I have been trying to write a simple Markdown -> docx parser/writer, but am completely stuck with the last part, which should be the easiest: i.e. compressing the folder into a .docx that Word, or any other .docx reader, will recognize.
My parser-writer is irrelevant really: I have this problem if I simply unzip any old Word-produced *.docx and then try to recompress it with the usual compression utilities, giving it the file-ending docx. Is there some mysterious header I should be adding, or do I need a special OPC compression utility, or what?
I don't so much want a tool that will do this, as to figure out what is supposed to be there. It seems to be independent of the WordprocessingML specification.
Needless to say I don't know anything about compression. Everything I can find via Google has to do with fancy utilities you can use in business, but I'm making a little executable that would be GPLd or something, and should work on anything.
The most common problem around manually zipping together Open XML documents is that it will not work if you zip the directory instead of the contents. In other words, the[content_types].xml file, and the word, docProps, and _rels directories need to reside at the root level of the zip file.
Here are steps to unzip my.docx and re-zip:
% mkdir unzipped
% cd unzipped/
% unzip ../my.docx
% zip -r ../rezipped.docx *
% open ../rezipped.docx
Further to what Mica said, the contents of the ZIP file are organised according to the Open Packaging Convention; cf. Microsoft's Essentials of the Open Packaging Convention.
You can use the .NET System.IO.Packaging to make and manipulate .docx files; this class is implemented in the Mono project.
The compression algorithm used is "Zip" (Base 64) compression.
7zip seems to offer this, though i have no tested it.
来源：https://stackoverflow.com/questions/1514052/how-to-zip-a-wordprocessingml-folder-into-readable-docx