J.A.R.V.I.S-py/scheduler_store.py

"""Proactive scheduler — port of `crates/jarvis-core/src/scheduler.rs`.

JSON store at <here>/schedule.json. Background daemon thread ticks every 30s
and fires due tasks via the speak callback.

Schedule formats (parsed by `parse_schedule`):
    "daily HH:MM"
    "at HH:MM"
    "every N minutes" / "every N hours" / "every N seconds"
    "in N minutes" / "in N hours" / "in N seconds"

Public API:
    init()                          # start the background thread, load state
    add(task_dict) -> str           # returns id
    remove(id) -> bool
    list_all() -> list[dict]
    remove_by_text(query) -> int
    clear() -> int
"""

import json
import os
import re
import threading
import time

_HERE = os.path.dirname(os.path.abspath(__file__))
_PATH = os.path.join(_HERE, 'schedule.json')
_TICK_SECS = 30

_speak_fn = print
_tasks: list[dict] = []
_lock = threading.Lock()
_thread_started = False


def set_speak(fn):
    global _speak_fn
    _speak_fn = fn


def _speak(text):
    try:
        _speak_fn(text)
    except Exception as exc:
        print(f"[scheduler] speak: {exc}")


# ── persistence ───────────────────────────────────────────────────────────

def _load():
    global _tasks
    if os.path.isfile(_PATH):
        try:
            with open(_PATH, encoding='utf-8') as f:
                data = json.load(f)
            _tasks = data.get('tasks', []) if isinstance(data, dict) else []
        except (OSError, json.JSONDecodeError) as exc:
            print(f"[scheduler] corrupt: {exc}")
            _tasks = []


def _save():
    tmp = _PATH + '.tmp'
    try:
        with open(tmp, 'w', encoding='utf-8') as f:
            json.dump({'tasks': _tasks}, f, ensure_ascii=False, indent=2)
        os.replace(tmp, _PATH)
    except OSError as exc:
        print(f"[scheduler] save: {exc}")


# ── parsing ───────────────────────────────────────────────────────────────

_RU_UNITS = {
    'минут': 60, 'минуту': 60, 'минуты': 60,
    'час': 3600, 'часа': 3600, 'часов': 3600,
    'секунд': 1, 'секунду': 1, 'секунды': 1,
}
_EN_UNITS = {
    'minute': 60, 'minutes': 60,
    'hour': 3600, 'hours': 3600,
    'second': 1, 'seconds': 1,
}


def _unit_to_secs(token: str) -> int | None:
    """Map a unit word to seconds. Tries Russian stems first."""
    low = token.lower()
    for stem, n in _RU_UNITS.items():
        if low.startswith(stem):
            return n
    for stem, n in _EN_UNITS.items():
        if low.startswith(stem):
            return n
    return None


def parse_schedule(spec: str) -> dict:
    """Returns a task-schedule dict, raises ValueError on bad input.

    Schedule shape:
        {'kind': 'daily', 'hour': H, 'minute': M}
        {'kind': 'interval', 'seconds': S}
        {'kind': 'once', 'at': ts}
    """
    s = spec.strip().lower()
    now = int(time.time())

    # daily HH:MM
    m = re.match(r'^daily\s+(\d{1,2}):(\d{2})$', s)
    if m:
        h, mn = int(m.group(1)), int(m.group(2))
        if not (0 <= h <= 23 and 0 <= mn <= 59):
            raise ValueError("hour/minute out of range")
        return {'kind': 'daily', 'hour': h, 'minute': mn}

    # at HH:MM
    m = re.match(r'^at\s+(\d{1,2}):(\d{2})$', s)
    if m:
        h, mn = int(m.group(1)), int(m.group(2))
        if not (0 <= h <= 23 and 0 <= mn <= 59):
            raise ValueError("hour/minute out of range")
        import datetime as dt
        target = dt.datetime.now().replace(hour=h, minute=mn, second=0, microsecond=0)
        if target.timestamp() <= now:
            target = target + dt.timedelta(days=1)
        return {'kind': 'once', 'at': int(target.timestamp())}

    # every N <unit>
    m = re.match(r'^every\s+(\d+)\s+(\S+)$', s)
    if m:
        n = int(m.group(1))
        secs_per = _unit_to_secs(m.group(2))
        if secs_per is None:
            raise ValueError(f"unknown unit: {m.group(2)}")
        return {'kind': 'interval', 'seconds': n * secs_per}

