I too stumbled into this problem.
Sadly it's not that easy to fix ( reasons are explained here - #377 (comment) ).

AFAIK there is some progress on this in the issue #85 and official patch will be released eventually, but if you need this working right now and content only getting text from hyperlinks (without address part), you can install package with this quick and dirty hack q210@336ed9f from my own fork (based on patch proposed by Brad-Python in #85 )

pip command is:
pip install git+git://github.com/q210/python-docx.git@336ed9fed27ff0460b674d91ba1646ded9cecb59

Running against the same wall. Would appreciate an official patch for this. Tested git above and that solves my problem

Thanks, /PA

I have written a brute force PoC just using LXML, which seems to work. I hope it helps

/PA

docx2txt.zip

I downloaded the below mentioned file and it is not working for me.

pip install git+git://github.com/q210/pythondocx.git@336ed9f

@kart8172 if you post some minimal example of your data here (perhaps .docx file and the script where you trying to use my fork) and describe what are your expectations in more details, I can try and debug the problem.

Hi I also meet this problem, and I try to use xml to fix it.
It works for me, you can try~

from docx.oxml.ns import qn

def find_hyperlink_indoc(doc):
    '''
    :param doc: doc file get by doc = Document('./xxxx.docx')
    :return: a list of all hyperlink item in doc.
    '''
    xml_e = doc.element
    hyperlink_list = xml_e.findall('.//' + qn("w:hyperlink"))
    return hyperlink_list


def get_hyperlink_text(hyperlink_item):
    text = hyperlink_item.findall('.//' + qn("w:t"))[0].text
    return text


def set_hyperlink_text(hyperlink_item, text):
    hyperlink_item.findall('.//' + qn("w:t"))[0].text = text

doc = Document('./test.docx')
hl_list = find_hyperlink_indoc(doc)
for item in hl_list:
    print(get_hyperlink_text(item))
    set_hyperlink_text(item, 'testtest')
doc.save('./test_out.docx')

Paragraph.text will include the text of hyperlinks in the forthcoming v1.0 due out in a week or so.