middleware.py 2.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475
  1. from collections import defaultdict, deque
  2. import logging
  3. import pprint
  4. from scrapy.exceptions import NotConfigured
  5. from scrapy.utils.misc import create_instance, load_object
  6. from scrapy.utils.defer import process_parallel, process_chain, process_chain_both
  7. logger = logging.getLogger(__name__)
  8. class MiddlewareManager(object):
  9. """Base class for implementing middleware managers"""
  10. component_name = 'foo middleware'
  11. def __init__(self, *middlewares):
  12. self.middlewares = middlewares
  13. self.methods = defaultdict(deque)
  14. for mw in middlewares:
  15. self._add_middleware(mw)
  16. @classmethod
  17. def _get_mwlist_from_settings(cls, settings):
  18. raise NotImplementedError
  19. @classmethod
  20. def from_settings(cls, settings, crawler=None):
  21. mwlist = cls._get_mwlist_from_settings(settings)
  22. middlewares = []
  23. enabled = []
  24. for clspath in mwlist:
  25. try:
  26. mwcls = load_object(clspath)
  27. mw = create_instance(mwcls, settings, crawler)
  28. middlewares.append(mw)
  29. enabled.append(clspath)
  30. except NotConfigured as e:
  31. if e.args:
  32. clsname = clspath.split('.')[-1]
  33. logger.warning("Disabled %(clsname)s: %(eargs)s",
  34. {'clsname': clsname, 'eargs': e.args[0]},
  35. extra={'crawler': crawler})
  36. logger.info("Enabled %(componentname)ss:\n%(enabledlist)s",
  37. {'componentname': cls.component_name,
  38. 'enabledlist': pprint.pformat(enabled)},
  39. extra={'crawler': crawler})
  40. return cls(*middlewares)
  41. @classmethod
  42. def from_crawler(cls, crawler):
  43. return cls.from_settings(crawler.settings, crawler)
  44. def _add_middleware(self, mw):
  45. if hasattr(mw, 'open_spider'):
  46. self.methods['open_spider'].append(mw.open_spider)
  47. if hasattr(mw, 'close_spider'):
  48. self.methods['close_spider'].appendleft(mw.close_spider)
  49. def _process_parallel(self, methodname, obj, *args):
  50. return process_parallel(self.methods[methodname], obj, *args)
  51. def _process_chain(self, methodname, obj, *args):
  52. return process_chain(self.methods[methodname], obj, *args)
  53. def _process_chain_both(self, cb_methodname, eb_methodname, obj, *args):
  54. return process_chain_both(self.methods[cb_methodname], \
  55. self.methods[eb_methodname], obj, *args)
  56. def open_spider(self, spider):
  57. return self._process_parallel('open_spider', spider)
  58. def close_spider(self, spider):
  59. return self._process_parallel('close_spider', spider)