private static void AppendSnippetReferences(SnippetDocumentControl snippet, HtmlAgilityPack.HtmlDocument doc, HtmlNode head)
{
// take care of javascript references
foreach (var reference in snippet.References)
{
if (reference.Type == ReferenceTypes.Css)
{
var s = doc.CreateElement("link");
s.SetAttributeValue("rel", "stylesheet");
s.SetAttributeValue("href", reference.Url);
head.AppendChild(s);
}
else if (reference.Type == ReferenceTypes.Javascript)
{
var s = doc.CreateElement("script");
s.SetAttributeValue("src", reference.Url);
s.SetAttributeValue("type", "text/javascript");
head.AppendChild(s);
}
else
{
throw new Exception($"Unsupported snippet reference type: {reference.Type}");
}
}
}
private void comboBox1_SelectedIndexChanged(object sender, EventArgs e)
{
if (comboBox1.SelectedItem != null)
{
checkedListBox1.Items.Clear();
checkedListBox1.Controls.Clear();
innerHtmlTextBox.Clear();
pathTextBox.Clear();
string[] arr;
document.Load(Application.StartupPath + "/tempHtml.txt");
string[] elem = comboBox1.SelectedItem.ToString().Split("<>".ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
node = document.CreateElement(elem[0]);
while (checkedListBox1.CheckedIndices.Count > 0)
{
checkedListBox1.SetItemChecked(checkedListBox1.CheckedIndices[0], false);
}
if (attr.AttrDict.ContainsKey(elem[0]))
{
arr = attr.AttrDict[elem[0]];
checkedListBox1.Items.AddRange(arr);
}
// Label lbl = new Label();
// lbl.AutoSize = true;
// lbl.Text = "Global Attributes";
// lbl.Location = new Point(0, arr.Length * 15);
// checkedListBox1.Controls.Add(lbl);
checkedListBox1.Items.AddRange(attr.AttrDict["Global"]);
checkedListBox1.Items.AddRange(attr.AttrDict["events"]);
}
else
{
return;
}
}
public static string DecodeHtml(string htmltext)
{
htmltext = htmltext.Replace("<p>", "").Replace("</p>", "\r\n\r\n");
string decoded = String.Empty;
if (htmltext.IndexOf('<') > -1 || htmltext.IndexOf('>') > -1 || htmltext.IndexOf('&') > -1)
{
try
{
HtmlDocument document = new HtmlDocument();
var decode = document.CreateElement("div");
htmltext = htmltext.Replace(".<", ". <").Replace("?<", "? <").Replace("!<", "! <").Replace("'", "'");
decode.InnerHtml = htmltext;
var allElements = decode.Descendants().ToArray();
for (int n = allElements.Length - 1; n >= 0; n--)
{
if (allElements[n].NodeType == HtmlNodeType.Comment || allElements[n].Name.EqualNoCase("style") || allElements[n].Name.EqualNoCase("script"))
{
allElements[n].Remove();
}
}
decoded = WebUtility.HtmlDecode(decode.InnerText);
}
catch { }
}
else
{
decoded = htmltext;
}
return decoded;
}
private static HtmlNode GetExternalReferencesListNode(
HtmlDocument document,
IEnumerable<ExternalReference> externalReferences)
{
var list = document.CreateElement("ol");
foreach (var reference in externalReferences)
{
var item = document.CreateElement("li");
string referenceId = GetExternalReferenceId(reference.Index);
var backLink = document.CreateElement("a");
backLink.InnerHtml = "<strong>^</strong>";
backLink.SetAttributeValue("href", "#" + referenceId + BackLinkReferenceIdSuffix);
var externalLink = document.CreateElement("a");
externalLink.InnerHtml = reference.Url;
externalLink.SetAttributeValue("href", reference.Url);
externalLink.SetAttributeValue("name", referenceId);
item.AppendChild(backLink);
item.AppendChild(document.CreateTextNode(" "));
item.AppendChild(externalLink);
list.AppendChild(item);
}
return list;
}
/// <summary>
/// Decodes the HTML.
/// </summary>
/// <param name="htmlText">The html text.</param>
/// <returns>String without any html tags.</returns>
public static string DecodeHtml(string htmlText)
{
htmlText = htmlText.Replace("<p>", "").Replace("</p>", "\r\n\r\n");
string decoded = String.Empty;
if (htmlText.IndexOf('<') > -1 || htmlText.IndexOf('>') > -1 || htmlText.IndexOf('&') > -1)
{
try
{
HtmlDocument document = new HtmlDocument();
var decode = document.CreateElement("div");
htmlText = htmlText.Replace(".<", ". <").Replace("?<", "? <").Replace("!<", "! <").Replace("'", "'");
decode.InnerHtml = htmlText;
decoded = WebUtility.HtmlDecode(decode.InnerText);
decoded = Regex.Replace(decoded, "<!--.*?-->", string.Empty, RegexOptions.Singleline);
}
catch { }
}
else
{
decoded = htmlText;
}
return decoded;
}
public void Parse(string input, string[] args = null)
{
var xxr = new XamlXmlReader(new StringReader(input), new XamlSchemaContext());
var graphReader = new XamlObjectWriter(xxr.SchemaContext);
while (xxr.Read())
graphReader.WriteNode(xxr);
var page = (Page)graphReader.Result;
// Map our generators
var g = new Generator();
g.Map<Page, PageGeneratorBlock>();
g.Map<Button, ButtonGeneratorBlock>();
g.Map<StackPanel, StackPanelGeneratorBlock>();
var doc = new HtmlDocument();
var html = doc.CreateElement("html");
g.Generate(html, page);
// HTML5 Doc type
doc.DocumentNode.AppendChild(doc.CreateComment("<!DOCTYPE html>"));
doc.DocumentNode.AppendChild(html);
doc.Save("test.htm");
var cssContents = g.GenerateStyles(page);
File.WriteAllText("XamlCore.css", cssContents);
}
private void WriteContent(HtmlDocument doc)
{
HtmlNode bodyContent = doc.GetElementbyId("bodyContent");
var heading = doc.CreateElement("h1");
heading.InnerHtml = "Heading";
bodyContent.AppendChild(heading);
HtmlNode table = bodyContent.CreateElement("table");
table.CreateAttributeWithValue("class", "table table-striped table-bordered");
var tableHead = table.CreateElement("thead");
var headerRow = tableHead.CreateElement("tr");
headerRow.CreateElementWithHtml("td", "First");
headerRow.CreateElementWithHtml("td", "Second");
headerRow.CreateElementWithHtml("td", "Third");
headerRow.CreateElementWithHtml("td", "Fourth");
headerRow.CreateElementWithHtml("td", "Fifth");
HtmlNode tableBody = table.CreateElement("tbody");
const string text = "Fi fa fo fum fi fa fo fum fi fa fo fum fi fa fo fum";
for (int i = 0; i < 10; i++)
{
HtmlNode bodyRow = tableBody.CreateElement("tr");
bodyRow.CreateElementWithHtml("td", i + " first " + text);
bodyRow.CreateElementWithHtml("td", i + " second " + text);
bodyRow.CreateElementWithHtml("td", i + " third " + text);
bodyRow.CreateElementWithHtml("td", i + " fourth " + text);
bodyRow.CreateElementWithHtml("td", i + " fifth " + text);
}
}
static Article GrabArticle(HtmlDocument doc)
{
Article article = new Article();
HtmlNode articleContent = doc.CreateElement("div");
ParagraphParentCollection paragraphParents = new ParagraphParentCollection(new ParagraphCollection(doc.DocumentNode.SelectNodes("//p")));
// Replace br tags with paragraph tags.
doc.DocumentNode.InnerHtml = Regex.Replace(doc.DocumentNode.InnerHtml, @"<br/?>[ \r\n\s]*<br/?>", @"</p><p>");
// TODO handle title.
