def __init__(self):
conf = ConfigCenter.FileConfig()
self.__path = conf
self.year, self.month, day = Time.now_ymd()
self.lasty, self.lastm, _ = Time.lastday_ymd()
if day == "01":
self._new_month_do_()
def __parse_html__(self, html_data):
html_data_iter = self.__html_data_iter__(html_data)
try:
next(html_data_iter)
except Exception:
raise ("Not a list.")
dict_data_list = list()
for html_data in html_data_iter:
hotkey_emoji = re.findall(
"""<a href=\"/weibo\?q=.+;Refer=top\" target=\"_blank\">(.+)</a>""",
html_data)
flag = re.findall("""<i class=\"icon-txt .+\">(.+)</i></td>""",
html_data)
amount = re.findall("""<span>([0-9]+)</span>""", html_data)
if len(hotkey_emoji) > 0 and "荐" not in flag:
hotkey = re.findall("(.+)<img", hotkey_emoji[0])
emoji = list()
if len(hotkey) == 0:
hotkey = hotkey_emoji[0]
else:
hotkey = hotkey[0]
emoji = re.findall("<img.+alt=\"\[(.+)\]\"",
hotkey_emoji[0])
dict_data_list.append(
dict(hotkey=hotkey,
flag=flag,
amount=amount[0],
emoji=emoji,
time=Time.now_timestamp()))
return dict_data_list
def push_data(self):
conf = ConfigCenter.GitConfig()
filepath = conf['filepath'].split(',')
msg = "Auto commit for %s." % (Time.now_date_str())
err = self.push(msg, *filepath)
if err is not None:
print(err)
# -*- coding: utf-8 -*-
# 监控机器人，当出现爆炸新闻或者出现监控词的时候会自动推送到群内。
from cronjob.crawler.hotkey_fetcher import HotkeyFetcher
from cronjob.datacenter.cache.cache_controller import CacheController
from util.configcenter.config_center import ConfigCenter
from cronjob.tool.qywx import QYWX, MSG_TYPE_MARKDOWN
from util.constant.enum import *
from util.base.time import Time
if __name__ == "__main__":
now = Time.now_timestamp()
fetcher = HotkeyFetcher()
data = fetcher.get_data(ENUM_DATATYPE_JSON, limit=10)
# cacheController = CacheController()
# cacheController.insert_cache(now, data)
print(now, data)