confluence
查看executeFileUpload方法
public String executeFileUpload()
throws Exception{
MultiPartRequestWrapper request = (MultiPartRequestWrapper)this._request;
// Struts请求包装过滤器
File file = null;
String fileName = null;
try{
ArrayList errors = (ArrayList)request.getErrors();
String maxDocSize;
if (errors != null) {
for (int i = 0; i < errors.size(); i++){
Matcher m = CONTENT_LENGTH_MESSAGE_PATTERN.matcher((String)errors.get(i));
if (m.matches()) {
maxDocSize = FileSize.format(Long.parseLong(m.group(2)));
throw new DocumentToLargeException("Total document size is too large. Reduce total document size to below " + maxDocSize);
}
}
}//当文件大小超过限制时报错
xsrfValidationResult = this.xsrfTokenService.validate(request);
if (xsrfValidationResult.isDefined()){
Message xsrfValidationErrorMessage = (Message)xsrfValidationResult.get();
super.addActionError(getText(xsrfValidationErrorMessage.getKey(), xsrfValidationErrorMessage.getArguments()));
return "error";
}//对异常请求进行校验
file = request.getFiles("filename")[0];
fileName = request.getFileNames("filename")[0];
this._pageTitle = fileName;
if (fileName.lastIndexOf('.') != -1) {
this._pageTitle = fileName.substring(0, fileName.lastIndexOf('.'));
}//获取文档名称
log.debug("Import word document [ {} ] ", fileName);
byte[] docBuf = buildDocumentTree(file, this._pageTitle);//建立文档树，具体过程后面会提到
this._importInfo.setTreeDepth(this._treeDepth);
this._session.put("wordImportFile", docBuf);
this._session.put("wordImportBookmarks", this._bookmarks);
this._session.put("wordImportNodes", this._orderedNodes);
this._session.put("wordImportRoot", this._treeRoot);//存入文档相关信息
}
catch (InvalidDocumentException e)
{
super.addActionError("The selected file is not a valid binary Word 97-2003 document");
return "error";
}
catch (DocumentToLargeException e)
{
Maybe<Message> xsrfValidationResult;
super.addActionError(e.getMessage());
return "error";
}
catch (Exception localException) {}finally
{
if ((file != null) &&
(!file.delete())) {
log.warn("Failed to delete uploaded file " + file.getAbsolutePath());
}
}
if ((file == null) || (fileName == null)) {
return "error";
}
String submitType = this._request.getParameter("submit");
if ((submitType != null) && (submitType.equals("Import")))
{
this._importInfo.setTitle(getPage().getTitle());
this._importInfo.setConflict(0);
this._importInfo.setLvl(0);
this._importInfo.setImportSpace(false);
//存入文档相关信息（用户空间，标题等）
return executeFileImport();
}
this._importInfo.setTitle(this._pageTitle);
return "input";
}
可以看出最后跳入excuteFileImport这个方法
public String executeFileImport()
throws Exception {
byte[] docBuf = (byte[])this._session.get("wordImportFile");//获取word文档输入流
this._bookmarks = ((BookmarkInfo)this._session.get("wordImportBookmarks"));
this._orderedNodes = ((ArrayList)this._session.get("wordImportNodes"));
this._treeRoot = ((DocumentTreeNode)this._session.get("wordImportRoot"));
this._pagesBeingDeleted = new ArrayList();
this._pagesBeingOverwritten = new ArrayList();
this._pagesBeingCreated = new ArrayList();
//获取文档的书签，列表节点，根节点,上面三个数组是将word文档拆分到多个页面时需要做记录。
User currentUser = AuthenticatedUserThreadLocal.getUser();
boolean failFast = permissionChecks(currentUser);
if (failFast) {
return "error";
}
boolean showWarning = shouldShowWarning();
if (showWarning) {
return "warning";
}
