python_scripts/twitch_video_info.py

import asyncio
import json
import httpx  # Switched from requests to prevent async loop freezing
import database
from twitchAPI.twitch import Twitch
from twitchAPI.helper import first
from twitchAPI.type import VideoType

SECRETS = None
TWITCH = None
USER = None
GAME_CACHE = {} # Local cache dictionary to store game_id -> game_name mapping

CHANNEL_NAME = "teampgp"

async def get_twitch():
    global SECRETS, TWITCH, USER
    if SECRETS is None:
        with open('twitch_secrets.json', 'r') as f:
            SECRETS = json.load(f)
    if TWITCH is None:
        TWITCH = await Twitch(SECRETS['client_id'], SECRETS['client_secret'])
    if USER is None:
        USER = await first(TWITCH.get_users(logins=[CHANNEL_NAME]))
        if not USER:
            print("User not found.")

async def get_game_name_by_id(game_id: str) -> str:
    """Helper function to fetch game names and cache them locally."""
    if not game_id:
        return "Unknown / No Category"
    if game_id in GAME_CACHE:
        return GAME_CACHE[game_id]
    try:
        game_generator = TWITCH.get_games(game_ids=[game_id])
        game = await first(game_generator)
        if game:
            GAME_CACHE[game_id] = game.name
            return game.name
    except Exception:
        pass
    return "Unknown Game"

async def get_vod_game_name(vod_id: str):
    """Asynchronously query Twitch GQL endpoint for VOD game metadata."""
    game_id = "0"
    game_name = "Unknown Game"
    url = "https://gql.twitch.tv/gql"

    headers = {
        "Client-ID": "kimne78kx3ncx6brgo4mv6wki5h1ko",
        "Content-Type": "application/json"
    }

    payload = [{
        "operationName": "VideoMetadata",
        "variables": {
            "channelLogin": "",
            "videoID": str(vod_id)
        },
        "extensions": {
            "persistedQuery": {
                "version": 1,
                "sha256Hash": "45111672eea2e507f8ba44d101a61862f9c56b11dee09a15634cb75cb9b9084d"
            }
        }
    }]

    # Using httpx async client to prevent blocking the asyncio event loop
    async with httpx.AsyncClient() as client:
        try:
            response = await client.post(url, headers=headers, json=payload)
            if response.status_code == 200:
                data = response.json()
                video_info = data[0].get('data', {}).get('video')
                if video_info and video_info.get('game'):
                    game_id = str(video_info['game']['id'])
                    game_name = video_info['game']['displayName']
                    print(f"GQL Found: {game_name} (ID: {game_id})")
                else:
                    print(f"No game information found in GQL for VOD {vod_id}.")
        except Exception as e:
            print(f"Error fetching GQL metadata for VOD {vod_id}: {e}")

    return game_id, game_name

async def get_streamer_vods():
    await get_twitch()
    
    print(f"Starting VOD extraction for {USER.display_name}...")
    vod_generator = TWITCH.get_videos(user_id=USER.id, first=100, video_type=VideoType.ALL)
 
    all_vods = []

    async for v in vod_generator:
        # Resolving game category safely without freezing the event loop
        game_id, game_name = await get_vod_game_name(v.id)
        
        # Safely convert game_id to integer if possible, otherwise default to 0
        try:
            clean_game_id = int(game_id)
        except ValueError:
            clean_game_id = 0

        vod_data = {
            "id": v.id,
            "title": v.title,
            "created_at": str(v.published_at),
            "view_count": int(v.view_count) if v.view_count else 0,
            "duration": v.duration,
            "url": v.url,
            "thumbnail_url": v.thumbnail_url,
            "game_id": clean_game_id,
            "game_name": game_name,
            "stream_id": str(v.stream_id) if v.stream_id else "0",
            "creator_name": CHANNEL_NAME,
            "clip_is": False,
        }
        all_vods.append(vod_data)
        print(f"Collected VOD: {v.title} | Category: {game_name} ({v.duration})")

    print(f"\nFinished extracting VODs. Total gathered: {len(all_vods)}")
    return all_vods

async def get_streamer_clips():
    await get_twitch()

    print(f"Starting clip extraction for {USER.display_name}...")
    clip_generator = TWITCH.get_clips(broadcaster_id=USER.id, first=100)
    all_clips = []

    async for c in clip_generator:
        game_name = await get_game_name_by_id(c.game_id)
        
        try:
            clean_game_id = int(c.game_id)
        except (ValueError, TypeError):
            clean_game_id = 0

        clip_data = {
            "id": c.id,
            "title": c.title,
            "created_at": str(c.created_at),             
            "view_count": int(c.view_count) if c.view_count else 0,
            "duration": c.duration,
            "url": c.url,
            "thumbnail_url": c.thumbnail_url,
            "game_id": clean_game_id,
            "game_name": game_name,
            "stream_id": "0",
            "creator_name": c.creator_name,
            "clip_is": True,
        }
        all_clips.append(clip_data)
        print(f"Collected clip: {c.title} | Category: {game_name} ({c.view_count} views)")

    print(f"\nFinished extracting clips. Total gathered: {len(all_clips)}")
    return all_clips

async def main():
    print("--- Script Started ---")
    
    db = database.Database()
    vods_list = await get_streamer_vods()
    
    for v in vods_list:
        db.insert_video_record(
            v['id'], v['title'], v['created_at'], v['view_count'], v['duration'], 
            v['url'], v['thumbnail_url'], v['game_id'], v['game_name'], 
            v['stream_id'], v['creator_name'], v['clip_is']
        )

    if vods_list:
        print(f"Recent VOD: '{vods_list[0]['title']}'")

    clips_list = await get_streamer_clips()
        
    for v in clips_list:
        db.insert_video_record(
            v['id'], v['title'], v['created_at'], v['view_count'], v['duration'], 
            v['url'], v['thumbnail_url'], v['game_id'], v['game_name'], 
            v['stream_id'], v['creator_name'], v['clip_is']
        )
    
    if clips_list:
        print(f"Recent VOD: '{clips_list[0]['title']}'")

if __name__ == "__main__":
    asyncio.run(main())