//采集
private bool GetTest(int testType)
{
string Pic = "/upload/images_collects/";
System.Text.Encoding enType = System.Text.Encoding.Default;
switch (WebEncode)
{
case "1":
enType = System.Text.Encoding.GetEncoding("gb2312");
break;
case "2":
enType = System.Text.Encoding.UTF8;
break;
case "3":
enType = System.Text.Encoding.Unicode;
break;
}
TPortalClass.Collection Cn = new TPortalClass.Collection();
string testList = Cn.GetHttpPageCode(ListUrl, enType);
if (testList == "$UrlIsFalse")
{
Response.Write("<script>alert('列表地址设置错误');window.location='list_collect.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testList == "$GetFalse")
{
Response.Write("<script>alert('无法连接列表页或连接超时');window.location='list_collect.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
testList = Cn.GetBody(testList, ListStart, ListEnd, true, true);
if (testList == "$StartFalse")
{
Response.Write("<script>alert('列表开始标记设置错误,请重新设置');window.location='list_collect.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testList == "$EndFalse")
{
Response.Write("<script>alert('列表结束记设置错误,请重新设置');window.location='list_collect.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testType == 0)
{
this.ltListTest.Text = testList;
}
else
{
ArrayList linkArray = Cn.GetLinkArray(testList, LinkStart, LinkEnd);
if (linkArray.Count == 0)
{
Response.Write("<script>alert('未取到链接,请检查链接设置');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
else
{
if (linkArray[0].ToString() == "$StartFalse")
{
Response.Write("<script>alert('链接开始标记设置错误,请重新设置111');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (linkArray[0].ToString() == "$EndFalse")
{
Response.Write("<script>alert('链接开始标记设置错误,请重新设置');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (linkArray[0].ToString() == "$NoneLink")
{
Response.Write("<script>alert('未取到链接,请检查链接设置222');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (IsDesc == 1)
{
linkArray.Reverse();
}
if (CollecNum > 0 && linkArray.Count > CollecNum)
{
linkArray.RemoveRange(CollecNum, linkArray.Count - CollecNum);
}
string linkStr = string.Empty;
if (testType == 1)//链接地址
{
for (int i = 0; i < linkArray.Count; i++)
{
linkStr = Cn.DefiniteUrl(linkArray[i].ToString(), WebUrl);
if (linkStr != "$False")
{
linkStr = "<a href='" + linkStr + "' target=_blank>" + linkStr + "</a><br>";
ltLinkTest.Text += linkStr;
}
}
}
if (testType == 2)//测试
{
linkStr = Cn.DefiniteUrl(linkArray[0].ToString(), WebUrl);
if (linkStr == "$False")
{
Response.Write("<script>alert('获取到的链接地址无效,请检查链接设置');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
string newsCode = Cn.GetHttpPageCode(linkStr, enType);
if (newsCode == "$UrlIsFalse")
{
Response.Write("<script>alert('获取到的链接地址无效,请检查链接设置');window.location='list_collect.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (newsCode == "$GetFalse")
{
Response.Write("<script>alert('无法连接内容页或连接超时');window.location='list_collect.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
string testTitle = Cn.GetBody(newsCode, TitleStart, TitleEnd, false, false);
string testKeyWord = Cn.GetBody(newsCode, KeyWordStart, KeyWordEnd, false, false);
string testDateTime = Cn.GetBody(newsCode, DateStart, DateEnd, false, false);
string testSource = Cn.GetBody(newsCode, SourceStart, SourceEnd, false, false);
//string testDateTime = Cn.GetRegValue(DateRegex, newsCode);
//string testSource = Cn.GetRegValue(SourceRegex, newsCode);
