With the simple awk program:
#!/bin/ksh
awk '
NR > 1 {for(i = length($2) - 5; i >= 1; i--)
c[substr($2, i, 6)]++
}
END { for(i in c)
print c, i | "sort -rn"
}' file
and with file containing the data you provided in message #1, the output produced is:
13 byebye
8 yebyeb
8 ebyeby
5 ypehel
5 typehe
5 pehell
5 llobye
5 hellob
5 elloby
5 ehello
5 catcat
4 tcatdo
4 ogcatc
4 gcatca
4 dogcat
4 catdog
4 atcatd
3 yedogc
3 ogbyeb
3 obyedo
3 lobyed
3 gbyeby
3 edogca
3 dogbye
3 byedog
2 tdogby
2 obyeby
2 lobyeb
2 atdogb
1 ttypeh
1 tdogdo
1 tcatty
1 ogdogb
1 gdogby
1 dogdog
1 cattyp
1 attype
1 atdogd
1 atcatt
so "byebye" is indeed the most common six character substring in the 2nd column of your input. But "catdog" and "typehe" don't even come close. To just print the most commonly occurring substring or (if more than one substring appears the same number of times as the first most common substring) substrings, try:
#!/bin/ksh
awk '
NR > 1 {for(i = length($2) - 5; i >= 1; i--)
c[substr($2, i, 6)]++
}
END { for(i in c)
print c, i | "sort -rn"
}' file | awk '
NR == 1 {c = $1}
$1 == c {print $2}'
which just prints:
byebye
If you want to try this on a Solaris/SunOS system, change awk from the default /usr/bin/awk to /usr/xpg4/bin/awk , /usr/xpg6/bin/awk , or nawk .
I use the Korn shell, but any shell that supports basic Bourne shell or POSIX standard shell syntax can be used instead.