ms word - How do I extract data from a doc/docx file using Python

Question

I know there are similar questions out there, but I couldn't find something that would answer my prayers. What I need is a way to access certain data from MS-Word files and save it in an XML file. Reading up on python-docx did not help, as it only seems to allow one to write into word documents, rather than read. To present my task exactly (or how i chose to approach my task): I would like to search for a key word or phrase in the document (the document contains tables) and extract text data from the table where the key word/phrase is found. Anybody have any ideas?

Answer 1

The docx is a zip file containing an XML of the document. You can open the zip, read the document and parse data using ElementTree.

The advantage of this technique is that you don't need any extra python libraries installed.

import zipfile
import xml.etree.ElementTree

WORD_NAMESPACE = '{http://schemas.openxmlformats.org/wordprocessingml/2006/main}'
PARA = WORD_NAMESPACE + 'p'
TEXT = WORD_NAMESPACE + 't'
TABLE = WORD_NAMESPACE + 'tbl'
ROW = WORD_NAMESPACE + 'tr'
CELL = WORD_NAMESPACE + 'tc'

with zipfile.ZipFile('<path to docx file>') as docx:
    tree = xml.etree.ElementTree.XML(docx.read('word/document.xml'))

for table in tree.iter(TABLE):
    for row in table.iter(ROW):
        for cell in row.iter(CELL):
            print ''.join(node.text for node in cell.iter(TEXT))

See my stackoverflow answer to How to read contents of an Table in MS-Word file Using Python? for more details and references.

In answer to a comment below, Images are not as clear cut to extract. I have created an empty docx and inserted one image into it. I then open the docx file as a zip archive (using 7zip) and looked at the document.xml. All the image information is stored as attributes in the XML not the CDATA like the text is. So you need to find the tag you are interested in and pull out the information that you are looking for.

For example adding to the script above:

IMAGE = '{http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing}' + 'docPr'

for image in tree.iter(IMAGE):
    print image.attrib

outputs:

{'id': '1', 'name': 'Picture 1'}

I'm no expert at the openxml format but I hope this helps.

I do note that the zip file contains a directory called media which contains a file called image1.jpeg that contains a renamed copy of my embedded image. You can look around in the docx zipfile to investigate what is available.