def capture_main_data():
print(sys.path)
sp = QQZoneSpider(use_redis=True, debug=True, download_small_image=False, download_big_image=False)
sp.login_with_qr_code()
sp.get_main_page_info()
sp.get_mood_list()
sp.user_info.save_user(sp.username)
def get_user_basic_info():
sp = QQZoneSpider(use_redis=True, debug=False, mood_begin=0, mood_num=-1,
stop_time='2015-06-01',
download_small_image=False, download_big_image=False,
download_mood_detail=True, download_like_detail=True,
download_like_names=True, recover=False, cookie_text=None)
return sp.user_info
def test_step(self):
sp = QQZoneSpider(use_redis=False, debug=False)
step1 = sp.find_best_step(1100, 5)
step2 = sp.find_best_step(1222, 5)
step3 = sp.find_best_step(2222, 10)
assert step1 == 220
assert step2 == 240
assert step3 == 220
def capture_main_data():
sp = QQZoneSpider(use_redis=True,
debug=False,
download_small_image=False,
download_big_image=False)
sp.login()
sp.get_main_page_info()
sp.get_mood_list()
sp.user_info.save_user(sp.username)
def test_capture_info(self):
sp = QQZoneSpider(use_redis=True,
debug=False,
mood_num=20,
download_mood_detail=False,
download_like_detail=True,
download_like_names=False)
sp.login()
sp.get_mood_list()
def test_download_image(self):
sp = QQZoneSpider(use_redis=False,
debug=False,
mood_num=100,
download_mood_detail=False,
download_like_detail=False,
download_like_names=False,
download_big_image=True,
download_small_image=True)
sp.login()
sp.get_mood_list()
print("Thread Wait:", sp.image_thread_pool.time_spend)
def test_step():
sp = QQZoneSpider(use_redis=True,
debug=True,
mood_begin=0,
mood_num=1000,
stop_time='-1',
download_small_image=False,
download_big_image=False,
download_mood_detail=True,
download_like_detail=True,
download_like_names=True,
recover=False,
cookie_text=None)
sp.find_best_step(1100, 5)
sp.find_best_step(1222, 5)
sp.find_best_step(2222, 10)
def test_extract_tid(self):
sp = QQZoneSpider(use_redis=False,
debug=False,
mood_num=20,
download_mood_detail=False,
download_like_detail=True,
download_like_names=False)
str1 = 'http://user.qzone.qq.com/1272082503/mood/4770d24b00acd95c13e00500.1<.>http://user.qzone.qq.com/1272082503/mood/4770d24b00acd95c13e00500.1'
str2 = 'http://user.qzone.qq.com/1272082503/mood/4770d24bbb83d95c05600900.1^||^https://b339.photo.store.qq.com/psb?/91b7c939-e3c7-41d8-a450-626a11d7ce19/4iKMKW.M8phXT8bghzPgKjz.rdkmby5vOiOJi4PqSlk!/b/dFMBAAAAAAAA&bo=7gI2Be4CNgURECc!^||^0'
str3 = 'http://user.qzone.qq.com/1272082503/mood/4770d24b5c91d65cf9af0b00.1<.>http://user.qzone.qq.com/1272082503/mood/4770d24b5c91d65cf9af0b00.1'
expect1 = '4770d24b00acd95c13e00500'
expect2 = '4770d24bbb83d95c05600900'
expect3 = '4770d24b5c91d65cf9af0b00'
assert expect1 == sp.extract_tid_from_unikey(str1)
assert expect2 == sp.extract_tid_from_unikey(str2)
assert expect3 == sp.extract_tid_from_unikey(str3)
def capture_data():
sp = QQZoneSpider(use_redis=True,
debug=True,
mood_begin=0,
mood_num=-1,
stop_time='-1',
download_small_image=False,
download_big_image=False,
download_mood_detail=True,
download_like_detail=True,
download_like_names=True,
recover=False,
cookie_text=None)
sp.login()
sp.get_main_page_info()
sp.get_mood_list()
sp.user_info.save_user(sp.username)
def test_get_main_page(self):
sp = QQZoneSpider(use_redis=False, debug=True)
sp.login()
sp.get_main_page_info()
def test_selenium_login(self):
sp = QQZoneSpider(use_redis=False, debug=False)
sp.login()
def test_simple_qr_login(self):
sp = QQZoneSpider(use_redis=False, debug=True, mood_num=200)
sp.login_with_qr_code()
def __init__(self):
self.sp = QQZoneSpider(use_redis=False,
debug=False,
from_client=True,
mood_num=20)
warm_tip = "****************************************\n" \
"**************QQ空间抽奖小程序***************\n" \
"****************************************"
self.output(warm_tip)
self.output("请输入获取最近访客的时间间隔，默认为60秒")
time_step = input()
try:
time_step = int(time_step)
except:
time_step = 60
visit_file_name = "最近访客" + get_full_time_from_mktime(int(
time.time())) + ".xlsx"
self.output("最近访客文件名:" + visit_file_name)
try:
self.sp.login_with_qr_code()
self.output("用户" + self.sp.username + "登陆成功！")
except BaseException:
self.output("用户登陆失败！请检查网络连接或稍后再试！")
exit(1)
visit_t = threading.Thread(target=self.sp.parse_recent_visit,
args=[visit_file_name, time_step])
visit_t.start()
self.output("正在获取最近的说说...")
