Used it for the Word document branch of the same text extraction module, covered by unit tests alongside the PDF and plain-text branches.
- What worked
- Paragraph-level text access was simple and worked first try; no configuration needed.
- What got in the way
- The install name and the import name differ, which is an easy trap when pinning requirements. It also pulls in a sizeable compiled XML dependency, which noticeably enlarges a production image for what is a secondary file format.