string testBody = Cn.GetBody(newsCode, ContentStart, ContentEnd, false, false);
if (testTitle == "$StartFalse")
{
Response.Write("<script>alert('标题开始标记设置错误,请重新设置');window.location='list_collect.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testTitle == "$EndFalse")
{
Response.Write("<script>alert('标题结束标记设置错误,请重新设置');window.location='list_collect.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testBody == "$StartFalse")
{
Response.Write("<script>alert('正文开始标记设置错误,请重新设置');window.location='list_collect.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testBody == "$EndFalse")
{
Response.Write("<script>alert('正文结束标记设置错误,请重新设置');window.location='list_collect.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (RemoveTitle != "")
{
testTitle = testTitle.Replace(RemoveTitle, "");
}
this.ltTestTitle.Text = testTitle;
this.ltKeyWord.Text = testKeyWord;
this.ltDateTime.Text = testDateTime;
this.ltSource.Text = testSource;
//------------------获取详细页内容的下一页开始-------------------
string NewsNextUrl = Cn.GetRegValue(NextPageRegex, newsCode);
int PageCount = 0;
while (NewsNextUrl.Length > 0 && PageCount < 20)
{
string NewsPaingNextCode = string.Empty;
string ContentTemp = string.Empty;
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl, WebUrl);
NewsPaingNextCode = Cn.GetHttpPageCode(NewsNextUrl, enType);
ContentTemp = Cn.GetBody(NewsPaingNextCode, ContentStart, ContentEnd, false, false);
testBody = testBody + "<br>$PageNext$<br>" + ContentTemp;
string NewsNextUrl_1 = Cn.GetRegValue(NextPageRegex, NewsPaingNextCode);
if (NewsNextUrl_1.Length > 0)
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl_1, WebUrl);
else
break;
PageCount++;
}
//-----------------------获取详细页内容的下一页结束-------------------
//-------过滤正文开始-----------------------
if (RemoveBodyStart != "" && RemoveBodyEnd != "")
{
string[] removeBodyStartArr = RemoveBodyStart.Split('$');
string[] removeBodyEndArr = RemoveBodyEnd.Split('$');
if (removeBodyStartArr.Length > 1 && removeBodyEndArr.Length > 1)
{
if (removeBodyStartArr.Length != removeBodyEndArr.Length)
{
Response.Write("<script>alert('少了$请注意');window.location='list_collect.aspx'</script");
return false;
}
else
{
for (int i = 0; i < removeBodyStartArr.Length; i++)
{
string remove = Cn.GetBody(testBody, removeBodyStartArr[i], removeBodyEndArr[i], true, true);
testBody = testBody.Replace(remove, "");
}
}
}
else
{
string remove = Cn.GetBody(testBody, RemoveBodyStart, RemoveBodyEnd, true, true);
testBody = testBody.Replace(remove, "");
}
}
//--------------过滤正文结束---------------------------
ArrayList testBodyArray = Cn.ReplaceSaveRemoteFile(testBody, Pic, WebUrl, "0");
ltTestContent.Text = testBodyArray[0].ToString();
if (testBodyArray.Count == 2)
{
ltPhotoUrl.Text = testBodyArray[1].ToString();
}
}
if (testType == 3)//采集
{
int falseNum = 0;
int successNum = 0;
SetProcessBar("从" + WebUrl + "采集信息", true);
for (int i = 0; i < linkArray.Count; i++)
{
int isImg = 0;
string photoUrl = string.Empty;
ProcessBar(i, linkArray.Count);
linkStr = Cn.DefiniteUrl(linkArray[i].ToString(), WebUrl);
if (linkStr == "$False$")
{
falseNum++;
continue;
}
string newsPageCode = Cn.GetHttpPageCode(linkStr, enType);
if (newsPageCode == "$UrlIsFalse" || newsPageCode == "$GetFalse")
{
falseNum++;
continue;
}
string cTitle = Cn.GetBody(newsPageCode, TitleStart, TitleEnd, false, false);
string cKeyWord = Cn.GetBody(newsPageCode, KeyWordStart, KeyWordEnd, false, false);
