From dc1ee09481c7655a8ebf77a75ccc965a4ba5400d Mon Sep 17 00:00:00 2001 From: Andrey Rakhmatullin Date: Fri, 27 Dec 2019 21:55:58 +0500 Subject: [PATCH] Rename ASYNCIO_ENABLED to ASYNCIO_REACTOR, change the logic accordingly. --- docs/topics/settings.rst | 14 +++---- scrapy/crawler.py | 17 +++++--- scrapy/settings/default_settings.py | 2 +- scrapy/utils/log.py | 5 ++- .../asyncio_enabled_no_reactor.py | 2 +- .../CrawlerProcess/asyncio_enabled_reactor.py | 2 +- tests/test_commands.py | 8 ++-- tests/test_crawler.py | 42 ++++++++----------- 8 files changed, 43 insertions(+), 49 deletions(-) diff --git a/docs/topics/settings.rst b/docs/topics/settings.rst index 62b2870b2..c02f877fc 100644 --- a/docs/topics/settings.rst +++ b/docs/topics/settings.rst @@ -160,26 +160,22 @@ to any particular component. In that case the module of that component will be shown, typically an extension, middleware or pipeline. It also means that the component must be enabled in order for the setting to have any effect. -.. setting:: ASYNCIO_ENABLED +.. setting:: ASYNCIO_REACTOR -ASYNCIO_ENABLED +ASYNCIO_REACTOR --------------- Default: ``False`` -Whether to support ``async def`` methods and callbacks which use code that -requires an asyncio loop. - -If an ``async def`` coroutine doesn't require the asyncio loop, it will work -even if this is set to ``False``. Coroutines that require the asyncio loop may -silently fail to run or raise errors unless this is set to ``True``. +Whether to install and require the Twisted reactor that uses the asyncio loop. When this option is set to ``True``, Scrapy will require :class:`~twisted.internet.asyncioreactor.AsyncioSelectorReactor`. It will install this reactor if no reactor is installed yet, such as when using the ``scrapy`` script or :class:`~scrapy.crawler.CrawlerProcess`. If you are using :class:`~scrapy.crawler.CrawlerRunner`, you need to install the correct reactor -manually. +manually. If a different reactor is installed outside Scrapy, it will raise an +exception. The default value for this option is currently ``False`` to maintain backward compatibility and avoid possible problems caused by using a different Twisted diff --git a/scrapy/crawler.py b/scrapy/crawler.py index a9443f7ac..f87e67d93 100644 --- a/scrapy/crawler.py +++ b/scrapy/crawler.py @@ -137,6 +137,7 @@ class CrawlerRunner(object): self._crawlers = set() self._active = set() self.bootstrap_failed = False + self._handle_asyncio_reactor() @property def spiders(self): @@ -230,6 +231,11 @@ class CrawlerRunner(object): while self._active: yield defer.DeferredList(self._active) + def _handle_asyncio_reactor(self): + if self.settings.getbool('ASYNCIO_REACTOR') and not is_asyncio_reactor_installed(): + raise Exception("ASYNCIO_REACTOR is on but the Twisted asyncio " + "reactor is not installed.") + class CrawlerProcess(CrawlerRunner): """ @@ -257,12 +263,6 @@ class CrawlerProcess(CrawlerRunner): def __init__(self, settings=None, install_root_handler=True): super(CrawlerProcess, self).