app.py

import gradio as gr
from youtube_transcript_api import YouTubeTranscriptApi, TranscriptsDisabled, NoTranscriptFound
from youtube_transcript_api.formatters import TextFormatter, JSONFormatter, WebVTTFormatter, SRTFormatter
import json

def get_transcript(video_id, languages, format_type, translate_to, preserve_formatting):
    try:
        # Set default languages if none provided
        if not languages:
            languages = ['en']
        else:
            languages = languages.split(',')

        # Get transcript
        transcript = YouTubeTranscriptApi.get_transcript(
            video_id,
            languages=languages,
            preserve_formatting=preserve_formatting
        )

        # Translate if requested
        if translate_to:
            transcript_list = YouTubeTranscriptApi.list_transcripts(video_id)
            base_transcript = transcript_list.find_transcript(languages)
            transcript = base_transcript.translate(translate_to).fetch()

        # Format transcript based on selected format
        formatter_map = {
            "Text": TextFormatter(),
            "JSON": JSONFormatter(),
            "WebVTT": WebVTTFormatter(),
            "SRT": SRTFormatter()
        }
        
        formatter = formatter_map[format_type]
        formatted_transcript = formatter.format_transcript(transcript)

        return formatted_transcript

    except TranscriptsDisabled:
        return "Error: Transcripts are disabled for this video"
    except NoTranscriptFound:
        return "Error: No transcript found for the specified languages"
    except Exception as e:
        return f"Unexpected error: {str(e)}"

def list_available_transcripts(video_id):
    try:
        transcript_list = YouTubeTranscriptApi.list_transcripts(video_id)
        transcripts_info = []
        
        for transcript in transcript_list:
            info = {
                "Language": transcript.language,
                "Code": transcript.language_code,
                "Is Generated": transcript.is_generated,
                "Is Translatable": transcript.is_translatable,
                "Translation Languages": transcript.translation_languages
            }
            transcripts_info.append(info)
        
        return json.dumps(transcripts_info, indent=2)
    except TranscriptsDisabled:
        return "Error: Transcripts are disabled for this video"
    except Exception as e:
        return f"Error: {str(e)}"

# Create Gradio interface
with gr.Blocks(title="YouTube Transcript Fetcher") as demo:
    gr.Markdown("# YouTube Transcript Fetcher")
    gr.Markdown("Retrieve transcripts from YouTube videos with various formatting options")
    
    with gr.Tab("Get Transcript"):
        with gr.Row():
            with gr.Column():
                video_id_input = gr.Textbox(label="YouTube Video ID", placeholder="e.g., dQw4w9WgXcQ")
                languages_input = gr.Textbox(
                    label="Languages (comma-separated)",
                    placeholder="e.g., en,de,es",
                    value="en"
                )
                format_dropdown = gr.Dropdown(
                    choices=["Text", "JSON", "WebVTT", "SRT"],
                    label="Output Format",
                    value="Text"
                )
                translate_dropdown = gr.Dropdown(
                    choices=["", "en", "de", "es", "fr", "it"],
                    label="Translate To (optional)",
                    value=""
                )
                preserve_formatting = gr.Checkbox(
                    label="Preserve Formatting",
                    value=False
                )
                submit_btn = gr.Button("Get Transcript")
            
            with gr.Column():
                output = gr.Textbox(label="Transcript", lines=20)
        
        submit_btn.click(
            fn=get_transcript,
            inputs=[video_id_input, languages_input, format_dropdown, translate_dropdown, preserve_formatting],
            outputs=output
        )
    
    with gr.Tab("List Available Transcripts"):
        with gr.Row():
            with gr.Column():
                list_video_id = gr.Textbox(label="YouTube Video ID", placeholder="e.g., dQw4w9WgXcQ")
                list_btn = gr.Button("List Transcripts")
            
            with gr.Column():
                list_output = gr.Textbox(label="Available Transcripts (JSON)", lines=20)
        
        list_btn.click(
            fn=list_available_transcripts,
            inputs=list_video_id,
            outputs=list_output
        )
    
    gr.Markdown("""
    ### Notes
    - Enter a valid YouTube video ID (found in the URL)
    - Specify languages as comma-separated codes (e.g., "en,de")
    - Choose output format from available options
    - Optional: Select a language to translate the transcript to
    - Preserve formatting keeps HTML tags if present
    """)

demo.launch()