import gradio as gr import subprocess import os from tqdm import tqdm import time from PIL import Image from concurrent.futures import ThreadPoolExecutor import torch def get_device(): if torch.cuda.is_available(): return torch.device('cuda') else: return torch.device('cpu') device = get_device() print(f"Using device: {device}") def get_unique_filename(base_name, extension): counter = 1 if not base_name.endswith(extension): unique_name = f"{base_name}{extension}" else: unique_name = base_name while os.path.exists(unique_name): unique_name = f"{base_name}_{counter}{extension}" counter += 1 return unique_name def get_temp_task_path(input_video_path, task_name): # Extract the base name of the input video file base_name = os.path.basename(input_video_path) print(f"Base name: {base_name}") # Debug print # Split the base name into words and trim the first 4 words words = base_name.split() if len(words) > 4: trimmed_name = '_'.join(words[4:]) else: trimmed_name = base_name # If there are less than 4 words, use the base name as is print(f"Trimmed name: {trimmed_name}") # Debug print # Create the new folder name folder_name = f"{task_name}_{trimmed_name}" print(f"Folder name: {folder_name}") # Debug print # Get the directory of the input video file input_dir = os.path.dirname(input_video_path) print(f"Input directory: {input_dir}") # Debug print # Create the full path for the temporary task folder temp_task_path = os.path.join(input_dir, folder_name) print(f"Temporary task path: {temp_task_path}") # Debug print # Create the directory if it doesn't exist os.makedirs(temp_task_path, exist_ok=True) print(f"Directory created: {temp_task_path}") # Debug print return temp_task_path def combine_videos(video_files, task_name): if not video_files: return "No video files selected.", None # Assuming the first video file is representative for the path input_dir = os.path.dirname(video_files[0]) # Use the input directory as the output directory output_file = os.path.join(input_dir, get_unique_filename("combined_video", ".mp4")) filelist_path = os.path.join(input_dir, "filelist.txt") with open(filelist_path, "w") as filelist: for video in video_files: video_path = os.path.abspath(video).replace('\\', '/') filelist.write(f"file '{video_path}'\n") command = [ "ffmpeg", "-f", "concat", "-safe", "0", "-i", filelist_path, "-c", "copy", output_file ] with ThreadPoolExecutor() as executor: future = executor.submit(subprocess.run, command, text=True, capture_output=True) result = future.result() # Waits for the command to complete and returns the result if result.returncode == 0: return f"Videos combined successfully into {output_file}", output_file else: return f"Error combining videos: {result.stderr}", None def combine_audios(audio_files): if not audio_files: return "No audio files selected.", None output_file = get_unique_filename("combined_audio", ".mp3") filelist_path = os.path.abspath("filelist.txt") with open(filelist_path, "w") as filelist: for audio in audio_files: filelist.write(f"file '{os.path.abspath(audio).replace('\\', '/')}'\n") command = [ "ffmpeg", "-f", "concat", "-safe", "0", "-i", filelist_path, "-c", "copy", output_file ] with ThreadPoolExecutor() as executor: future = executor.submit(subprocess.run, command, text=True, capture_output=True) result = future.result() # Waits for the command to complete and returns the result if result.returncode == 0: return f"Audios combined successfully into {output_file}", output_file else: return f"Error combining audios: {result.stderr}", None def combine_images(image_files): if not image_files: return "No image files selected.", None output_file = get_unique_filename("combined_image", ".mp4") command = ["convert"] + image_files + [output_file] process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True) total_time = 0 with tqdm(total=100, desc="Combining Images") as pbar: while True: output = process.stderr.readline() if output == '' and process.poll() is not None: break if output: total_time += 1 pbar.update(1) time.sleep(0.1) return f"Images combined successfully into {output_file}", output_file def split_video(input_file, chunk_size): input_dir = os.path.dirname(input_file) base_name = os.path.splitext(os.path.basename(input_file))[0] chunk_files = [] command = [ "ffmpeg", "-i", input_file, "-c", "copy", "-map", "0", "-segment_time", str(chunk_size), "-f", "segment", os.path.join(input_dir, f"{base_name}_chunk_%03d.mp4") ] subprocess.run(command, check=True) for file in os.listdir(input_dir): if file.startswith(base_name) and file.endswith(".mp4") and "chunk" in file: chunk_files.append(os.path.join(input_dir, file)) return chunk_files def adjust_speed_chunked(media_file, speed, chunk_size=60): if not media_file: return "No media file selected.", None chunk_files = split_video(media_file, chunk_size) processed_files = [] def process_chunk(chunk_file): output_file = get_unique_filename(f"adjusted_speed_{os.path.basename(chunk_file)}", ".mp4") codec = get_gpu_codec() or "libx264" command = [ "ffmpeg", "-hwaccel", "auto", "-i", chunk_file, "-filter:v", f"setpts={1/speed}*PTS", "-filter:a", f"atempo={speed}", "-c:v", codec, "-preset", "fast", output_file ] subprocess.run(command, check=True) return output_file with ThreadPoolExecutor() as executor: futures = [executor.submit(process_chunk, chunk) for chunk in chunk_files] for future in futures: processed_files.append(future.result()) combined_output_file = os.path.join(os.path.dirname(media_file), get_unique_filename("adjusted_speed_combined", ".mp4")) with open("filelist.txt", "w") as filelist: for processed_file in processed_files: filelist.write(f"file '{processed_file}'\n") command = [ "ffmpeg", "-f", "concat", "-safe", "0", "-i", "filelist.txt", "-c", "copy", combined_output_file ] subprocess.run(command, check=True) return f"Speed adjusted successfully to {speed}x in {combined_output_file}", combined_output_file def adjust_speed(media_file, speed): if not media_file: return "No media file selected.", None output_file = get_unique_filename(f"adjusted_speed_{os.path.basename(media_file)}", ".mp4") # Determine the codec based on available hardware codec = get_gpu_codec() if not codec: codec = "libx264" # Fallback to a widely supported software codec command = [ "ffmpeg", "-hwaccel", "auto", "-i", media_file, "-filter:v", f"setpts={1/speed}*PTS", "-filter:a", f"atempo={speed}", "-c:v", codec, "-preset", "fast", output_file ] process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True) total_time = 0 with tqdm(total=100, desc="Adjusting Speed") as pbar: while True: output = process.stderr.readline() if output == '' and process.poll() is not None: break if output: total_time += 1 pbar.update(1) time.sleep(0.1) return f"Speed adjusted successfully to {speed}x in {output_file}", output_file def adjust_speed_by_length(media_file, desired_length_hhmmss): if not media_file: return "No media file selected.", None original_length = get_media_length(media_file) desired_length = hhmmss_to_seconds(desired_length_hhmmss) speed = original_length / desired_length return adjust_speed(media_file, speed) def adjust_speed_combined(media_file, speed, hours, minutes, seconds, adjust_speed_checkbox, compress_checkbox): if not adjust_speed_checkbox and not compress_checkbox: return "Error: At least one of 'Adjust Speed' or 'Compress Video' must be checked.", None output_file = media_file output_message = "" if adjust_speed_checkbox: if hours or minutes or seconds: desired_length = f"{int(hours):02}:{int(minutes):02}:{int(seconds):02}" output_message, output_file = adjust_speed_by_length(media_file, desired_length) else: output_message, output_file = adjust_speed_chunked(media_file, speed) if compress_checkbox and output_file: compressed_output_file = get_unique_filename(f"compressed_{os.path.basename(output_file)}", ".mp4") compress_message, output_file = compress_video(output_file, compressed_output_file) output_message = f"{output_message} and {compress_message}" return output_message, output_file def hhmmss_to_seconds(hhmmss): h, m, s = map(int, hhmmss.split(':')) return h * 3600 + m * 60 + s def get_media_length(media_file): result = subprocess.run( ["ffprobe", "-v", "error", "-show_entries", "format=duration", "-of", "default=noprint_wrappers=1:nokey=1", media_file], stdout=subprocess.PIPE, stderr=subprocess.STDOUT ) return float(result.stdout) def auto_color_correct(video_file): if not video_file: return "No video file selected.", None input_dir = os.path.dirname(video_file) output_file = os.path.join(input_dir, get_unique_filename(f"color_corrected_{os.path.basename(video_file)}", "")) command = [ "ffmpeg", "-i", video_file, "-vf", "eq=brightness=0.06:saturation=1.5", output_file ] subprocess.run(command, check=True) return f"Auto color correction applied successfully to {output_file}", output_file def extract_audio(video_file): if not video_file: return "No video file selected.", None output_file = get_unique_filename(f"extracted_audio_{os.path.basename(video_file)}", ".mka") command = [ "ffmpeg", "-i", video_file, "-vn", "-acodec", "copy", output_file ] try: subprocess.run(command, check=True) except subprocess.CalledProcessError as e: return f"Error extracting audio: {e}. Command: {' '.join(command)}", None return f"Audio extracted successfully into {output_file}", output_file def compress_video(input_file, output_file): codec = get_gpu_codec() if not codec: print("No supported GPU codec found. Falling back to software encoding.") codec = "libx264" # Fallback to a widely supported software codec # Adjust these parameters based on your specific needs bitrate = "1M" # Lower bitrate for smaller file size maxrate = "1.2M" # Max bitrate to limit peaks in bitrate bufsize = "2M" # Buffer size for bitrate control preset = "fast" # Faster preset for quicker compression command = [ "ffmpeg", "-hwaccel", "auto", "-i", input_file, "-c:v", codec, "-preset", preset, "-b:v", bitrate, "-maxrate", maxrate, "-bufsize", bufsize, "-profile:v", "high", "-level", "4.1", output_file ] try: subprocess.run(command, check=True) return f"Video compressed successfully into {output_file}", output_file except subprocess.CalledProcessError as e: return f"Error compressing video: {e}", None def add_watermark(video_file, watermark_type, watermark_text, watermark_image, opacity, position_x, position_y, font_size, font_color): if not video_file: return "No video file selected.", None output_file = get_unique_filename(f"watermarked_{os.path.basename(video_file)}", ".mp4") drawtext = f"drawtext=text='{watermark_text}':x={position_x}:y={position_y}:fontsize={font_size}:fontcolor={font_color}@{opacity}" if watermark_type == "text" else "" overlay = f"overlay={position_x}:{position_y}" if watermark_type == "image" else "" command = [ "ffmpeg", "-i", video_file, "-vf", f"{drawtext if watermark_type == 'text' else ''}{overlay if watermark_type == 'image' else ''}", output_file ] process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True) total_time = 0 with tqdm(total=100, desc="Adding Watermark") as pbar: while True: output = process.stderr.readline() if output == '' and process.poll() is not None: break if output: total_time += 1 pbar.update(1) time.sleep(0.1) return f"Watermark added successfully to {output_file}", output_file def compress_image(input_image): if not input_image: return "No image file selected.", None output_file = get_unique_filename(f"compressed_{os.path.basename(input_image)}", ".jpg") with Image.open(input_image) as img: img.save(output_file, "JPEG", quality=95) # Adjust quality as needed return f"Image compressed successfully into {output_file}", output_file def compress_image_lossless(input_image): if not input_image: return "No image file selected.", None output_file = get_unique_filename(f"compressed_lossless_{os.path.basename(input_image)}", ".png") with Image.open(input_image) as img: img.save(output_file, "PNG", optimize=True) # PNG is lossless return f"Image compressed losslessly into {output_file}", output_file def mp3_to_video(mp3_file, image_file): if not mp3_file or not image_file: return "MP3 file or image file not selected.", None output_file = get_unique_filename(f"{os.path.splitext(os.path.basename(mp3_file))[0]}", ".mp4") command = [ "ffmpeg", "-loop", "1", "-i", image_file, "-i", mp3_file, "-c:v", "libx264", "-c:a", "aac", "-b:a", "192k", "-shortest", output_file ] subprocess.run(command, check=True) return f"MP3 converted to video successfully into {output_file}", output_file def video_to_mp3(video_file): if not video_file: return "No video file selected.", None output_file = get_unique_filename(f"{os.path.splitext(os.path.basename(video_file))[0]}", ".mp3") command = [ "ffmpeg", "-i", video_file, "-q:a", "0", "-map", "a", output_file ] subprocess.run(command, check=True) return f"Video converted to MP3 successfully into {output_file}", output_file def convert_image_format(input_image, output_format): if not input_image: return "No image file selected.", None output_file = get_unique_filename(f"{os.path.splitext(os.path.basename(input_image))[0]}", f".{output_format}") with Image.open(input_image) as img: img.save(output_file, output_format.upper()) return f"Image converted to {output_format} format successfully into {output_file}", output_file def get_gpu_codec(): # Prioritize Intel QSV codecs for Intel GPUs codecs = ["h264_qsv", "hevc_qsv", "mpeg2_qsv"] # Use the -encoders flag to get a list of all available encoders result = subprocess.run(["ffmpeg", "-hide_banner", "-encoders"], capture_output=True, text=True) available_encoders = result.stdout print("Available encoders:", available_encoders) # Debug: print available encoders for codec in codecs: # Check if each codec is in the list of available encoders if codec in available_encoders: print(f"Using GPU codec: {codec}") # Debug: confirm which codec is used return codec print("No supported GPU codec found.") return None def interface(): with gr.Blocks() as demo: gr.Markdown("### Media Combiner Tool") with gr.Tab("Combine Videos"): with gr.Row(): video_files = gr.File(label="Select Video Files", type="filepath", file_count="multiple") task_name = gr.Textbox(label="Task Name") video_submit = gr.Button("Combine Videos") video_output = gr.Textbox(label="Output") video_download = gr.File(label="Download Combined Video") video_submit.click(combine_videos, inputs=[video_files, task_name], outputs=[video_output, video_download]) with gr.Tab("Combine Audios"): with gr.Row(): audio_files = gr.File(label="Select Audio Files", type="filepath", file_count="multiple") audio_submit = gr.Button("Combine Audios") audio_output = gr.Textbox(label="Output") audio_download = gr.File(label="Download Combined Audio") audio_submit.click(combine_audios, inputs=[audio_files], outputs=[audio_output, audio_download]) with gr.Tab("Combine Images"): with gr.Row(): image_files = gr.File(label="Select Image Files", type="filepath", file_count="multiple") image_submit = gr.Button("Combine Images") image_output = gr.Textbox(label="Output") image_download = gr.File(label="Download Combined Image") image_submit.click(combine_images, inputs=[image_files], outputs=[image_output, image_download]) with gr.Tab("Adjust Speed"): with gr.Row(): media_file = gr.File(label="Select Media File", type="filepath") speed = gr.Slider(label="Speed", minimum=0.5, maximum=2.0, step=0.1, value=1.0) with gr.Row(): hours = gr.Number(label="Hours", value=0, precision=0) minutes = gr.Number(label="Minutes", value=0, precision=0) seconds = gr.Number(label="Seconds", value=0, precision=0) with gr.Row(): adjust_speed_checkbox = gr.Checkbox(label="Adjust Speed", value=True) compress_checkbox = gr.Checkbox(label="Compress Video", value=False) speed_submit = gr.Button("Submit") speed_output = gr.Textbox(label="Output") speed_download = gr.File(label="Download Adjusted Media") speed_submit.click(adjust_speed_combined, inputs=[media_file, speed, hours, minutes, seconds, adjust_speed_checkbox, compress_checkbox], outputs=[speed_output, speed_download]) with gr.Tab("Auto Color Correction"): with gr.Row(): video_file = gr.File(label="Select Video File", type="filepath") color_submit = gr.Button("Apply Color Correction") color_output = gr.Textbox(label="Output") color_download = gr.File(label="Download Color Corrected Video") color_submit.click(auto_color_correct, inputs=[video_file], outputs=[color_output, color_download]) with gr.Tab("Extract Audio"): with gr.Row(): video_file = gr.File(label="Select Video File", type="filepath") extract_submit = gr.Button("Extract Audio") extract_output = gr.Textbox(label="Output") extract_download = gr.File(label="Download Extracted Audio") extract_submit.click(extract_audio, inputs=[video_file], outputs=[extract_output, extract_download]) with gr.Tab("Add Watermark"): with gr.Row(): video_file = gr.File(label="Select Video File", type="filepath") watermark_type = gr.Radio(label="Watermark Type", choices=["text", "image"], value="text") with gr.Row(): watermark_text = gr.Textbox(label="Watermark Text", visible=True) watermark_image = gr.File(label="Watermark Image", type="filepath", visible=False) with gr.Row(): opacity = gr.Slider(label="Opacity", minimum=0.0, maximum=1.0, step=0.1, value=1.0) position_x = gr.Slider(label="Position X", minimum=0, maximum=1920, step=1, value=0) position_y = gr.Slider(label="Position Y", minimum=0, maximum=1080, step=1, value=0) with gr.Row(): font_size = gr.Slider(label="Font Size", minimum=10, maximum=100, step=1, value=24, visible=True) font_color = gr.ColorPicker(label="Font Color", value="#FFFFFF", visible=True) watermark_submit = gr.Button("Add Watermark") watermark_output = gr.Textbox(label="Output") watermark_download = gr.File(label="Download Watermarked Video") def update_visibility(watermark_type): return { watermark_text: gr.update(visible=watermark_type == "text"), watermark_image: gr.update(visible=watermark_type == "image"), font_size: gr.update(visible=watermark_type == "text"), font_color: gr.update(visible=watermark_type == "text") } watermark_type.change(update_visibility, inputs=[watermark_type], outputs=[watermark_text, watermark_image, font_size, font_color]) watermark_submit.click(add_watermark, inputs=[video_file, watermark_type, watermark_text, watermark_image, opacity, position_x, position_y, font_size, font_color], outputs=[watermark_output, watermark_download]) with gr.Tab("Compress Image Losslessly"): with gr.Row(): image_file = gr.File(label="Select Image File", type="filepath") compress_submit = gr.Button("Compress Image Losslessly") compress_output = gr.Textbox(label="Output") compress_download = gr.File(label="Download Compressed Image") compress_submit.click(compress_image_lossless, inputs=[image_file], outputs=[compress_output, compress_download]) with gr.Tab("Convert MP3 to Video"): with gr.Row(): mp3_file = gr.File(label="Select MP3 File", type="filepath") image_file = gr.File(label="Select Image File", type="filepath") convert_submit = gr.Button("Convert MP3 to Video") convert_output = gr.Textbox(label="Output") convert_download = gr.File(label="Download Converted Video") convert_submit.click(mp3_to_video, inputs=[mp3_file, image_file], outputs=[convert_output, convert_download]) with gr.Tab("Convert Video to MP3"): with gr.Row(): video_file = gr.File(label="Select Video File", type="filepath") convert_submit = gr.Button("Convert Video to MP3") convert_output = gr.Textbox(label="Output") convert_download = gr.File(label="Download Converted MP3") convert_submit.click(video_to_mp3, inputs=[video_file], outputs=[convert_output, convert_download]) return demo if __name__ == "__main__": demo = interface() demo.launch(share=True)