import os
import time
from pathlib import Path
import logging

def autoclean_debug_material(data_dir="data", max_age_hours=48):
    """
    Delete debug files (PNGs, HTMLs, etc.) in data/ and subfolders older than max_age_hours.
    Does NOT delete listings, applications, state, or CSV/JSON/LOG files.
    """
    logger = logging.getLogger()
    now = time.time()
    max_age = max_age_hours * 3600
    # File extensions considered debug material
    debug_exts = {".png", ".html"}
    # Always skip these files (listing, state, applications, logs, csv, json, ttf, etc.)
    safe_exts = {".json", ".csv", ".log", ".ttf"}
    safe_names = {"listings.json", "applications.json", "state.json", "wgcompany_listings.json", "wgcompany_times.csv", "listing_times.csv"}
    data_path = Path(data_dir)
    deleted = []
    for root, dirs, files in os.walk(data_path):
        for fname in files:
            fpath = Path(root) / fname
            ext = fpath.suffix.lower()
            if ext in debug_exts and ext not in safe_exts and fname not in safe_names:
                try:
                    mtime = fpath.stat().st_mtime
                    if now - mtime > max_age:
                        fpath.unlink()
                        deleted.append(str(fpath))
                except Exception as e:
                    logger.warning(f"Could not delete {fpath}: {e}")
    logger.info(f"Autocleaned {len(deleted)} debug files older than {max_age_hours}h.")
    return deleted