/// <summary>
///
/// </summary>
public void start()
{
//new 深圳().start();
//new 广州().start();
new 海().start();
new 北京().start();
//new 贵阳().start();
//new 哈尔滨().start();
//new 海口().start();
//new 合肥().start();
//new 呼和浩特().start();
//new 兰州().start();
//new 南宁().start();
//new 石家庄().start();
//new 太原().start();
//new 西宁().start();
//new 银川().start();
//new 长春().start();
//new 郑州().start();
//new 重庆().start();
//new 昆明().start();
new 天津().start();
List <VIEW_网站爬取配置_城市表_网站表> list = SpiderWebConfigManager.获取新浪二手房下所有城市爬取配置();
foreach (VIEW_网站爬取配置_城市表_网站表 _view in list)
{
NewDataRum exists = NewDataRumList.Find(
delegate(NewDataRum _newDataRum) { return(_newDataRum.CityName.Equals(_view.城市名称)); });
if (exists == null)
{
new 其他城市(_view.城市名称).start(_view.域名, _view.列表页链接, _view.详细页面爬取频率, _view.列表页面爬取频率);
}
}
}
public void start()
{
NowPageEncoding = "gbk";
regex_lpm.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_lpm", 网站名称, "NewDataSpider2"));
regex_xzq.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_xzq", 网站名称, "NewDataSpider2"));
regex_pq.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_pq", 网站名称, "NewDataSpider2"));
regex_mj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_mj", 网站名称, "NewDataSpider2"));
regex_dj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_dj", 网站名称, "NewDataSpider2"));
regex_zj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zj", 网站名称, "NewDataSpider2"));
regex_szlc.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_szlc", 网站名称, "NewDataSpider2"));
regex_zlc.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zlc", 网站名称, "NewDataSpider2"));
regex_hx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_hx", 网站名称, "NewDataSpider2"));
regex_cx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_cx", 网站名称, "NewDataSpider2"));
regex_zx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zx", 网站名称, "NewDataSpider2"));
regex_jznd.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_jznd", 网站名称, "NewDataSpider2"));
regex_title.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_title", 网站名称, "NewDataSpider2"));
regex_phone.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_phone", 网站名称, "NewDataSpider2"));
regex_infUrl.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_infUrl", 网站名称, "NewDataSpider2"));
regex_address.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_address", 网站名称, "NewDataSpider2"));
regex_nextPage.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_nextPage", 网站名称, "NewDataSpider2"));
网站爬取配置 obj = SpiderWebConfigManager.根据城市获取新浪二手房爬取配置("上海");
if (obj != null)
{
NewDataRum newDataRum = new NewDataRum("上海", obj.域名, obj.列表页链接, obj.详细页面爬取频率, obj.列表页面爬取频率);
newDataRum.start(this);
}
}
public void start()
{
网站爬取配置 obj = SpiderWebConfigManager.根据城市获取新浪二手房爬取配置("银川");
if (obj != null)
{
NewDataRum newDataRum = new NewDataRum("银川", obj.域名, obj.列表页链接, obj.详细页面爬取频率, obj.列表页面爬取频率);
newDataRum.start(this);
}
}
public void start()
{
List <网站爬取配置> objlist = SpiderWebConfigManager.根据城市获取搜房网爬取配置("海口");
foreach (网站爬取配置 obj in objlist)
{
NewDataRum newDataRum = new NewDataRum("海口", obj.域名, obj.列表页链接, obj.详细页面爬取频率, obj.列表页面爬取频率, obj.规则编号, obj.主要用途, obj.主要案例类型);
newDataRum.start(this);
}
}
public void start()
{
List <VIEW_网站爬取配置_城市表_网站表> list = SpiderWebConfigManager.获取常州房产网下所有城市爬取配置();
foreach (VIEW_网站爬取配置_城市表_网站表 _view in list)
{
NewDataRum exists = NewDataRumList.Find(
delegate(NewDataRum _newDataRum) { return(_newDataRum.CityName.Equals(_view.城市名称)); });
if (exists == null)
{
new 其他城市(_view.城市名称).start(_view.域名, _view.列表页链接, _view.详细页面爬取频率, _view.列表页面爬取频率);
}
}
}
public void start()
{
//GetHouseByUrl("http://lw.58.com/ershoufang/16131733324033x.shtml", "");
