public IActionResult Document(Models.EditDocumentRequestModel requestModel)
{
requestModel.AccountId = HttpContext.Session.GetInt32("AccountId").GetValueOrDefault(0);
if (requestModel.Title.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档标题"));
}
if (requestModel.Contents.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档内容"));
}
var repository = _unitOfWork.GetRepository <Entity.m_Docs>();
Entity.m_Docs model = repository.Query().Where(q => q.DocsId == requestModel.DocsId).FirstOrDefault();
if (model == null)
{
return(APIReturnMethod.ReturnFailed("您要编辑的文档内容信息不存在!"));
}
if (model.AccountId != requestModel.AccountId)
{
return(APIReturnMethod.ReturnFailed("您无权对当前的数据进行编辑操作!"));
}
model.Contents = HtmlFilter.SanitizeHtml(requestModel.Contents);
model.LastTime = DateTime.Now;
model.Title = HtmlFilter.StripHtml(requestModel.Title);
model.ShortTitle = HtmlFilter.StripHtml(requestModel.ShortTitle);
repository.Update(model);
var resultCount = _unitOfWork.SaveChanges();
return(resultCount > 0 ? APIReturnMethod.ReturnSuccess() : APIReturnMethod.ReturnFailed());
}
protected void Submit_Click(object sender, EventArgs args)
{
if (!IsAuthorized)
{
cvAuthorized.IsValid = false;
return;
}
HtmlFilter filter = Engine.Resolve <HtmlFilter>();
Items.WikiArticle article = CurrentPage;
if (IsNew)
{
article = Engine.Resolve <ContentActivator>().CreateInstance <Items.WikiArticle>(CurrentPage);
article.Title = filter.StripHtml(CurrentArguments);
article.Name = filter.CleanUrl(CurrentArguments);
}
else
{
Engine.Resolve <IVersionManager>().SaveVersion(article);
}
article["SavedDate"] = DateTime.Now;
article["SavedByAddress"] = Request.UserHostAddress;
article["Syndicatable"] = CurrentPage.WikiRoot["Syndicatable"];
article.Text = filter.FilterHtml(txtText.Text);
Engine.Persister.Save(article);
Response.Redirect(article.Url);
}
public IActionResult Post([FromBody] Models.ThemeCreateRequestModel requestModel)
{
if (requestModel.Title.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档主题标题"));
}
if (requestModel.Contents.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档主题内容"));
}
Entity.m_DocsTheme model = new Entity.m_DocsTheme();
model.AppendTime = DateTime.Now;
model.Contents = HtmlFilter.SanitizeHtml(requestModel.Contents);
model.IsShow = true;
model.LastTime = DateTime.Now;
model.PlusCount = 0;
model.ReadCount = 0;
model.Tags = "";
model.Title = HtmlFilter.StripHtml(requestModel.Title);
model.AccountId = requestModel.AccountId;
model.VersionText = "";
var repository = _unitOfWork.GetRepository <Entity.m_DocsTheme>();
repository.Insert(model);
var resultCount = _unitOfWork.SaveChanges();
return(resultCount > 0 ? APIReturnMethod.ReturnSuccess() : APIReturnMethod.ReturnFailed());
}
public IActionResult Put([FromBody] Models.ThemeEditRequestModel requestModel)
{
if (requestModel.Title.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档主题标题"));
}
if (requestModel.Contents.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档主题内容"));
}
var repository = _unitOfWork.GetRepository <Entity.m_DocsTheme>();
Entity.m_DocsTheme model = repository.Query().Where(q => q.ThemeId == requestModel.ThemeId).FirstOrDefault();
if (model == null)
{
return(APIReturnMethod.ReturnFailed("您要编辑的文档主题信息不存在!"));
}
