How to add line numbers in every line using shell command?

Question

My file,

PSS-A  (Primary A)
PSS-B  (Primary B)
PSS-C  (Primary C)
PSS-D  (Primary D)
PSS-E  (Primary E)
PSS-F  (Primary F)
PSS-G  (Primary G)
PSS-H  (Primary H)
PSS-I  (Primary I)
SPARE  (SPARE)

Output file,

 1> PSS-A  (Primary A)
 2> PSS-B  (Primary B)
 3> PSS-C  (Primary C)
 4> PSS-D  (Primary D)
 5> PSS-E  (Primary E)
 6> PSS-F  (Primary F)
 7> PSS-G  (Primary G)
 8> PSS-H  (Primary H)
 9> PSS-I  (Primary I)
10> SPARE  (SPARE)

Solution 1

If you want the same format that you have specified

awk '{print NR  "> " $s}' inputfile > outputfile

otherwise, though not standard, most implementations of the cat command can print line numbers for you (numbers padded to width 6 and followed by TAB in at least the GNU, busybox, Solaris and FreeBSD implementations).

cat -n inputfile > outputfile

Or you can use grep -n (numbers followed by :) with a regexp like ^ that matches any line:

grep -n '^' inputfile > outputfile

Solution 2

The right tool for this job is nl:

nl -w2 -s'> ' file

You may want to tune width option according to the total number of lines in the file (if you want numbers to be aligned nicely).

Output:

 1> PSS-A  (Primary A)
 2> PSS-B  (Primary B)
 3> PSS-C  (Primary C)
 4> PSS-D  (Primary D)
 5> PSS-E  (Primary E)
 6> PSS-F  (Primary F)
 7> PSS-G  (Primary G)
 8> PSS-H  (Primary H)
 9> PSS-I  (Primary I)
10> SPARE  (SPARE)

Solution 3

In Linux/Unix, there are almost always multiple ways to do common tasks. Just for completeness, here are some other ways you can do it besides the obvious:

    pr -t -n [file]

From an old command to format text to send to a line printer. The '-t' will omit header and footer information that are not relevant to a terminal.

Here's a cute sed method that prints the line number on every other line. We use 'paste' to fold them into a single line:

    sed = /etc/passwd | paste - -

Or, we can use the one true editor, ed:

    echo '1,$n' | ed -s [file]

Or, ex, vi's non-cursor-addressing predecessor:

    printf 'set number\ng/^/p\n' | ex /etc/passwd

And one final complicated answer, requiring ksh93 or bash (and the seq command. Using the .. range and an eval statement is left as an exercise):

    paste <(seq $(wc -l < [file])) [file]

Tested on Debian Linux, FreeBSD, and Solaris 10 (the last fails there because of no 'seq').

Comments

[Nainita]

Yes...both the commands are working.... but in cat command its printing the line numbers... but not exactly what I wanted.... but awk '{print NR "> " $s}' inputfile > outputfile gives me the desired output.... :-) @amit kumar

[vinc17]

Note also that cat -n is not portable. Only the -u option is specified in POSIX cat.

[myrdd]

@StéphaneChazelas indeed, thanks a lot! Note that $'...' syntax is bash-specific.

[Sridhar Sarnobat]

My heart sank when I saw that seq didn't do it. Thank god for nl

[Stéphane Chazelas]

@myrdd, $'...' comes from ksh93 and is also supported by zsh, mksh, busybox sh, FreeBSD sh and bash at least. It's not standard yet, but is planned for inclusion in the next major POSIX version.

[myrdd]

@StéphaneChazelas thanks. for reference, there's a question on $'...' (ANSI-C Quoting) portability: unix.stackexchange.com/questions/371827/…

[Onnonymous]

I like how nl can set a starting number with -v.

[xeruf]

please add an explanation, I'd love to do this with sed, but I don't understand the command :)