Пример #1
0
    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):
        """
        :param use_redis: 是否使用redis
        :param debug: 是否开启debug模式
        :param analysis: 如果为true, 会执行爬虫程序,再执行分析程序,如果为false,只执行分析程序
        """
        QQZoneSpider.__init__(self,
                              use_redis,
                              debug,
                              recover=recover,
                              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,
                              pool_flag=pool_flag,
                              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)

        if self.g_tk == 0 and analysis == False:
            self.login()
        self.friend_detail = []
        self.friend_list = []
        self.friend_df = pd.DataFrame()
        self.re = self.connect_redis()
        self.friend_thread_list = []
        self.export_excel = export_excel
        self.export_csv = export_csv
        self.error_friend_num = 0
Пример #2
0
    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'):
        """
        :param use_redis: 是否使用redis
        :param debug: 是否开启debug模式
        :param analysis: 如果为true, 会执行爬虫程序,再执行分析程序,如果为false,只执行分析程序
        """
        QQZoneSpider.__init__(self,
                              use_redis,
                              debug,
                              recover=recover,
                              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,
                              pool_flag=pool_flag)

        if self.g_tk == 0 and analysis == False:
            self.login()
        USER_BASE_DIR = BASE_DIR + self.username + '/'
        util.check_dir_exist(USER_BASE_DIR)
        FRIEND_DIR_HEAD = USER_BASE_DIR + 'friend/'
        self.FRIEND_LIST_FILE_NAME = FRIEND_DIR_HEAD + 'friend_list.json'
        self.FRIEND_DETAIL_FILE_NAME = FRIEND_DIR_HEAD + 'friend_detail.json'
        self.FRIEND_DETAIL_LIST_FILE_NAME = FRIEND_DIR_HEAD + 'friend_detail_list.csv'
        self.FRIEND_DETAIL_EXCEL_FILE_NAME = FRIEND_DIR_HEAD + 'friend_detail_list.xlsx'
        # 头像下载到web的static文件夹,以便在web中调用
        self.FRIEND_HEADER_IMAGE_PATH = BASE_PATH + '/src/web/static/image/header/' + self.username + '/'
        util.check_dir_exist(USER_BASE_DIR + 'friend/')
        util.check_dir_exist(self.FRIEND_HEADER_IMAGE_PATH)
        self.friend_detail = []
        self.friend_list = []
        self.friend_df = pd.DataFrame()
        self.re = self.connect_redis()
        self.friend_thread_list = []
        self.export_excel = export_excel
        self.export_csv = export_csv
Пример #3
0
 def __init__(self, use_redis=False, debug=False, 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):
     QQZoneSpider.__init__(self, use_redis=use_redis, debug=debug,
                           mood_begin=mood_begin, mood_num=mood_num, stop_time=stop_time,
                           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,
                           recover=recover)
     self.friend_name_list = self.get_friend_username()
     self.base_dir = ''