if (!isValidTitle()) {
return "error";
}
return doFileImport();//对用户校验后转入doFileImport
}
doFileImport方法（方法太长摘取了一些片段）：
byte[] docBuf = (byte[])this._session.get("wordImportFile");
this._bookmarks = ((BookmarkInfo)this._session.get("wordImportBookmarks"));//获取文档标签（插入图片或者其他内容时会用到）
this._orderedNodes = ((ArrayList)this._session.get("wordImportNodes"));//获取节点数组
this._treeRoot = ((DocumentTreeNode)this._session.get("wordImportRoot"));//获取文档根节点
this._pagesBeingDeleted = new ArrayList();
Page oldPage = this._treeRoot.getOldPage();
Page rootPage;
if (oldPage != null){
Page rootPage = oldPage;
oldPage = (Page)rootPage.clone();
this._treeRoot.setOldPage(oldPage);
}
else{
rootPage = new Page();
rootPage.setSpace(getSpace());
Page homePage = super.getSpace().getHomePage();
if (homePage != null) {
homePage.addChild(rootPage);
}
rootPage.setCreator(currentUser);
}
rootPage.setTitle(this._treeRoot.getText());
this._importInfo.setTitle(this._treeRoot.getText());
//判断是覆盖以前的页面还是保存为新页面，如果是新页面，作为主页的孩子节点加入进去
try{
Document doc = createNewDocument(docBuf);
Doc2Wiki converter = new DocSplitter(new DefaultSplitImportContext(rootPage, this._treeRoot.getOldPage(), currentUser, this.pageManager, this._attachmentManager, this._bookmarks, this._orderedNodes, this._importInfo.getLvl(), this.ocSettingsManager, this.wikiMarkupToXhtmlConverter), this.ocSettingsManager.isDoFootnotes());
doc.accept(converter);
}
else {
StringBuffer out = new StringBuffer();
Document doc = createNewDocument(docBuf);
BookmarkCollection bkmks = doc.getRange().getBookmarks();
Doc2Wiki converter = new Doc2Wiki(out, new DefaultImportContext(this.pageManager, rootPage, this._attachmentManager, bkmks, this.ocSettingsManager), true, this.ocSettingsManager.isDoFootnotes());
doc.accept(converter); rootPage.setBodyAsString(this.wikiMarkupToXhtmlConverter.convertToXhtml(out.toString()));
if (this._treeRoot.getOldPage() == null) {
this.pageManager.saveContentEntity(rootPage, null);
} else {
this.pageManager.saveContentEntity(rootPage, this._treeRoot.getOldPage(), null);
}
}
}//利用转换器将word文档格式转换为包含html标签的格式。然后保存入mysql。
导入成功后转跳转请求,
找到viewaction，这时候jd-gui出问题了源代码出不来，百度了一下说是加了壳，没办过果断换一个反编译器，Luyten，这下源代码出来了。
将存入数据库中的html代码和加上前台页面的html组合成一个页面保存在pagexhtmlcontent这个字符串中，然后跳转前台
输出html代码，完成页面渲染。这就是word导入的整个流程。
来看看confluence如何解析word文档。
private byte[] buildDocumentTree(final File file, final String pageTitle) throws Exception {
this._treeRoot = new DocumentTreeNode(pageTitle, 0);
final DocumentTreeBuilder builder = new DocumentTreeBuilder(this._treeRoot);//创建文档树构造器
final FileInputStream in = new FileInputStream(file);
final ByteArrayOutputStream out = new ByteArrayOutputStream((int)file.length());//获取输出流
final byte[] buf = new byte[4096];
for (int read = in.read(buf); read != -1; read = in.read(buf)) {
out.write(buf, 0, read);
}//word文档写如输出流
final byte[] docBuf = out.toByteArray();//转换为字节数组
file.delete();
Document doc = null;
try {
doc = this.createNewDocument(docBuf);//通过字节数组建立文档
}
catch (Throwable t) {
throw new InvalidDocumentException();
}
doc.accept(builder);
this._orderedNodes = new ArrayList();
this.normalizeDocTree(this._treeRoot);
this._bookmarks = builder.getBookmarks();
return docBuf;
}
这里文档的建立采用了aspose插件包（不过confluence在aspose.word基础上做了修改）。查询api文档可以看到document这个类的构造方法有如下几种，
confluence采用了最后一个构造方法。
