Fixed the error caused by file format when docx loading files

[iCanDoAllThingszz]

Description

When uploading a doc file to the knowledge base, docx will occur an error when loading a file with a specific format. The reason is that docx does not parse the file element correctly and returns a null value.

The error message is :

2024-10-12 15:00:56 | ERROR | dbgpt.serve.rag.service.service | document embedding, failed:海外市场相关研究-补充材料.docx, "There is no item named 'NULL' in the archive"

sample file:
海外市场相关研究-补充材料.docx

sample code:

import docx
from docx.document import Document

doc = docx.Document('/Users/zhaoyu/Desktop/海外市场相关研究-补充材料.docx')

How To Fix

This problem has been solved in the docx issues on the git
We just need to rewrite the load_from_xml_v2 method, and when the target element to be parsed is NULL, do not throw an exception and continue parsing.
This method allows users to upload doc files in a specific format without being bothered by errors.

[Aries-ckt]

LGTM.

[Aries-ckt]

LGTM.

[csunny]

r+