| 100 | |
| 101 | |
| 102 | class ExecutionEngine: |
| 103 | _SLOT_HEARTBEAT_INTERVAL: float = 5.0 |
| 104 | |
| 105 | def __init__( |
| 106 | self, |
| 107 | crawler: Crawler, |
| 108 | spider_closed_callback: Callable[ |
| 109 | [Spider], Coroutine[Any, Any, None] | Deferred[None] | None |
| 110 | ], |
| 111 | ) -> None: |
| 112 | self.crawler: Crawler = crawler |
| 113 | self.settings: Settings = crawler.settings |
| 114 | self.signals: SignalManager = crawler.signals |
| 115 | assert crawler.logformatter |
| 116 | self.logformatter: LogFormatter = crawler.logformatter |
| 117 | self._slot: _Slot | None = None |
| 118 | self.spider: Spider | None = None |
| 119 | self.running: bool = False |
| 120 | self._starting: bool = False |
| 121 | self._stopping: bool = False |
| 122 | self.paused: bool = False |
| 123 | self._spider_closed_callback: Callable[ |
| 124 | [Spider], Coroutine[Any, Any, None] | Deferred[None] | None |
| 125 | ] = spider_closed_callback |
| 126 | self.start_time: float | None = None |
| 127 | self._start: AsyncIterator[Any] | None = None |
| 128 | self._closewait: Deferred[None] | None = None |
| 129 | self._start_request_processing_awaitable: ( |
| 130 | asyncio.Future[None] | Deferred[None] | None |
| 131 | ) = None |
| 132 | downloader_cls: type[Downloader] = load_object(self.settings["DOWNLOADER"]) |
| 133 | try: |
| 134 | self.scheduler_cls: type[BaseScheduler] = self._get_scheduler_class( |
| 135 | crawler.settings |
| 136 | ) |
| 137 | self.downloader: Downloader = downloader_cls(crawler) |
| 138 | self._downloader_fetch_needs_spider: bool = argument_is_required( |
| 139 | self.downloader.fetch, "spider" |
| 140 | ) |
| 141 | if self._downloader_fetch_needs_spider: |
| 142 | warnings.warn( |
| 143 | f"The fetch() method of {global_object_name(downloader_cls)} requires a spider argument," |
| 144 | f" this is deprecated and the argument will not be passed in future Scrapy versions.", |
| 145 | ScrapyDeprecationWarning, |
| 146 | stacklevel=2, |
| 147 | ) |
| 148 | |
| 149 | self.scraper: Scraper = Scraper(crawler) |
| 150 | except Exception: |
| 151 | if hasattr(self, "downloader"): |
| 152 | self.downloader.close() |
| 153 | raise |
| 154 | |
| 155 | def _get_scheduler_class(self, settings: BaseSettings) -> type[BaseScheduler]: |
| 156 | scheduler_cls: type[BaseScheduler] = load_object(settings["SCHEDULER"]) |
| 157 | if not issubclass(scheduler_cls, BaseScheduler): |
| 158 | raise TypeError( |
| 159 | f"The provided scheduler class ({settings['SCHEDULER']})" |
no outgoing calls