Capitalize strings in sed or awk - bash

Question

I have three types of strings that I'd like to capitalize in a bash script. I figured sed/awk would be my best bet, but I'm not sure. What's the best way given the following requirements?
single word
e.g. taco -> Taco
multiple words separated by hyphens
e.g. my-fish-tacos -> My-Fish-Tacos
multiple words separated by underscores
e.g. my_fish_tacos -> My_Fish_Tacos

Answer 1

There's no need to use capture groups (although & is a one in a way):
echo "taco my-fish-tacos my_fish_tacos" | sed 's/[^ _-]*/\u&/g'
The output:
Taco My-Fish-Tacos My_Fish_Tacos
The escaped lower case "u" capitalizes the next character in the matched sub-string.

Answer 2

Using awk:
echo 'test' | awk '{
for ( i=1; i <= NF; i++) {
sub(".", substr(toupper($i), 1,1) , $i);
print $i;
# or
# print substr(toupper($i), 1,1) substr($i, 2);
}
}'

Answer 3

Try the following:
sed 's/\([a-z]\)\([a-z]*\)/\U\1\L\2/g'
It works for me using GNU sed, but I don't think BSD sed supports \U and \L.

Answer 4

Here is a solution that does not use the \u, that is not common to all seds.
Save this file into capitalize.sed, then run sed -i -f capitalize.sed FILE
s:^:.:
h
y/qwertyuiopasdfghjklzxcvbnm/QWERTYUIOPASDFGHJKLZXCVBNM/
G
s:$:\n:
:r
/^.\n.\n/{s:::;p;d}
/^[^[:alpha:]][[:alpha:]]/ {
s:.\(.\)\(.*\):x\2\1:
s:\n\(..\):\nx:
tr
}
/^[[:alpha:]][[:alpha:]]/ {
s:\n.\(.\)\(.*\)$:\nx\2\1:
s:..:x:
tr
}
/^[^\n]/ {
s:^.\(.\)\(.*\)$:.\2\1:
s:\n..:\n.:
tr
}

Answer 5

alinsoar's mind-blowing solution doesn't work at all in Plan9 sed, or correctly in busybox sed. But you should still try to figure out how it's supposed to do its thing: you will learn a lot about sed.
Here's a not-as-clever but easier to understand version which works in at least Plan9, busybox, and GNU sed (and probably BSD and MacOS). Plan9 sed needs backslashes removed in the match part of the s command.
#! /bin/sed -f
y/PYFGCRLAOEUIDHTNSQJKXBMWVZ/pyfgcrlaoeuidhtnsqjkxbmwvz/
s/\(^\|[^A-Za-z]\)a/\1A/g
s/\(^\|[^A-Za-z]\)b/\1B/g
s/\(^\|[^A-Za-z]\)c/\1C/g
s/\(^\|[^A-Za-z]\)d/\1D/g
s/\(^\|[^A-Za-z]\)e/\1E/g
s/\(^\|[^A-Za-z]\)f/\1F/g
s/\(^\|[^A-Za-z]\)g/\1G/g
s/\(^\|[^A-Za-z]\)h/\1H/g
s/\(^\|[^A-Za-z]\)i/\1I/g
s/\(^\|[^A-Za-z]\)j/\1J/g
s/\(^\|[^A-Za-z]\)k/\1K/g
s/\(^\|[^A-Za-z]\)l/\1L/g
s/\(^\|[^A-Za-z]\)m/\1M/g
s/\(^\|[^A-Za-z]\)n/\1N/g
s/\(^\|[^A-Za-z]\)o/\1O/g
s/\(^\|[^A-Za-z]\)p/\1P/g
s/\(^\|[^A-Za-z]\)q/\1Q/g
s/\(^\|[^A-Za-z]\)r/\1R/g
s/\(^\|[^A-Za-z]\)s/\1S/g
s/\(^\|[^A-Za-z]\)t/\1T/g
s/\(^\|[^A-Za-z]\)u/\1U/g
s/\(^\|[^A-Za-z]\)v/\1V/g
s/\(^\|[^A-Za-z]\)w/\1W/g
s/\(^\|[^A-Za-z]\)x/\1X/g
s/\(^\|[^A-Za-z]\)y/\1Y/g
s/\(^\|[^A-Za-z]\)z/\1Z/g

Answer 6

This might work for you (GNU sed):
echo "aaa bbb ccc aaa-bbb-ccc aaa_bbb_ccc aaa-bbb_ccc" | sed 's/\<.\|_./\U&/g'
Aaa Bbb Ccc Aaa-Bbb-Ccc Aaa_Bbb_Ccc Aaa-Bbb_Ccc