if (model.AccountId != requestModel.AccountId)
{
return(APIReturnMethod.ReturnFailed("您无权对当前的数据进行编辑操作!"));
}
model.Contents = HtmlFilter.SanitizeHtml(requestModel.Contents);
model.LastTime = DateTime.Now;
model.Title = HtmlFilter.StripHtml(requestModel.Title);
model.VersionText = "";
repository.Update(model);
var resultCount = _unitOfWork.SaveChanges();
return(resultCount > 0 ? APIReturnMethod.ReturnSuccess() : APIReturnMethod.ReturnFailed());
}
public IActionResult Index(Models.ReleaseRequestModel requestModel)
{
requestModel.AccountId = HttpContext.Session.GetInt32("AccountId").GetValueOrDefault(0);
if (string.IsNullOrEmpty(requestModel.Title) || requestModel.Title == "")
{
return(APIReturnMethod.ReturnFailed("标题不能为空"));
}
if (string.IsNullOrEmpty(requestModel.Contents) || requestModel.Contents == "")
{
return(APIReturnMethod.ReturnFailed("内容不能为空"));
}
//
Entity.m_CmsContents entity = new Entity.m_CmsContents();
entity.Contents = HtmlFilter.SanitizeHtml(requestModel.Contents);
entity.ImgUrl = string.Empty;
entity.StateCode = 1;
entity.PostTime = DateTime.Now;
entity.PlusCount = 0;
entity.LastTime = DateTime.Now;
entity.Tags = "";
entity.ReadCount = 0;
entity.Title = requestModel.Title;
entity.AccountId = requestModel.AccountId;
entity.AnswerCount = 0;
entity.ChannelId = requestModel.ChannelId;
var repository = _unitOfWork.GetRepository <Entity.m_CmsContents>();
repository.Insert(entity);
int resultCount = _unitOfWork.SaveChanges();
return(resultCount > 0 ? APIReturnMethod.ReturnSuccess() : APIReturnMethod.ReturnFailed());
}
public IActionResult Document(Models.DocsContentsCreateRequestModel requestModel)
{
requestModel.AccountId = HttpContext.Session.GetInt32("AccountId").GetValueOrDefault(0);
if (requestModel.Title.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档标题"));
}
if (requestModel.Contents.Trim().Length <= 0)
{
return(APIReturnMethod.ReturnFailed("请输入文档内容"));
}
Entity.m_Docs model = new Entity.m_Docs();
model.AppendTime = DateTime.Now;
model.Contents = HtmlFilter.SanitizeHtml(requestModel.Contents);
model.IsShow = true;
model.LastTime = DateTime.Now;
model.PlusCount = 0;
model.ReadCount = 0;
model.Tags = "";
model.Title = HtmlFilter.StripHtml(requestModel.Title);
model.AccountId = requestModel.AccountId;
model.VersionText = "";
model.ThemeId = requestModel.ThemeId;
model.ShortTitle = HtmlFilter.StripHtml(requestModel.ShortTitle);
model.IsAudit = true;
var repository = _unitOfWork.GetRepository <Entity.m_Docs>();
repository.Insert(model);
var resultCount = _unitOfWork.SaveChanges();
return(resultCount > 0 ? APIReturnMethod.ReturnSuccess() : APIReturnMethod.ReturnFailed());
}
public static void BubbleDeletedPopUp(Control control, String DeletedName)
{
DeletedName = HtmlFilter.StipTags(DeletedName);
DeletedName += (String)HttpContext.GetGlobalResourceObject("Global", "BubbleDeletedString_Resource") + " " + "<a href = \"\"> X</a>";
ScriptManager.RegisterStartupScript(control, typeof(string), "DeleteBubblePopup", String.Format("bubbleNotify('{0}');", DeletedName), true);
}
public void testAllowedTags()
{
String input = "这是多种<span>内容1</span>标签和<strong>内容2</strong>风格，还有<img src=\"\" />图片与<a href=\"abc.html\">链接</a>对吗？";
String result = HtmlFilter.Filter(input, "strong");
