def __init__(self,
use_redis=False,
debug=False,
analysis=True,
recover=False,
username='',
mood_begin=0,
mood_num=-1,
stop_time='-1',
from_web=False,
nickname='',
no_delete=True,
cookie_text='',
export_excel=False,
export_csv=True,
pool_flag='127.0.0.1',
download_small_image=False,
download_big_image=False,
download_mood_detail=True,
download_like_detail=True,
download_like_names=True):
QQZoneFriendSpider.__init__(self,
use_redis=use_redis,
debug=debug,
username=username,
export_csv=export_csv,
mood_begin=mood_begin,
mood_num=mood_num,
stop_time=stop_time,
from_web=from_web,
download_small_image=download_small_image,
download_big_image=download_big_image,
download_mood_detail=download_mood_detail,
download_like_detail=download_like_detail,
download_like_names=download_like_names,
nickname=nickname,
no_delete=no_delete,
cookie_text=cookie_text,
recover=recover,
export_excel=export_excel,
pool_flag=pool_flag,
analysis=analysis)
if not self.from_web:
self.friend_name_list = self.get_friend_username()
else:
self.friend_name_list = []
self.base_dir = ''
def __init__(self,
use_redis=False,
debug=False,
username='',
analysis_friend=False,
mood_begin=0,
mood_num=-1,
stop_time='-1',
from_web=True,
nickname='',
no_delete=True,
cookie_text='',
pool_flag='127.0.0.1'):
QQZoneFriendSpider.__init__(self,
use_redis,
debug,
recover=False,
username=username,
mood_num=mood_num,
mood_begin=mood_begin,
stop_time=stop_time,
from_web=from_web,
nickname=nickname,
no_delete=no_delete,
cookie_text=cookie_text,
analysis=True,
export_excel=True,
export_csv=False,
pool_flag=pool_flag)
self.mood_data = []
self.mood_data_df = pd.DataFrame()
self.like_detail_df = []
self.like_list_names_df = []
self.file_name_head = username
self.analysis_friend = analysis_friend
self.has_clean_data = False
self.friend_dir = BASE_DIR + self.file_name_head + '/friend/' + 'friend_detail_list.csv'
self.history_like_agree_file_name = BASE_DIR + self.file_name_head + '/friend/' + 'history_like_list.json'
self.av = Average(use_redis=False,
file_name_head=username,
analysis=True,
debug=debug)
self.init_analysis_path()
def __init__(self, file_name_head=""):
QQZoneFriendSpider.__init__(self, analysis=True)
if file_name_head != "":
self.file_name_head = file_name_head
self.FRIEND_DETAIL_LIST_FILE_NAME = 'friend/' + self.file_name_head + '_friend_detail_list.csv'
self.data = pd.read_csv(self.FRIEND_DETAIL_LIST_FILE_NAME)
self.remove_waste_index()
print("获取文件：", self.FRIEND_DETAIL_LIST_FILE_NAME)
self.SOURCE_TARGET_FILE_NAME = '../data/' + self.file_name_head + '_source_targets.csv'
self.NODE_FILE_NAME = '../data/' + self.file_name_head + '_node.csv'
self.RELATION_FILE_NAME = '../data/' + self.file_name_head + '_relation.csv'
name = self.data['common_group_names']
group_name = pd.DataFrame([name])
self.people_name = self.data['nick_name'].values
self.gravity = self.data['common_friend_num'].values
self.add_friend_time = self.data['add_friend_time'].values
self.common_group_names = group_name.T
self.group_name_set = set()
self.edge_list = []
self.change_image_url()
self.calculate()
self.format_output()
def test_init(self):
QQZoneFriendSpider()
def test_download_friend_header(self):
fs = QQZoneFriendSpider(use_redis=True, analysis=True)
fs.download_head_image()
print("spend time to wait:", fs.image_thread_pool.time_spend)
def test_get_first_friend_info(self):
fs = QQZoneFriendSpider(use_redis=False, analysis=True)
fs.get_first_friend_info()
print(fs.user_info.first_friend, fs.user_info.first_friend_time)
def test_clean_friend_data(self):
fs = QQZoneFriendSpider(use_redis=False,
analysis=True,
export_csv=False)
fs.clean_friend_data()
print("friend df:", fs.friend_df.shape)
def test_get_friend_detail(self):
fs = QQZoneFriendSpider(use_redis=False, analysis=False, debug=True)
fs.get_friend_detail()
def test_get_friend_list(self):
fs = QQZoneFriendSpider(use_redis=False, analysis=False)
fs.get_friend_list()