Question

I have a CSV file, like this:

ACDB,this is a sentence
BECD,this is another sentence
BCAB,this is yet another

Each character in the first column corresponds to a word in the second column, e.g., in the first column, A corresponds with “this”, C with “is”, D with “a”, and B, with sentence.

Given the variable character, which can be set to any of the characters appearing in the first column, I need to isolate the word which corresponds to the selected letter, e.g., if I set character="B", then the output of the above would be:

sentence
this
this another

If I set `character=”C”, then the output of the above would be:

is
another
is

How can I output only those words which correspond to the position of the selected letter?

• The file contains many UTF-8 characters.
• For every character in column 1, there is always an equal number of words in column 2.
• The words in column 2 are separated by spaces.

Here is the code I have so far:

while read line
do
    characters="$(echo $line | awk -F, '{print $1}')"
    words="$(echo $line | awk -F, '{print $2}')"
    character="B"
done < ./file.csv

Answer 1

This might work for you:

x=B                                                      # set wanted key variable
sed '
:a;s/^\([^,]\)\(.*,\)\([^ \n]*\) *\(.*\)/\2\4\n\1 \3/;ta # pair keys with values
s/,//                                                    # delete ,
s/\n[^'$x'] [^\n]*//g                                    # delete unwanted keys/values
s/\n.//g                                                 # delete wanted keys
s/ //                                                    # delete first space
/^$/d                                                    # delete empty lines
' file
sentence
this
this another

or in awk:

awk -F, -vx=B '{i=split($1,a,"");split($2,b," ");c=s="";for(n=1;n<=i;n++)if(a[n]==x){c=c s b[n];s=" "} if(length(c))print c}' file
sentence
this
this another