article.Title = doc.DocumentNode.SelectSingleNode("//title") == null ? null : doc.DocumentNode.SelectSingleNode("//title").InnerText;
foreach (ParagraphParent parent in paragraphParents)
{
foreach (HtmlAttribute att in parent.Node.Attributes.AttributesWithName("class"))
{
if (Regex.IsMatch(att.Name, @"/(comment|meta|footer|footnote)/"))
parent.Score -= 50;
else if (Regex.IsMatch(att.Name, @"/((^|\\s)(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)(\\s|$))/"))
parent.Score += 25;
break;
}
foreach (HtmlAttribute att in parent.Node.Attributes.AttributesWithName("id"))
{
if (Regex.IsMatch(att.Name, @"/(comment|meta|footer|footnote)/"))
parent.Score -= 50;
else if (Regex.IsMatch(att.Name, @"/^(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)$/"))
parent.Score += 25;
}
foreach (Paragraph paragraph in parent.Paragraphs)
{
if (paragraph.Node.InnerText.Length > 10)
parent.Score++;
parent.Score += GetCharCount(paragraph.Node);
}
}
ParagraphParent winner = paragraphParents.OrderByDescending(a => a.Score).FirstOrDefault();
// TODO cleanup.
winner.Clean("style");
winner.KillDivs();
winner.KillBreaks();
winner.Clean("form");
winner.Clean("object");
winner.Clean("table", 250);
winner.Clean("h1");
winner.Clean("h2");
winner.Clean("iframe");
winner.Clean("script");
article.Content.DocumentNode.AppendChild(winner.Node);
return article;
}
public IHtmlElement AddElement(IHtmlContainer container, int index, string name)
{
var element = _document.CreateElement(name);
AddNode(container, index, element);
return(element.AsElement());
}
public void when_html_document_doesnt_contain_rss_feed()
{
var document = new HtmlDocument();
document.CreateElement("html");
var parser = new Scraper(document);
var actual = parser.GetRssFeedUrl();
Assert.IsNull(actual);
}
public override void Execute(IEmailItem emailItem = null, int? lastExitCode = null)
{
if (AppliesTo(emailItem, lastExitCode))
{
if (BodyFormat.Text == emailItem.Message.Body.BodyFormat)
{
StringBuilder sb = new StringBuilder();
sb.Append(emailItem.Message.GetBody());
sb.AppendLine();
sb.AppendLine();
sb.Append(Text);
emailItem.Message.SetBody(sb.ToString());
}
if (BodyFormat.Rtf == emailItem.Message.Body.BodyFormat)
{
var messageRtf = new RtfDocument(emailItem.Message.GetBody());
var mergeRtf = new RtfDocument(Rtf);
if (messageRtf.Merge(mergeRtf))
{
emailItem.Message.SetBody(messageRtf.Content);
}
}
if (BodyFormat.Html == emailItem.Message.Body.BodyFormat)
{
var messageDoc = new HtmlDocument();
messageDoc.LoadHtml(emailItem.Message.GetBody());
var disclaimerDoc = new HtmlDocument();
disclaimerDoc.LoadHtml(Html);
var messageBodyNode = messageDoc.DocumentNode.SelectSingleNode("//body");
var disclaimerBodyNode = disclaimerDoc.DocumentNode.SelectSingleNode("//body");
var brNode = messageDoc.CreateElement("br");
messageBodyNode.AppendChild(brNode);
messageBodyNode.AppendChildren(disclaimerBodyNode.ChildNodes);
emailItem.Message.SetBody(messageDoc.DocumentNode.InnerHtml);
}
if (null != Handlers && Handlers.Count > 0)
{
foreach (IHandler handler in Handlers)
{
handler.Execute(emailItem, lastExitCode);
}
}
}
}
public static void Save(SnippetDocumentControl snippet, string filePath)
{
var doc = new HtmlAgilityPack.HtmlDocument();
// use newlines with Stringbuilder so the document formats better than using @""
var node = HtmlAgilityPack.HtmlNode.CreateNode(
new System.Text.StringBuilder()
.AppendLine("<html>")
.AppendLine("<head></head>")
.AppendLine($"<body>{snippet.Html}</body>")
.AppendLine("</html>")
.ToString()
);
doc.DocumentNode.AppendChild(node);
var head = doc.DocumentNode.SelectSingleNode("//head");
var body = doc.DocumentNode.SelectSingleNode("//body");
// save CSS
if (!string.IsNullOrWhiteSpace(snippet.CSS))
{
var s = doc.CreateElement("style");
s.SetAttributeValue("type", "text/css");
s.AppendChild(doc.CreateTextNode(snippet.CSS));
head.AppendChild(s);
}
// save javascript
if (!string.IsNullOrWhiteSpace(snippet.Javascript))
{
var s = doc.CreateElement("script");
s.SetAttributeValue("type", "text/javascript");
s.AppendChild(doc.CreateTextNode(snippet.Javascript));
body.AppendChild(s);
}
AppendSnippetReferences(snippet, doc, head);
doc.Save(filename: filePath);
}
public string TransformImgToPicture(string content)
{
try
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(content);
// - activating"do no harm"-mode
doc.OptionFixNestedTags = false;
doc.OptionAutoCloseOnEnd = false;
doc.OptionCheckSyntax = false;
HtmlNodeCollection collection = doc.DocumentNode.SelectNodes("//img[@src]");
if (collection == null)
return content;
var changed = false;
foreach (HtmlNode img in collection)
{
HtmlAttribute src = img.Attributes["src"];
HtmlAttribute cls = img.Attributes["class"];
if ((src != null && src.Value.IndexOf("slimmage=true", StringComparison.OrdinalIgnoreCase) > -1)
|| (cls != null && cls.Value.IndexOf("slimmage", StringComparison.OrdinalIgnoreCase) > -1))
{
// - append fallback image
HtmlNode container = doc.CreateElement("noscript");
container.SetAttributeValue("data-slimmage", "true");
//copy attributes for IE6/7/8 support
foreach (var a in img.Attributes)
{
container.SetAttributeValue("data-img-" + a.Name, a.Value);
}
//Place 'img' inside 'noscript'
img.ParentNode.InsertBefore(container, img);
img.Remove();
container.AppendChild(img);
changed = true;
}
}
//Don't modify the DOM unless you actually edited the HTML
return changed ? doc.DocumentNode.OuterHtml : content;
}
catch(Exception ex)
{
Trace.TraceWarning("SlimResponse failed to parse HTML: " + ex.ToString() + ex.StackTrace);
// - better that nothing(tm) ...