Assert.AreEqual("这是多种内容1标签和<strong>内容2</strong>风格，还有图片与链接对吗？", result);
result = HtmlFilter.Filter(input, "img");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有<img src=\"\" />图片与链接对吗？", result);
result = HtmlFilter.Filter(input, "a");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有图片与<a href=\"abc.html\">链接</a>对吗？", result);
result = HtmlFilter.Filter(input, "img,a");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有<img src=\"\" />图片与<a href=\"abc.html\">链接</a>对吗？", result);
// 对换行的测试
result = HtmlFilter.Filter("这是多种内容1标签和<br>内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br>内容2风格", result);
// TODO 这个测试不能通过
//result = HtmlFilter.Filter( "这是多种内容1标签和<br/>内容2风格", "br" );
//Assert.AreEqual( "这是多种内容1标签和<br/>内容2风格", result );
result = HtmlFilter.Filter("这是多种内容1标签和<br >内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br >内容2风格", result);
result = HtmlFilter.Filter("这是多种内容1标签和<br />内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br />内容2风格", result);
result = HtmlFilter.Filter("这是<strong>多种</strong>内容1标签和<br>内容2风格", "strong");
Assert.AreEqual("这是<strong>多种</strong>内容1标签和内容2风格", result);
}
public void testStyle()
{
String input = @"<style id=""abc"">
.h1 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 22pt; MARGIN: 17pt 0cm 16.5pt; LINE-HEIGHT: 240%; TEXT-ALIGN: justify
}
.h2 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 16pt; MARGIN: 13pt 0cm; LINE-HEIGHT: 173%; TEXT-ALIGN: justify
}
.h3 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 16pt; MARGIN: 13pt 0cm; LINE-HEIGHT: 173%; TEXT-ALIGN: justify
}
.h1 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 22pt; MARGIN: 17pt 0cm 16.5pt; LINE-HEIGHT: 240%; TEXT-ALIGN: justify
}
.h2 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 16pt; MARGIN: 13pt 0cm; LINE-HEIGHT: 173%; TEXT-ALIGN: justify
}
.h3 {
FONT-WEIGHT: bold; TEXT-JUSTIFY: inter-ideograph; FONT-SIZE: 16pt; MARGIN: 13pt 0cm; LINE-HEIGHT: 173%; TEXT-ALIGN: justify
}
</style>" + @"<P> <FONT face=楷体_GB2312>从1990年至2000年期间，人文社会科学领域学术期刊的价格涨幅高达185.9%，科技和医学领域学术期刊价格涨幅分别达178.3%和184.3%。全球范围内出现的“期刊危机”，倒逼科技界提出一项全新的出版理念——“开放获取”。 </FONT></P>
<P><FONT face=楷体_GB2312> 所谓“开放获取”(Open Access)，是指科学研究信息在网络环境中，免费供公众自由获取。开放获取有两种实现形式：一为开放出版，即期刊或会议论文出版后立即开放获取；二是开放存储，即论文出版后存储到相关知识库，一段时间后开放获取。</FONT></P> ";
String result = HtmlFilter.Filter(input);
Console.WriteLine(result);
}
public static void BubbleSavePopUp(Control control, String SavedName, int delayTime, int fadeOutTime)
{
SavedName = HtmlFilter.StipTags(SavedName);
SavedName += (String)HttpContext.GetGlobalResourceObject("Global", "BubbleAddedString_Resource") + " " + "<a href = \"\"> X</a>";
ScriptManager.RegisterStartupScript(control, typeof(string), "SaveBubblePopup", String.Format("bubbleNotifyCustomDelay('{0}', {1}, {2});", SavedName, delayTime, fadeOutTime), true);
}
public void testTableTag()
{
String input = "<table></table>";
String str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
input = "<table><tr><td>abc</td></tr></table>";
str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
input = "<table ><tr ><td > abc </td></tr></table>";
str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
input = "<table border=\"5\"><tr ><td > abc </td></tr></table>";
