Write values with embedded commas from arrays into CSV file using BASH

Question

I am trying to write different arrays into a csv file ($file) columns.
Each array has the same number of values. For example, arr1 and arr2 have 3 values each

arr1=( 23 45 34 )
arr2=( "abc,d"  ef  g )

I tried the following code but I'm getting the wrong results

for i in "${!arr1[@]}"; do
echo  "${arr1[i]}, ${arr2[i]}"  >> $file
done

I'm getting the following where col1, col2 and col3 are 3 columns.

col1    col2    col3
23      "abc    d"
45       ef 
34       g

but the desired result is

col1    col2   
23      "abc,d"
45       ef 
34       g

It seems like the embedded coma in "abc,d" creates a problem. Anyone knows a way around this or any better way to do this?

Thank you in advance!

Answer 1

It's not clear from your question exactly how you are looking at the file (perhaps with a spreadsheet program?) however the shell is likely stripping off the quotes in your array definition: if you want embedded commas to be handled in a CSV-complient way, you should add them as literal quotes ex.

arr1=( 23 45 34 )
arr2=( \"abc,d\"  ef  g )

so that the shell's internal representation looks like

$ declare -p arr1 arr2
declare -a arr1=([0]="23" [1]="45" [2]="34")
declare -a arr2=([0]="\"abc,d\"" [1]="ef" [2]="g")

Then for example

for i in "${!arr1[@]}"; do 
  printf '%s,%s\n' "${arr1[i]}" "${arr2[i]}"
done > "$file"

results in

$ cat "$file"
23,"abc,d"
45,ef
34,g

Another option is to quote all fields ex.

arr1=( 23 45 34 )
arr2=( "abc,d"  ef  g )

for i in "${!arr1[@]}"; do 
  printf '"%s","%s"\n' "${arr1[i]}" "${arr2[i]}"
done > "$file"

giving

$ cat "$file"
"23","abc,d"
"45","ef"
"34","g"

Answer 2

This answer assumes that your data contains no literal tabs.

Start by creating a simple column for each array (with header), and combine them using paste, creating a TSV data stream:

#!/bin/bash

arr1=( 23 45 34 )
arr2=( abc,d ef g )

paste <( echo col1; printf '%s\n' "${arr1[@]}" ) \
      <( echo col2; printf '%s\n' "${arr2[@]}" )

The output of this will read

col1    col2
23      abc,d
45      ef
34      g

You can then use Miller to convert it into properly quoted CSV:

#!/bin/bash

arr1=( 23 45 34 )
arr2=( abc,d ef g )

paste <( echo col1; printf '%s\n' "${arr1[@]}" ) \
      <( echo col2; printf '%s\n' "${arr2[@]}" ) | mlr --t2c cat

This would generate

col1,col2
23,"abc,d"
45,ef
34,g

... which a CSV-aware program would read as

col1	col2
23	abc,d
45	ef
34	g

A CSV-aware spreadsheet program would likely also be able to read that TSV data, so Miller would not be strictly needed.