"""Simple in-memory LRU cache for TTS WAV output.

The CLI client maintains its own disk-based WAV cache, so this server-side
cache is a secondary layer.  A small default (50 entries, ~12-25 MB for
typical notification phrases) is sufficient; raise via POLYTTS_CACHE_MAX
if the server handles many unique phrases without a client cache in front.
"""

import os
from collections import OrderedDict

MAX_ENTRIES = int(os.environ.get("POLYTTS_CACHE_MAX", "50"))

_store: OrderedDict[tuple, bytes] = OrderedDict()


def get(key: tuple) -> bytes | None:
    """Return cached WAV bytes and promote to most-recent, or None."""
    if key in _store:
        _store.move_to_end(key)
        return _store[key]
    return None


def put(key: tuple, wav_bytes: bytes):
    """Store WAV bytes, evicting oldest if over capacity."""
    _store[key] = wav_bytes
    _store.move_to_end(key)
    while len(_store) > MAX_ENTRIES:
        _store.popitem(last=False)