public static void SplitDocumentToPages(string docName)
{
string folderName = Path.GetDirectoryName(docName);
string fileName = Path.GetFileNameWithoutExtension(docName);
string extensionName = Path.GetExtension(docName);
string outFolder = Path.Combine(folderName, "_out_");
Document doc = new Document(docName);
// Create and attach collector to the document before page layout is built.
LayoutCollector layoutCollector = new LayoutCollector(doc);
// This will build layout model and collect necessary information.
doc.UpdatePageLayout();
// Split nodes in the document into separate pages.
DocumentPageSplitter splitter = new DocumentPageSplitter(layoutCollector);
HtmlSaveOptions option = new HtmlSaveOptions();
option.ExportHeadersFootersMode = ExportHeadersFootersMode.PerSection;
option.PrettyFormat = true;
option.UseAntiAliasing = true;
// Save each page to the disk as a separate document.
for (int page = 1; page <= doc.PageCount; page++)
{
Document pageDoc = splitter.GetDocumentOfPage(page);
NodeCollection paras = pageDoc.GetChildNodes(NodeType.Paragraph, true);
foreach (Paragraph para in paras)
{
if (para.GetText().Contains("Evaluation Only. Created with Aspose.Words. Copyright 2003-2016 Aspose Pty Ltd."))
{
paras.Remove(para);
pageDoc.UpdatePageLayout();
}
}
var name = Path.Combine(outFolder, string.Format("{0}{1}{2}", fileName, page, ".html"));
pageDoc.Save(name, option);
}
// Detach the collector from the document.
layoutCollector.Document = null;
}
/// <summary>
/// word生成html字符串
/// </summary>
/// <param name="path">word路径</param>
/// <returns></returns>
public override string ToHTML(string path)
{
//openxml引用
//string htmlPath=WmlToHtmlConverterHelper.ConvertToHtml(path, this.FilePath);
//aspose引用 api网址http://www.aspose.com
if (!File.Exists(path))
{
return(string.Empty);
}
//PageNumberFinder
Document doc = new Document(path);
var paperSize = doc.FirstSection.PageSetup.PaperSize.ToString(); //纸张大小
var pageCount = doc.PageCount; //获得页数
doc.MailMerge.UseNonMergeFields = false;
doc.MailMerge.UseWholeParagraphAsRegion = true;
//doc.MailMergeSettings.ViewMergedData = false;
//doc.MailMergeSettings.LinkToQuery = true;
//doc.MailMergeSettings.MailAsAttachment = true;
//html设置
ws.HtmlSaveOptions option = new ws.HtmlSaveOptions(SaveFormat.Html);
option.ExportHeadersFootersMode = ws.ExportHeadersFootersMode.PerSection;
option.PrettyFormat = true;
option.UseAntiAliasing = true;
option.ExportTocPageNumbers = false;
option.ExportRoundtripInformation = true;
//option.ExportListLabels = ExportListLabels.AsInlineText;
string name = Path.GetFileNameWithoutExtension(path);
//生成图片保存路径
option.ImagesFolderAlias = MapPathReverse(path);
//Path.Combine(Path.GetDirectoryName(path), name);
//string.Format("/Resource/emw/UserFile/{0}/", name);
var savePath = Path.Combine(this.FilePath, name + "\\");
if (!Directory.Exists(savePath))
{
Directory.CreateDirectory(savePath);
}
LayoutCollector layoutCollector = new LayoutCollector(doc);
// 构建布局模型,收集必要的信息。
doc.UpdatePageLayout();
// 节点在文档分割成单独的页面。
DocumentPageSplitter splitter = new DocumentPageSplitter(layoutCollector);
//for (int page = 1; page <= doc.PageCount; page++)
//{
// Document pageDoc = splitter.GetDocumentOfPage(page);
// NodeCollection paras = pageDoc.GetChildNodes(NodeType.Paragraph, true);
// foreach (Paragraph para in paras)
// {
// if (para.GetText().Contains("Evaluation Only. Created with Aspose.Words. Copyright 2003-2016 Aspose Pty Ltd."))
// {
// paras.Remove(para);
// }
// }
// var pageFileName = Path.Combine(savePath, string.Format("{0}-page{1}{2}", name, page, ".html"));
// pageDoc.Save(pageFileName,option);
//}
// 从文档中分离收集器。
layoutCollector.Document = null;
//html保存地址
string htmlPath = savePath + name + ".html";
doc.Save(htmlPath, option);
XElement xhtml = XElement.Load(htmlPath);
var idCounter = 1000000;
foreach (var d in xhtml.Descendants())
{
if (d.Name.LocalName == "head" || d.Name.LocalName == "meta" || d.Name.LocalName == "title" || d.Name.LocalName == "body")
{
continue;
}
string idName = d.Name.LocalName + "-" + idCounter.ToString().Substring(1);
XAttribute id = new XAttribute("data-uniqid", idName);
if (d.Attribute("id") == null)
{
d.Add(id);
}
idCounter++;
}
File.WriteAllText(htmlPath, xhtml.ToString(), Encoding.UTF8);
string htmlString = xhtml.ToString();// File.ReadAllText(htmlPath);
string jsonStr = JsonConvert.SerializeObject(new { pageCount = pageCount, paperSize = paperSize, htmlString = htmlString, path = path });
return(jsonStr);
}
