CODEWITHSUNDEEP
macosgitterminalms-wordinstallation
Kouichi C. Nakamura
Kouichi C. Nakamura

Reputation: 1108

Version control Word .docx files with docx2txt with Git on Mac OS X

On Git web site there is a detailed instruction for version controling Microsoft Word .doc files with catdoc.

http://git-scm.com/book/en/Customizing-Git-Git-Attributes

However, I realized that this doesn't work for .docx files. It seems that you need either docx2txt or unoconv instead of catdoc (found here). I decided to go with docx2txt for no reason, but I was stuck at the installation of docx2txt into Mac OS X.

This sort of illustrates the steps. In my understanding, all you need is docx2txt.pl at somewhere sensible. I thought /usr/local/bin/ would do. I copied it there. Then, according to the instruction, I tried the following:

$ cd /usr/local/bin/
$ echo '#!/bin/bash
docx2txt.pl "$1" -' > docx2txt

When I try this:

$ docx2txt

I got

Can't read docx file <>!

so, docx2txt seems to be in the path.

Then I edited .gitattributes in the repository folder (ASCII, LF) to add the following line:

*.docx diff=wordx

Then, I also edited .git/config file in the repository as follows:

[diff "wordx"]
    binary = true
    textconv = docx2txt

Because the repository is already in use, I didn't do git init. I edited a .docx Word file in the repository and then typed git diff in the Terminal. But result was not successful.

Binary files a/foo/foo.docx and b/foo/foo.docx differ

Could anyone have any suggestions?

Upvotes: 3

Views: 3002

Answers (2)

user2473131
user2473131

Reputation: 1

I download the docx2txt using brew on my Mac M1. I search the path where original files were downloaded /opt/homebrew/Cellar/docx2txt/1.4/bin there I found the files. In the bin directory there are two executable .sh and .pl extension files. I run the docx2txt.sh file and build the docx2txt file. And during file conversion you don't have to give lt and gt sign like some instruction mentioned and I think if you are using linux/Unix all should be same. Last command should be simple like this (docx2txt input.docx output.txt).

Upvotes: -1

Kouichi C. Nakamura
Kouichi C. Nakamura

Reputation: 1108

Thanks to klang, I made it. Now I can diff .docx files in Terminal.app in Mac OS X (10.9). But this one doesn't seamlessly work with SourceTree GUI. Below is basically the same as klang's but with minor corrections.

Download and install the docx2txt converter from http://docx2txt.sourceforge.net/

wget -O doc2txt.tar.gz http://docx2txt.cvs.sourceforge.net/viewvc/docx2txt/?view=tar
tar zxf doc2txt.tar.gz
cd docx2txt/docx2txt/
sudo make

Then make a small wrapper script to make docx2txt output to STDOUT

echo '#!/bin/bash
docx2txt.pl "$1" -' > /usr/local/bin/docx2txt
chmod +x /usr/local/bin/docx2txt

Git attributes for (Word) .docx diffing in your repository

echo "*.docx diff=wordx" >> .gitattributes
git config diff.wordx.textconv docx2txt

Use .git/info/attributes if the setting should not be committed with the project.

Git attributes for (Word) .doc diffing

echo "*.doc diff=word" >> .gitattributes
git config diff.word.textconv strings

Upvotes: 3

Related Questions