str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
// 不在白名单的属性会被删除
String newInput = "<table border=\"5\" data-new=\"88\"><tr ><td > abc </td></tr></table>";
str = HtmlFilter.Filter(newInput);
Assert.AreEqual(input, str);
}
protected void btnSubmit_Click(object sender, EventArgs e)
{
HtmlFilter filter = Engine.Resolve <HtmlFilter>();
string name = filter.CleanUrl(txtName.Text);
string url = Url.Parse(CurrentPage.WikiRoot.Url).AppendSegment(name);
Response.Redirect(url);
}
public static void RegisterGlobalFilters(GlobalFilterCollection filters)
{
filters.Add(new HandleErrorAttribute());
XDocument myXDoc = XDocument.Load(HttpContext.Current.Server.MapPath("~/htmlfilter.xml"));
XElement ele = myXDoc.Element("config");
FilterXmlConfig filterXmlConfig = new FilterXmlConfig(ele);
htmlFilter = new HtmlFilter(filterXmlConfig);
}
public void HtmlFilter()
{
IHootFilter _filter = new HtmlFilter();
String _result;
_filter.InitializeFilter(null);
_result = _filter.FilterText(_htmlSource);
Assert.AreEqual(_result.Replace("\r", ""), _htmlResult.Replace("\r", ""));
}
public void testFilter()
{
String input = "这是脚本<script>引用了</script>内容";
String result = HtmlFilter.Filter(input);
Assert.AreEqual("这是脚本内容", result);
Console.WriteLine(result);
}
public void testAllowedFlash()
{
String input = "<object CLASSID=\"clsid:D27CDB6E-AE6D-11cf-96B8-444553540000\" data=\"http://player.youku.com/player.php/sid/XMTg0MTI0NjU2/v.swf\" type=\"application/x-shockwave-flash\" width=\"300\" height=\"255\"><param name=\"movie\" value =\"http://player.youku.com/player.php/sid/XMTg0MTI0NjU2/v.swf\" /></object>";
String str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
input = "<object classid=\"clsid:D27CDB6E-AE6D-11cf-96B8-444553540000\" codebase=\"http://download.macromedia.com/pub/shockwave/cabs/flash/swflash.cab#version=9,0,28,0\" width=\"95\" height=\"360\"> <param name=\"movie\" value=\"a.swf\" /> <param name=\"quality\" value=\"high\" /> <embed src=\"a.swf\" quality=\"high\" pluginspage=\"http://www.adobe.com/shockwave/download/download.cgi?P1_Prod_Version=ShockwaveFlash\" type=\"application/x-shockwave-flash\" width=\"300\" height=\"255\"></embed></object>";
str = HtmlFilter.Filter(input);
Assert.AreEqual(input, str);
}
public void testLink()
{
String str = "aaa<a href=\"xxx.html\">ccc</a>zzz";
string val = HtmlFilter.Filter(str);
Assert.AreEqual(val, str);
str = "AAA<a href=\"XXX.html\">CCC</a>ZZZ";
val = HtmlFilter.Filter(str);
Assert.AreEqual(val, str);
}
/// <summary>
/// 在默认白名单的基础上，允许 allowedTags 中指定的tag
/// </summary>
/// <param name="postItem"></param>
/// <param name="allowedTags"></param>
/// <returns></returns>
public String PostHtmlAppendTags(String postItem, String allowedTags)
{
String val = _context.post(postItem);
if (val != null)
{
val = strUtil.TrimHtml(val);
val = HtmlFilter.FilterAppendTags(val, allowedTags);
}
return(val);
}
/// <summary>
/// Removes non allowed tags from the string. Non allowed tags are defined in the <see cref="DefaultFilterRules"/> class.
/// In this scenario we add the ability to allow the element img to have width and height
/// </summary>
/// <param name="input">A string that may contain html-tags.</param>
/// <returns>
/// The input washed from non allowed html-tags.