__init__(settings) - if self.settings.getbool('ASYNCIO_ENABLED'): - install_asyncio_reactor() - if not is_asyncio_reactor_installed(): - raise Exception("ASYNCIO_ENABLED is on but the Twisted asyncio " - "reactor is not installed, this is not supported.") - install_shutdown_handlers(self._signal_shutdown) configure_logging(self.settings, install_root_handler) log_scrapy_info(self.settings) @@ -333,6 +333,11 @@ class CrawlerProcess(CrawlerRunner): except RuntimeError: # raised if already stopped or in shutdown stage pass + def _handle_asyncio_reactor(self): + if self.settings.getbool('ASYNCIO_REACTOR'): + install_asyncio_reactor() + super()._handle_asyncio_reactor() + def _get_spider_loader(settings): """ Get SpiderLoader instance from settings """ diff --git a/scrapy/settings/default_settings.py b/scrapy/settings/default_settings.py index a7792e248..d03fd37b0 100644 --- a/scrapy/settings/default_settings.py +++ b/scrapy/settings/default_settings.py @@ -19,7 +19,7 @@ from os.path import join, abspath, dirname AJAXCRAWL_ENABLED = False -ASYNCIO_ENABLED = False +ASYNCIO_REACTOR = False AUTOTHROTTLE_ENABLED = False AUTOTHROTTLE_DEBUG = False diff --git a/scrapy/utils/log.py b/scrapy/utils/log.py index 6179e1bd1..e4cf0196b 100644 --- a/scrapy/utils/log.py +++ b/scrapy/utils/log.py @@ -11,6 +11,7 @@ from twisted.python import log as twisted_log import scrapy from scrapy.settings import Settings from scrapy.exceptions import ScrapyDeprecationWarning +from scrapy.utils.asyncio import is_asyncio_reactor_installed from scrapy.utils.versions import scrapy_components_versions @@ -148,8 +149,8 @@ def log_scrapy_info(settings): {'versions': ", ".join("%s %s" % (name, version) for name, version in scrapy_components_versions() if name != "Scrapy")}) - if settings.getbool('ASYNCIO_ENABLED'): - logger.debug("Asyncio support enabled") + if is_asyncio_reactor_installed(): + logger.debug("Asyncio reactor is installed") class StreamLogger(object): diff --git a/tests/CrawlerProcess/asyncio_enabled_no_reactor.py b/tests/CrawlerProcess/asyncio_enabled_no_reactor.py index dfe028ef4..db1b75931 100644 --- a/tests/CrawlerProcess/asyncio_enabled_no_reactor.py +++ b/tests/CrawlerProcess/asyncio_enabled_no_reactor.py @@ -10,7 +10,7 @@ class NoRequestsSpider(scrapy.Spider): process = CrawlerProcess(settings={ - 'ASYNCIO_ENABLED': True, + 'ASYNCIO_REACTOR': True, }) process.crawl(NoRequestsSpider) diff --git a/tests/CrawlerProcess/asyncio_enabled_reactor.py b/tests/CrawlerProcess/asyncio_enabled_reactor.py index 7a172ea28..cec3c9c25 100644 --- a/tests/CrawlerProcess/asyncio_enabled_reactor.py +++ b/tests/CrawlerProcess/asyncio_enabled_reactor.py @@ -15,7 +15,7 @@ class NoRequestsSpider(scrapy.Spider): process = CrawlerProcess(settings={ - 'ASYNCIO_ENABLED': True, + 'ASYNCIO_REACTOR': True, }) process.crawl(NoRequestsSpider) diff --git a/tests/test_commands.py b/tests/test_commands.py index 197d80217..6024af71c 100644 --- a/tests/test_commands.py +++ b/tests/test_commands.py @@ -296,12 +296,12 @@ class BadSpider(scrapy.Spider): self.assertIn("badspider.py", log) def test_asyncio_enabled_true(self): - log = self.get_log(self.debug_log_spider, args=['-s', 'ASYNCIO_ENABLED=True']) - self.assertIn("DEBUG: Asyncio support enabled", log) + log = self.get_log(self.debug_log_spider, args=['-s', 'ASYNCIO_REACTOR=True']) + self.assertIn("DEBUG: Asyncio reactor is installed", log) def test_asyncio_enabled_false(self): - log = self.get_log(self.debug_log_spider, args=['-s', 'ASYNCIO_ENABLED=False']) - self.assertNotIn("DEBUG: Asyncio support enabled", log) + log = self.get_log(self.debug_log_spider, args=['-s', 'ASYNCIO_REACTOR=False']) + self.assertNotIn("DEBUG: Asyncio reactor is installed", log) class BenchCommandTest(CommandTest): diff --git a/tests/test_crawler.py b/tests/test_crawler.py index a2865fcd1..fce60ca37 100644 --- a/tests/test_crawler.py +++ b/tests/test_crawler.py @@ -13,7 +13,6 @@ import scrapy from scrapy.crawler