text, voice=kokoro_voice, speed=0.9, split_pattern=r'\n+')
@@ -500,46 +359,28 @@ def generate_tts(text, voice):
         return file_path
     except Exception as e:
         print(f"Error with Kokoro TTS: {e}")
-        try:
-            print("Falling back to gTTS...")
-            tts = gTTS(text=text, lang='en')
-            mp3_path = os.path.join(TEMP_FOLDER, f"tts_{safe_text}.mp3")
-            tts.save(mp3_path)
-            audio = AudioSegment.from_mp3(mp3_path)
-            audio.export(file_path, format="wav")
-            os.remove(mp3_path)
-            print(f"Fallback TTS saved to {file_path} (gTTS)")
-            return file_path
-        except Exception as fallback_error:
-            print(f"Both TTS methods failed: {fallback_error}")
-            return generate_silent_audio(duration=max(3, len(text.split()) * 0.5))
+        return None
 
 def apply_kenburns_effect(clip, target_resolution, effect_type=None):
-    """Apply a smooth Ken Burns effect with a single movement pattern."""
     target_w, target_h = target_resolution
     clip_aspect = clip.w / clip.h
     target_aspect = target_w / target_h
-
     if clip_aspect > target_aspect:
         new_height = target_h
         new_width = int(new_height * clip_aspect)
     else:
         new_width = target_w
         new_height = int(new_width / clip_aspect)
-
     clip = clip.resize(newsize=(new_width, new_height))
     base_scale = 1.15
     new_width = int(new_width * base_scale)
     new_height = int(new_height * base_scale)
     clip = clip.resize(newsize=(new_width, new_height))
-
     max_offset_x = new_width - target_w
     max_offset_y = new_height - target_h
-
     available_effects = ["zoom-in", "zoom-out", "pan-left", "pan-right", "up-left"]
     if effect_type is None or effect_type == "random":
         effect_type = random.choice(available_effects)
-
     if effect_type == "zoom-in":
         start_zoom = 0.9
         end_zoom = 1.1
@@ -567,7 +408,6 @@ def apply_kenburns_effect(clip, target_resolution, effect_type=None):
         end_center = (target_w / 2, target_h / 2)
     else:
         raise ValueError(f"Unsupported effect_type: {effect_type}")
-
     def transform_frame(get_frame, t):
         frame = get_frame(t)
         ratio = t / clip.duration if clip.duration > 0 else 0
@@ -586,15 +426,12 @@ def apply_kenburns_effect(clip, target_resolution, effect_type=None):
         cropped_frame = cv2.getRectSubPix(frame, (crop_w, crop_h), (current_center_x, current_center_y))
         resized_frame = cv2.resize(cropped_frame, (target_w, target_h), interpolation=cv2.INTER_LANCZOS4)
         return resized_frame
-
     return clip.fl(transform_frame)
 
 def resize_to_fill(clip, target_resolution):
-    """Resize and crop a clip to fill the target resolution while maintaining aspect ratio."""
     target_w, target_h = target_resolution
     clip_aspect = clip.w / clip.h
     target_aspect = target_w / target_h
-
     if clip_aspect > target_aspect:
         clip = clip.resize(height=target_h)
         crop_amount = (clip.w - target_w) / 2
@@ -603,57 +440,35 @@ def resize_to_fill(clip, target_resolution):
         clip = clip.resize(width=target_w)
         crop_amount = (clip.h - target_h) / 2
         clip = clip.crop(x1=0, x2=clip.w, y1=crop_amount, y2=clip.h - crop_amount)
-
     return clip
 