//return content;
return ex.ToString();
}
}
public string ReplaceDeletedImageBySpanCoupons(string htmlContent)
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
foreach (HtmlNode img in doc.DocumentNode.SelectNodes("//img[@class='" + hdnImageClass.Value + "']"))
{
string value = img.Attributes.Contains("value") ? img.Attributes["value"].Value : " ";
HtmlNode lbl = doc.CreateElement("span");
lbl.Attributes.Add("class", hdnImageClass.Value);
lbl.InnerHtml = value;
img.ParentNode.ReplaceChild(lbl, img);
}
return doc.DocumentNode.OuterHtml;
}
public string ReplaceDeletedIframeBySpan(string htmlContent)
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(htmlContent);
foreach (HtmlNode img in doc.DocumentNode.SelectNodes("//iframe"))
{
HtmlNode lbl = doc.CreateElement("span");
lbl.Attributes.Add("class", "IF");
lbl.Attributes.Add("height", "187");
lbl.Attributes.Add("width", "240");
img.ParentNode.ReplaceChild(lbl, img);
}
return doc.DocumentNode.OuterHtml;
}
public override TaskExecutionDetails Execute(string Value)
{
TaskExecutionDetails d = new TaskExecutionDetails();
HtmlDocument doc = new HtmlDocument();
doc.Load(IO.IOHelper.MapPath(SystemDirectories.Masterpages) + "/" + TargetFile);
//if (doc.DocumentNode.SelectSingleNode(string.Format("//link [@href = '{0}']", Value)) == null)
//{
HtmlNode target = doc.DocumentNode.SelectSingleNode(string.IsNullOrEmpty(TargetSelector) ? "//head" : TargetSelector.ToLower());
if (target != null)
{
HtmlNode s = doc.CreateElement("link");
//s.Name = "link";
s.Attributes.Append("rel", "stylesheet");
s.Attributes.Append("type", "text/css");
s.Attributes.Append("href", Value);
if (!string.IsNullOrEmpty(Media))
s.Attributes.Append("media", Media);
target.AppendChild(s);
doc.Save(IO.IOHelper.MapPath(SystemDirectories.Masterpages) + "/" + TargetFile);
d.TaskExecutionStatus = TaskExecutionStatus.Completed;
d.NewValue = Value;
}
else
d.TaskExecutionStatus = TaskExecutionStatus.Cancelled;
//}
//else
// d.TaskExecutionStatus = TaskExecutionStatus.Cancelled;
return d;
}
public static HtmlNode CreateNode(
HtmlDocument document,
string name,
string className = null,
HtmlNode parentNode = null)
{
var node = document.CreateElement(name);
if (!string.IsNullOrWhiteSpace(className))
{
node.SetAttributeValue("class", className);
}
if (parentNode != null)
{
parentNode.AppendChild(node);
}
return node;
}
public void requestUpdatedDinoTable(string search)
{
//Create the HTML table for the DINOSAURS
var doc = new HtmlAgilityPack.HtmlDocument();
var node = HtmlNode.CreateNode("<html><head></head><body></body></html>");
doc.DocumentNode.AppendChild(node);
HtmlNode table = doc.CreateElement("table");
table.Attributes.Add("class", "mdl-data-table mdl-js-data-table mdl-data-table--selectable mdl-shadow--2dp dino_table_inner");
foreach (ArkDino dino in StaticVars.dinos)
{
if (search == "" || dino.name.ToLower().Contains(search.ToLower()))
{
HtmlNode entry = doc.CreateElement("tr");
entry.AddClass("dino_list_element");
var icon = doc.CreateElement("td");
icon.AddClass("mdl-data-table__cell--non-numeric");
var iconImg = doc.CreateElement("img");
iconImg.Attributes.Add("src", dino.img);
icon.AppendChild(iconImg);
entry.AppendChild(icon);
//name
var name = doc.CreateElement("td");
name.AddClass("mdl-data-table__cell--non-numeric");
name.InnerHtml = dino.name;
entry.AppendChild(name);
//Buttons
var btns = doc.CreateElement("td");
btns.AddClass("mdl-data-table__cell--non-numeric");
btns.AddClass("dino_list_element_btns");
string dinoString = Convert.ToBase64String(Encoding.ASCII.GetBytes(RpTools.SerializeObject(dino)));
btns.InnerHtml = "<button class=\"mdl-button mdl-js-button mdl-button--raised\" onclick=\"SpawnCharacter('" + dinoString + "');\"> Spawn </button>";
entry.AppendChild(btns);
//Add this to the table.
table.AppendChild(entry);
}
}
//Convert this to the html.
string html = table.WriteTo();
//Send this back in base64 because it's a lot of effort to escape it.
var plainTextBytes = System.Text.Encoding.UTF8.GetBytes(html);
StaticVars.browser.ExecuteScriptAsync("FinishRefreshTable('" + System.Convert.ToBase64String(plainTextBytes) + "');");
}
public static string GetPreview(string html, string replacmentHtml, Guid productID)
{
var doc = new HtmlDocument();
doc.LoadHtml(string.Format("<html>{0}</html>", htmlTags.Replace(html, string.Empty)));
var nodes = doc.DocumentNode.SelectNodes("//div[translate(@class,'ABCDEFGHIJKLMNOPQRSTUVWXYZ','abcdefghijklmnopqrstuvwxyz')='asccut']");
if (nodes != null)
{
foreach (var node in nodes)
{
var newNode = doc.CreateElement("div");
var styleAttr = doc.CreateAttribute("style");
styleAttr.Value = "display:inline;";
newNode.Attributes.Append(styleAttr);
newNode.InnerHtml = replacmentHtml ?? string.Empty;
node.ParentNode.ReplaceChild(newNode, node);
}
}
ProcessCustomTags(doc, productID);
return htmlTags.Replace(doc.DocumentNode.InnerHtml, string.Empty);
}
public void Download(HtmlDocument document, string path)
{
Directory.CreateDirectory(ResourcePath);
//Descargar todos los recursos
_downloadImages(document);
_downloadResources(document, "link", "href", true);
_downloadResources(document, "script", "src");
//Añadir cabeceras meta para indificar codificación UTF8
HtmlNode meta = document.CreateElement("meta");
meta.Attributes.Add("charset", "utf-8");
document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta);
/*HtmlNode meta2 = document.CreateElement("meta");
meta2.Attributes.Add("http-equiv", "Content-Type");
meta2.Attributes.Add("content", "Type=text/html; charset=utf-8");
document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta2);*/
File.WriteAllText(path, document.DocumentNode.WriteTo(), new UTF8Encoding(true));
}
private void downloadProductImages(string productUrl, string folderName)
{
Regex urlRegex = new Regex(@"(?:^|\?|&)id=(\d*)");
Match m = urlRegex.Match(productUrl);
string mvalue = m.Value;
string productId = m.Value.Substring(mvalue.IndexOf("id=") + 3);
TmallClient client = new TmallClient();
string description = client.getProductDesc(productId);
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
HtmlNode productDesc = doc.CreateElement("description");
productDesc.InnerHtml = description;
IEnumerable<HtmlNode> imageList = productDesc.Descendants("img");
string path = tbImageFolder.Text + "\\" + folderName;
foreach (HtmlNode imageNode in imageList)
{
string imageUrl = imageNode.GetAttributeValue("src", "");
string imageName = imageUrl.Substring(imageUrl.LastIndexOf("/") + 1);
saveImage(path, imageName, imageUrl);
}
}
public void Download(HtmlDocument document, string path)
{
Directory.CreateDirectory(ResourcePath);
//Descargar todos los recursos
_downloadImages(document);
_downloadResources(document, "link", "href", true);
_downloadResources(document, "script", "src");
//Añadir cabeceras meta para indificar codificación UTF8
HtmlNode meta = document.CreateElement("meta");
meta.Attributes.Add("charset", "utf-8");
document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta);
/*HtmlNode meta2 = document.CreateElement("meta");
* meta2.Attributes.Add("http-equiv", "Content-Type");
