def __init__(self, solr_url, reconnect_maxcnt=2):
""" eg : http://192.168.201.63:10000/solr/select? """
self.solr_url = solr_url
self.timeHelper = TimeHelper()
self.reconnect_cnt = 0
self.reconnect_maxcnt = reconnect_maxcnt
self.addressType = self.addressIsList()
def accessSolrByMonth(self, query, month, prows, raw_fq=[], raw_param={}):
"""
#提供按月搜索新闻,使搜到的新闻尽可能的分散
# @param query :
# @param month : 搜索的月数
# @param prows : 每个月搜索量
# 其余参数同上
# return ： 同上 list
"""
reslist = []
nowtime = time.time()
month_day = 30
timeHelper = TimeHelper()
for x in xrange(month):
end_time = nowtime - x * month_day
current_end_day = timeHelper.getDateTimeFromSeconds(
end_time, "%Y-%m-%d")
flag, curreslist = self.process(query,
rows=prows,
last_day=month_day,
end_day=current_end_day,
fq=raw_fq,
solr_param=raw_param)
if flag:
reslist.extend(curreslist)
return reslist
def getPos(self):
return self.offset
pass
def getAverageTime(self):
pass_time = time.time() - self.start_time
result_cnt = self.offset - self.org_offset
if result_cnt == 0:
return 0
return pass_time / (1. * result_cnt)
pass
if __name__ == "__main__":
#设置搜索时间
time_helper = TimeHelper()
time_start = time_helper.getSecondsFromDateTime("2011-03-01 00:00")
time_end = time.mktime(time.localtime())
config_dict = {"url":"http://192.168.201.87:8888/search", \
"channel":"report", \
"other_param":{"sde":"0"}, \
"solr":{ "UID":"", "contentsize":"1"},\
"desc":{"title":"", "content":""}, \
"time_start": int(time_start), \
"time_end": int(time_end),\
"max_number":500,\
"q": "烟台万华" }
online_news_reader = OnlineTimeReader()
online_news_reader.open(config_dict)
result_dict = [] #存放结果 每个结果是一个字典，key是 solr 和 desc指定的内容
next_meta = {}
