|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +import asyncio |
| 4 | +import logging |
| 5 | +import typing as tp |
| 6 | +import uuid |
| 7 | + |
| 8 | +import ydb # type: ignore[import-untyped] |
| 9 | +import ydb.aio # type: ignore[import-untyped] |
| 10 | +from pydantic import ValidationError |
| 11 | +from taskiq import ScheduleSource |
| 12 | +from taskiq.scheduler.scheduled_task import ScheduledTask |
| 13 | +from taskiq.serializers import PickleSerializer |
| 14 | + |
| 15 | +from taskiq_ydb.exceptions import DatabaseConnectionError |
| 16 | + |
| 17 | + |
| 18 | +if tp.TYPE_CHECKING: |
| 19 | + from taskiq.abc.broker import AsyncBroker |
| 20 | + |
| 21 | + |
| 22 | +logger = logging.getLogger(__name__) |
| 23 | + |
| 24 | + |
| 25 | +class YdbScheduleSource(ScheduleSource): |
| 26 | + """Schedule source that uses YDB to store schedules in YDB database.""" |
| 27 | + |
| 28 | + def __init__( |
| 29 | + self, |
| 30 | + broker: AsyncBroker, |
| 31 | + driver_config: ydb.aio.driver.DriverConfig, |
| 32 | + table_name: str = 'taskiq_schedules', |
| 33 | + pool_size: int = 5, |
| 34 | + connection_timeout: int = 5, |
| 35 | + ) -> None: |
| 36 | + """ |
| 37 | + Construct new schedule source. |
| 38 | +
|
| 39 | + Args: |
| 40 | + broker: The TaskIQ broker instance to use for finding and managing tasks. |
| 41 | + driver_config: YDB driver configuration. |
| 42 | + table_name: Table name for storing task results. |
| 43 | + pool_size: YDB session pool size. |
| 44 | + connection_timeout: Timeout for connection to database during startup. |
| 45 | + """ |
| 46 | + self._broker: tp.Final = broker |
| 47 | + self._driver = ydb.aio.Driver(driver_config=driver_config) |
| 48 | + self._table_name: tp.Final = table_name |
| 49 | + self._pool_size: tp.Final = pool_size |
| 50 | + self._pool: ydb.aio.SessionPool |
| 51 | + self._connection_timeout: tp.Final = connection_timeout |
| 52 | + self._serializer: tp.Final = PickleSerializer() |
| 53 | + |
| 54 | + |
| 55 | + async def startup(self) -> None: |
| 56 | + """ |
| 57 | + Initialize the result backend. |
| 58 | +
|
| 59 | + Construct new connection pool |
| 60 | + and create new table for results if not exists. |
| 61 | + """ |
| 62 | + try: |
| 63 | + logger.debug('Waiting for YDB driver to be ready') |
| 64 | + await self._driver.wait(fail_fast=True, timeout=self._connection_timeout) |
| 65 | + except (ydb.issues.ConnectionLost, asyncio.exceptions.TimeoutError) as exception: |
| 66 | + raise DatabaseConnectionError from exception |
| 67 | + self._pool = ydb.aio.SessionPool(self._driver, size=self._pool_size) |
| 68 | + session = await self._pool.acquire() |
| 69 | + |
| 70 | + table_path = f'{self._driver._driver_config.database}/{self._table_name}' # noqa: SLF001 |
| 71 | + try: |
| 72 | + logger.debug('Checking if table %s exists', self._table_name) |
| 73 | + existing_table = await session.describe_table(table_path) |
| 74 | + except ydb.issues.SchemeError: |
| 75 | + existing_table = None |
| 76 | + if not existing_table: |
| 77 | + logger.debug('Table %s does not exist, creating...', self._table_name) |
| 78 | + await session.create_table( |
| 79 | + table_path, |
| 80 | + ydb.TableDescription() |
| 81 | + .with_column(ydb.Column('id', ydb.PrimitiveType.UUID)) |
| 82 | + .with_column(ydb.Column('task_name', ydb.PrimitiveType.Utf8)) |
| 83 | + .with_column(ydb.Column('schedule', ydb.PrimitiveType.String)) |
| 84 | + .with_primary_key('id'), |
| 85 | + ) |
| 86 | + logger.debug('Table %s created', self._table_name) |
| 87 | + else: |
| 88 | + logger.debug('Table %s already exists', self._table_name) |
| 89 | + |
| 90 | + # Load existing schedules from labels in tasks |
| 91 | + schedule_tasks = self._extract_scheduled_tasks_from_broker() |
| 92 | + for task in await self.get_schedules(): |
| 93 | + await self.delete_schedule(task.schedule_id) |
| 94 | + for schedule_task in schedule_tasks: |
| 95 | + await self.add_schedule(schedule_task) |
| 96 | + |
| 97 | + async def shutdown(self) -> None: |
| 98 | + """Close the connection pool.""" |
| 99 | + await asyncio.to_thread(self._driver.topic_client.close) |
| 100 | + if hasattr(self, '_pool'): |
| 101 | + await self._pool.stop(timeout=10) |
| 102 | + await self._driver.stop(timeout=10) |
| 103 | + |
| 104 | + def _extract_scheduled_tasks_from_broker(self) -> list[ScheduledTask]: |
| 105 | + """ |
| 106 | + Extract schedules from tasks that were registered in broker. |
| 107 | +
|
| 108 | + Returns: |
| 109 | + A list of ScheduledTask instances extracted from the task's labels. |