* meta2.Attributes.Add("content", "Type=text/html; charset=utf-8");
* document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta2);*/
File.WriteAllText(path, document.DocumentNode.WriteTo(), new UTF8Encoding(true));
}
protected virtual HtmlNode GetBodyNode(HtmlDocument htmlDoc)
{
var bodyNode = htmlDoc.CreateElement("body");
var headNode = htmlDoc.DocumentNode.SelectSingleNode("//head");
if (headNode != null)
{
var linkNodes = headNode.SelectNodes("//link[@href] | //script[@src]");
bodyNode.AppendChildren(linkNodes);
}
var rawBodyNode = htmlDoc.DocumentNode.SelectSingleNode("//body");
if (rawBodyNode == null)
{
bodyNode.AppendChild(htmlDoc.DocumentNode);
}
else
{
bodyNode.AppendChildren(rawBodyNode.ChildNodes);
}
return bodyNode;
}
static private void start_Operation()
{
while (true)
{
if (file_Paths.Count == 0)
{
System.Threading.Thread.Sleep(1000);
}
else
{
try
{
//System.Threading.Interlocked.Increment(ref word_Application_Count);
while (file_Paths.Count > 0)
{
string temp_String = "";
if (file_Paths.TryDequeue(out temp_String))
{
string[] temp_Collection = temp_String.Split(new string[] { "|^|" }, StringSplitOptions.None);
string file_Path = temp_Collection[0];
string web_Page_Base_Address = temp_Collection[1];
string folder_Path = file_Path.Substring(0, file_Path.LastIndexOf("\\"));
string file_Name = file_Path.Substring(file_Path.LastIndexOf("\\") + 1);
if (!System.IO.File.Exists(file_Path))
{
return;// continue;
}
string file_Name_Htm = folder_Path + "\\HTML\\" + System.IO.Path.ChangeExtension(file_Name, "htm");
if (System.IO.File.Exists(file_Name_Htm))
{
return;// continue;
}
Microsoft.Office.Interop.Word._Document doc = app.Documents.Open(file_Path, Visible: false);
//Console.WriteLine(file_Name_Htm);
doc.SaveAs(FileName: file_Name_Htm, FileFormat: Microsoft.Office.Interop.Word.WdSaveFormat.wdFormatHTML);
System.Threading.ThreadPool.QueueUserWorkItem((state) => {
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
// There are various options, set as needed
htmlDoc.OptionFixNestedTags = true;
// filePath is a path to a file containing the html
//string filePath = @"C:\Documents and Settings\prabhakaran\Desktop\rv_3b72167e18a54fe28a330c5fbecc222f.htm";
string filePath = file_Name_Htm;
while (true)
{
try
{
htmlDoc.Load(filePath);
break;
}
catch (Exception excp)
{
}
}
// Use: htmlDoc.LoadXML(xmlString); to load from a string
// ParseErrors is an ArrayList containing any errors from the Load statement
if (htmlDoc.ParseErrors != null && htmlDoc.ParseErrors.Count() > 0)
{
// Handle any parse errors as required
}
else
{
if (htmlDoc.DocumentNode != null)
{
HtmlAgilityPack.HtmlNode bodyNode = htmlDoc.DocumentNode.SelectSingleNode("//body");
if (bodyNode != null)
{
bodyNode.SetAttributeValue("onload", "var temp_Element_Pdf = document.getElementById('PDF Link');temp_Element_Pdf.style.width=screen.width/2;temp_Element_Pdf.style.height='50';var temp_Element_Doc = document.getElementById('DOC Link');temp_Element_Doc.style.width=screen.width/2;temp_Element_Doc.style.height='50';temp_Element_Doc.style.position='absolute';");
HtmlNode pdf_Link = htmlDoc.CreateElement("button"); //new HtmlNode(HtmlNodeType.Element, htmlDoc, 0);
pdf_Link.SetAttributeValue("value", "PDF");
pdf_Link.SetAttributeValue("id", "PDF Link");
string pdf_Path = web_Page_Base_Address + "Local_Database/PDF/" + System.IO.Path.ChangeExtension(file_Name, "pdf");
pdf_Link.SetAttributeValue("onclick", "window.open('" + pdf_Path + "','_blank','location=no,menubar=no,status=no,toolbar=no');");
pdf_Link.InnerHtml = "PDF";
pdf_Link.SetAttributeValue("style", "background-color:#262626;color:#747474;border-style:groove;border-color:#121212;-moz-appearance:none;");
pdf_Link.SetAttributeValue("onmouseover", "this.style.backgroundColor='#262626';this.style.color='#adbdde';this.style.fontWeight='bold';this.style.fontSize='normal';this.style.borderColor='#afafaf';"); //#afafaf
pdf_Link.SetAttributeValue("onmouseout", "this.style.backgroundColor='#262626';this.style.color='#747474';this.style.fontWeight='normal';this.style.fontSize='normal';this.style.borderColor='#121212';");
bodyNode.ChildNodes.Add(pdf_Link);
HtmlNode doc_Link = htmlDoc.CreateElement("button"); //new HtmlNode(HtmlNodeType.Element, htmlDoc, 0);
doc_Link.SetAttributeValue("value", "DOC");
doc_Link.SetAttributeValue("id", "DOC Link");
string doc_Path = web_Page_Base_Address + "Local_Database/DOC/" + System.IO.Path.ChangeExtension(file_Name, "doc");
doc_Link.SetAttributeValue("onclick", "window.open('" + doc_Path + "','_blank','location=no,menubar=no,status=no,toolbar=no');");
doc_Link.InnerHtml = "DOC";
doc_Link.SetAttributeValue("style", "background-color:#262626;color:#747474;border-style:groove;border-color:#121212;-moz-appearance:none;");
doc_Link.SetAttributeValue("onmouseover", "this.style.backgroundColor='#262626';this.style.color='#adbdde';this.style.fontWeight='bold';this.style.fontSize='normal';this.style.borderColor='#afafaf';"); //#afafaf
doc_Link.SetAttributeValue("onmouseout", "this.style.backgroundColor='#262626';this.style.color='#747474';this.style.fontWeight='normal';this.style.fontSize='normal';this.style.borderColor='#121212';");
bodyNode.ChildNodes.Add(doc_Link);
//htmlDoc.Save(@"C:\Documents and Settings\prabhakaran\Desktop\new.html");
htmlDoc.Save(file_Name_Htm);
// Do something with bodyNode
}
}
}// ***
});
string file_Name_Pdf = folder_Path + "\\PDF\\" + System.IO.Path.ChangeExtension(file_Name, "pdf");
//Console.WriteLine(file_Name_Pdf);
doc.SaveAs(FileName: file_Name_Pdf, FileFormat: Microsoft.Office.Interop.Word.WdSaveFormat.wdFormatPDF);
//Console.WriteLine(folder_Path + "\\DOC\\" + file_Name);
((Microsoft.Office.Interop.Word._Document)doc).Close(SaveChanges: Microsoft.Office.Interop.Word.WdSaveOptions.wdDoNotSaveChanges);
while (true)
{
try
{
System.IO.File.Move(file_Path, folder_Path + "\\DOC\\" + file_Name);
break;
}
catch (Exception excp)
{
}
}
}
}
//System.Threading.Interlocked.Decrement(ref word_Application_Count);
}
catch (Exception excp)
{
Console.WriteLine(excp.Message);
}
}
}
}
public static void GetQuoteAndReply ( HtmlNode t , Post p , HtmlDocument doc )
{
HtmlNode div;
HtmlNodeCollection quotes;
div = t.SelectSingleNode ( ".//td[@class='alt1']" ).SelectSingleNode ( "./div[@id]" );
if ( div == null )
{
p.SetFullPost ( t.SelectSingleNode ( ".//td[@class='alt1']" ).InnerText.Trim () );
}
else
{
//replace <br> with new line
HAP.ReplaceNewLine ( div );
//if post have color, remove it
HAP.RemovePostColor ( div );
//set emo link
HAP.SetEmoLink ( div , p );
if ( UserData.settings.GetValueOrDefault ( AppSettings.keyEmo , true ) == false )
HAP.ReplaceEmoWithText ( div );
//edit link
HAP.EditLink ( div , p );
//replace img
HAP.ReplaceImage ( div , p );
//check quotes
quotes = div.SelectNodes ( "./div[@style='margin:20px; margin-top:5px; ']" );
if ( quotes != null )
{
foreach ( HtmlNode quote in quotes )
{
string s = HAP.ProcessQuote ( quote , doc );
HtmlNode newNode = doc.CreateElement ( "title" );
newNode.InnerHtml = HtmlDocument.HtmlEncode ( s );