string cDateTime = Cn.GetBody(newsPageCode, DateStart, DateEnd, false, false);
string cSource = Cn.GetBody(newsPageCode, SourceStart, SourceEnd, false, false);
//string cDateTime = Cn.GetRegValue(DateRegex, newsPageCode);
//string cSource = Cn.GetRegValue(SourceRegex, newsPageCode);
string cBody = Cn.GetBody(newsPageCode, ContentStart, ContentEnd, false, false);
if (cTitle == "$StartFalse" || cBody == "$StartFalse" || cTitle == "$EndFalse" || cBody == "$EndFalse")
{
falseNum++;
continue;
}
//--------获取详细页内容的下一页开始---------------
string NewsNextUrl = Cn.GetRegValue(NextPageRegex, newsPageCode);
int PageCount = 0;
while (NewsNextUrl.Length > 0 && PageCount < 20)
{
string NewsPaingNextCode = string.Empty;
string ContentTemp = string.Empty;
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl, WebUrl);
NewsPaingNextCode = Cn.GetHttpPageCode(NewsNextUrl, enType);
ContentTemp = Cn.GetBody(NewsPaingNextCode, ContentStart, ContentEnd, false, false);
cBody = cBody + "<br>$PageNext$<br>" + ContentTemp;
string NewsNextUrl_1 = Cn.GetRegValue(NextPageRegex, NewsPaingNextCode);
if (NewsNextUrl_1.Length > 0)
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl_1, WebUrl);
else
break;
PageCount++;
}
//---------获取详细页内容的下一页结束--------------
//string url_ss = dt.Rows[0]["ListUrl"].ToString();
string dq_url = "/"+ListUrl.Substring(20,4);
string ls_newurl = WebUrl + dq_url;
Response.Write(dq_url + "||");
ArrayList bodyArray = Cn.ReplaceSaveRemoteFile(cBody, Pic, ls_newurl, IsSaveImg.ToString());
if (bodyArray.Count == 2)
{
isImg = 1;
photoUrl = bodyArray[1].ToString();
}
cBody = bodyArray[0].ToString();
cBody = cBody.Replace("'", "");
//-------过滤正文开始-----------------------
if (RemoveBodyStart != "" && RemoveBodyEnd != "")
{
string[] removeBodyStartArr = RemoveBodyStart.Split('$');
string[] removeBodyEndArr = RemoveBodyEnd.Split('$');
if (removeBodyStartArr.Length > 1 && removeBodyEndArr.Length > 1)
{
for (int j = 0; j < removeBodyStartArr.Length; j++)
{
string remove = Cn.GetBody(cBody, removeBodyStartArr[j], removeBodyEndArr[j], true, true);
cBody = cBody.Replace(remove, "");
}
}
else
{
string remove = Cn.GetBody(cBody, RemoveBodyStart, RemoveBodyEnd, true, true);
cBody = cBody.Replace(remove, "");
}
}
//--------------过滤正文结束---------------------------
//---------------过滤开始-----------IFRAME---------
if (Script_Iframe == 1)
cBody = Cn.ScriptHtml(cBody, "Iframe", 2);
if (Script_Object == 1)
cBody = Cn.ScriptHtml(cBody, "Object", 2);
if (Script_Script == 1)
cBody = Cn.ScriptHtml(cBody, "Script", 2);
if (Script_Div == 1)
cBody = Cn.ScriptHtml(cBody, "Div", 3);
if (Script_Table == 1)
cBody = Cn.ScriptHtml(cBody, "Table", 2);
if (Script_Span == 1)
cBody = Cn.ScriptHtml(cBody, "Span", 3);
if (Script_Img == 1)
cBody = Cn.ScriptHtml(cBody, "Img", 3);
if (Script_Font == 1)
cBody = Cn.ScriptHtml(cBody, "Font", 3);
if (Script_A == 1)
cBody = Cn.ScriptHtml(cBody, "A", 3);
if (Script_Html == 1)
cBody = Cn.HtmlScript(cBody);
//-------------------过滤结束-------------------------------
//采集的数据保存至JpArticle
//关键字
string ls_str = cKeyWord.Trim();
string ls_date = System.DateTime.Now.ToString();
string ls_ly = "本网站";
if (cDateTime != "")
{
ls_date = cDateTime;
}
//if (cSource != "")
//{
// ls_ly = cSource;
//}
if (RemoveTitle != "")
{
cTitle = cTitle.Replace(RemoveTitle, "");
}
Int32 ld_aid = -1;
TPortalClass.JpArticle art = new TPortalClass.JpArticle(ld_aid);
