diff --git a/New_College.Tasks/HtmlAgSpider/NationWideNewsAgHelper.cs b/New_College.Tasks/HtmlAgSpider/NationWideNewsAgHelper.cs index a3e73a6..bd0fed9 100644 --- a/New_College.Tasks/HtmlAgSpider/NationWideNewsAgHelper.cs +++ b/New_College.Tasks/HtmlAgSpider/NationWideNewsAgHelper.cs @@ -11,6 +11,7 @@ using System.Threading.Tasks; using System.Text.Json; using New_College.Common.Helper; using System.Threading; +using System.Text.RegularExpressions; namespace New_College.Tasks { public class NationWideNewsAgHelper @@ -35,7 +36,7 @@ namespace New_College.Tasks { resultlist.msg.ForEach(o => { - Thread.Sleep(100); + Thread.Sleep(100); string newsUrl = $"{itemUrl}{o.uri}"; HtmlDocument doc = webClient.Load(newsUrl); var inntertitle = doc.DocumentNode.SelectSingleNode("//*[@id=\"app\"]/div[2]/h2").InnerText; @@ -54,19 +55,19 @@ namespace New_College.Tasks //if (!HtmlHelper.ReplaceHtmlTag(innerhtml).Contains("浏览器")) //{ - if (innerhtml.Contains(".pdf") || innerhtml.Contains(".doc") || innerhtml.Contains(".docx") || innerhtml.Contains(".xls") || innerhtml.Contains(".xlsx")) - { - innerhtml = $"{innerhtml}\n若有附件详情,请至本省招生考试院下载附件!!!"; - } + if (innerhtml.Contains(".pdf") || innerhtml.Contains(".doc") || innerhtml.Contains(".docx") || innerhtml.Contains(".xls") || innerhtml.Contains(".xlsx")) + { + innerhtml = $"{innerhtml}\n若有附件详情,请至本省招生考试院下载附件!!!"; + } - list.Add(new NewsModels() - { - title = inntertitle, - author = author, - pubtime = Convert.ToDateTime(createtime), - detail = innerhtml.Replace("src=\"", "src=\"https://gaokao.chsi.com.cn") - }); - // } + list.Add(new NewsModels() + { + title = inntertitle, + author = author, + pubtime = Convert.ToDateTime(createtime), + detail = Regex.Replace(innerhtml, @"src=""(?!https?:\/\/)(.*?)""", @"src=""https://gaokao.chsi.com.cn$1""") + }); + // } }); } }