ingestion/classifiers.py

import os
from urllib.parse import urlparse

def classify_source(source: str):

    if not source:
        raise Exception("Empty source")

    if os.path.exists(source):
        return "local"

    if source.endswith(".mp4"):
        return "direct"

    domain = urlparse(source).netloc.lower()

    if "youtube" in domain or "youtu.be" in domain:
        return "youtube"

    if "tiktok" in domain:
        return "tiktok"

    if "instagram" in domain:
        return "instagram"

    if "facebook" in domain:
        return "facebook"

    return "unknown"