How do I replace all occourences of a symbol except the last one using sed?

Question

For example, I got a line like this:

44567743346_567864_56788_5677_3

I want to change all the _ into :, except for the last one would change into \. The expected output is:

44567743346:567864:56788:5677\3

Will it always end in a 3? – Michael Bailey Aug 15 '15 at 07:10 — Michael Bailey, Aug 15 '15 at 07:10
Why do change your question in a comment? – A.B. Aug 15 '15 at 12:14 — A.B., Aug 15 '15 at 12:14

John1024 · Accepted Answer · 2015-08-15T07:48:48.277

6

$ echo '44567743346_567864_56788_5677_3' | sed -r 's|_|:|g; s|:([^:]*)$|\\\1|'
44567743346:567864:56788:5677\3

This uses two sed substitution commands. The first, s|_|:|g changes all underlines to colons. The second, s|:([^:]*)$|\\\1| finds the last colon and changes it to a backslash.

Modifying every fourth line of a file this way

sed -r '1~4{s|_|:|g; s|:([^:]*)$|\\\1|;}' file

To modify the file in-place:

sed -i -r '1~4{s|_|:|g; s|:([^:]*)$|\\\1|;}' file

edited Aug 15 '15 at 07:48

answered Aug 15 '15 at 07:21

John1024

13,687
43
51

In my actual script, I write sed -i '1~4s|_|:|g; 1~4s|:([^:]*)$|\\1|' ${1}, since I need to change all the first lines of every 4 lines, but it didnot wrok:( – Lucas Lau Aug 15 '15 at 07:41
@LucasLau Please add the -r option to enable extended regex and try again. – John1024 Aug 15 '15 at 07:49
Thx man! Really help – Lucas Lau Aug 15 '15 at 08:01

score 2 · Answer 2 · answered Aug 15 '15 at 16:33

If you know the count of the last occurrence of _ e.g. in your example the last _ is the 4th _ in the line, you can do :

sed 's/_/\\/4; s/_/:/g' file.txt

Here we have first converted the 4th (last) occurrence of _ to \, then the remaining occurrences of _ are converted to :.

For example :

$ echo '44567743346_567864_56788_5677_3' | sed 's/_/\\/4; s/_/:/g'
44567743346:567864:56788:5677\3

On the other hand if you don't know the position of the last occurrence of _, you can greedily match upto the last occurrence of _, replace it with \ and then replace the rest with : :

sed -r 's/^(.*)_/\1\\/; s/_/:/g' file.txt

For example :

$ echo '44567743346_567864_56788_5677_3' | sed -r 's/^(.*)_/\1\\/; s/_/:/g'
44567743346:567864:56788:5677\3

Devesh Saini · Answer 3 · 2017-06-10T11:15:15.047

A Simple One Using sed

echo "44567743346_567864_56788_5677_3" | rev | sed "s/_/:/2g" | sed 's/_/\\/' | rev

Now let's see how it works:-

1) rev reverses the string,

it becomes 3_7765_88765_468765_64334776544

2) sed "s/_/:/2g replaces all occurrence of _ to : except the first occurrence in reversed string,

it becomes, 3_7765:88765:468765:64334776544

3) Next sed i.e. sed 's/_/\\/' replaces just the first occurrence of _ to \.

It becomes, 3\7765:88765:468765:64334776544

4) rev reverses this processed string

Now, final string is:-

44567743346:567864:56788:5677\3

Now, its done!

Note: It doesn't give much efficiency. Its here just because its simple

score 0 · Answer 4 · answered Apr 18 '21 at 13:06

0

Works in any line length:

$ echo "44567743346_567864_56788_5677_3" | sed -E 's/^(.*)_/\1\\/; s/_/:/g'
44567743346:567864:56788:5677\3

answered Apr 18 '21 at 13:06

storenth

101

Can you explain what these regexes do and why they work? – Keefer Rourke Apr 29 '21 at 21:43

A.B. · Answer 5 · 2015-08-15T12:17:08.697

Why sed if it's simpler using awk?

awk -F_ '{printf "%s",$1; for(i=2;i<NF;i++){printf ":%s",$i}; printf "\\%s\n",$NF}' your_input_file

And as you said in a comment? you need only every fourth line

awk -F_ '(NR % 4) == 1 {printf "%s",$1; for(i=2;i<NF;i++){printf ":%s",$i}; printf "\\%s\n",$NF; next} {print}'  your_input_file

Example

Input file

% cat foo
1_44567743346_567864_56788_5677_3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1_44567743346_567864_56788_5677_3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1_44567743346_567864_56788_5677_3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1_44567743346_567864_56788_5677_3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22

Command output to change every fourth line

% awk -F_ '(NR % 4) == 1 {printf "%s",$1; for(i=2;i<NF;i++){printf ":%s",$i}; printf "\\%s\n",$NF; next} {print}' foo
1:44567743346:567864:56788:5677\3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1:44567743346:567864:56788:5677\3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1:44567743346:567864:56788:5677\3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22
1:44567743346:567864:56788:5677\3
2_908798_08098_878_24332_24_24244_22
3_44567743346_567864_56788_5677_3
4_908798_08098_878_24332_24_24244_22

Command output to change each line

% awk -F_ '{printf "%s",$1; for(i=2;i<NF;i++){printf ":%s",$i}; printf "\\%s\n",$NF}' foo                               
1:44567743346:567864:56788:5677\3
2:908798:08098:878:24332:24:24244\22
3:44567743346:567864:56788:5677\3
4:908798:08098:878:24332:24:24244\22
1:44567743346:567864:56788:5677\3
2:908798:08098:878:24332:24:24244\22
3:44567743346:567864:56788:5677\3
4:908798:08098:878:24332:24:24244\22
1:44567743346:567864:56788:5677\3
2:908798:08098:878:24332:24:24244\22
3:44567743346:567864:56788:5677\3
4:908798:08098:878:24332:24:24244\22
1:44567743346:567864:56788:5677\3
2:908798:08098:878:24332:24:24244\22
3:44567743346:567864:56788:5677\3
4:908798:08098:878:24332:24:24244\22

How do I replace all occourences of a symbol except the last one using sed?

5 Answers5

Modifying every fourth line of a file this way