art.cid = ClassId.ToString(); //栏目
art.title = cTitle; //标题
art.subhead = ""; //小标题
art.ht_content = cBody; //正文
art.summary = ""; //摘要
art.author = "系统管理员";
art.status = "发布";
art.crtime = ls_date;
art.modtime = ls_date;
art.pubtime = ls_date;
art.endtime = System.DateTime.Now.AddYears(20).ToString();
art.hits = 1; //点击率
art.ifindexdisplay ="0"; //是否首页显示
art.indexdisplaypicpath = "0";//首页显示图片上传路径
art.seotitle = "";
art.seokeywords = "";
art.seodescription = "";
art.asort = 50 + successNum; //排序号
art.iftop = "0"; //是否置顶
art.topcreatetime = ls_date; //置顶时间
art.topendtime = ls_date; //置顶结束时间
art.ifcomment = "0"; //是否启用评论
art.source = ls_ly; //来源
art.role_userid = "";
art.role_username = "";
art.shtml_created = 0; //是否已生成静态页面
art.GUID = System.Guid.NewGuid().ToString(); //随机数
art.map_aid = 0;
art.input = "admin";
art.targettype = "";
//art.hasimg = 0;
//art.set("artcontent", cBody);
//是否包含图片
//String artcontent = "";
//artcontent =
//String imgsrc = TPortalClass.JpRegx.parseIMG(artcontent);
//if (imgsrc.ToLower().IndexOf("<img") >= 0)
// art.hasimg = 1;
//else
// art.hasimg = 0;
art.Insert();
successNum++;
}
Response.Write("<script>alert('采集完成,成功 " + successNum + "条 ,失败 " + falseNum + "');window.location='list_collect.aspx'</script");
}
}
}
return true;
}
//采集
private bool GetTest(int testType)
{
string Pic = "/upload/images_collects/";
System.Text.Encoding enType = System.Text.Encoding.Default;
switch (WebEncode)
{
case "1":
enType = System.Text.Encoding.GetEncoding("gb2312");
break;
case "2":
enType = System.Text.Encoding.UTF8;
break;
case "3":
enType = System.Text.Encoding.Unicode;
break;
}
TPortalClass.Collection Cn = new TPortalClass.Collection();
string testList = Cn.GetHttpPageCode(ListUrl, enType);
if (testList == "$UrlIsFalse")
{
Response.Write("<script>alert('列表地址设置错误');window.location='list_collect_box.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testList == "$GetFalse")
{
Response.Write("<script>alert('无法连接列表页或连接超时');window.location='list_collect_box.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
testList = Cn.GetBody(testList, ListStart, ListEnd, true, true);
if (testList == "$StartFalse")
{
Response.Write("<script>alert('列表开始标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testList == "$EndFalse")
{
Response.Write("<script>alert('列表结束记设置错误,请重新设置');window.location='list_collect_box.aspx?action=list&id=" + this.txtId.Text + "'</script");
return false;
}
if (testType == 0)
{
this.ltListTest.Text = testList;
}
else
{
ArrayList linkArray = Cn.GetLinkArray(testList, LinkStart, LinkEnd);
if (linkArray.Count == 0)
{
Response.Write("<script>alert('未取到链接,请检查链接设置');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
else
{
if (linkArray[0].ToString() == "$StartFalse")
{
Response.Write("<script>alert('链接开始标记设置错误,请重新设置111');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (linkArray[0].ToString() == "$EndFalse")
{
Response.Write("<script>alert('链接开始标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (linkArray[0].ToString() == "$NoneLink")
{
Response.Write("<script>alert('未取到链接,请检查链接设置222');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (IsDesc == 1)
{
linkArray.Reverse();
}
if (CollecNum > 0 && linkArray.Count > CollecNum)
{
linkArray.RemoveRange(CollecNum, linkArray.Count - CollecNum);
}
string linkStr = string.Empty;
if (testType == 1)//链接地址
{
for (int i = 0; i < linkArray.Count; i++)
