__init__.py 1.9 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556
  1. from os.path import realpath, dirname, splitext, join
  2. from os import listdir
  3. from imp import load_source
  4. import grequests
  5. engine_dir = dirname(realpath(__file__))
  6. engines = {}
  7. for filename in listdir(engine_dir):
  8. modname = splitext(filename)[0]
  9. if filename.startswith('_') or not filename.endswith('.py'):
  10. continue
  11. filepath = join(engine_dir, filename)
  12. engine = load_source(modname, filepath)
  13. if not hasattr(engine, 'request') or not hasattr(engine, 'response'):
  14. continue
  15. engines[modname] = engine
  16. def default_request_params():
  17. return {'method': 'GET', 'headers': {}, 'data': {}, 'url': ''}
  18. def make_callback(engine_name, results, callback):
  19. def process_callback(response, **kwargs):
  20. for result in callback(response):
  21. result['engine'] = engine_name
  22. results.append(result)
  23. return process_callback
  24. def search(query, request, selected_engines):
  25. global engines
  26. requests = []
  27. results = []
  28. user_agent = request.headers.get('User-Agent', '')
  29. for ename, engine in engines.items():
  30. if ename not in selected_engines:
  31. continue
  32. headers = default_request_params()
  33. headers['User-Agent'] = user_agent
  34. request_params = engine.request(query, headers)
  35. callback = make_callback(ename, results, engine.response)
  36. if request_params['method'] == 'GET':
  37. req = grequests.get(request_params['url']
  38. ,headers=headers
  39. ,hooks=dict(response=callback)
  40. )
  41. else:
  42. req = grequests.post(request_params['url']
  43. ,data=request_params['data']
  44. ,headers=headers
  45. ,hooks=dict(response=callback)
  46. )
  47. requests.append(req)
  48. grequests.map(requests)
  49. return results