scheduler.py 文件源码

def __init__(self, original_url=None, timeout=2, search=(), ssl_context=None, url_check=True,
                 similarity_threshold=0.6,
                 display_number=100):
        # Parse chinese to ascii and delete parameters.
        self.url = original_url
        if self.url:
            self.url_check = url_check
            self._check_url()
            self.host = self.url.split('//')[0] + '//' + self.url.split('//')[1].split('/')[0]
        # Create ssl context.
        if ssl_context:
            self.ssl_context = ssl_context
        else:
            self.ssl_context = _create_unverified_context()
        # Initialization parameters.
        self.temp_file_name = 'mini-spider.temp'
        self.timeout = timeout
        self.similarity_threshold = similarity_threshold
        self.pattern_list = []
        self.search_list = self._initialize_search(search)
        self.display_number = display_number
        self.result = []
        self.http_flag = 0