json yeid_如何控制Scrapy yield的顺序

我想从一个json接口爬点数据 http://china.fathom.info/data/data.json，比如看这个代码，如果yield放在parse里，我想当然的认为应该是for循环完yield Request再yield group_item，然而这样parse_member里的东西就不会存进group_item了，但如果yield group_item放在parse_member里，数据又会变成

{'A':1, 'members':[{'id':11}]}, {'A':1, 'members':[{'id':22}]}

而不是

{'A':1, 'members':[{'id':11}, {'id':22}]}

请问该如何解决这个问题。如果用urllib2顺序写一下比较好理解，但如果非得用scrapy呢？感觉自己对scrapy的request的请求顺序理解得很混乱。

start_urls = [

"http://china.fathom.info/data/data.json"

]

def parse(self, response):

groups = json.loads(response.body)['group_members']

for i in groups:

group_item = GroupItem()

group_item['name'] = groups[i]['name']

group_item['chinese'] = groups[i]['chinese']

group_item['members'] = []

members = groups[i]['members']

for member in members:

yield Request(self.person_url % member['id'], meta={'group_item': group_item, 'member': member},

callback=self.parse_member, priority=100)

yield group_item

def parse_member(self, response):

group_item = response.meta['group_item']

member = response.meta['member']

person = json.loads(response.body)

ego = person['ego']

group_item['members'].append({

'id': ego['id'],

'name': ego['name'],

'chinese': ego['chinese'],

'role': member['role']

})