UNIX Bash: Number each line of a file with a specific format

Question

I'm trying to number each line of a file with this format: /1/,/2/,...

For example

From

UK
Australia
Newzealand
Brazil
America

To

/1/ UK
/2/ Australia
/3/ Newzealand
/4/ Brazil
/5/ America

I think about

grep -n file.txt // but the format is 1: , 2: , ...
nl file.txt  // but the format is 1 , 2 , 3

Can i give a specific format to nl command?

Maybe `Perl` is acceptable to the professor... `perl -pe 'print "/$./ "' file` — Mark Setchell, Jan 09 '17 at 13:01

Inian · Answer 1 · 2017-01-09T13:27:26.123

4

Use the printf function in Awk, with the characters / around the integer part.

awk '{printf("/%d/ %s\n", NR, $0)}' file
/1/ UK
/2/ Australia
/3/ Newzealand
/4/ Brazil
/5/ America

NR is a special variable in awk which tracks the record number of each line as it is processed, and the line as such being $0. The above command is POSIX compatible since awk in all major platforms support printf with format specifiers.

A proper way to do it in bash alone, without any third-party tools.

#!/bin/bash

count=1
while IFS= read -r line
do
  printf "/%d/ %s\n" "$count" "$line"
  ((count++))
done <input-file

AFAIK, from the man nl page, there are options you can use, but NONE to wrap the line numbers around the characters you want, you can of-course use it one side, with -s flag

The closest I could do with it is,

nl -nrz -s/ -w1 file
1/UK
2/Australia
3/Newzealand
4/Brazil
5/America

here, -nrz for right justification and -s for the special character to use after the numbering done. In this case, I have a added a / and -w controls the width of the numbering part.

edited Jan 09 '17 at 13:27

answered Jan 09 '17 at 11:59

Inian

62,560
7
92
110

1

note: there is also FNR, which is a separate counter per file. – Karoly Horvath Jan 09 '17 at 12:02
Your answer is working perfectly. But for an academical purpose i have to do it without the `awk` command. Does it have another method? – Francis Ngueukam Jan 09 '17 at 12:03
1

With small simplifications (without `count=1`): `while IFS= read -r line; do printf "/%d/ %s\n" "$((++count))" "$line"; done – Cyrus Jan 09 '17 at 12:27
This solution ist great. But unfortunately i wanted a simple method without `awk`. Thanks for help. +1 – Francis Ngueukam Jan 09 '17 at 13:11
2

Why is a pipeline involving `cat` and `sed` preferable to a single `awk` command? – chepner Jan 09 '17 at 13:49
@chepner: Yup ,sounded weird, stopped my argument when I saw _because of my Professor. I have to do it without awk_ in one of the comments above. :) – Inian Jan 09 '17 at 13:50
1

Alternative version `awk '{print "/"NR"/", $0}' file` – Zlemini Jan 09 '17 at 15:29

score 2 · Accepted Answer · edited Jun 20 '20 at 09:12

2

With GNU cat and GNU sed:

cat -n file | sed 's|^ *|/|;s|\t|/ |'

Output:

/1/ UK
/2/ Australia
/3/ Newzealand
/4/ Brazil
/5/ America

I used two s commands with this format: s|regexp|replacement|.

s|^ *|/|: This replaces in a line starting (^) with zero or more (*) blanks by a /

s|\t|/ |: This Replaces a tab (\t) by a slash followed by a blank.

See: The Stack Overflow Regular Expressions FAQ

edited Jun 20 '20 at 09:12

Community

1
1

answered Jan 09 '17 at 12:30

Cyrus

69,405
13
65
117

In the same lines as yours, `cat -n file | while read number line; do printf "/%d/ %s\n" "$number" "$line"; done` – Inian Jan 09 '17 at 12:39
I like this answer. Can you please explain me how this substitution is working here `^ *` and `\t` in stream editor command? – Francis Ngueukam Jan 09 '17 at 12:48
@Inian Unfortunately yes, because of my Professor. I have to do it without awk, but with commands like sed, nl, ... – Francis Ngueukam Jan 09 '17 at 12:54
Thankyou so much for the explanations. I now understand. There are many others alternatives, but this is my best. +1 – Francis Ngueukam Jan 09 '17 at 13:09

UNIX Bash: Number each line of a file with a specific format

2 Answers2