/// </returns>
public static string ToSafeString(this string input)
{
using (var inputStream = new StringReader(input))
using (var outputStrean = new StringWriter())
{
DefaultFilterRules defaultRules = new DefaultFilterRules(elem => GetAttributesForElement(elem), new DefaultFilterRules());
HtmlFilter filter = new HtmlFilter(defaultRules);
filter.FilterHtml(inputStream, outputStrean);
return(outputStrean.ToString());
}
}
/// <summary>
/// 获取客户端 post 的 html，结果已被过滤，只允许 allowedTags 中指定的 tag
/// </summary>
/// <param name="postItem"></param>
/// <param name="allowedTags">允许的tag，包括属性列表</param>
/// <returns></returns>
public String PostHtml(String postItem, Dictionary <String, String> allowedTags)
{
String val = _context.post(postItem);
if (val != null)
{
val = strUtil.TrimHtml(val);
val = HtmlFilter.Filter(val, allowedTags);
}
return(val);
}
public static void BubbleNotify(Control control, string description, bool addCloseButton)
{
description = HtmlFilter.StipTags(description);
// add close button
if (addCloseButton)
{
description = string.Format("{0} <a href = \"\">X</a>", description);
}
ScriptManager.RegisterStartupScript(control, typeof(string), "bubbleNotification",
string.Format("bubbleNotify('{0}');", description), true);
}
public void testAllowedTags()
{
String input = "这是多种<span>内容1</span>标签和<strong>内容2</strong>风格，还有<img src=\"\" />图片与<a href=\"abc.html\">链接</a>对吗？";
String result = HtmlFilter.Filter(input, "strong");
Assert.AreEqual("这是多种内容1标签和<strong>内容2</strong>风格，还有图片与链接对吗？", result);
result = HtmlFilter.Filter(input, "img");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有<img src=\"\" />图片与链接对吗？", result);
result = HtmlFilter.Filter(input, "a");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有图片与<a href=\"abc.html\">链接</a>对吗？", result);
result = HtmlFilter.Filter(input, "img,a");
Assert.AreEqual("这是多种内容1标签和内容2风格，还有<img src=\"\" />图片与<a href=\"abc.html\">链接</a>对吗？", result);
// 对换行的测试
result = HtmlFilter.Filter("这是多种内容1标签和<br>内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br>内容2风格", result);
result = HtmlFilter.Filter("这是多种内容1标签和<br >内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br >内容2风格", result);
result = HtmlFilter.Filter("这是多种内容1标签和<br />内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br />内容2风格", result);
result = HtmlFilter.Filter("这是多种内容1标签和<br/>内容2风格", "br");
Assert.AreEqual("这是多种内容1标签和<br/>内容2风格", result);
result = HtmlFilter.Filter("这是<strong>多种</strong>内容1标签和<br>内容2风格", "strong");
Assert.AreEqual("这是<strong>多种</strong>内容1标签和内容2风格", result);
//-----------------------------------------
input = "文字<style>div{color:red;}</style>内容";
result = HtmlFilter.Filter(input, "style");
Assert.AreEqual(input, result);
Dictionary <String, String> dic = new Dictionary <String, String>();
dic.Add("script", "src");
input = "文字<script src=\"aa\">some code</script>内容";
result = HtmlFilter.Filter(input, dic);
Assert.AreEqual(input, result);
dic = new Dictionary <String, String>();
dic.Add("script", "src"); // 没有的属性会被过滤
input = "文字<script src=\"aa\" xdata=99>some code</script>内容";
result = HtmlFilter.Filter(input, dic);
Assert.AreEqual("文字<script src=\"aa\">some code</script>内容", result);
}
public void testBase()
{
String input = "这是脚本<script>引用了</script>内容";