{
linkStr = Cn.DefiniteUrl(linkArray[i].ToString(), WebUrl);
if (linkStr != "$False")
{
linkStr = "<a href='" + linkStr + "' target=_blank>" + linkStr + "</a><br>";
ltLinkTest.Text += linkStr;
}
}
}
if (testType == 2)//测试
{
linkStr = Cn.DefiniteUrl(linkArray[0].ToString(), WebUrl);
if (linkStr == "$False")
{
Response.Write("<script>alert('获取到的链接地址无效,请检查链接设置');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
string newsCode = Cn.GetHttpPageCode(linkStr, enType);
if (newsCode == "$UrlIsFalse")
{
Response.Write("<script>alert('获取到的链接地址无效,请检查链接设置');window.location='list_collect_box.aspx?action=link&id=" + this.txtId.Text + "'</script");
return false;
}
if (newsCode == "$GetFalse")
{
Response.Write("<script>alert('无法连接内容页或连接超时');window.location='list_collect_box.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
string testTitle = Cn.GetBody(newsCode, TitleStart, TitleEnd, false, false);
string testKeyWord = Cn.GetBody(newsCode, KeyWordStart, KeyWordEnd, false, false);
string testDateTime = Cn.GetBody(newsCode, DateStart, DateEnd, false, false);
string testSource = Cn.GetBody(newsCode, SourceStart, SourceEnd, false, false);
//string testDateTime = Cn.GetRegValue(DateRegex, newsCode);
//string testSource = Cn.GetRegValue(SourceRegex, newsCode);
string testBody = Cn.GetBody(newsCode, ContentStart, ContentEnd, false, false);
if (testTitle == "$StartFalse")
{
Response.Write("<script>alert('标题开始标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testTitle == "$EndFalse")
{
Response.Write("<script>alert('标题结束标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testBody == "$StartFalse")
{
Response.Write("<script>alert('正文开始标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (testBody == "$EndFalse")
{
Response.Write("<script>alert('正文结束标记设置错误,请重新设置');window.location='list_collect_box.aspx?action=content&id=" + this.txtId.Text + "'</script");
return false;
}
if (RemoveTitle != "")
{
testTitle = testTitle.Replace(RemoveTitle, "");
}
this.ltTestTitle.Text = testTitle;
this.ltKeyWord.Text = testKeyWord;
this.ltDateTime.Text = testDateTime;
this.ltSource.Text = testSource;
//------------------获取详细页内容的下一页开始-------------------
string NewsNextUrl = Cn.GetRegValue(NextPageRegex, newsCode);
int PageCount = 0;
while (NewsNextUrl.Length > 0 && PageCount < 20)
{
string NewsPaingNextCode = string.Empty;
string ContentTemp = string.Empty;
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl, WebUrl);
NewsPaingNextCode = Cn.GetHttpPageCode(NewsNextUrl, enType);
ContentTemp = Cn.GetBody(NewsPaingNextCode, ContentStart, ContentEnd, false, false);
testBody = testBody + "<br>$PageNext$<br>" + ContentTemp;
string NewsNextUrl_1 = Cn.GetRegValue(NextPageRegex, NewsPaingNextCode);
if (NewsNextUrl_1.Length > 0)
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl_1, WebUrl);
else
break;
PageCount++;
}
//-----------------------获取详细页内容的下一页结束-------------------
//-------过滤正文开始-----------------------
if (RemoveBodyStart != "" && RemoveBodyEnd != "")
{
string[] removeBodyStartArr = RemoveBodyStart.Split('$');
string[] removeBodyEndArr = RemoveBodyEnd.Split('$');
if (removeBodyStartArr.Length > 1 && removeBodyEndArr.Length > 1)
{
if (removeBodyStartArr.Length != removeBodyEndArr.Length)
{
Response.Write("<script>alert('少了$请注意');window.location='list_collect_box.aspx'</script");
return false;
}
else
{
for (int i = 0; i < removeBodyStartArr.Length; i++)
