• python-docx could parse the DOCX files
• Create required data classes. Nested classes fro AST, but need not store all details library provides
• python-docx result need to be parsed recursive descentant manner, for converting to data classes
• On data classes, do transformations like converting to Markdown, trimming spaces etc.
• Also provide convertion from AST back to DOCX
• In addition to AST <—> DOCX, AST <—> JSON is useful for visualizing in browser & editing in text mode
• Serialize/deserialzie data class to JSON. Nested deserialization requires some custom handling

Thoughts

• Probably python-docx has a similar AST internally