Hi Vincent, When I implement the XWikiDocumentConvertOutput, I meet some pb. I understood that the XWikiDocumentConvertOutput contain a xwiki document, and it will store the result html to the xwikidocument as content, store the images to the xwikidocument as the attachments, and save this xwiki document to the xwiki. But we should separate the officeconverter from xwiki core, so XWikiDocumentConvertOutput can't get any information about xwiki contex, right? Problems with this design: 1) when I use the saveDocument() method to save the inputstream to document content, I need convert inputstream to html. But I can't get the encoding of the xwiki because I can't access the xwiki context. 2) I can't set the attachments' author, because have no idea about the author without context. 3) I can't save the xwikidocument to the xwiki without the xwiki context. 4) Maybe we can just save everything in the xwiki document and get it from ConverterOutput and set author, then save it to the xwiki. If so, we need a method getDocument() in ConverterOutput to get the document. 5) Actually, not all the accessory output is image. So the saveImage interface is not appropriate. Why we use ConverterOutput? I think because: 1) the output of conversion(especially convert to html) have no only one output, so we just use ConverterOutput to save all this outputs and their names for later use. 2) Maybe the special ConverterOutput like XWikiDocumentConvertOutput can handle the coversion result to the special destination in special format. I think 2) is not right, because it make the office convert connected to xwiki context. So I proposal ConverterOutput is just a data structure to store the output results and their name. My proposal is: public interface ConverterOutput { /** * Save the target contents in {@link InputStream} to the output destination. And save the document's name as the * main output file's name. * * @param documentName the name of the source content * @param documentData the source data */ void saveDocument(String documentName, InputStream documentData); /** * Save the other output files like images, other htmls to the output destination. In most case, this method will be * used when convert office convert to html. * * @param accessoryName the accessory name * @param accessoryData the accessory content */ void saveAccessory(String accessoryName, InputStream accessoryData); /** * @return the main document */ InputStream getDocument(); /** * @return return the main document's name */ String getDocumentName(); /** * @return the accessories of this document. The map's key is the accessory's name. */ Map<String, InputStream> getAccessories(); } I implement the office converter with this interface. Please review the code at your convenience. -- Thanks Wang Ning