String result = HtmlFilter.Filter(input);
Assert.AreEqual("这是脚本内容", result);
Console.WriteLine(result);
input = "<P> <FONT face=楷体_GB2312>从1990年至2000年期间 </FONT></P> ";
result = HtmlFilter.Filter(input); // 过滤之后，会给属性加上引号
Assert.AreEqual("<P> <FONT face=\"楷体_GB2312\">从1990年至2000年期间 </FONT></P> ", result);
}
/// <summary>
/// 获取客户端 post 的 html，结果已被过滤，只有在白名单中的 tag 才被允许
/// </summary>
/// <param name="postItem"></param>
/// <returns></returns>
public String PostHtml(String postItem)
{
String val = _context.post(postItem);
if (val != null)
{
if (this.viewer != null && this.viewer.IsAdministrator())
{
return(val);
}
val = strUtil.TrimHtml(val);
val = HtmlFilter.Filter(val);
}
return(val);
}
public void testComment()
{
// 默认是允许注释的
String input = "<ul><!-- BEGIN list --><li>这是注释</li><!-- END list --></ul>";
String result = HtmlFilter.Filter(input);
Assert.AreEqual("<ul><!-- BEGIN list --><li>这是注释</li><!-- END list --></ul>", result);
input = "<!--abcd<script>alert( 'abcd' );</script>123-->";
result = HtmlFilter.Filter(input);
Assert.AreEqual("<!--abcd123-->", result);
}
protected string getMatchedBody(string page, SpiderTemplate s, StringBuilder sb)
{
Match match = Regex.Match(page, s.GetDetailPattern(), RegexOptions.Singleline);
if (match == null || !match.Success || string.IsNullOrEmpty(match.Value))
{
logInfo("error=没有匹配的页面内容:" + _url, this._url, s, sb);
return(null);
}
page = match.Groups[1].Value;
String fpage = HtmlFilter.Filter(page); // 过滤广告
return(fpage);
}
private async void BtnCompare_Click(object sender, RoutedEventArgs e)
{
try
{
// First load the html into a document objects we can manipulate
string html1 = await File.ReadAllTextAsync("Resources/bbc1.html");
HtmlDocument doc1 = new HtmlDocument();
doc1.LoadHtml(html1);
string html2 = await File.ReadAllTextAsync("Resources/bbc2.html");
HtmlDocument doc2 = new HtmlDocument();
doc2.LoadHtml(html2);
// Filter the html to remove parts that are not very interesting
HtmlFilter filter = new HtmlFilter();
filter.CleanHtml(doc1.DocumentNode);
filter.CleanHtml(doc2.DocumentNode);
// convert the html into lines
HtmlConverter converter = new HtmlConverter();
var lines1 = converter.GetLines(doc1.DocumentNode, ConvertOptions.Default);
var lines2 = converter.GetLines(doc2.DocumentNode, ConvertOptions.Default);
// find the difference between these 2 sets of lines
LineDiffer differ = new LineDiffer();
var diffResult = differ.GetDiff(lines1, lines2, true);
// load the original html into a document again
HtmlDocument originalDoc = new HtmlDocument();
originalDoc.LoadHtml(html1);
// highlight the changed parts of the document
Highlighter.ApplyHighlights(originalDoc, "#FFFF99", diffResult.Added.Select(i => i.Line.XPath));
string htmlHighlighted = originalDoc.Save();
webBrowserDiff.NavigateToString(htmlHighlighted);
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
}
public void testScriptAndIFrame()
{
string val = HtmlFilter.Filter("国灭亡论，是中国人就得看！！！——日本宣称将于2015年灭亡中国。本文并非在下所写，只是基于一个爱国者的心，将此文贴于此。（这是一个在日本人气极旺的帖子，感谢一位不知名的网友，翻译过来。本着“奇文共欣赏，疑义相与析”的原则，转载过来是给大家一个对“日本”这个禽兽国度清醒的");
Assert.AreEqual(val, target);
val = HtmlFilter.Filter("国灭亡论，是中国人就得看！！！——日本宣称将于2015年灭亡中国。本文并非在下所写，只是基于一个爱国者的心，将此文贴于此。（这是<script> alert( 'kkkkk'); </script>一个在日本人气极旺的帖子，感谢一位不知名的网友，翻译过来。本着“奇文共欣赏，疑义相与析”的原则，转载过来是给大家一个对“日本”这个禽兽国度清醒的");