-def find_mp3_files():
-    """Search for any MP3 files in the current directory and subdirectories."""
-    mp3_files = []
-    for root, dirs, files in os.walk('.'):
-        for file in files:
-            if file.endswith('.mp3'):
-                mp3_path = os.path.join(root, file)
-                mp3_files.append(mp3_path)
-                print(f"Found MP3 file: {mp3_path}")
-    return mp3_files[0] if mp3_files else None
-
-def add_background_music(final_video, bg_music_volume=0.08):
-    """Add background music to the final video using any MP3 file found."""
-    try:
-        bg_music_path = find_mp3_files()
-        if bg_music_path and os.path.exists(bg_music_path):
-            print(f"Adding background music from: {bg_music_path}")
-            bg_music = AudioFileClip(bg_music_path)
-            if bg_music.duration < final_video.duration:
-                loops_needed = math.ceil(final_video.duration / bg_music.duration)
-                bg_segments = [bg_music] * loops_needed
-                bg_music = concatenate_audioclips(bg_segments)
-            bg_music = bg_music.subclip(0, final_video.duration)
-            bg_music = bg_music.volumex(bg_music_volume)
-            video_audio = final_video.audio
-            mixed_audio = CompositeAudioClip([video_audio, bg_music])
-            final_video = final_video.set_audio(mixed_audio)
-            print("Background music added successfully")
-        else:
-            print("No MP3 files found, skipping background music")
-        return final_video
-    except Exception as e:
-        print(f"Error adding background music: {e}")
-        print("Continuing without background music")
-        return final_video
+def add_background_music(final_video, music_file, bg_music_volume=0.08):
+    if music_file and os.path.exists(music_file):
+        print(f"Adding background music from: {music_file}")
+        bg_music = AudioFileClip(music_file)
+        if bg_music.duration < final_video.duration:
+            loops_needed = math.ceil(final_video.duration / bg_music.duration)
+            bg_segments = [bg_music] * loops_needed
+            bg_music = concatenate_audioclips(bg_segments)
+        bg_music = bg_music.subclip(0, final_video.duration)
+        bg_music = bg_music.volumex(bg_music_volume)
+        video_audio = final_video.audio
+        mixed_audio = CompositeAudioClip([video_audio, bg_music])
+        final_video = final_video.set_audio(mixed_audio)
+        print("Background music added successfully")
+    else:
+        print("No music file provided, skipping background music")
+    return final_video
 
 def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, narration_text=None, segment_index=0):
-    """Create a video clip with synchronized subtitles and narration."""
     try:
         print(f"Creating clip #{segment_index} with asset_type: {asset_type}, media_path: {media_path}")
         if not os.path.exists(media_path) or not os.path.exists(tts_path):
             print("Missing media or TTS file")
             return None
-
         audio_clip = AudioFileClip(tts_path).audio_fadeout(0.2)
         audio_duration = audio_clip.duration
         target_duration = audio_duration + 0.2
-
         if asset_type == "video":
             clip = VideoFileClip(media_path)
             clip = resize_to_fill(clip, TARGET_RESOLUTION)
@@ -673,7 +488,6 @@ def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, n
             clip = clip.fadein(0.3).fadeout(0.3)
         else:
             return None
-
         if narration_text and CAPTION_COLOR != "transparent":
             try:
                 words = narration_text.split()
@@ -686,11 +500,9 @@ def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, n
                         current_chunk = []
                 if current_chunk:
                     chunks.append(' '.join(current_chunk))
-
                 chunk_duration = audio_duration / len(chunks)
                 subtitle_clips = []
                 subtitle_y_position = int(TARGET_RESOLUTION[1] * 0.70)
-
                 for i, chunk_text in enumerate(chunks):
                     start_time = i * chunk_duration
                     end_time = (i + 1) * chunk_duration
@@ -708,7 +520,6 @@ def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, n
                     ).set_start(start_time).set_end(end_time)
                     txt_clip = txt_clip.set_position(('center', subtitle_y_position))
                     subtitle_clips.append(txt_clip)
-
                 clip = CompositeVideoClip([clip] + subtitle_clips)
             except Exception as sub_error:
                 print(f"Subtitle error: {sub_error}")
@@ -720,7 +531,6 @@ def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, n
                     size=(TARGET_RESOLUTION[0] * 0.7, None)
                 ).set_position(('center', int(TARGET_RESOLUTION[1] / 3))).set_duration(clip.duration)
                 clip = CompositeVideoClip([clip, txt_clip])
-
         clip = clip.set_audio(audio_clip)
         print(f"Clip created: {clip.duration:.1f}s")
         return clip
@@ -729,7 +539,6 @@ def create_clip(media_path, asset_type, tts_path, duration=None, effects=None, n
         return None
 
 def fix_imagemagick_policy():
-    """Fix ImageMagick security policies."""
     try:
         print("Attempting to fix ImageMagick security policies...")
         policy_paths = [
@@ -753,55 +562,33 @@ def fix_imagemagick_policy():
         print(f"Error fixing policies: {e}")
         return False
 