//new 其他城市("深圳").start("http://sz.58.com", "http://sz.58.com/ershoufang/", 2000, 1000);
//return;
List <VIEW_网站爬取配置_城市表_网站表> list = SpiderWebConfigManager.获取五八同城下所有城市爬取配置();
foreach (VIEW_网站爬取配置_城市表_网站表 _view in list)
{
NewDataRum exists = NewDataRumList.Find(
delegate(NewDataRum _newDataRum) { return(_newDataRum.CityName.Equals(_view.城市名称)); });
if (exists == null)
{
new 其他城市(_view.城市名称).start(_view.域名, _view.列表页链接, _view.详细页面爬取频率, _view.列表页面爬取频率);
}
}
}
public void start(string 域名, string 列表页链接, int 详细页面爬取频率, int 列表页面爬取频率)
{
NewDataRum newDataRum = new NewDataRum(CityName, 域名, 列表页链接, 详细页面爬取频率, 列表页面爬取频率);
newDataRum.start(this);
}
public void start()
{
NowPageEncoding = "gbk";
//regex_lpm.RegexInfoList.Add(new RegexInfo("<div class=\"approve_price\">小区名称：<strong>(<a [^<>]*>([^<>]*)</a>|([^<>]+))</strong>", "$2$3"));
//regex_xzq.RegexInfoList.Add(new RegexInfo("<div class=\"approve_price\">小区名称：<strong>(?:(?!</strong>).)*</strong> （<strong>([^<>\\-]*)\\-\\-[^<>\\-]*</strong>）", "$1"));
//regex_pq.RegexInfoList.Add(new RegexInfo("<div class=\"approve_price\">小区名称：<strong>(?:(?!</strong>).)*</strong> （<strong>[^<>\\-]*\\-\\-([^<>\\-]*)</strong>）", "$1"));
//regex_mj.RegexInfoList.Add(new RegexInfo("<li class=\"w400\"><strong>面积：([\\d\\.]*)㎡</strong></li>", "$1"));
//regex_dj.RegexInfoList.Add(new RegexInfo("<li class=\"w240\"><strong>单价：([\\d\\.]*)元/㎡</strong>", "$1"));
//regex_zj.RegexInfoList.Add(new RegexInfo("<li class=\"w240\"><strong>总价：([\\d\\.]*)万元</strong></li>", "$1"));
//regex_szlc.RegexInfoList.Add(new RegexInfo("<li class=\"w240\">楼层：第([\\d]*)层（共[\\d]*层）</li>", "$1"));
//regex_zlc.RegexInfoList.Add(new RegexInfo("<li class=\"w240\">楼层：第[\\d\\-]*层（共([\\d]*)层）</li>", "$1"));
//regex_hx.RegexInfoList.Add(new RegexInfo("<li class=\"w400\"><strong>户型：([^<>]*)</strong></li>", "$1"));
//regex_cx.RegexInfoList.Add(new RegexInfo("<li class=\"w240\">朝向：([^<>]*)</li>", "$1"));
//regex_zx.RegexInfoList.Add(new RegexInfo("<li class=\"w400\">装修：([^<>]*)</li>", "$1"));
//regex_jznd.RegexInfoList.Add(new RegexInfo("<li class=\"w400\">建筑年代：(\\d*)</li>", "$1"));
//regex_title.RegexInfoList.Add(new RegexInfo("<div class=\"approvecot_titbg\"><span>([^<>]*)</span></div>", "$1"));
//regex_phone.RegexInfoList.Add(new RegexInfo("<span class=\"telbg\"><strong>([^<>]*)</strong></span>", "$1"));
//regex_infUrl.RegexInfoList.Add(new RegexInfo("<div class=\"search_font_line1_tit_list\"><a href=\"([^\"]*)\" target=\"_blank\" title=\"[^\"]*\">[^<>]*</a></div>", "$1"));
//regex_address.RegexInfoList.Add(new RegexInfo("<li class=\"w650\">地址：([^<>]*)</li>", "$1"));
//regex_nextPage.RegexInfoList.Add(new RegexInfo("<a href=\"([^\"]+)\" class=\"nextpage\">下一页</a>", "$1"));
regex_lpm.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_lpm", 网站名称, cityName: "NewDataSpider2"));
regex_xzq.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_xzq", 网站名称, cityName: "NewDataSpider2"));
regex_pq.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_pq", 网站名称, cityName: "NewDataSpider2"));
regex_mj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_mj", 网站名称, cityName: "NewDataSpider2"));
regex_dj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_dj", 网站名称, cityName: "NewDataSpider2"));
regex_zj.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zj", 网站名称, cityName: "NewDataSpider2"));
regex_szlc.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_szlc", 网站名称, cityName: "NewDataSpider2"));
regex_zlc.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zlc", 网站名称, cityName: "NewDataSpider2"));
regex_hx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_hx", 网站名称, cityName: "NewDataSpider2"));
