public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.HtmlDoc, "args.HtmlDoc");
Assert.ArgumentNotNull(
args.HtmlDoc.DocumentNode,
"args.HtmlDoc.DocumentNode");
Assert.ArgumentNotNullOrEmpty(args.Name, "args.Name");
Assert.ArgumentNotNull(args.Path, "args.Path");
this.Process(args.HtmlDoc.DocumentNode, args);
}
//TODO: ensure adding things to body, not root
public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.HtmlDoc, "args.HtmlDoc");
Assert.ArgumentNotNull(args.HtmlDoc.DocumentNode, "args.HtmlDoc.DocumentNode");
//TODO: this should not be necessary; may be due to no <html> element or wrong method for adding elements
args.HtmlDoc.LoadHtml(args.HtmlDoc.DocumentNode.WriteTo());
foreach (HtmlNode child in args.HtmlDoc.DocumentNode.ChildNodes)
{
if (child.NodeType != HtmlNodeType.Element || child.Name != "body")
{
continue;
}
foreach (HtmlNode grandchild in child.ChildNodes)
{
if (grandchild.NodeType != HtmlNodeType.Element)
{
continue;
}
ProcessWordNodeTreeArgs innerArgs = new ProcessWordNodeTreeArgs(grandchild);
CorePipeline.Run("processWordNodeTree", innerArgs);
HtmlNode next = innerArgs.NextElement;
if (innerArgs.TokenTable != null)
{
args.TokenTable = innerArgs.TokenTable;
}
while (next != null)
{
ProcessWordNodeTreeArgs innerInnerArgs = new ProcessWordNodeTreeArgs(next);
CorePipeline.Run("processWordNodeTree", innerInnerArgs);
if (innerInnerArgs.TokenTable != null)
{
Assert.IsNull(args.TokenTable, "multiple token tables");
args.TokenTable = innerInnerArgs.TokenTable;
}
next = innerInnerArgs.NextElement;
}
break;
}
}
}
public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.WordMediaItem, "args.WordMediaItem");
if (args.WordMediaItem.Publishing.NeverPublish)
{
return;
}
using (new EditContext(args.WordMediaItem))
{
args.WordMediaItem.Publishing.NeverPublish = true;
}
}
private void Process(HtmlNode node, RenderWordArgs args)
{
Assert.ArgumentNotNull(node, "node");
if (node.NodeType == HtmlNodeType.Element)
{
ProcessWordElementArgs innerArgs = new ProcessWordElementArgs(
args.Database,
node,
args.Name,
args.Path);
CorePipeline.Run("processWordElement", innerArgs);
}
foreach (HtmlNode child in node.ChildNodes)
{
this.Process(child, args);
}
}
public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.HtmlDoc, "args.HtmlDoc");
Assert.ArgumentNotNull(args.HtmlDoc.DocumentNode, "args.HtmlDoc.DocumentNode");
Assert.ArgumentNotNull(args.TitleField, "args.TitleField");
Assert.ArgumentNotNull(args.BodyField, "args.BodyField");
Assert.ArgumentNotNull(args.TokenTableField, "args.TokenTableField");
using (new EditContext(args.BodyField.Item))
{
args.BodyField.Value = Regex.Replace(
args.HtmlDoc.DocumentNode.WriteTo(),
"</?(body|sitecorebulletedlistchar)>",
string.Empty).Replace("<code>\r\n", "<code>");
}
if ((!string.IsNullOrEmpty(args.Title)) && args.Title != args.TitleField.Value)
{
using (new EditContext(args.TitleField.Item))
{
args.TitleField.Value = args.Title;
}
}
string value = string.Empty;
//TODO: if token table is empty, should clear token table?
if (args.TokenTable != null)
{
foreach (string key in args.TokenTable.Keys)
{
value += key + "=" + args.TokenTable[key].Trim() + "&";
}
using (new EditContext(args.TokenTableField.Item))
{
args.TokenTableField.Value = value.TrimEnd('&');
}
}
}
public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.WordDoc, "args.WordDoc");
// create a root <body> element for an XML structure
args.HtmlDoc = new HtmlDocument();
HtmlNode htmlBody = args.HtmlDoc.CreateElement("body");
args.HtmlDoc.DocumentNode.ChildNodes.Add(htmlBody);
Body body = args.WordDoc.MainDocumentPart.Document.Body;
// for each element in the Word document
foreach (OpenXmlElement current in body.ChildElements)
{
Paragraph paragraph = current as Paragraph;
// Word treats almost everything (list items, etc.) as paragraphs.
if (paragraph != null)
{
// the name of the style, which can also identify what should be <li>
// or could be used as or to determine values for the class attribute
string style = null;
if (paragraph.ParagraphProperties != null && paragraph.ParagraphProperties.ParagraphStyleId != null)
{
style = paragraph.ParagraphProperties.ParagraphStyleId.Val.ToString().ToLower();
}
// the text of the "paragraph"
string text = string.Empty;
// for each thing in the paragraph
foreach (var child in paragraph.ChildElements)
{
// ignore paragraph properties.
if (child is ParagraphProperties)
{
continue;
}
// some paragraphs may consist of a single run in a single style
SdtRun sdtRun = child as SdtRun;
if (sdtRun != null)
{
text += HttpUtility.HtmlEncode(sdtRun.InnerText);
if (sdtRun.HasChildren)
{
//TODO: process images and runs?