| 110 | + """ |
| 111 | + scheduled_tasks_for_creation: list[ScheduledTask] = [] |
| 112 | + for task_name, task in self._broker.get_all_tasks().items(): |
| 113 | + if 'schedule' not in task.labels: |
| 114 | + logger.debug('Task %s has no schedule, skipping', task_name) |
| 115 | + continue |
| 116 | + if not isinstance(task.labels['schedule'], list): |
| 117 | + logger.warning( |
| 118 | + 'Schedule for task %s is not a list, skipping', |
| 119 | + task_name, |
| 120 | + ) |
| 121 | + continue |
| 122 | + for schedule in task.labels['schedule']: |
| 123 | + try: |
| 124 | + new_schedule = ScheduledTask.model_validate( |
| 125 | + { |
| 126 | + 'task_name': task_name, |
| 127 | + 'labels': schedule.get('labels', {}), |
| 128 | + 'args': schedule.get('args', []), |
| 129 | + 'kwargs': schedule.get('kwargs', {}), |
| 130 | + 'schedule_id': str(uuid.uuid4()), |
| 131 | + 'cron': schedule.get('cron', None), |
| 132 | + 'cron_offset': schedule.get('cron_offset', None), |
| 133 | + 'time': schedule.get('time', None), |
| 134 | + }, |
| 135 | + ) |
| 136 | + scheduled_tasks_for_creation.append(new_schedule) |
| 137 | + except ValidationError: # noqa: PERF203 |
| 138 | + logger.exception( |
| 139 | + 'Schedule for task %s is not valid, skipping', |
| 140 | + task_name, |
| 141 | + ) |
| 142 | + continue |
| 143 | + return scheduled_tasks_for_creation |
| 144 | + |
| 145 | + async def get_schedules(self) -> list[ScheduledTask]: |
| 146 | + """Get list of taskiq schedules.""" |
| 147 | + query = f""" |
| 148 | + SELECT schedule FROM {self._table_name}; |
| 149 | + """ # noqa: S608 |
| 150 | + session = await self._pool.acquire() |
| 151 | + result_sets = await session.transaction().execute( |
| 152 | + await session.prepare(query), |
| 153 | + commit_tx=True, |
| 154 | + ) |
| 155 | + await self._pool.release(session) |
| 156 | + scheduled_tasks = [] |
| 157 | + for result_set in result_sets: |
| 158 | + rows = result_set.rows |
| 159 | + for row in rows: |
| 160 | + scheduled_tasks.append( # noqa: PERF401 |
| 161 | + self._serializer.loadb(row.schedule), |
| 162 | + ) |
| 163 | + return scheduled_tasks |
| 164 | + |
| 165 | + async def add_schedule( |
| 166 | + self, |
| 167 | + schedule: ScheduledTask, |
| 168 | + ) -> None: |
| 169 | + """ |
| 170 | + Add a new schedule. |
| 171 | +
|
| 172 | + This function is used to add new schedules. |
| 173 | + It's a convenient helper for people who want to add new schedules |
| 174 | + for the current source. |
| 175 | +
|
| 176 | + Args: |
| 177 | + schedule: schedule to add. |
| 178 | + """ |
| 179 | + schedule_id = uuid.UUID(schedule.schedule_id) |
| 180 | + query = f""" |
| 181 | + DECLARE $id AS Uuid; |
| 182 | + DECLARE $task_name AS Utf8; |
| 183 | + DECLARE $schedule AS String; |
| 184 | +
|
| 185 | + UPSERT INTO {self._table_name} (id, task_name, schedule) |
| 186 | + VALUES ($id, $task_name, $schedule); |
| 187 | + """ |
| 188 | + session = await self._pool.acquire() |
| 189 | + await session.transaction().execute( |
| 190 | + await session.prepare(query), |
| 191 | + { |
| 192 | + '$id': schedule_id, |
| 193 | + '$task_name': schedule.task_name, |
| 194 | + '$schedule': self._serializer.dumpb(schedule), |
| 195 | + }, |
| 196 | + commit_tx=True, |
| 197 | + ) |
| 198 | + await self._pool.release(session) |
| 199 | + |
| 200 | + async def delete_schedule(self, schedule_id: str) -> None: |
| 201 | + """ |
| 202 | + Method to delete schedule by id. |
| 203 | +
|
| 204 | + This is useful for schedule cancelation. |
| 205 | +
|
| 206 | + Args: |
| 207 | + schedule_id: id of schedule to delete. |
| 208 | + """ |
| 209 | + schedule_id_uuid = uuid.UUID(schedule_id) |
| 210 | + query = f""" |
| 211 | + DECLARE $id AS Uuid; |
| 212 | +
|
| 213 | + DELETE FROM {self._table_name} |
| 214 | + WHERE id = $id; |
| 215 | + """ # noqa: S608 |
| 216 | + session = await self._pool.acquire() |
| 217 | + await session.transaction().execute( |
| 218 | + await session.prepare(query), |
| 219 | + { |
| 220 | + '$id': schedule_id_uuid, |
| 221 | + }, |
| 222 | + commit_tx=True, |
| 223 | + ) |
| 224 | + await self._pool.release(session) |
| 225 | + |
| 226 | + async def post_send( |
| 227 | + self, |
| 228 | + task: ScheduledTask, |
| 229 | + ) -> None: |
| 230 | + """ |
| 231 | + Delete schedule if it was one-time task. |
| 232 | +
|
| 233 | + Args: |
| 234 | + task: task that just have sent |
| 235 | + """ |
| 236 | + if task.time: |
| 237 | + await self.delete_schedule(task.schedule_id) |
0 commit comments