quote.ParentNode.ReplaceChild ( newNode , quote );
}
}
p.content = HtmlEntity.DeEntitize ( div.InnerText.Trim () );
p.ReplaceImage ();
p.ReplaceNewLine ();
p.ReplaceLink ();
p.SetFullPost ( p.content );
}
}
public static string ProcessQuote ( HtmlNode quote , HtmlDocument doc )
{
RemoveViewPost ( quote );
string quoteContent = "";
quoteContent = Model.Post.GetQuoteBorderTop ();
HtmlNode td = quote.SelectSingleNode ( "./table" ).SelectSingleNode ( ".//td[@class='alt2']" );
if ( td != null )
{
HtmlNodeCollection quotes = td.SelectNodes ( "./div[@style='margin:20px; margin-top:5px; ']" );
if ( quotes != null )
{
foreach ( HtmlNode q in quotes )
{
string s = ProcessQuote ( q , doc );
HtmlNode newNode = doc.CreateElement ( "title" );
newNode.InnerHtml = HtmlDocument.HtmlEncode ( s );
q.ParentNode.ReplaceChild ( newNode , q );
}
}
//check if quote has user and link
HtmlNode strongUserName = td.SelectSingleNode ( ".//strong" );
if ( strongUserName != null )//user quote exist
{
string quoteUser = "******" + HtmlEntity.DeEntitize ( strongUserName.InnerText.Trim () ) + "</b>";
quoteContent += quoteUser + "<br>";
td.RemoveChild ( td.Element ( "div" ) );
}
quoteContent += HtmlEntity.DeEntitize ( td.InnerText.Trim () );
quoteContent += Model.Post.GetQuoteBorderBottom ();
}
else
quoteContent += HtmlEntity.DeEntitize ( td.InnerText.Trim () ).Trim ();
return quoteContent;
}
private static void ProcessZoomImages(HtmlDocument doc, Guid productID)
{
HtmlNodeCollection nodes = doc.DocumentNode.SelectNodes("//img[@_zoom]");
HtmlNode hrefNode;
HtmlAttribute borderAttribute, hrefAttribute, srcAttribute, zoomAttribute;
string imgSrc = string.Empty;
if (nodes == null)
return;
foreach (HtmlNode node in nodes)
{
srcAttribute = node.Attributes["src"];
if (srcAttribute == null || string.IsNullOrEmpty(srcAttribute.Value))
continue;
zoomAttribute = node.Attributes["_zoom"];
if (zoomAttribute == null || string.IsNullOrEmpty(zoomAttribute.Value))
continue;
borderAttribute = node.Attributes["border"];
if (borderAttribute == null)
{
borderAttribute = doc.CreateAttribute("border");
node.Attributes.Append(borderAttribute);
}
borderAttribute.Value = "0";
imgSrc = srcAttribute.Value;
if (!rxNumeric.IsMatch(zoomAttribute.Value))
{
imgSrc = zoomAttribute.Value;
}
if (node.ParentNode != null)
{
hrefNode = doc.CreateElement("a");
hrefAttribute = doc.CreateAttribute("href");
hrefAttribute.Value = imgSrc;
hrefNode.Attributes.Append(hrefAttribute);
hrefAttribute = doc.CreateAttribute("class");
hrefAttribute.Value = "fancyzoom";
hrefNode.Attributes.Append(hrefAttribute);
/*
hrefAttribute = doc.CreateAttribute("onclick");
hrefAttribute.Value = string.Format(@"javascript:if(typeof(popimgFckup) == 'function')popimgFckup('{0}');", srcAttribute.Value);
hrefNode.Attributes.Append(hrefAttribute);*/
node.ParentNode.ReplaceChild(hrefNode, node);
hrefNode.AppendChild(node);
}
}
}
/// <summary>
/// Replaces the base href for some HTML supplied in a string.
/// </summary>
/// <param name="content">HTML content to have its base href replaced.</param>
/// <param name="baseUri">New URI for the base href.</param>
/// <returns>The content with a base element pointing at baseUri.</returns>
public static string SetBaseUri(string content, string baseUri)
{
// Quick and dirty method: Just do a search and replace on a <head> element.
// if (content.IndexOf("<head>") >= 0)
// return content.Replace("<head>", String.Format("<head><base href=\"{0}\" />", response.ResponseUri.AbsoluteUri));
//
// if (content.IndexOf("<HEAD>") >= 0)
// return content.Replace("<HEAD>", String.Format("<HEAD><base href=\"{0}\" />", response.ResponseUri.AbsoluteUri));
// Load the content into an HTML document
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(content);
// find the head element (if there is one)
// first, look for /html/head
HtmlNode head = doc.DocumentNode.SelectSingleNode("/html/head");
if (head == null)
// now look for /head
head = doc.DocumentNode.SelectSingleNode("/head");
if (head == null)
{
// still no head element - create one and insert it
head = doc.CreateElement("head");
doc.DocumentNode.PrependChild(head);
}
if (head != null)
{
// look for a base element in the head
HtmlNode baseElement = head.SelectSingleNode("base");
if (baseElement == null)
{
// not there - create one and insert
baseElement = doc.CreateElement("base");
head.PrependChild(baseElement);
}
// set the base element's href attribute to the new baseUri
baseElement.SetAttributeValue("href", baseUri);
}
else
log.Warn("Could not find or create a head element.");
// convert the document back to a string and return
return doc.DocumentNode.OuterHtml;
}
public void ParseTagForInnerHTMLTest()
{
// arrange
string uriPrefix = string.Empty;
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(this.htmlFileContents);
doc.OptionUseIdAttribute = true;
HtmlNode node = doc.CreateElement("span");
node.InnerHtml = "Access Control";
HtmlParser target = new HtmlParser(uriPrefix);
string expected = "Access Control";
string actual;
// act
actual = target.InnerHTML(node);
// assert
Assert.AreEqual(expected, actual);
}
public static HtmlString Sanitize(this string html)
{
// Run it through Markdown first
var md = new Markdown();
html = md.Transform(html);
// Linkify images if they are shown as resized versions (only relevant for new Markdown comments)
var doc = new HtmlDocument();
doc.LoadHtml(html);
var root = doc.DocumentNode;
if (root != null)
{
var images = root.SelectNodes("//img");
if (images != null)
{
foreach (var image in images)
{
var src = image.GetAttributeValue("src", "");
var orgSrc = src.Replace("rs/", "");
if (src == orgSrc || image.ParentNode.Name == "a") continue;
var a = doc.CreateElement("a");
a.SetAttributeValue("href", orgSrc);
a.SetAttributeValue("target", "_blank");
a.AppendChild(image.Clone());
image.ParentNode.ReplaceChild(a, image);
}
}
// Any links not going to an "approved" domain need to be marked as nofollow
var links = root.SelectNodes("//a");
if (links != null)
{
foreach (var link in links)
{
if (link.Attributes["href"] != null && (SpamChecker.CountValidLinks(link.Attributes["href"].Value, 0) == 0))
{
if (link.Attributes["rel"] != null)
{
link.Attributes.Remove("rel");
}
link.Attributes.Add("rel", "nofollow");
}
}
}
// Remove styles from all elements
var elementsWithStyleAttribute = root.SelectNodes("//@style");
if (elementsWithStyleAttribute != null)
{
foreach (var element in elementsWithStyleAttribute)
{
element.Attributes.Remove("style");
}
}
using (var writer = new StringWriter())
{
doc.Save(writer);
html = writer.ToString();
}
}
return new HtmlString(Utils.Sanitize(html));
}
/// <summary>
/// This method needs heavy revision.
///
/// Since element on which was module dropped is COM interface IHTMLElement
/// and that seems like it's not supporting own module tag (it gets removed on whatever operation),
/// this method adds id to IHTMLElement(if needed), to be able to identify it in HtmlAgilityPack HtmlDocument
/// then adds preview output as end to that element, then converts document back to COM IHTMLDocument2
///
/// Doh.
