Regular Expression in sed for multiple replacements in one statement

Question

I want to sanitise some input and replace several characters with acceptable input, e.g. a Danish 'å' with 'aa'.

This is easily done using several statements, e.g. /æ/ae/, /å/aa/, /ø/oe/, but due to tool limitations, I want to be able to do this in a single regular expression.

I can catch all of the relevant cases (/[(æ)(ø)(å)(Æ)(Ø)(Å)]/) but I replacement does not work as I want it to (but probably completely as intended):

 $ temp="RødgrØd med flæsk"

 $ echo $temp
 RødgrØd med flæsk

 $ echo $temp | sed 's/[(æ)(ø)(å)(Æ)(Ø)(Å)]/(ae)(oe)(aa)(Ae)(Oe)(Aa)/g'
 R(ae)(oe)(aa)(Ae)(Oe)(Aa)dgr(ae)(oe)(aa)(Ae)(Oe)(Aa)d med fl(ae)(oe)(aa)(Ae)(Oe)(Aa)sk

(first echo line is to show that it isn't an encoding issue)

Just as an aside, the tool issue is that I should like to also use the same regex in a Sublime Text 2 snippet.

Anyone able to discern what is wrong with my regex statement?

Thanks in advance.

Is `sed` the only possibility? What about `awk`, `perl`, `ruby`, `raku`, etc.? — jubilatious1, Jun 13 '23 at 18:14

score 64 · Accepted Answer · answered Jan 03 '13 at 08:03

64

Split it up into several sed statements, separated by ;:

sed 's/æ/ae/g;s/ø/oe/g;s/å/aa/g;s/Æ/Ae/g;s/Ø/Oe/g;s/Å/Aa/g'

answered Jan 03 '13 at 08:03

Anders Johansson

3,926
19
19

Blarg, of course. I should have emphasized the Sublime Text snippet usage more, but this is the completely correct answer to the main question. – Jan Jan 03 '13 at 08:07
2

@Jan, thanks. I don't think you can write a single regex that replaces one of several possible matches with one of several replacements, especially not with `sed`. – Anders Johansson Jan 03 '13 at 08:27

score 11 · Answer 2 · answered Jan 03 '13 at 08:04

11

With

sed -e 's/Find/Replace/g;s/Find/Replace/g;[....];/Find/Replace/g'

you'll do the trick.

So, translate into what you need

sed -e 's/æ/ae/g;s/ø/oe/g;s/å/aa/g;s/Æ/Ae/g;s/Ø/Oe/g;s/Å/Aa/g'

answered Jan 03 '13 at 08:04

DonCallisto

29,419
9
72
100

Blarg, of course. I should have emphasized the Sublime Text snippet usage more, but this is the completely correct answer to the main question. – Jan Jan 03 '13 at 08:06

potong · Answer 3 · 2013-01-03T13:00:29.777

3

This might work for you (GNU sed):

sed -r 's/$/\næaeøoeåaaÆAeØOeÅAa/;:a;s/([æøåÆØÅ])(.*\n.*\1(..))/\3\2/;ta;s/\n.*//' file

It works by adding a lookup table to the end of the line, looping until all keys are replaced then removes the lookup table.

edited Jan 03 '13 at 13:00

answered Jan 03 '13 at 09:29

potong

55,640
6
51
83

Regular Expression in sed for multiple replacements in one statement

3 Answers3

Linked