How do I compare two text files while ignoring fields that change from one instance to the next?

Question

I have plain text college transcripts... these are meant to used a fixed-width font and have ascii columns and so forth. It's difficult for me to notice visually if a change to the program that generates them has introduced defects, my eyes glaze over. However, each time you generate even the same document, a timestamp within it will change. Is there a way to force diff (or a similar tool) to ignore a particular line, or even a particular range of characters within that line?

If I use the -I "regexp" switch, this will ignore differences in the entire line, even though I only wish to ignore changes to the date.

Is there a better tool for this? Can diff be made to do this with some bash fu?

See my answer at https://stackoverflow.com/questions/44019/an-easy-way-to-diff-log-files-ignoring-the-time-stamps — Chris Koknat, Sep 18 '20 at 20:36

score 1 · Answer 1 · answered Apr 23 '14 at 16:16

1

Check out his webpage, maybe it will do what you want. http://ifdeflinux.blogspot.com/2011/08/diff-two-files-ignoring-certain-fields.html

Paul

answered Apr 23 '14 at 16:16

pcantalupo

2,212
17
27

score 1 · Answer 2 · answered Apr 23 '14 at 16:24

You can use process substitution for this, passing the output of some other commands in place of the files themselves. Example:

$ diff file1 file2
2c2
< example: 123
---
> example: 456

$ diff <(sed -r 's/(.*: )[0-9]+/\1/' file1) <(sed -r 's/(.*: )[0-9]+/\1/' file2)
[files are the same]

In the example the sed command is just removing any digits after a colon - you will need to come up with a command appropriate for your input data which strips out your timestamps.

How do I compare two text files while ignoring fields that change from one instance to the next?

2 Answers2