app.py

import re
import requests
import gradio as gr
import json

def retrieve_transcript(video_url):
    # Extract video ID from the URL
    video_id_match = re.search(r"v=([a-zA-Z0-9_-]+)", video_url)
    if not video_id_match:
        return "Invalid YouTube URL. Please ensure it contains a valid video ID."

    video_id = video_id_match.group(1)

    # Regular expression to extract ytInitialPlayerResponse
    YT_INITIAL_PLAYER_RESPONSE_RE = re.compile(
        r"ytInitialPlayerResponse\s*=\s*({.+?})\s*;\s*(?:var\s+(?:meta|head)|</script|\n)"
    )

    # Fetch the YouTube page
    response = requests.get(f"https://www.youtube.com/watch?v={video_id}")
    if response.status_code != 200:
        return "Unable to fetch video page. Please check the URL or your connection."

    # Extract ytInitialPlayerResponse from the page
    match = YT_INITIAL_PLAYER_RESPONSE_RE.search(response.text)
    if not match:
        return "Unable to parse player response. The video may not have captions available."

    # Parse the player response JSON using json.loads
    player = match.group(1)
    try:
        player = json.loads(player)  # Use json.loads for secure parsing
    except json.JSONDecodeError:
        return "Error decoding YouTube response. The response format may have changed."

    # Check if videoDetails exists
    if "videoDetails" not in player:
        return "The video details could not be found. The video might be private, restricted, or unavailable."

    # Extract metadata
    metadata = {
        "title": player["videoDetails"].get("title", "Unknown Title"),
        "duration": player["videoDetails"].get("lengthSeconds", "Unknown Duration"),
        "author": player["videoDetails"].get("author", "Unknown Author"),
        "views": player["videoDetails"].get("viewCount", "Unknown Views"),
    }

    # Get the tracks and sort them by priority
    tracks = player.get("captions", {}).get("playerCaptionsTracklistRenderer", {}).get("captionTracks", [])
    if not tracks:
        return f"Title: {metadata['title']}\n\nNo captions available for this video."

    # Sort tracks by priority
    tracks.sort(compare_tracks)

    # Fetch the transcript
    transcript_response = requests.get(tracks[0]["baseUrl"] + "&fmt=json3")
    if transcript_response.status_code != 200:
        return "Unable to fetch transcript. Please try again later."

    transcript = transcript_response.json()

    # Parse the transcript
    parsed_transcript = (
        " ".join(
            [
                " ".join([seg["utf8"] for seg in event["segs"]])
                for event in transcript["events"]
                if "segs" in event
            ]
        )
        .replace("\u200B", "")
        .replace("\u200C", "")
        .replace("\u200D", "")
        .replace("\uFEFF", "")
        .replace("\s+", " ")
    )

    return f"Title: {metadata['title']}\nAuthor: {metadata['author']}\nViews: {metadata['views']}\nDuration: {metadata['duration']} seconds\n\nTranscript:\n{parsed_transcript}"


def compare_tracks(track1, track2):
    lang_code1 = track1.get("languageCode", "")
    lang_code2 = track2.get("languageCode", "")

    # Sort by priority: English > Non-English, Non-ASR > ASR
    if lang_code1 == "en" and lang_code2 != "en":
        return -1
    elif lang_code1 != "en" and lang_code2 == "en":
        return 1
    elif track1.get("kind") != "asr" and track2.get("kind") == "asr":
        return -1
    elif track1.get("kind") == "asr" and track2.get("kind") != "asr":
        return 1
    return 0  # Preserve order if both have same priority


# Gradio Interface
def gradio_interface(video_url):
    return retrieve_transcript(video_url)


# Create Gradio UI
interface = gr.Interface(
    fn=gradio_interface,
    inputs="text",
    outputs="text",
    title="YouTube Transcript Extractor",
    description="Enter a YouTube video URL to extract the transcript. The video must have captions available.",
)

# Launch the app
if __name__ == "__main__":
    interface.launch()