///
/// TODO: Could use hard refactoring, probably own customized html editor
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
void theSite_drop(DataObject sender, DragEventArgs e)
{
CFormController.Instance.mainForm.setStatus("Module(s) Added");
if (sender.GetData("System.Windows.Forms.ListView+SelectedListViewItemCollection", false) != null)
{
// Get module preview (multiple modules can be dragged)
String input = String.Empty;
ListView.SelectedListViewItemCollection listViewItemModules = (ListView.SelectedListViewItemCollection)sender.GetData("System.Windows.Forms.ListView+SelectedListViewItemCollection", false);
foreach (ListViewItem listViewItemModule in listViewItemModules)
{
input += CXMLParser.Instance.GetPreviewFromProjectXML(CXMLParser.Instance.GetNodeFromModule(CModuleReader.Instance.GetModuleInstanceFromName(listViewItemModule.Text)).OuterHtml);
}
// Get relative drop location
Point htmlEditorCorner = htmlEditor1.PointToScreen(new Point(0, 0));
int X = e.X - htmlEditorCorner.X;
int Y = e.Y - htmlEditorCorner.Y;
// Get element on which module was dropped
IHTMLElement hoverElem = htmlEditor1.HtmlDocument2.ElementFromPoint(X, Y);
IHTMLElement moduleElem = null;
// If it gets dropped on module, pass its parent element instead
if (CRestrictedEditDesigner.isModule(hoverElem, out moduleElem))
{
hoverElem = moduleElem;
}
if (hoverElem.tagName.Equals("BODY"))
{
Debug.WriteLine("dropped on body");
if (hoverElem.innerText == null && hoverElem.innerHTML == null)
{
htmlEditor1.LoadDocument("<body>" + input + "</body>");
}
else
{
htmlEditor1.LoadDocument("<body>" + hoverElem.innerHTML + input + "</body>");
}
}
else
{
Debug.WriteLine("dropped on " + hoverElem.tagName);
//Mshtml deletes <module> in element load,
//uhm so it has to be converted to HtmlAgilityPack.HtmlDocument
//and then back
String guid = Guid.NewGuid().ToString();
Boolean idChanged;
if (hoverElem.id == null)
{
hoverElem.id = guid;
idChanged = true;
}
else
{
guid = hoverElem.id;
idChanged = false;
}
// Get wanted element and modify its content
HtmlAgilityPack.HtmlDocument htmlDoc = HTMLDocumentConverter.mshtmlDocToAgilityPackDoc(htmlEditor1.HtmlDocument2);
HtmlAgilityPack.HtmlNode node = htmlDoc.GetElementbyId(guid);
// Dont remove id if it was there before
if (idChanged)
{
node.Attributes.Remove("id");
}
// Need to create element, because HtmlNode dont have OuterHtml settable
HtmlNode addedModulesNode = htmlDoc.CreateElement("div");
addedModulesNode.InnerHtml = input;
try
{
// Well, this sometimes fails.. god knows why
htmlDoc.DocumentNode.InsertAfter(addedModulesNode, node);
}
catch (Exception)
{
// So if it fails, add module in the end of parent module
node.ParentNode.InnerHtml += input;
}
// And back to IHTMLDocument
htmlEditor1.LoadDocument("<body>" + htmlDoc.DocumentNode.InnerHtml + "</body>");
}
}
}
private static List<HelpCenterItem> ParseHelpCenterHtml(string html, string helpLinkBlock)
{
var helpCenterItems = new List<HelpCenterItem>();
if (string.IsNullOrEmpty(html)) return helpCenterItems;
var doc = new HtmlDocument();
doc.LoadHtml(html);
var urlHelp = CommonLinkUtility.GetHelpLink(false);
var mainContent = doc.DocumentNode.SelectSingleNode("//div[@class='MainHelpCenter GettingStarted']");
if (mainContent == null) return helpCenterItems;
var blocks = (mainContent.SelectNodes(".//div[@class='gs_content']"))
.Where(r => r.Attributes["id"] != null)
.Select(x => x.Attributes["id"].Value).ToList();
foreach (var block in mainContent.SelectNodes(".//div[@class='gs_content']"))
{
var hrefs = block.SelectNodes(".//a[@href]")
.Where(r =>
{
var value = r.Attributes["href"].Value;
return r.Attributes["href"] != null
&& !string.IsNullOrEmpty(value)
&& !value.StartsWith("mailto:")
&& !value.StartsWith("http");
});
foreach (var href in hrefs)
{
var value = href.Attributes["href"].Value;
if (value.IndexOf("#", StringComparison.Ordinal) != 0 && value.Length > 1)
{
href.Attributes["href"].Value = urlHelp + value.Substring(1);
href.SetAttributeValue("target", "_blank");
}
else
{
if (!blocks.Contains(value.Substring(1))) continue;
href.Attributes["href"].Value = helpLinkBlock + blocks.IndexOf(value.Substring(1)).ToString(CultureInfo.InvariantCulture);
}
}
var images = block.SelectNodes(".//img");
if (images != null)
{
foreach (var img in images.Where(img => img.Attributes["src"] != null))
{
img.Attributes["src"].Value = GetInternalLink(urlHelp + img.Attributes["src"].Value);
}
foreach (var screenPhoto in images.Where(img =>
img.Attributes["class"] != null && img.Attributes["class"].Value.Contains("screenphoto")
&& img.Attributes["target"] != null && img.ParentNode != null))
{
var bigphotoScreenId = screenPhoto.Attributes["target"].Value;
var bigphotoScreen = images.FirstOrDefault(img =>
img.Attributes["id"] != null && img.Attributes["id"].Value == bigphotoScreenId
&& img.Attributes["class"] != null && img.Attributes["class"].Value.Contains("bigphoto_screen")
&& img.Attributes["src"] != null);
if (bigphotoScreen == null) continue;
var hrefNode = doc.CreateElement("a");
var hrefAttribute = doc.CreateAttribute("href");
hrefAttribute.Value = bigphotoScreen.Attributes["src"].Value;
hrefNode.Attributes.Append(hrefAttribute);
hrefAttribute = doc.CreateAttribute("class");
hrefAttribute.Value = "screenzoom";
hrefNode.Attributes.Append(hrefAttribute);
hrefAttribute = doc.CreateAttribute("rel");
hrefAttribute.Value = "imageHelpCenter";
hrefNode.Attributes.Append(hrefAttribute);
screenPhoto.ParentNode.ReplaceChild(hrefNode, screenPhoto);
hrefNode.AppendChild(screenPhoto);
}
}
var titles = block.SelectSingleNode(".//h2");
var contents = block.SelectSingleNode(".//div[@class='PortalHelp']");
if (titles != null && contents != null)
{
helpCenterItems.Add(new HelpCenterItem { Title = titles.InnerText, Content = contents.InnerHtml });
}
}
return helpCenterItems;
}
private static void ProcessZoomImages(HtmlDocument doc)
{
var nodes = doc.DocumentNode.SelectNodes("//img[@_zoom]");
if (nodes == null) return;
foreach (var node in nodes)
{
var srcAttribute = node.Attributes["src"];
if (srcAttribute == null || string.IsNullOrEmpty(srcAttribute.Value)) continue;
var zoomAttribute = node.Attributes["_zoom"];
if (zoomAttribute == null || string.IsNullOrEmpty(zoomAttribute.Value)) continue;
var borderAttribute = node.Attributes["border"];
if (borderAttribute == null)
{
borderAttribute = doc.CreateAttribute("border");
node.Attributes.Append(borderAttribute);
}
borderAttribute.Value = "0";
var imgSrc = srcAttribute.Value;
if (!RxNumeric.IsMatch(zoomAttribute.Value))
{
imgSrc = zoomAttribute.Value;
}
if (node.ParentNode != null)
{
var hrefNode = doc.CreateElement("a");
var hrefAttribute = doc.CreateAttribute("href");
hrefAttribute.Value = imgSrc;
hrefNode.Attributes.Append(hrefAttribute);
hrefAttribute = doc.CreateAttribute("class");
hrefAttribute.Value = "screenzoom";
hrefNode.Attributes.Append(hrefAttribute);
