I realize that this has been covered numerous times in the past, but I can't seem to get any of the previous suggestions to work. I've tried various things with grep and sed.

I am trying to replace a string of text "line-width = 6 \in"
with the string "line-width = 7.8 \in"
in 500+ lilypond files so that they display better on my nook. The problem seems to be the backslash, but trying it with a double backslash doesn't improve things.
Thank you.

Replace 6 with 7.8

Sometimes you need to break the problem into pieces. You can use fgrep without the backslash but with double quotes to find the pattern in your files. You can then use sed with backslashes to esccape the lines and make them come out right:

The -l tells fgrep to simply list files that match. The loop will make the change for every file found by fgrep.

1 members found this post helpful.

The best thing to do is try to avoid the problem area entirely. You don't necessarily need to use the whole string in the sed command.

This first finds lines that contain the phrase "line-width", then replaces "= 6" with "= 7.8" in those lines.

In fact, if the text you want to change is all on a single line, with nothing else that needs to be preserved, why not just replace the entire line?

Edit: BTW, you can use globbing patterns to feed sed multiple files at once, and the -i option to edit them in place. Be sure not to enable it until you've thoroughly tested your solution, of course.

1 members found this post helpful.

You oughta know by now...Don't Read Lines With For!

Even though I shouldn't have read the line with 'for' it solved my problem.
Thank you.

for has worked for years for me and no matter how often you post this I'll likely continue doing it.

i do that too, but:would handle white spaces better.

Well, I wouldn't have to post it so often it if certain posters would stop suggesting it. Because actually, it doesn't work, at least not consistently.

With loops like the above, the user will find his script breaking at the very first file with spaces in the name. That's why you should always use a while+read loop, preferably with null separators, when working with filename lists.

The safest form of the loop would be this, using bash and gnu grep:
Or optionally, you could store the list in an array first, then the for loop would be safe to use, as long as you expanded it with the "[@]" index and quoteed it properly. (notice too how grep is unnecessary)

You can even bypass the array and just use the globbing form directly as the loop feed.


And as I mentioned, as long as you're using globbing anyway (and the resulting file list) isn't overly huge, you don't even need to use a loop at all. sed can handle multiple file inputs directly.

Clearly it DOES work as it least one other person said they did it.

Preferring a number 2 pencil does NOT mean a pen doesn't work.

That's why I added the caveat "not consistently". It only works when the conditions are exactly right, and the input doesn't contain whitespace or globbing patterns that would force breaks in undesirable locations.

Perhaps your filenames are always clean, but on an open board like this you can't assume that the person you are giving advice to has the same setup. Almost certainly some naive newbie is going to try to use it and wonder why it's not working for him, at the least, if not have some data destroyed because of it.

As long as you, or anyone else, keeps posting such poor code, I'm going to keep correcting it.

(BTW, I expect nothing but the same in return. If I post something that can potentially break things or especially cause data loss, it needs to be corrected, whatever the cost to my own ego.)