sed find and replace with curly braces

Question

I am trying to use this command:

sed -i 's#\{test1\}#test2#' /example/myfile.txt

To replace instances of {test1} with test2.

I get the error:

sed: -e expression #1, char 17: Invalid preceding regular expression

Am I not escaping the curly braces correctly?

Hmm.. I removed the back slashes and it seems to work fine. – atdev Feb 09 '12 at 05:26 — atdev, Feb 09 '12 at 05:26

score 48 · Accepted Answer · answered Feb 09 '12 at 05:21

48

sed -i 's#{test1}#test2#' /example/myfile.txt

You don't need escape {}

answered Feb 09 '12 at 05:21

kev

155,172
47
273
272

score 28 · Answer 2 · answered Feb 09 '12 at 06:03

28

You aren't escaping the curly braces at all. In sed, the default regular expressions are BREs, where \{ and \} indicate a range expression. Since test1 isn't a range, your BRE is incorrect.

To fix it, you can either drop the backslashes (braces aren't special in BREs) or keep it the same and tell sed to use EREs (-r flag with GNU sed, -E flag with BSD/MacOSX sed).

answered Feb 09 '12 at 06:03

Michael J. Barber

24,518
9
68
88

TIL BRE ERE. BREs are pretty handy in the `sed` world. – ericsoco Dec 29 '16 at 20:37
4

TIL: BRE: Basic Regular Expression, ERE: Extended Regular Expression; TIL: Today I Learned :) – Generic Ratzlaugh Jan 17 '21 at 16:33

sed find and replace with curly braces

2 Answers2

Linked