Match single digits only sed

Question

I have to make a regex to match one digit only. it should match 7 and a7b but not 77. I made this but it doesn`t seem to work in sed.

(?<![\d])(?<![\S])[1](?![^\s.,?!])(?!^[\d]) 
(?<![\d])(?<!^[\a-z])\d(?![^a-z])(?!^[\d])

What am I doing wrong?

Edit:

I need to replace only 1-digit numbers with something like

sed 's/regex/@/g' file //regex to match "1"

file content

1 2 3 4 5 11 1
agdse1tg1xw 
6 97 45 12

Should become

 @ 2 3 4 5 11 @ 
 agdse@tg@xw 
 6 97 45 12

sed only support BRE and ERE. But you might be able to use `grep -oP` or `grep -P` to use PCRE. — Andreas Louv, May 23 '16 at 19:12
Why don't 2, 3, 4, 5 and 6 in your example get converted to `@`s? Are you trying to convert 1-digit numbers or the digit 1? — Ed Morton, May 24 '16 at 03:45

sjsam · Answer 1 · 2016-05-23T19:57:18.160

1

Input

a77
a7b
2ab
882
9
abcfg9
9fg
ab9

Script

sed -En '/^[^[:digit:]]*[[:digit:]]{1}[^[:digit:]]*$/p' filename

Output

a7b
2ab
9
abcfg9
9fg
ab9

edited May 23 '16 at 19:57

answered May 23 '16 at 19:04

sjsam

21,411
5
55
102

Worth to mention that you should use `-E` on OSX and that GNU sed should support `-E` as well, even though it's not mentioned in `man sed` – Andreas Louv May 23 '16 at 19:06
`*` will make both digits optional matching `77` – rock321987 May 23 '16 at 19:06
Single digit is not matched with this pattern. – SLePort May 23 '16 at 19:07
@andlrc : Thanks for the update. :-) – sjsam May 23 '16 at 19:59
@Kenavoz : It is, now ;) – sjsam May 23 '16 at 19:59
@rock321987 : Pasted the wrong stuff.. Updated. – sjsam May 23 '16 at 20:02

score 0 · Answer 2 · answered May 23 '16 at 19:16

0

sed only supports BRE and ERE, but you can enable PCRE with grep -P:

% printf 'a77\na7b\n2ab\n82\n' | grep -P '(?<!\d)\d(?!\d)'
a7b
2ab

grep will as demonstrated print matched lines, but have an option to print the match only:

% printf 'a77\na7b\n2ab\n82\n' | grep -oP '(?<!\d)\d(?!\d)'
7
2

answered May 23 '16 at 19:16

Andreas Louv

46,145
13
104
123

this is what i need but it kinda has to be made with sed because i have to replace the digit – Cosma Aurelian May 23 '16 at 20:23
@Restal How do you feel about adding Perl to your dependencies: `perl -pe s/(?<!\d)\d(?!\d)/replacement/g input_file` – Andreas Louv May 23 '16 at 20:25
unfortunately is not acceptable as i have to make it using sed – Cosma Aurelian May 23 '16 at 20:29
@Restal You can modify Kenavoz answer to match your needs. Just make use of capturing groups: `\(...\)` which will store the captured value in `\1`, `\2`, ... – Andreas Louv May 23 '16 at 20:31

Ed Morton · Accepted Answer · 2016-05-24T04:01:41.270

0

To do what you show in the Example in your question is:

$ sed -r 's/(^|[^0-9])1([^0-9]|$)/\1@\2/g' file
@ 2 3 4 5 11 @
agdse@tg@xw
6 97 45 12

but that only works because you didn't have 1 1 in your data. If you did you'd need 2 passes:

$ echo '1 1' | sed -r 's/(^|[^0-9])[0-9]([^0-9]|$)/\1@\2/g'
@ 1

$ echo '1 1' | sed -r 's/(^|[^0-9])[0-9]([^0-9]|$)/\1@\2/g; s/(^|[^0-9])[0-9]([^0-9]|$)/\1@\2/g'
@ @

and if you wanted to do that for any single digit it would be:

$ sed -r 's/(^|[^0-9])[0-9]([^0-9]|$)/\1@\2/g; s/(^|[^0-9])[0-9]([^0-9]|$)/\1@\2/g' file
@ @ @ @ @ 11 @
agdse@tg@xw
@ 97 45 12

edited May 24 '16 at 04:01

answered May 24 '16 at 03:52

Ed Morton

188,023
17
78
185

1

this is what i was looking for. Thank you! – Cosma Aurelian May 24 '16 at 05:18

Match single digits only sed

3 Answers3