}
continue;
}
// outer for-each invokes this for each run in the paragraph
Run run = child as Run;
if (run != null)
{
// if the run has a specific style, add it as an element name around the text
if (run.RunProperties != null && run.RunProperties.RunStyle != null)
{
string name = run.RunProperties.RunStyle.Val.ToString().ToLower();
text += "<" + name + ">" + HttpUtility.HtmlEncode(run.InnerText) + "</" + name + ">";
}
else
{
text += HttpUtility.HtmlEncode(run.InnerText);
}
foreach (Drawing drawing in run.Descendants <Drawing>())
{
string alt = null;
string title = null;
// path to image file on disk
string file = null;
string width = null;
string height = null;
// for each drawing in the run
foreach (DocumentFormat.OpenXml.Drawing.Extents extent
in drawing.Descendants <DocumentFormat.OpenXml.Drawing.Extents>())
{
if (extent.Cx != null)
{
width = this.GetPixels(extent.Cx.ToString()).ToString(CultureInfo.InvariantCulture);
}
if (extent.Cy != null)
{
height = this.GetPixels(extent.Cy.ToString()).ToString(CultureInfo.InvariantCulture);
}
}
// there is probably only one DocProperties per image
// retrieve alt text, title, and part of file name if possible
foreach (DocProperties docProps in drawing.Descendants <DocProperties>())
{
if (!string.IsNullOrEmpty(docProps.Title))
{
title = docProps.Title;
}
if (!string.IsNullOrEmpty(docProps.Description))
{
alt = docProps.Description;
}
if (!string.IsNullOrEmpty(docProps.Name))
{
file = docProps.Name;
}
if (string.IsNullOrEmpty(alt))
{
if (string.IsNullOrEmpty(title))
{
if (!string.IsNullOrEmpty(file))
{
alt = file;
}
}
else
{
alt = title;
}
}
}
string unique = null;
foreach (Blip blip
in drawing.Descendants <Blip>())
{
OpenXmlAttribute attribute = blip.GetAttribute(
"embed",
"http://schemas.openxmlformats.org/officeDocument/2006/relationships");
if (attribute != null)
{
Assert.IsNull(unique, "unique");
unique = attribute.Value;
ImagePart imagePart = (ImagePart)args.WordDoc.MainDocumentPart.GetPartById(attribute.Value);
Assert.IsNotNull(imagePart, "imagePart");
}
else
{
Log.Info("no embed", this);
}
}
if (!string.IsNullOrEmpty(unique))
{
string path = this.WriteImageFile(args.WordDoc, "C:\\temp", unique, file).Replace("\\", "\\\\");
text += "<img src=\"" + path + "\" height=\"" + height + "\" width=\"" + width + "\" alt=\"" + alt + "\" title=\"" + title + "\" />";
}
}
continue;
}
// ignore these types of elements for now
if (child is ProofError ||
child is BookmarkStart ||
child is BookmarkEnd)
{
continue;
}
// identify any new types of elements
throw new Exception("Unknown child element: " + child.GetType());
}
string html = "<p";
if (!string.IsNullOrEmpty(style))
{
html += " class=\"" + style + "\"";
}
HtmlNode para = HtmlNode.CreateNode(html + ">" + text + "</p>");
htmlBody.ChildNodes.Add(para);
continue;
}
Table table = current as Table;
if (table != null)
{
HtmlNode tableNode = args.HtmlDoc.CreateElement("table");
HtmlNode tableBody = args.HtmlDoc.CreateElement("tbody");
tableNode.ChildNodes.Add(tableBody);
foreach (TableRow row in table.Elements <TableRow>())
{
HtmlNode tableRow = args.HtmlDoc.CreateElement("tr");
tableBody.ChildNodes.Add(tableRow);
foreach (TableCell cell in row.Elements <TableCell>())
{
HtmlNode tableCell = args.HtmlDoc.CreateElement("td");
tableRow.ChildNodes.Add(tableCell);
foreach (Paragraph para in cell.Elements <Paragraph>())
{
//TODO: process runs?
tableCell.ChildNodes.Add(HtmlNode.CreateNode(para.InnerText));
}
}
}
htmlBody.ChildNodes.Add(tableNode);
continue;
}
// ignore these types of elements
if (current is SectionProperties ||
current is BookmarkEnd)
{
continue;
}
throw new Exception("Unknown element: " + current.GetType());
}
}
public override void Process(RenderWordArgs args)
{
Assert.ArgumentNotNull(args, "args");
Assert.ArgumentNotNull(args.WordDoc, "args.WordDoc");
args.Title = args.WordDoc.PackageProperties.Title;
}