__init__.py 2.1 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859
  1. from os.path import realpath, dirname, splitext, join
  2. from os import listdir
  3. from imp import load_source
  4. import grequests
  5. from itertools import izip_longest, chain
  6. engine_dir = dirname(realpath(__file__))
  7. engines = {}
  8. for filename in listdir(engine_dir):
  9. modname = splitext(filename)[0]
  10. if filename.startswith('_') or not filename.endswith('.py'):
  11. continue
  12. filepath = join(engine_dir, filename)
  13. engine = load_source(modname, filepath)
  14. if not hasattr(engine, 'request') or not hasattr(engine, 'response'):
  15. continue
  16. engines[modname] = engine
  17. def default_request_params():
  18. return {'method': 'GET', 'headers': {}, 'data': {}, 'url': ''}
  19. def make_callback(engine_name, results, callback):
  20. def process_callback(response, **kwargs):
  21. cb_res = []
  22. for result in callback(response):
  23. result['engine'] = engine_name
  24. cb_res.append(result)
  25. results[engine_name] = cb_res
  26. return process_callback
  27. def search(query, request, selected_engines):
  28. global engines
  29. requests = []
  30. results = {}
  31. user_agent = request.headers.get('User-Agent', '')
  32. for ename, engine in engines.items():
  33. if ename not in selected_engines:
  34. continue
  35. headers = default_request_params()
  36. headers['User-Agent'] = user_agent
  37. request_params = engine.request(query, headers)
  38. callback = make_callback(ename, results, engine.response)
  39. if request_params['method'] == 'GET':
  40. req = grequests.get(request_params['url']
  41. ,headers=headers
  42. ,hooks=dict(response=callback)
  43. )
  44. else:
  45. req = grequests.post(request_params['url']
  46. ,data=request_params['data']
  47. ,headers=headers
  48. ,hooks=dict(response=callback)
  49. )
  50. requests.append(req)
  51. grequests.map(requests)
  52. return list(filter(None, chain(*izip_longest(*results.values()))))