url_mood = self.sp.get_mood_url()
url_mood = url_mood + '&uin=' + str(self.sp.username)
self.content_list = self.get_content_list(url_mood, 0)
self.content_list += self.get_content_list(url_mood, 20)
self.content_list += self.get_content_list(url_mood, 40)
self.output('------------------------')
self.output("最近的60条说说:")
for item in self.content_list:
content = item['content']
if len(content) > 20:
content = content[0:20] + '。。。'
item_str = '|' + str(item['order']) + '|' + content
self.output(item_str)
while True:
try:
self.output('------------------------')
self.output("**以下输入请全部只输入数字！按回车键结束输入！**")
self.output("**输入Q退出本程序**")
is_digit = False
while not is_digit:
self.output('请输入您要选择的说说序号:')
mood_order = input()
is_digit = self.check_input(mood_order)
mood_order = int(mood_order)
if mood_order > len(self.content_list):
pos = (math.ceil(mood_order / 20) - 1) * 20
mood_order = mood_order % 20
t1 = threading.Thread(target=self.get_all,
args=(url_mood, pos, mood_order))
t1.setDaemon(True)
t1.start()
else:
unikey = self.content_list[mood_order - 1]
key = unikey['unikey']
tid = unikey['tid']
t2 = threading.Thread(target=self.start_like_cmt_thread,
args=(key, tid))
t2.setDaemon(True)
t2.start()
is_digit = False
while not is_digit:
self.output('请输入抽奖的类型，1-点赞；2-评论；(其它)-我全都要！ ：')
type = input()
is_digit = self.check_input(type)
is_digit = False
while not is_digit:
self.output('请选择抽奖的用户数量：')
user_num = input()
is_digit = self.check_input(user_num)
# 等待线程运行完
try:
t1.join()
t2.join()
self.like_t.join()
self.cmt_t.join()
except:
pass
self.type = int(type)
self.user_num = int(user_num)
self.file_name = self.content_list[mood_order - 1]['content']
if len(self.file_name) > 10:
self.file_name = self.file_name[:10]
self.file_name = re.sub('[^\w\u4e00-\u9fff]+', '',
self.file_name)
if len(self.file_name) <= 0:
self.file_name = str(mood_order)
print("说说:", self.file_name)
self.do_raffle()
cmt_df = pd.DataFrame(self.cmt_list)
like_df = pd.DataFrame(self.like_list)
cmt_df.to_excel("评论-" + self.file_name + '.xlsx', index=False)
like_df.to_excel("点赞-" + self.file_name + '.xlsx', index=False)
except BaseException as e:
if str(e) == '2':
exit(2)
else:
self.output('----------------------')
print(e)
self.output("未知错误，请重新尝试")
self.output('----------------------')
def test_get_first_mood(self):
sp = QQZoneSpider(use_redis=False, debug=True)
sp.login_with_qr_code()
mood_num = sp.get_mood_num()
sp.get_first_mood(mood_num)
def test_capture_info_parallel(self):
sp = QQZoneSpider(use_redis=True, debug=False, mood_num=200)
sp.login()
sp.get_mood_list()
def test_login_with_qr_code(self):
sp = QQZoneSpider(use_redis=True, debug=True, mood_num=200)
sp.login_with_qr_code()
print("Login success")
sp.get_main_page_info()
sp.get_mood_list()
def test_capture_main_data(self):
sp = QQZoneSpider(use_redis=True, debug=False)
sp.login()
sp.get_main_page_info()
sp.get_mood_list()
def test_login_with_qr_code(self):
sp = QQZoneSpider(use_redis=True, debug=True)
sp.login_with_qr_code()