DOCX conversion fails with IndexError on unmatched field marks; fixed in mammoth 1.12.2, but MarkItDown pins mammoth~=1.11.0 #2646

A Word document that has a complex-field separator or end mark without a matching start fails to convert:

FileConversionException: File conversion failed after 1 attempts:
 - DocxConverter threw IndexError with message: pop from empty list

Other users have hit this in real documents (mwilliamson/python-mammoth#168 and mwilliamson/python-mammoth#172). The attached document has one paragraph: "Reset value ", a field end mark, then "is 0x0.". Expected output: Reset value is 0x0.

markitdown-unmatched-field-end.docx

mammoth fixed this in 1.12.2 (mwilliamson/python-mammoth#168). That release also avoids excessive backtracking on unterminated strings in crafted documents. MarkItDown has pinned mammoth~=1.11.0 in the docx and all extras since #1452, so it can't pick up the fix. The latest MarkItDown, 0.1.8, still has that pin; the latest mammoth is 1.13.0.

Environment: markitdown 0.1.8 (main 4cc9fa1), mammoth 1.11.0, Python 3.14.8, Ubuntu 24.04.5 LTS.