docx2txt is a tool that attempts to generate equivalent text files from (even corrupted) Microsoft .docx documents, preserving some formatting and document information (which MS text conversion drops) along with appropriate character conversions for a good (ASCII) text experience.
It is a platform independent solution consisting of (core) Perl and (wrapper) Unix/Windows shell scripts and a configuration file to control the output text appearance to fair extent. It depends upon a commandline unzipping program (like unzip, 7z, pkzipc, or wzunzip) that can silently extract single files from zip archives to console/standard output/pipe.
It can very conveniently be used to build a Web based docx document conversion service. Some Makefiles and Windows batch files are provided for easy installation of the scripts. With unzippers like CakeCmd that can deal with corrupt Zip archives, this tool can extract text from corrupt docx documents in many cases, where MS word processor fails to even open them.