mirror of
https://github.com/kuhyx/steam-backlog-enforcer.git
synced 2026-11-11 13:47:31 +01:00
Krzysztof kuhy Rudnicki
eb47067617
The enforce loop now waits on an inotify watch of log.json's directory (_file_wake, ctypes, no new dependency) instead of a plain sleep, so a credited workout moves the gaming budget within the tick. Earner answers (workout, LeetCode, reading) are memoised with the source file's (inode, mtime_ns, size) stamp, so a rewritten file re-asks at once instead of after the 60 s TTL. Any inotify setup or read failure warns and falls back to the plain sleep; the stamp still catches the change next tick. Tests: conftest stubs the wake factory and points workout_log_path at a tmp path, so no test ever watches the real screen-locker tree. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018RHwVPyiiQMPh4M3JQwQku
72 lines
2.6 KiB
Python
72 lines
2.6 KiB
Python
"""Whether a paper-book reading session was credited today, worth an hour.
|
|
|
|
book-guard (``~/src/book-guard``) publishes reading through its signed ledger,
|
|
the same shape as leetcode-guard's, and screen-locker reads the same fact for a
|
|
shutdown hour. Shaped like :mod:`steam_backlog_enforcer._leetcode_bonus` where
|
|
it matters:
|
|
|
|
**Fail closed.** Bad JSON, a missing entries array or an unreadable key yields
|
|
``None`` -- no bonus, logged at warning. Failing open would let the coupling be
|
|
defeated by deleting a file.
|
|
|
|
**Independent.** Read separately and summed by ``resolve_budget``; nothing here
|
|
can move the workout or LeetCode terms.
|
|
|
|
**One transport.** book-guard has no status server, so there is no fallback;
|
|
and no incident either -- until the first book is registered the ledger simply
|
|
does not exist, and that is not worth waking anyone about.
|
|
|
|
The contract (book-guard's ``_quiz.record_verdict``): only ``credit`` rows with
|
|
``detail["bonus"] == "1"`` count -- a passed quiz on a 20+ page, 20+ minute
|
|
session -- dated by ``detail["ended_at"]``, when the reading happened.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import logging
|
|
from pathlib import Path
|
|
from typing import TYPE_CHECKING, Final
|
|
|
|
import earned_time
|
|
|
|
from steam_backlog_enforcer._leetcode_ledger import HMAC_KEY_FILE
|
|
from steam_backlog_enforcer._status_api import AnswerCache, file_stamp
|
|
|
|
if TYPE_CHECKING:
|
|
from steam_backlog_enforcer.config import Config
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
_cache = AnswerCache()
|
|
_CACHE_KEY: Final = "book-guard-ledger"
|
|
|
|
|
|
def reset_cache() -> None:
|
|
"""Drop the cached answer, forcing the next call to re-read."""
|
|
_cache.clear()
|
|
|
|
|
|
def read_ledger_read_today(path: Path) -> bool | None:
|
|
"""Whether a verified, bonus-eligible reading session ended today.
|
|
|
|
Returns:
|
|
True or False when the ledger could be read, ``None`` when it could
|
|
not -- which is never "did not read". A ledger that does not exist
|
|
yet (no book registered) is an honest, silent "no".
|
|
"""
|
|
return earned_time.done_today(earned_time.READING, path, HMAC_KEY_FILE)
|
|
|
|
|
|
def read_today(config: Config) -> bool | None:
|
|
"""Today's reading answer, memoised like the other earners."""
|
|
path = Path(config.book_ledger_path).expanduser()
|
|
# Stat before reading, so a credit landing mid-read is re-read next tick.
|
|
stamp = file_stamp(path)
|
|
cached = _cache.fresh(_CACHE_KEY, stamp=stamp)
|
|
if cached is not None:
|
|
return cached
|
|
answer = read_ledger_read_today(path)
|
|
if answer is None:
|
|
return None
|
|
return _cache.store(_CACHE_KEY, answer=answer, stamp=stamp)
|