    # in N <unit>
    m = re.match(r'^in\s+(\d+)\s+(\S+)$', s)
    if m:
        n = int(m.group(1))
        secs_per = _unit_to_secs(m.group(2))
        if secs_per is None:
            raise ValueError(f"unknown unit: {m.group(2)}")
        return {'kind': 'once', 'at': now + n * secs_per}

    raise ValueError(f"unknown schedule spec: {spec}")


def _next_fire(schedule: dict, last_fired: int | None, now: int) -> int | None:
    kind = schedule.get('kind')
    if kind == 'once':
        if last_fired is not None:
            return None
        return schedule.get('at')
    if kind == 'interval':
        anchor = last_fired if last_fired is not None else now
        return anchor + int(schedule['seconds'])
    if kind == 'daily':
        import datetime as dt
        h, mn = int(schedule['hour']), int(schedule['minute'])
        today = dt.datetime.now().replace(hour=h, minute=mn, second=0, microsecond=0)
        today_ts = int(today.timestamp())
        if today_ts > now and (last_fired is None or
                               dt.datetime.fromtimestamp(last_fired).date() != today.date()):
            return today_ts
        return today_ts + 86400
    return None


# ── public API ────────────────────────────────────────────────────────────

def _gen_id() -> str:
    import uuid as uuidlib
    return f"task-{uuidlib.uuid4().hex[:12]}"


def add(task: dict) -> str:
    """Add task. Required keys: name, schedule (dict or string), action.
    Action is a dict with at least {'type': 'speak', 'text': '...'}.
    """
    sched = task.get('schedule')
    if isinstance(sched, str):
        sched = parse_schedule(sched)
    if not isinstance(sched, dict):
        raise ValueError("missing or invalid schedule")

    tid = task.get('id') or _gen_id()
    record = {
        'id': tid,
        'name': task.get('name', 'task'),
        'schedule': sched,
        'action': task.get('action', {'type': 'speak', 'text': 'Напоминание.'}),
        'last_fired': None,
        'enabled': task.get('enabled', True),
        'created_at': int(time.time()),
    }
    with _lock:
        _tasks[:] = [t for t in _tasks if t['id'] != tid]
        _tasks.append(record)
        _save()
    return tid


def remove(task_id: str) -> bool:
    with _lock:
        before = len(_tasks)
        _tasks[:] = [t for t in _tasks if t['id'] != task_id]
        if len(_tasks) != before:
            _save()
            return True
    return False


def list_all() -> list[dict]:
    with _lock:
        return [dict(t) for t in _tasks]


def clear() -> int:
    with _lock:
        n = len(_tasks)
        _tasks.clear()
        _save()
    return n


def remove_by_text(query: str) -> int:
    nq = query.strip().lower()
    if not nq:
        return 0
    with _lock:
        before = len(_tasks)
        def matches(t):
            if nq in t['name'].lower():
                return True
            action = t.get('action', {})
            if action.get('type') == 'speak':
                if nq in (action.get('text') or '').lower():
                    return True
            return False
        _tasks[:] = [t for t in _tasks if not matches(t)]
        removed = before - len(_tasks)
        if removed:
            _save()
    return removed


# ── background tick ────────────────────────────────────────────────────────

def _fire_action(action: dict) -> None:
    atype = action.get('type', 'speak')
    if atype == 'speak':
        _speak(action.get('text', '...'))
    else:
        print(f"[scheduler] unknown action type: {atype}")


def _tick():
    now = int(time.time())
    fired_ids = []
    expired_once = []
    with _lock:
        for task in list(_tasks):
            if not task.get('enabled', True):
                continue
            nf = _next_fire(task['schedule'], task.get('last_fired'), now)
            if nf is not None and nf <= now:
                fired_ids.append(task['id'])
                if task['schedule'].get('kind') == 'once':
                    expired_once.append(task['id'])

    # Fire outside the lock — speak() can block.
    for task in list_all():
        if task['id'] in fired_ids:
            print(f"[scheduler] firing {task['id']} '{task['name']}'")
            _fire_action(task['action'])

    if fired_ids:
        with _lock:
            for task in _tasks:
                if task['id'] in fired_ids:
                    task['last_fired'] = now
            _tasks[:] = [t for t in _tasks if t['id'] not in expired_once]
            _save()


def _tick_loop():
    while True:
        try:
            time.sleep(_TICK_SECS)
            _tick()
        except Exception as exc:
            print(f"[scheduler] tick err: {exc}")


def init():
    """Load tasks + start the background thread (idempotent)."""
    global _thread_started
    _load()
    if _thread_started:
        return
    _thread_started = True
    threading.Thread(target=_tick_loop, name='jarvis-scheduler',
                     daemon=True).start()
    print(f"[scheduler] thread started ({len(_tasks)} tasks loaded)")