Python.Scrapy.14-scrapy-source-code-analysis-part-4
Scrapy 源代码分析系列－4 scrapy.commands 子包
子包scrapy.commands定义了在命令scrapy中使用的子命令(subcommand): bench, check, crawl, deploy, edit, fetch,
genspider, list, parse, runspider, settings, shell, startproject, version, view。 所有的子命令模块都定义了一个继承自
类ScrapyCommand的子类Command。
首先来看一下子命令crawl, 该子命令用来启动spider。
1. crawl.py
关注的重点在方法run(self, args, opts):
1 def run(self, args, opts): 2 if len(args) < 1: 3 raise UsageError() 4 elif len(args) > 1: 5 raise UsageError("running 'scrapy crawl' with more than one spider is no longer supported") 6 spname = args[0] 7 8 crawler = self.crawler_process.create_crawler() # A 9 spider = crawler.spiders.create(spname, **opts.spargs) # B 10 crawler.crawl(spider) ＃ C 11 self.crawler_process.start() # D
那么问题来啦，run接口方法是从哪里调用的呢? 让我们回到 Python.Scrapy.11-scrapy-source-code-analysis-part-1
中 "1.2 cmdline.py command.py" 关于"_run_print_help() "的说明。
A: 创建类Crawler对象crawler。在创建Crawler对象时, 同时将创建Crawler对象的实例属性spiders(SpiderManager)。如下所示:
1 class Crawler(object): 2 3 def __init__(self, settings): 4 self.configured = False 5 self.settings = settings 6 self.signals = SignalManager(self) 7 self.stats = load_object(settings['STATS_CLASS'])(self) 8 self._start_requests = lambda: () 9 self._spider = None 10 # TODO: move SpiderManager to CrawlerProcess 11 spman_cls = load_object(self.settings['SPIDER_MANAGER_CLASS']) 12 self.spiders = spman_cls.from_crawler(self) # spiders 的类型是: SpiderManager
Crawler对象对应一个SpiderManager对象，而SpiderManager对象管理多个Spider。
B: 获取Sipder对象。
C: 为Spider对象安装Crawler对象。(为蜘蛛安装爬行器)
D: 类CrawlerProcess的start()方法如下:
1 def start(self): 2 if self.start_crawling(): 3 self.start_reactor() 4 5 def start_crawling(self): 6 log.scrapy_info(self.settings) 7 return self._start_crawler() is not None 8 9 def start_reactor(self): 10 if self.settings.getbool('DNSCACHE_ENABLED'): 11 reactor.installResolver(CachingThreadedResolver(reactor)) 12 reactor.addSystemEventTrigger('before', 'shutdown', self.stop) 13 reactor.run(installSignalHandlers=False) # blocking call 14 15 def _start_crawler(self): 16 if not self.crawlers or self.stopping: 17 return 18 19 name, crawler = self.crawlers.popitem() 20 self._active_crawler = crawler 21 sflo = log.start_from_crawler(crawler) 22 crawler.configure() 23 crawler.install() 24 crawler.signals.connect(crawler.uninstall, signals.engine_stopped) 25 if sflo: 26 crawler.signals.connect(sflo.stop, signals.engine_stopped) 27 crawler.signals.connect(self._check_done, signals.engine_stopped) 28 crawler.start() ＃ 调用类Crawler的start()方法 29 return name, crawler
类Crawler的start()方法如下:
1 def start(self): 2 yield defer.maybeDeferred(self.configure) 3 if self._spider: 4 yield self.engine.open_spider(self._spider, self._start_requests()) # 和Engine建立了联系 （ExecutionEngine） 5 yield defer.maybeDeferred(self.engine.start)
关于类ExecutionEngine将在子包scrapy.core分析涉及。
2. startproject.py
3. subcommand是如何加载的
在cmdline.py的方法execute()中有如下几行代码:
1 inproject = inside_project() 2 cmds = _get_commands_dict(settings, inproject) 3 cmdname = _pop_command_name(argv)
_get_commands_dict():
1 def _get_commands_dict(settings, inproject): 2 cmds = _get_commands_from_module('scrapy.commands', inproject) 3 cmds.update(_get_commands_from_entry_points(inproject)) 4 cmds_module = settings['COMMANDS_MODULE'] 5 if cmds_module: 6 cmds.update(_get_commands_from_module(cmds_module, inproject)) 7 return cmds
_get_commands_from_module():
1 def _get_commands_from_module(module, inproject): 2 d = {} 3 for cmd in _iter_command_classes(module): 4 if inproject or not cmd.requires_project: 5 cmdname = cmd.__module__.split('.')[-1] 6 d[cmdname] = cmd() 7 return d
To Be Continued
接下来解析settings相关的逻辑。Python.Scrapy.15-scrapy-source-code-analysis-part-5
