you may already have one. McCusker on planned fragility: “It’s likely true Office uses such inflexible code that feature changes do need file format changes…. [Aaron Swartz: The Weblog]
Many years ago, in the days of Word 4 & 5.1 (both used almost the same file format), I had their file format documentation and I wrote code for Scriptor to read Word files. I don’t remember the details, and I no longer have the documents, but the format was extremely *ugly*, with the file broken up into blocks with pointers to the next block of the same type. We didn’t even attempt to read files saved with “fast save”, which worked by just appending blocks & rewriting pointers elsewhere in the file to the newly appended blocks.