/*
hrefAttribute = doc.CreateAttribute("onclick");
hrefAttribute.Value = string.Format(@"javascript:if(typeof(popimgFckup) == 'function')popimgFckup('{0}');", srcAttribute.Value);
hrefNode.Attributes.Append(hrefAttribute);*/
node.ParentNode.ReplaceChild(hrefNode, node);
hrefNode.AppendChild(node);
}
}
}
public static string SetBaseUri(string content, string baseUri)
{
// if (content.IndexOf("<head>") >= 0)
// return content.Replace("<head>", String.Format("<head><base href=\"{0}\" />", response.ResponseUri.AbsoluteUri));
//
// if (content.IndexOf("<HEAD>") >= 0)
// return content.Replace("<HEAD>", String.Format("<HEAD><base href=\"{0}\" />", response.ResponseUri.AbsoluteUri));
// Tidy HTML and rebase
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(content);
HtmlNode head = doc.DocumentNode.SelectSingleNode("/html/head");
if (head == null)
head = doc.DocumentNode.SelectSingleNode("/head");
if (head == null)
{
head = doc.CreateElement("head");
doc.DocumentNode.PrependChild(head);
}
if (head != null)
{
HtmlNode baseElement = head.SelectSingleNode("base");
if (baseElement == null)
{
baseElement = doc.CreateElement("base");
head.PrependChild(baseElement);
}
baseElement.SetAttributeValue("href", baseUri);
}
return doc.DocumentNode.OuterHtml;
}
private void CreatePlayer(ClipData clip, string folder)
{
MetaDataData[] metadata = mediamanager.GetClipMetaData(clip.ID);
List<MetaDataData> indices = metadata
.Where(x => x.TimeStamp > 0 && x.Payload.GetType().Name == "AgendaItem" )
.OrderBy(x => x.TimeStamp)
.ToList<MetaDataData>();
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.Load(folder + @"\play.html");
foreach (HtmlNode div in doc.DocumentNode.SelectNodes("//span[@id='title']"))
{
div.InnerHtml = clip.Name;
}
foreach (HtmlNode div in doc.DocumentNode.SelectNodes("//ul[@id='jumpto']"))
{
indices.ForEach(index =>
{
HtmlNode li = doc.CreateElement("li");
div.AppendChild(li);
HtmlNode link = doc.CreateElement("a");
link.Attributes.Add("data-rel", index.TimeStamp.ToString());
link.Attributes.Add("href", "#");
link.InnerHtml = index.Name;
li.AppendChild(link);
});
}
doc.Save(folder+ @"\play.html");
}
public static string Sanitize(string html)
{
// Run it through Markdown first
var md = new Markdown();
html = md.Transform(html);
// Linkify images if they are shown as resized versions (only relevant for new Markdown comments)
var doc = new HtmlDocument();
doc.LoadHtml(html);
var root = doc.DocumentNode;
if (root != null)
{
var images = root.SelectNodes("//img");
if (images != null)
{
var replace = false;
foreach (var image in images)
{
var src = image.GetAttributeValue("src", "");
var orgSrc = src.Replace("rs/", "");
if (src == orgSrc || image.ParentNode.Name == "a") continue;
var a = doc.CreateElement("a");
a.SetAttributeValue("href", orgSrc);
a.SetAttributeValue("target", "_blank");
a.AppendChild(image.Clone());
image.ParentNode.ReplaceChild(a, image);
replace = true;
}
if (replace)
{
html = root.OuterHtml;
}
}
}
return Utills.Sanitize(html);
}
private static void ProcessZoomImages(HtmlDocument doc)
{
var nodes = doc.DocumentNode.SelectNodes("//img[@_zoom]");
if (nodes == null) return;
foreach (var node in nodes)
{
if (node.ParentNode != null && (node.ParentNode.Name ?? "").ToLower() == "a") continue;
var srcAttribute = node.Attributes["src"];
if (srcAttribute == null || string.IsNullOrEmpty(srcAttribute.Value)) continue;
var zoomAttribute = node.Attributes["_zoom"];
if (zoomAttribute == null || string.IsNullOrEmpty(zoomAttribute.Value)) continue;
var borderAttribute = node.Attributes["border"];
if (borderAttribute == null)
{
borderAttribute = doc.CreateAttribute("border");
node.Attributes.Append(borderAttribute);
}
borderAttribute.Value = "0";
var imgSrc = srcAttribute.Value;
if (!RxNumeric.IsMatch(zoomAttribute.Value))
{
imgSrc = zoomAttribute.Value;
}
if (node.ParentNode != null)
{
var hrefNode = doc.CreateElement("a");
var hrefAttribute = doc.CreateAttribute("href");
hrefAttribute.Value = imgSrc;
hrefNode.Attributes.Append(hrefAttribute);
hrefAttribute = doc.CreateAttribute("class");
hrefAttribute.Value = "screenzoom";
hrefNode.Attributes.Append(hrefAttribute);
string title = null;
var titleAttribute = node.Attributes["title"];
if (titleAttribute != null)
{
title = titleAttribute.Value;
}
else
{
var altAttribute = node.Attributes["alt"];
if (altAttribute != null)
{
title = altAttribute.Value;
}
}
if (!string.IsNullOrEmpty(title))
{
hrefAttribute = doc.CreateAttribute("title");
hrefAttribute.Value = title;
hrefNode.Attributes.Append(hrefAttribute);
}
node.ParentNode.ReplaceChild(hrefNode, node);
hrefNode.AppendChild(node);
}
}
}
private void HtmlEdit(string path, bool DoTransfer, bool ToTraidional)
{
// string fullpath = @"temp\unzipping\" + OpfBasedPath + @"\" + url;
FileStream xhtmlloder = new FileStream(path, FileMode.Open, FileAccess.Read);
StreamReader sr = new StreamReader(xhtmlloder);
string source = sr.ReadToEnd();
source = HttpUtility.HtmlDecode(source);
sr.Close();
xhtmlloder.Close();
var doc = new HtmlAgilityPack.HtmlDocument();
// doc.OptionOutputAsXml = true;
bool hasCSSStylesheet = false;
doc.OptionWriteEmptyNodes = true;
doc.LoadHtml(source);
foreach (HtmlNode node in doc.DocumentNode.ChildNodes)
{
if (node.Name == "html")
{
foreach (HtmlNode bodynode in node.ChildNodes)
{
if (bodynode.Name == "head")
{
foreach (HtmlNode item in bodynode.ChildNodes)
{
if (item.Name == "link")
{
if (item.Attributes["rel"].Value == "stylesheet" && item.Attributes["type"].Value == "text/css")
{
hasCSSStylesheet = true;
}
}
}
if (hasCSSStylesheet == false)
{
HtmlNode verticalstyle = doc.CreateElement("style");
verticalstyle.InnerHtml = ChoHoe.Properties.Resources.VerticalStyle;
bodynode.AppendChild(verticalstyle);
}
}
if (bodynode.Name == "body")
{
foreach (HtmlNode body_child_Node in bodynode.ChildNodes)
{
RecursivelyReplaceText(body_child_Node, ToTraidional, DoTransfer);
if (isRemoveCss)
{
RemoveHtmlStyle(body_child_Node);
}
}
}
}
}
//if (ToTradictional == true)
//{
// after = ChineseConverter.Convert(before, ChineseConversionDirection.SimplifiedToTraditional);
//}
//else
//{
// after = ChineseConverter.Convert(before, ChineseConversionDirection.TraditionalToSimplified);
//}
//node.ParentNode.ReplaceChild(HtmlTextNode.CreateNode(after), node);
//node.InnerText.Replace(node.InnerText,after);
}
// Console.WriteLine(doc.DocumentNode.InnerHtml);
FileStream sw = new FileStream(path, FileMode.Create);
// doc.DocumentNode.InnerHtml= HttpUtility.HtmlEncode(doc.ParsedText);
doc.Save(sw, System.Text.Encoding.UTF8);
sw.Close();
}
/// <summary>
/// 下载html
/// </summary>
public void DownLoadHtml()
{
string sqlStr = "select Url from IcookMenu where IsDownload = 0";
object[] menuUrlObj = sh.GetField(sqlStr);
string title, fullFoldPath = string.Empty;
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
