bash - trim input after string manipulation - bash

Question

This is how my input looks like:
string="a 1,a 2,a 3"
This is how I generate list out of the input:
sed -e 's/[^,]*/"&"/g' <<< ${string}
Above command gives me the desired output as:
"a 1","a 2","a 3"
How do I trim each element so that if the input is " a 1, a 2, a 3", my output still comes back as "a 1","a 2","a 3"?

Answer 1

I think it is important to understand that in bash, the double quotes have a special meaning.
string="a 1,a 2,a 3" represents the string a 1,a 2,a 3 (no quotes)
sed -e 's/[^,]*/"&"/g' <<< ${string} is equivalent to the variable out='"a 1","a 2","a 3"'
To accomplish what you want, you can do:
$ string=" a 1, a 2, a 3 "
$ echo "\"$(echo ${string//*( ),*( )/\",\"})\""
"a 1","a 2","a 3"
This is only using bash builtin operations.
replace all combinations of multiple spaces and commas by the quoted comma ${string//*( ),*( )/\",\"}
use word splitting to remove all leading and trailing blanks $(echo ...) (note: this is a bit ugly and will fail on cases like a 1 , a 2 as it will remove the double space between a and 1)
print two extra double-quotes at the beginning and end of the string.
A better way is to use a double substitution:
$ string=" a 1, a 2, a 3 "
$ foobar="\"${string//,/\",\"}\""
$ echo "${foobar//*( )\"*( )/\"}"
"a 1","a 2","a 3"
note: here we make use of KSH-globs which can be enabled with the extglob setting (shopt -s extglob)

Answer 2

Here an answer which extends your sed command with some basic preprocessing that removes unwanted spaces:
sed -E -e 's/ *(^|,) */\1/g;s/[^,]*/"&"/g' <<< ${string}
The -E option enables extended regular expression which saves some \.

Answer 3

EDIT: Since OP told to wrap output in " so adding it now.
echo "$string" | sed -E 's/^ +/"/;s/$/"/;s/, +/"\,"/g'
Output will be as follows.
echo "$string" | sed -E 's/^ +/"/;s/$/"/;s/, +/"\,"/g'
"a 1","a 2","a 3"
Could you please try following and let me know if this helps you.
awk '{sub(/^\" +/,"\"");gsub(/, +/,"\",\"")} 1' Input_file
In case you want to save output into same Input_file itself append > temp_file && mv temp_file Input_file.
Solution 2nd: Using sed.
sed -E 's/^" +/"/;s/, +/"\,"/g' Input_file

Answer 4

Instead of complex sed pattern, you can use grep -ow option.
[nooka#lori ~]$ string1="a 1,a 2,a 3"
[nooka#lori ~]$ string2=" a 1, a 2, a 3, a 4"
[nooka#lori ~]$
nooka#lori ~]$ echo $(echo $string1 | grep -ow "[a-zA-Z] [0-9]"|sed "s/^/\"/;s/$/\"/")|sed "s/\" /\",/g"
"a 1","a 2","a 3"
[nooka#lori ~]$ echo $(echo $string2 | grep -ow "[a-zA-Z] [0-9]"|sed "s/^/\"/;s/$/\"/")|sed "s/\" /\",/g"
"a 1","a 2","a 3","a 4"
1) use grep -ow to get only those words as per the pattern defined above. You can tweak the pattern per your needs (for ex: [a-zA-Z] [0-9][0-9]* etc) for more patterns cases.
2) Then you wrap the output (a 1 or a 2 etc) with a " using the first sed cmd.
3) Then you just put , between 2 " and you get what you wanted. This assumes you pattern always follows a single space between string and number value.

Answer 5

Ok. Finally got it working.
string=" a 1 b 2 c 3 , something else , yet another one with spaces , totally working "
trimmed_string=$(awk '{gsub(/[[:space:]]*,[[:space:]]*/,",")}1' <<< $string)
echo ${trimmed_string}
a 1 b 2 c 3,something else,yet another one with spaces,totally working
string_as_list=$(sed -e 's/[^,]*/"&"/g' <<< ${trimmed_string})
echo ${string_as_list}
"a 1 b 2 c 3","something else","yet another one with spaces","totally working"
This whole thing had to be done because terraform expects list
variables to be passed like that. They must be surrounded by double quotes (" "),
delimited by comma( , ) inside square brackets([ ]).