#!/usr/bin/env python3
# trim-captures — prune old VOICE capture audio from /tmp/airband, keeping the durable detection log.
# Safe by design: heard.jsonl (the detections Activity reads — channel + time) is SEPARATE and never
# touched; this only deletes bulky .mp3 audio AFTER it's logged. It forces an ingest first (GET /api/heard,
# which appends any new clip to heard.jsonl) and ABORTS if the dashboard is unreachable — so a not-yet-logged
# clip can never be deleted. --dry-run to preview.
import os, glob, time, sys, urllib.request

CAP = os.environ.get("AIRBAND_DIR", "/tmp/airband")
KEEP_HOURS = float(os.environ.get("CAP_KEEP_HOURS", "24"))
DASH = os.environ.get("DASH_URL", "http://127.0.0.1:8157")
DRY = "--dry-run" in sys.argv

def main():
    try:                                            # log every current clip to heard.jsonl first
        urllib.request.urlopen(DASH + "/api/heard", timeout=10).read()
    except Exception as e:
        print(f"[trim] dashboard unreachable ({e}) — NOT trimming (clips may be unlogged)"); return
    cutoff = time.time() - KEEP_HOURS * 3600
    mp3s = glob.glob(os.path.join(CAP, "*.mp3"))
    old = [m for m in mp3s if os.path.getmtime(m) < cutoff]
    freed = 0
    for m in old:
        freed += os.path.getsize(m)
        if not DRY:
            try: os.remove(m)
            except OSError: pass
            # whisper sidecar (2026-07-10) rides with its clip — no orphan .txt buildup
            try: os.remove(m[:-4] + ".txt")
            except OSError: pass
    print(f"[trim{' DRY' if DRY else ''}] {len(mp3s)} clips on disk, deleted {len(old)} older than "
          f"{KEEP_HOURS:.0f}h, freed {freed//1024} KB; heard.jsonl untouched")

if __name__ == "__main__":
    main()