Exemplo n.º 1
0
def start(config_dir):
    """ Start all the process
    """
    websites = config.load_websites(os.path.join(config_dir, "websites.json"))

    for website in websites:
        webconfigs = config.load_pages(os.path.join(config_dir, website["config"]))
        for webconfig in webconfigs:
            if not database.check_table(webconfig["model"]):
                database.initialize_table(webconfig["model"])
        #get_data_link.delay(website["start_url"], website["start_model"], webconfigs)
        get_data_link(website["start_url"], website["start_model"], webconfigs)
    pass
Exemplo n.º 2
0
 def testTasks(self):
     pages_config = config.load_pages("../config/skyscanner.json")
     model = "skyscanner_flight"
     test_page = "http://www.skyscanner.net/flights-to/lond/cheap-flights-to-london.html"
     tasks.get_data_link(test_page, model, pages_config)