{
string remove = Cn.GetBody(testBody, removeBodyStartArr[i], removeBodyEndArr[i], true, true);
testBody = testBody.Replace(remove, "");
}
}
}
else
{
string remove = Cn.GetBody(testBody, RemoveBodyStart, RemoveBodyEnd, true, true);
testBody = testBody.Replace(remove, "");
}
}
//--------------过滤正文结束---------------------------
ArrayList testBodyArray = Cn.ReplaceSaveRemoteFile(testBody, Pic, WebUrl, "0");
ltTestContent.Text = testBodyArray[0].ToString();
if (testBodyArray.Count == 2)
{
ltPhotoUrl.Text = testBodyArray[1].ToString();
}
}
if (testType == 3)//采集
{
int falseNum = 0;
int successNum = 0;
SetProcessBar("从" + WebUrl + "采集信息", true);
for (int i = 0; i < linkArray.Count; i++)
{
int isImg = 0;
string photoUrl = string.Empty;
ProcessBar(i, linkArray.Count);
linkStr = Cn.DefiniteUrl(linkArray[i].ToString(), WebUrl);
if (linkStr == "$False$")
{
falseNum++;
continue;
}
string newsPageCode = Cn.GetHttpPageCode(linkStr, enType);
if (newsPageCode == "$UrlIsFalse" || newsPageCode == "$GetFalse")
{
falseNum++;
continue;
}
string cTitle = Cn.GetBody(newsPageCode, TitleStart, TitleEnd, false, false); //标题
string cKeyWord = Cn.GetBody(newsPageCode, KeyWordStart, KeyWordEnd, false, false); //摘要
string cDateTime = Cn.GetBody(newsPageCode, DateStart, DateEnd, false, false); //发布日期
string cSource = Cn.GetBody(newsPageCode, SourceStart, SourceEnd, false, false); //来源
//string cDateTime = Cn.GetRegValue(DateRegex, newsPageCode);
//string cSource = Cn.GetRegValue(SourceRegex, newsPageCode);
string cBody = Cn.GetBody(newsPageCode, ContentStart, ContentEnd, false, false);
if (cTitle == "$StartFalse" || cBody == "$StartFalse" || cTitle == "$EndFalse" || cBody == "$EndFalse")
{
falseNum++;
continue;
}
//--------获取详细页内容的下一页开始---------------
string NewsNextUrl = Cn.GetRegValue(NextPageRegex, newsPageCode);
int PageCount = 0;
while (NewsNextUrl.Length > 0 && PageCount < 20)
{
string NewsPaingNextCode = string.Empty;
string ContentTemp = string.Empty;
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl, WebUrl);
NewsPaingNextCode = Cn.GetHttpPageCode(NewsNextUrl, enType);
ContentTemp = Cn.GetBody(NewsPaingNextCode, ContentStart, ContentEnd, false, false);
cBody = cBody + "<br>$PageNext$<br>" + ContentTemp;
string NewsNextUrl_1 = Cn.GetRegValue(NextPageRegex, NewsPaingNextCode);
if (NewsNextUrl_1.Length > 0)
NewsNextUrl = Cn.DefiniteUrl(NewsNextUrl_1, WebUrl);
else
break;
PageCount++;
}
//---------获取详细页内容的下一页结束--------------
//string url_ss = dt.Rows[0]["ListUrl"].ToString();
//string dq_url = "/"+ListUrl.Substring(20,4);
//string ls_newurl = WebUrl + dq_url;
//Response.Write(dq_url + "||");
ArrayList bodyArray = Cn.ReplaceSaveRemoteFile(cBody, Pic, WebUrl, IsSaveImg.ToString());
if (bodyArray.Count == 2)
{
isImg = 1;
photoUrl = bodyArray[1].ToString();
}
cBody = bodyArray[0].ToString();
cBody = cBody.Replace("'", "");
//-------过滤正文开始-----------------------
if (RemoveBodyStart != "" && RemoveBodyEnd != "")
{
string[] removeBodyStartArr = RemoveBodyStart.Split('$');
string[] removeBodyEndArr = RemoveBodyEnd.Split('$');
if (removeBodyStartArr.Length > 1 && removeBodyEndArr.Length > 1)
{
for (int j = 0; j < removeBodyStartArr.Length; j++)
{
string remove = Cn.GetBody(cBody, removeBodyStartArr[j], removeBodyEndArr[j], true, true);
cBody = cBody.Replace(remove, "");
}
}
else
{
string remove = Cn.GetBody(cBody, RemoveBodyStart, RemoveBodyEnd, true, true);
