Column to row translation using sed

Question

I'm writing a bash script to get data from procmail and produce an output CSV file.

At the end, I need to translate from this:

---label1: 123456
---label2: 654321
---label3: 246810
---label4: 135791
---label5: 101010

to this:

label1,label2,label3,label4,label5
123456,654321,246810,135791,101010

I could do this easily with a ruby script but I don't want to call another script other than the bash script. So I've though of doing it with sed.

I can extract the data I want like this:

sed -nr 's/^---(\S+): (\S+)$/\1,\2/p'

But I don't know how to transpose it. Can you help me?

Answer 1

If you're running sed I'd argue that you are calling another script. So why not write it in Ruby if that's easier to write and maintain?

If you're worried about having multiple files could embed the Ruby code in the bash script as a here document (assuming Ruby can read a script from the Standard Input).

Answer 2

Perhaps this is what you are looking for? (from unix.com)

num=$(awk -F"," 'NR==1 { print NF }' data)
print $num

i=1
while (( $i > tmpdata
(( i = i + 1 ))
done
mv tmpdata data

Answer 3

Finally I resolved it using ruby, as suggested buy Dave Webb but as a one-liner not a here document with the following script:

ruby -ne 'BEGIN{@head=[];@data=[]}; @head << $1 && @data << $2 if $_.match(/^---(\S+): (\S+)$/); END{puts @head.join(",");puts @data.join(",")}' $FILE

I didn't know that I could use BEGIN, END blocks to set the variables and output the results.

Answer 4

you said you needed bash script..

#!/bin/bash
labels=`awk '/---/{printf("%s,",$1)}' file.txt `
values=`awk '/---/{printf("%s,",$2)}' file.txt `
labels=`echo $labels|sed 's/---//g;s/\://g;s/\,$//'`
values=`echo $values|sed 's/\,$//`
echo $labels
echo $values

Answer 5

You can do something like this from within the shell script:

cat demo | awk -F':' '{print $1}' | sed -e s/'---'// | tr '\n' ',' > csv_file

cat demo | awk '{print $2}' | tr '\n' ',' >> csv_file

Answer 6

you can just do everything in awk

awk 'BEGIN{
    FS=": "
}
{ 
  gsub("---","")
  label[++c] = $1
  num[++d] =$2
}
END{
    for(i=1;i<c;i++){
        printf label[i]","
    }
    print label[c]
    for(i=1;i<d;i++){
        printf num[i]","
    }
    print num[d]    
}' file

output

# ./test.sh
label1,label2,label3,label4,label5
123456,654321,246810,135791,101010

redirect the output to csv file as needed