Assert.AreEqual(val, target);
val = HtmlFilter.Filter("国灭亡论，是中国人就得看！！！——日本宣称将于2015年灭亡中国。本文并非在下所写，只是基于一个爱国者的心，将此文贴于此。（这是<iframe src=abc.php width=500> </iframe>一个在日本人气极旺的帖子，感谢一位不知名的网友，翻译过来。本着“奇文共欣赏，疑义相与析”的原则，转载过来是给大家一个对“日本”这个禽兽国度清醒的");
Assert.AreEqual(val, target);
val = HtmlFilter.Filter("国灭亡论，是中国人就得看！！！——日本宣称将于2015年灭亡中国。本文并非在下所写，只是基于一个爱国者的心，将此文贴于此。（这是<scRIPT> alert( 'kkkkk'); </SCRIPT>一个在日本人气极旺的帖子，感谢一位不知名的网友，翻译过来。本着“奇文共欣赏，疑义相与析”的原则，转载过来是给大家一个对“日本”这个禽兽国度清醒的");
Assert.AreEqual(val, target);
val = HtmlFilter.Filter("国灭亡论，是中国人就得看！！！——日本宣称将于2015年灭亡中国。本文并非在下所写，只是基于一个爱国者的心，将此文贴于此。（这是<ifRAme src=abc.php width=500> </IfRAMe>一个在日本人气极旺的帖子，感谢一位不知名的网友，翻译过来。本着“奇文共欣赏，疑义相与析”的原则，转载过来是给大家一个对“日本”这个禽兽国度清醒的");
Assert.AreEqual(val, target);
}
public static bool ValidateFilter(this HtmlFilter filter, out string error)
{
error = null;
switch (filter.Type)
{
case HtmlFilterType.Regex:
try
{
// ReSharper disable once ObjectCreationAsStatement
new Regex(filter.Filter);
return(true);
}
catch (Exception ex)
{
error = ex.Message;
}
return(false);
case HtmlFilterType.XPath:
try
{
XPathExpression.Compile(filter.Filter);
return(true);
}
catch (Exception ex)
{
error = ex.Message;
}
return(false);
default:
throw new NotImplementedException("Invalid filter type");
}
}
public ActionResult GetContent(HtmlFilter hf, int?page)
{
Html html = new Html();
if (hf.DivId == null)
{
hf.DivId = "list_middle";
}
if (page == 1 || page == 0 || page == null)
{
html.contentList = GetFilteredContent(hf.URL, hf.Keyword, hf.DivId);
}
else if (page > 0 && page > 1)
{
html.contentList = GetFilteredContent(hf.URL, hf.Keyword, hf.DivId);
for (int i = 1; i <= page; i++)
{
hf.URL = "http://www.sydneytoday.com/job_information?page=" + i;
html.contentList.AddRange(GetFilteredContent(hf.URL, hf.Keyword, hf.DivId));
}
}
return(View(html));
}
public static string Filter(this HtmlFilter filter, string content)
{
switch (filter.Type)
{
case HtmlFilterType.Regex:
var filteredContents = new List <string>();
var regex = new Regex(filter.Filter, RegexOptions.Singleline);
var matches = regex.Matches(content);
for (int i = 0; i < matches.Count; i++)
{
var match = matches[i];
if (match.Success == false)
{
continue;
}
for (int j = 1; j < match.Groups.Count; j++)
{
filteredContents.Add(match.Groups[j].Value);
}
}
if (filteredContents.Any() == false)
{
return(null);
}
content = string.Join("\r\n", filteredContents);
break;
case HtmlFilterType.XPath:
try
{
HtmlDocument htmlDoc = new HtmlDocument();
htmlDoc.LoadHtml(content);
filteredContents = htmlDoc.DocumentNode
.SelectNodes(filter.Filter)
?.Select(n => n.OuterHtml)
.ToList();
// ReSharper disable once ConditionIsAlwaysTrueOrFalse
if (filteredContents == null || filteredContents.Any() == false)
{
return(null);
}
content = string.Join("\r\n", filteredContents);
}
catch (Exception ex)
{
LogTo.Error(ex, $"Invalid XPath filter '{filter}'");
throw;
}
break;
default:
throw new NotImplementedException($"Feed filter type {filter.Type} not implemented");
}
if (filter.Children.Any())
{
var subContents = filter.Children.Select(f => f.Filter(content)).ToList();
if (subContents.Any() == false)
{
return(null);
}
return(string.Join("\r\n", subContents));
}
return(content);
}
