Пример #1
0
 public void GetJobInfoParserTest()
 {
     Category catalog = new Category(); // TODO: 初始化为适当的值
     Get1010Jobs target = new Get1010Jobs(catalog); // TODO: 初始化为适当的值
     string url = "http://sh.1010jz.com/html/shanghai/011_1522688.html"; // TODO: 初始化为适当的值
     Job expected = null; // TODO: 初始化为适当的值
     Job actual;
     ATag node = new ATag();
     node.Link = url;
     actual = target.GetJobInfoParser(url);
     Console.WriteLine("title:{0}", actual.title);
     Console.WriteLine("email:{0}", actual.poster_email);
 }
Пример #2
0
        public void Run()
        {
            bool sign = true;

            IList<Category> catalogs = null;
            IList<City> citys = null;

            while (sign)
            {
                try
                {
                    citys = CategoryMap.GetCitys();
                    catalogs = CategoryMap.GetCategorys();
                    sign = false;
                }
                catch (Exception ex)
                {
                    SpiderEventLog.WriteWarningLog("开始加载目录地址出现数据库错误:"+ex.ToString());
                    sign = true;
                }
            }

            if (catalogs == null)
            {
                return;
            }

            foreach (City c in citys)
            {
                foreach (Category catalog in catalogs)
                {
                    Category newCatalog = new Category();

                    newCatalog.id = catalog.id;
                    newCatalog.name = catalog.name;
                    newCatalog.sp1010 = catalog.sp1010.Replace("sh.", c.sub_domain + ".");
                    newCatalog.city = c;

                    new Get1010Jobs(newCatalog).StartSpider();
                }
            }
        }
Пример #3
0
 public Get1010Jobs(Category catalog)
 {
     Catalog = catalog;
 }