regex_cx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_cx", 网站名称, cityName: "NewDataSpider2"));
regex_zx.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_zx", 网站名称, cityName: "NewDataSpider2"));
regex_jznd.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_jznd", 网站名称, cityName: "NewDataSpider2"));
regex_title.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_title", 网站名称, cityName: "NewDataSpider2"));
regex_phone.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_phone", 网站名称, cityName: "NewDataSpider2"));
regex_infUrl.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_infUrl", 网站名称, cityName: "NewDataSpider2"));
regex_address.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_address", 网站名称, cityName: "NewDataSpider2"));
regex_nextPage.RegexInfoList.Add(SpiderRegexInfoHelp.GetRegexInfoByXmlName("regex_nextPage", 网站名称, cityName: "NewDataSpider2"));
#region 生成xml
//StringBuilder stest = new StringBuilder();
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_lpm.RegexInfoList[regex_lpm.RegexInfoList.Count-1], "regex_lpm", "楼盘名"));
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_xzq.RegexInfoList[regex_xzq.RegexInfoList.Count - 1], "regex_xzq", "行政区")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_pq.RegexInfoList[regex_pq.RegexInfoList.Count - 1], "regex_pq", "片区")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_hx.RegexInfoList[regex_hx.RegexInfoList.Count - 1], "regex_hx", "户型")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_mj.RegexInfoList[regex_mj.RegexInfoList.Count - 1], "regex_mj", "面积")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_dj.RegexInfoList[regex_dj.RegexInfoList.Count - 1], "regex_dj", "单价")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_zj.RegexInfoList[regex_zj.RegexInfoList.Count - 1], "regex_zj", "总价")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_jznd.RegexInfoList[regex_jznd.RegexInfoList.Count - 1], "regex_jznd", "建筑年代")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_cx.RegexInfoList[regex_cx.RegexInfoList.Count - 1], "regex_cx", "朝向")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_szlc.RegexInfoList[regex_szlc.RegexInfoList.Count - 1], "regex_szlc", "所在楼层")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_zlc.RegexInfoList[regex_zlc.RegexInfoList.Count - 1], "regex_zlc", "总楼层")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_zx.RegexInfoList[regex_zx.RegexInfoList.Count - 1], "regex_zx", "装修")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_title.RegexInfoList[regex_title.RegexInfoList.Count - 1], "regex_title", "信息")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_phone.RegexInfoList[regex_phone.RegexInfoList.Count - 1], "regex_phone", "电话")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_address.RegexInfoList[regex_address.RegexInfoList.Count - 1], "regex_address", "地址")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_infUrl.RegexInfoList[regex_infUrl.RegexInfoList.Count - 1], "regex_infUrl", "url")); ;
//stest.Append(SpiderRegexInfoHelp.GetRegexInfoXmlByObj(regex_nextPage.RegexInfoList[regex_nextPage.RegexInfoList.Count - 1], "regex_nextPage", "下一页正则"));
//string str = stest.ToString();
#endregion
网站爬取配置 obj = SpiderWebConfigManager.根据城市获取新浪二手房爬取配置("北京");
if (obj != null)
{
NewDataRum newDataRum = new NewDataRum("北京", obj.域名, obj.列表页链接, obj.详细页面爬取频率, obj.列表页面爬取频率);
newDataRum.start(this);
}
}
public void start(object runObj, string 域名, string 列表页链接, int 详细页面爬取频率, int 列表页面爬取频率, int?_RegexNumber, int?_BasePurposeCode, int?_BaseCaseTypeCode)
{
NewDataRum newDataRum = new NewDataRum(CityName, 域名, 列表页链接, 详细页面爬取频率, 列表页面爬取频率, _RegexNumber, _BasePurposeCode, _BaseCaseTypeCode);
newDataRum.start(runObj);
}