import Crawler, CrawlerRunner, CrawlerProcess from scrapy.settings import Settings, default_settings from scrapy.spiderloader import SpiderLoader -from scrapy.utils.asyncio import is_asyncio_reactor_installed from scrapy.utils.log import configure_logging, get_scrapy_root_handler from scrapy.utils.spider import DefaultSpider from scrapy.utils.misc import load_object @@ -209,14 +208,6 @@ class NoRequestsSpider(scrapy.Spider): return [] -class AsyncioSpider(scrapy.Spider): - name = 'asyncio' - - def start_requests(self): - self.logger.info('Asyncio support: %s', is_asyncio_reactor_installed()) - return [] - - @mark.usefixtures('reactor_pytest') class CrawlerRunnerHasSpider(unittest.TestCase): @@ -261,31 +252,32 @@ class CrawlerRunnerHasSpider(unittest.TestCase): self.assertEqual(runner.bootstrap_failed, True) + def test_crawler_runner_asyncio_enabled_true(self): + if self.reactor_pytest == 'asyncio': + runner = CrawlerRunner(settings={'ASYNCIO_REACTOR': True}) + else: + msg = "ASYNCIO_REACTOR is on but the Twisted asyncio reactor is not installed" + with self.assertRaisesRegex(Exception, msg): + runner = CrawlerRunner(settings={'ASYNCIO_REACTOR': True}) + @defer.inlineCallbacks def test_crawler_process_asyncio_enabled_true(self): with LogCapture(level=logging.DEBUG) as log: if self.reactor_pytest == 'asyncio': - runner = CrawlerProcess(settings={'ASYNCIO_ENABLED': True}) + runner = CrawlerProcess(settings={'ASYNCIO_REACTOR': True}) yield runner.crawl(NoRequestsSpider) - self.assertIn("Asyncio support enabled", str(log)) + self.assertIn("Asyncio reactor is installed", str(log)) else: - msg = "ASYNCIO_ENABLED is on but the Twisted asyncio reactor is not installed" + msg = "ASYNCIO_REACTOR is on but the Twisted asyncio reactor is not installed" with self.assertRaisesRegex(Exception, msg): - runner = CrawlerProcess(settings={'ASYNCIO_ENABLED': True}) + runner = CrawlerProcess(settings={'ASYNCIO_REACTOR': True}) @defer.inlineCallbacks def test_crawler_process_asyncio_enabled_false(self): - runner = CrawlerProcess(settings={'ASYNCIO_ENABLED': False}) + runner = CrawlerProcess(settings={'ASYNCIO_REACTOR': False}) with LogCapture(level=logging.DEBUG) as log: yield runner.crawl(NoRequestsSpider) - self.assertNotIn("Asyncio support enabled", str(log)) - - @defer.inlineCallbacks - def test_crawler_runner_asyncio_supported(self): - runner = CrawlerRunner() - with LogCapture() as log: - yield runner.crawl(AsyncioSpider) - log.check_present(('asyncio', 'INFO', 'Asyncio support: %s' % (self.reactor_pytest == 'asyncio'))) + self.assertNotIn("Asyncio reactor is installed", str(log)) class CrawlerProcessSubprocess(unittest.TestCase): @@ -302,14 +294,14 @@ class CrawlerProcessSubprocess(unittest.TestCase): def test_simple(self): log = self.run_script('simple.py') self.assertIn('Spider closed (finished)', log) - self.assertNotIn("DEBUG: Asyncio support enabled", log) + self.assertNotIn("DEBUG: Asyncio reactor is installed", log) def test_asyncio_enabled_no_reactor(self): log = self.run_script('asyncio_enabled_no_reactor.py') self.assertIn('Spider closed (finished)', log) - self.assertIn("DEBUG: Asyncio support enabled", log) + self.assertIn("DEBUG: Asyncio reactor is installed", log) def test_asyncio_enabled_reactor(self): log = self.run_script('asyncio_enabled_reactor.py') self.assertIn('Spider closed (finished)', log) - self.assertIn("DEBUG: Asyncio support enabled", log) + self.assertIn("DEBUG: Asyncio reactor is installed", log)