youtube_tool.py

from urllib.parse import parse_qs, urlparse
from llama_index.core.tools import FunctionTool
from youtube_transcript_api import YouTubeTranscriptApi, TranscriptsDisabled, NoTranscriptFound, VideoUnavailable

#loader = YoutubeTranscriptReader()
yt_ap = YouTubeTranscriptApi()


def extract_video_id(url: str) -> str:
    """
    Extracts the video ID from a YouTube URL.
    Args:
        url (str): The full YouTube video URL.
    Returns:
        str: The extracted video ID or raises ValueError.
    """
    patterns = [
        r"youtube\.com/watch\?v=([a-zA-Z0-9_-]{11})",
        r"youtu\.be/([a-zA-Z0-9_-]{11})"
    ]
    for pattern in patterns:
        match = re.search(pattern, url)
        if match:
            return match.group(1)
    raise ValueError("Invalid YouTube URL or unable to extract video ID.")

def get_youtube_transcript(url: str) -> str:
    """
    Fetches the transcript text for a given YouTube video.
    Args:
        url (str): The YouTube video URL.
    Returns:
        str: Combined transcript text or an error message.
    """
    try:
        video_id = extract_video_id(url)
        transcript_list = YouTubeTranscriptApi.get_transcript(video_id)
        full_text = " ".join([entry["text"] for entry in transcript_list])
        return full_text.strip()[:2000]  # Truncate to 2000 chars to prevent token overflow
    except TranscriptsDisabled:
        return "This video has transcripts disabled."
    except NoTranscriptFound:
        return "No transcript was found for this video."
    except Exception as e:
        return f"Transcript error: {str(e)}"
    
youtube_tool = FunctionTool.from_defaults(get_youtube_transcript)


def extract_video_id(url: str) -> str:
    """
    Handles typical YouTube URLs:
    - https://www.youtube.com/watch?v=VIDEO_ID
    - https://youtu.be/VIDEO_ID
    - with extra query params
    """
    parsed = urlparse(url)
    if parsed.hostname in {"www.youtube.com", "youtube.com"}:
        qs = parse_qs(parsed.query)
        if "v" in qs:
            return qs["v"][0]
    # fallback for youtu.be or raw IDs
    return parsed.path.lstrip("/")

def fetch_youtube_transcript(video_url: str) -> str:
    """
    Fetch YouTube transcript text for the given URL.
    In English language.
    """
    video_id = extract_video_id(video_url)

    try:
        # ✅ call on the class, NOT an instance
        transcript_data = yt_ap.fetch(
            video_id=video_id,
            languages=["en"], #You can add as many languages, use yt_ap.list(video_id) function to get the langauges 
        )

        #FROM TRANSCRIPT DATA, YOU CAN CREATE A OBJECT OF TRANSCRIPT SNIPET AND TIME
        arr = [snippet.text for snippet in transcript_data]
        return " ".join(arr)
        #return " ".join(entry["text"] for entry in arr)
    except Exception as e:
        return f"Error fetching video details: {str(e)}"
    
def fetch_youtube_transcript_snippets(video_url: str) -> str:
    """
    Fetch YouTube transcript snippets for the given URL.
    It gets the start-time, end-time and duration of each snippet.
    """
    video_id = extract_video_id(video_url)

    try:
        # ✅ call on the class, NOT an instance
        transcript_data = yt_ap.fetch(
            video_id=video_id,
            languages=["en"], #You can add as many languages, use yt_ap.list(video_id) function to get the langauges 
        )
        arr = [
            {"text": snippet.text, "duration": snippet.duration, "start": snippet.start}
            for snippet in transcript_data
        ]
        return " ".join(f"Text: {entry['text']} Duration: {entry['duration']} StartTime: {entry['start']} <End>" for entry in arr)
    except Exception as e:
        return f"Error fetching video details: {str(e)}"

youtube_transcript_tool = FunctionTool.from_defaults(fetch_youtube_transcript)
youtube_transcript_snippet_tool = FunctionTool.from_defaults(fetch_youtube_transcript_snippets)