cBody = cBody.Replace(remove, "");
}
}
//--------------过滤正文结束---------------------------
//---------------过滤开始-----------IFRAME---------
if (Script_Iframe == 1)
cBody = Cn.ScriptHtml(cBody, "Iframe", 2);
if (Script_Object == 1)
cBody = Cn.ScriptHtml(cBody, "Object", 2);
if (Script_Script == 1)
cBody = Cn.ScriptHtml(cBody, "Script", 2);
if (Script_Div == 1)
cBody = Cn.ScriptHtml(cBody, "Div", 3);
if (Script_Table == 1)
cBody = Cn.ScriptHtml(cBody, "Table", 2);
if (Script_Span == 1)
cBody = Cn.ScriptHtml(cBody, "Span", 3);
if (Script_Img == 1)
cBody = Cn.ScriptHtml(cBody, "Img", 3);
if (Script_Font == 1)
cBody = Cn.ScriptHtml(cBody, "Font", 3);
if (Script_A == 1)
cBody = Cn.ScriptHtml(cBody, "A", 3);
if (Script_Html == 1)
cBody = Cn.HtmlScript(cBody);
//-------------------过滤结束-------------------------------
//采集的数据保存至JpArticle
//关键字
string ls_str = cKeyWord.Trim();
string ls_date = System.DateTime.Now.ToString();
string ls_ly = "本网站";
if (cDateTime != "")
{
if (cDateTime.Length > 19)
{
ls_date = DateTime.Parse(cDateTime.Substring(0, 19)).ToString("yyyy-MM-dd HH:mm:ss");
}
//ls_date = cDateTime;
}
string ls_xm = "", ls_tel = "", ls_sfz = "", ls_addtime = System.DateTime.Now.ToString(), ls_dept = "", ls_mail = "", ls_type = "";
if (cSource != "")
{
string[] lv_sourse = cSource.Split('|');
ls_xm = lv_sourse[0];
ls_dept = lv_sourse[1];
//if (lv_sourse[1] == "01")
//{
// ls_dept = "办公室";
//}
//else if (lv_sourse[1] == "02")
//{
// ls_dept = "养老保险处";
//}
//else if (lv_sourse[1] == "03")
//{
// ls_dept = "养老保险综合管理中心";
//}
//else if (lv_sourse[1] == "04")
//{
// ls_dept = "医疗工伤处";
//}
//else if (lv_sourse[1] == "05")
//{
// ls_dept = "医疗工伤保险管理中心";
//}
//else if (lv_sourse[1] == "06")
//{
// ls_dept = "计算机信息中心";
//}
//else if (lv_sourse[1] == "07")
//{
// ls_dept = "社会保险稽核处";
//}
//else if (lv_sourse[1] == "08")
//{
// ls_dept = "计划财务处";
//}
//else if (lv_sourse[1] == "09")
//{
// ls_dept = "离（退）休人员管理处";
//}
//else if (lv_sourse[1] == "10")
//{
// ls_dept = "新城管理中心";
//}
//else if (lv_sourse[1] == "11")
//{
// ls_dept = "定海社保";
//}
//else if (lv_sourse[1] == "12")
//{
// ls_dept = "普陀社保";
//}
//else if (lv_sourse[1] == "13")
//{
// ls_dept = "岱山社保";
//}
//else if (lv_sourse[1] == "14")
//{
// ls_dept = "嵊泗社保";
//}
//else
//{
// ls_dept = "";
//}
if (lv_sourse[2].Length > 19)
{
ls_addtime = DateTime.Parse(lv_sourse[2].Substring(0, 19)).ToString("yyyy-MM-dd HH:mm:ss");
}
ls_tel = lv_sourse[3];
ls_mail = lv_sourse[4];
ls_sfz = lv_sourse[5];
if (lv_sourse[6] == "1")
{
ls_type = "局长信箱";
}
if (lv_sourse[6] == "3")
{
ls_type = "投诉咨询";
}
}
if (RemoveTitle != "")
{
cTitle = cTitle.Replace(RemoveTitle, "");
}
//Response.Write(ls_date);
TPortalClass.JpBox JpBox = new TPortalClass.JpBox();
JpBox.lyusername = ls_xm;
JpBox.tel = ls_tel;
JpBox.mail = ls_mail;
JpBox.content = cBody; //正文2
JpBox.addtime = ls_addtime;
JpBox.ip = "127.0.0.1";
string ls_cxm = System.Guid.NewGuid().ToString();
JpBox.cxm = ls_cxm;
JpBox.yb = ls_sfz; //sfz
JpBox.nrzt = cTitle; //标题1
JpBox.jtzz = "公开"; //是否公开
JpBox.type = ls_type;
JpBox.ifhf = "1"; //是否回复 默认为未回复 0
JpBox.ifshow = "1"; //是否前台显示 默认为未前台显示 0
JpBox.hfsj = ls_date; //回复时间
JpBox.hfcontent = ls_str; //回复内容
JpBox.by4 = ls_dept;
JpBox.InsertLS();
successNum++;
}
Response.Write("<script>alert('采集完成,成功 " + successNum + "条 ,失败 " + falseNum + "');window.location='list_collect_box.aspx'</script");
}
}
}
return true;
}