ArrayList mainAList = hh.GetHtmlData(mainUrl, cookie);
int htmlCount = 0;
foreach (var menuUrl in menuUrlObj)
{
try
{
ArrayList menuResList = hh.GetHtmlData(menuUrl.ToString(), cookie);
doc.LoadHtml(menuResList[1].ToString());
HtmlNode h1Node = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-header-title']/h1[@class='title']");
title = myUtils.StringConvert(h1Node.InnerText).Trim();//标题
title = myUtils.FilterPath(title);
fullFoldPath = outPath + title + @"\";
if (!Directory.Exists(fullFoldPath)) //判断是否存在
{
Directory.CreateDirectory(fullFoldPath); //创建新路径
}
HtmlNode headerNode = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-header recipe-details-block']");
HtmlNode headerChild = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-header recipe-details-block']/div[@class='header-row center-row']");
HtmlNode rightChild = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-header recipe-details-block']/div[@class='header-row center-row']/div[@class='header-col right-col']");
headerChild.RemoveChild(rightChild);//删除右边
HtmlNode headerImgParentNode = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-header recipe-details-block']/div[@class='header-row center-row']/div[@class='header-col left-col']/div[@class='recipe-cover']");
HtmlNode headerImgChild = doc.DocumentNode.SelectSingleNode("//img[@class='main-pic']");
string headerImgSrc = headerImgChild.GetAttributeValue("src", "");
myUtils.DownLoadImage(headerImgSrc, fullFoldPath + @"图片1.jpg", cookie);
headerImgParentNode.RemoveAllChildren();
HtmlNode newheaderImgNode = doc.CreateElement("div");
newheaderImgNode.InnerHtml = $"图片{1}";
headerImgParentNode.AppendChild(newheaderImgNode);
string headerHtml = headerNode.InnerHtml;//头部内容
HtmlNode mainNode = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-main']");
HtmlNode mainChild = doc.DocumentNode.SelectSingleNode("//div[@class='recipe-details-main']/div[@class='recipe-ad-placeholder']");
mainNode.RemoveChild(mainChild);
HtmlNodeCollection imgParentNodeList = doc.DocumentNode.SelectNodes("//div[@class='step-cover']");
int totalImgCount = 0, index = 0;
if (imgParentNodeList != null)
{
totalImgCount = imgParentNodeList.Count();
for (int i = 1; i < imgParentNodeList.Count + 1; i++)
{
try
{
HtmlNode imgChildNode = imgParentNodeList[i - 1].SelectSingleNode("a");
string imgUrl = imgChildNode.GetAttributeValue("href", "").Replace("medium_", "large_");
myUtils.DownLoadImage(imgUrl, fullFoldPath + $"图片{i + 1}.jpg", cookie);
imgParentNodeList[i - 1].RemoveAllChildren();
HtmlNode newImgNode = doc.CreateElement("div");
newImgNode.InnerHtml = $"图片{i + 1}";
imgParentNodeList[i - 1].AppendChild(newImgNode);
index++;
}
catch (Exception ex)
{
myUtils.WriteLog(ex);
}
}
}
string mainStr = mainNode.InnerHtml;//主题内容
string allStr = headerHtml + mainStr;
// sqlStr = $"UPDATE IcookMenu SET Title = '{title}', Html = '{allStr}' WHERE Url = '{menuUrl}'";
sqlStr = $"UPDATE IcookMenu SET Title = '{title}' WHERE Url = '{menuUrl}'";
sh.RunSql(sqlStr);
bool isOk = false;
if (totalImgCount == index)
{
if (myUtils.TransToWord(allStr, title, fullFoldPath))
{
if (myUtils.InsertPictureToWord(outPath, title))
{
sqlStr = $"UPDATE IcookMenu SET IsDownload = 1 WHERE Url = '{menuUrl}'";
sh.RunSql(sqlStr);
htmlCount++;
myUtils.UpdateLabel(label3, htmlCount);
myUtils.UpdateListBox(listBox1, title);
isOk = true;
}
}
}
if (!isOk)
{
if (Directory.Exists(fullFoldPath))
{
Directory.Delete(fullFoldPath, true);
}
}
}
catch (Exception e)
{
myUtils.WriteLog(e);
}
}
}
/// <summary>
/// 下载html
/// </summary>
public void DownLoadHtml()
{
string sqlStr = "select Url from TeeprNewsUrl where IsDownload = 0";
object[] newsUrlObj = sh.GetField(sqlStr);
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
ArrayList httpList = hh.GetHtmlData(mainUrl, cookie);
string title, htmlStr, fullFoldPath = string.Empty;
int htmlCount = 0;
foreach (var newsUrl in newsUrlObj)
{
try
{
httpList = hh.GetHtmlData(newsUrl.ToString(), cookie);
doc.LoadHtml(httpList[1].ToString());
HtmlNode titleNode = doc.DocumentNode.SelectSingleNode("//h1[@class='title single-title']");
title = titleNode.InnerText;
title = myUtils.StringConvert(title).Trim();
title = myUtils.FilterPath(title);
fullFoldPath = outPath + title + @"\";
if (!Directory.Exists(fullFoldPath))
{
Directory.CreateDirectory(fullFoldPath);
}
HtmlNode mianContentNode = doc.DocumentNode.SelectSingleNode("//div[@class='post-single-wrapper']");
HtmlNodeCollection adNodeList = doc.DocumentNode.SelectNodes("//div[@class='post-single-wrapper']/div[@class='mid-post-ad-2']");
foreach (HtmlNode adNode in adNodeList)
{
try
{
mianContentNode.RemoveChild(adNode);
}
catch (Exception ex)
{
myUtils.WriteLog(ex);
}
}
HtmlNodeCollection imgDivNodeList = doc.DocumentNode.SelectNodes("//div[@class='post-single-wrapper']/div");
HtmlNodeCollection imgList = doc.DocumentNode.SelectNodes("//div[@class='post-single-wrapper']/div/img");
string imgUrl = string.Empty;
int totalImg = 0, imgCount = 0;
if (imgDivNodeList != null)
{
totalImg = imgDivNodeList.Count();
for (int i = 0; i < imgDivNodeList.Count(); i++)
{
try
{
HtmlNode imgNode = imgDivNodeList[i].FirstChild;
if (imgNode != null)
{
imgUrl = imgNode.GetAttributeValue("src", "");
if ((!string.IsNullOrEmpty(imgUrl) || imgUrl.Contains(".jpg") ||
imgUrl.Contains(".png") || imgUrl.Contains(".JPEG")) &&
(!imgUrl.Contains("video") && !imgUrl.Contains("width")))
{
myUtils.DownLoadImage(imgUrl, fullFoldPath + $"图片{imgCount + 1}.jpg", cookie);
imgDivNodeList[i].RemoveChild(imgNode);
HtmlNode newImgNode = doc.CreateElement("div");
newImgNode.InnerHtml = $"图片{imgCount + 1}";
imgDivNodeList[i].AppendChild(newImgNode);
imgCount++;
}
}
}
catch (Exception e)
{
myUtils.WriteLog(e);
}
}
}
htmlStr = mianContentNode.InnerHtml;
sqlStr = $"UPDATE TeeprNewsUrl SET Title = '{title}' WHERE Url = '{newsUrl}'";
sh.RunSql(sqlStr);
bool isOk = false;
if (myUtils.TransToWord(htmlStr, title, fullFoldPath))
{
if (myUtils.InsertPictureToWord(outPath, title))
{
sqlStr = $"UPDATE TeeprNewsUrl SET IsDownload = 1 WHERE Url = '{newsUrl}'";
sh.RunSql(sqlStr);
htmlCount++;
myUtils.UpdateLabel(label3, htmlCount);
myUtils.UpdateListBox(listBox1, title);
isOk = true;
}
}
if (!isOk)
{
if (Directory.Exists(fullFoldPath))
{
Directory.Delete(fullFoldPath, true);
}
}
}
catch (Exception ew)
{
myUtils.WriteLog(ew);
}
}
}
public void ParseTagForAttributeValueTest_input()
{
// arrange
string uriPrefix = string.Empty;
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(this.htmlFileContents);
doc.OptionUseIdAttribute = true;
HtmlNode node = doc.CreateElement("input");
node.SetAttributeValue("value", "NSACSSEA");
HtmlParser target = new HtmlParser(uriPrefix);
string returnAttributeName = "value";
string expected = "NSACSSEA";
string actual;
// act
actual = target.ParseTagForAttributeValue(node, returnAttributeName);
// assert
Assert.AreEqual(expected, actual);
}