-# ---------------- Main Function with Gradio Integration ---------------- #
-def generate_video(user_input, resolution, caption_option):
-    """Generate a video based on user input via Gradio."""
+# Main Video Generation Function with Script Input
+def generate_video_from_script(script, resolution, caption_option, music_file):
     global TARGET_RESOLUTION, CAPTION_COLOR, TEMP_FOLDER
-    import shutil
-
-    # Set resolution
     if resolution == "Full":
         TARGET_RESOLUTION = (1920, 1080)
     elif resolution == "Short":
         TARGET_RESOLUTION = (1080, 1920)
     else:
-        TARGET_RESOLUTION = (1920, 1080)  # Default
-
-    # Set caption color
+        TARGET_RESOLUTION = (1920, 1080)
     CAPTION_COLOR = "white" if caption_option == "Yes" else "transparent"
-
-    # Create a unique temporary folder
     TEMP_FOLDER = tempfile.mkdtemp()
-
-    # Fix ImageMagick policy
     fix_success = fix_imagemagick_policy()
     if not fix_success:
         print("Will use alternative methods if needed")
-
-    print("Generating script from API...")
-    script = generate_script(user_input)
-    if not script:
-        print("Failed to generate script.")
-        shutil.rmtree(TEMP_FOLDER)
-        return None
-    print("Generated Script:\n", script)
     elements = parse_script(script)
     if not elements:
         print("Failed to parse script into elements.")
         shutil.rmtree(TEMP_FOLDER)
         return None
-    print(f"Parsed {len(elements)//2} script segments.")
-
     paired_elements = []
     for i in range(0, len(elements), 2):
         if i + 1 < len(elements):
             paired_elements.append((elements[i], elements[i + 1]))
-
     if not paired_elements:
         print("No valid script segments found.")
         shutil.rmtree(TEMP_FOLDER)
         return None
-
     clips = []
     for idx, (media_elem, tts_elem) in enumerate(paired_elements):
         print(f"\nProcessing segment {idx+1}/{len(paired_elements)} with prompt: '{media_elem['prompt']}'")
@@ -826,40 +613,48 @@ def generate_video(user_input, resolution, caption_option):
             clips.append(clip)
         else:
             print(f"Clip creation failed for segment {idx+1}.")
-
     if not clips:
         print("No clips were successfully created.")
         shutil.rmtree(TEMP_FOLDER)
         return None
-
     print("\nConcatenating clips...")
     final_video = concatenate_videoclips(clips, method="compose")
-    final_video = add_background_music(final_video, bg_music_volume=0.08)
-
+    final_video = add_background_music(final_video, music_file, bg_music_volume=0.08)
     print(f"Exporting final video to {OUTPUT_VIDEO_FILENAME}...")
     final_video.write_videofile(OUTPUT_VIDEO_FILENAME, codec='libx264', fps=60, preset='veryfast')
     print(f"Final video saved as {OUTPUT_VIDEO_FILENAME}")
-
-    # Clean up
-    print("Cleaning up temporary files...")
     shutil.rmtree(TEMP_FOLDER)
     print("Temporary files removed.")
-
     return OUTPUT_VIDEO_FILENAME
 
-# ---------------- Gradio Interface ---------------- #
-iface = gr.Interface(
-    fn=generate_video,
-    inputs=[
-        gr.Textbox(label="Video Concept", placeholder="Enter your video concept here..."),
-        gr.Radio(["Full", "Short"], label="Resolution", value="Short"),
-        gr.Radio(["Yes", "No"], label="Captions", value="No")
-    ],
-    outputs=gr.Video(label="Generated Video"),
-    title="AI Documentary Video Generator",
-    description="Create a funny documentary-style video based on your concept. Note: Generation may take several minutes on CPU."
-)
-
-# Launch the interface
-iface.launch(share=True)
+# Gradio Blocks Interface
+with gr.Blocks(title="AI Documentary Video Generator") as demo:
+    gr.Markdown("### Create a Funny Documentary Video")
+    gr.Markdown("Enter a concept, generate a script, edit it, upload music, and generate your video!")
+    gr.Markdown("**Note**: Generation may take several minutes on CPU.")
+    
+    concept = gr.Textbox(label="Video Concept", placeholder="Enter your video concept here...")
+    resolution = gr.Radio(["Full", "Short"], label="Resolution", value="Short")
+    captions = gr.Radio(["Yes", "No"], label="Captions", value="No")
+    script = gr.Textbox(label="Script", lines=10, placeholder="Generated script will appear here...")
+    music = gr.Audio(label="Background Music (Optional)", type="filepath")
+    
+    generate_script_btn = gr.Button("Generate Script")
+    generate_video_btn = gr.Button("Generate Video")
+    
+    video_output = gr.Video(label="Generated Video")
+    
+    def on_generate_script(concept):
+        script_text = generate_script(concept)
+        return script_text if script_text else "Failed to generate script."
+    
+    def on_generate_video(script, resolution, captions, music):
+        if not script or "Failed" in script:
+            return None
+        video_path = generate_video_from_script(script, resolution, captions, music)
+        return video_path
+    
+    generate_script_btn.click(on_generate_script, inputs=concept, outputs=script)
+    generate_video_btn.click(on_generate_video, inputs=[script, resolution, captions, music], outputs=video_output)
 
+demo.launch(share=True)