[SOLVED] Avoid splitting string by space in output stored in bash array

Issue

The array A stores the output of this awk script:

A=(`awk 'BEGIN{
        z[1]=4;z[2]=7;z[3]=5;
        w[1]="how are you?"; w[2]="fine and you?"; w[3]="fine, thank you"; 
        
        for (i = 1; i <=3; i++ ) {
            printf("%s|%s\n",w[i],z[i])
        }
    }'`)   

Array A is supposed to be storing these 3 lines:

A[1] = "how are you?|4"
A[2] = "fine and you?|7"
A[3] = "fine, thank you|5"

But when I try to split the content of each line within A, is taking the space as field separator even I set the IFS ='|', because when I send this for loop script

n=${#A[*]}

for ((i=0; i<=n; i++)); do
    IFS='|' read -ra B <<< "${A[$i]}"
    
    echo "String: " ${B[0]} ", value: " ${B[1]}
done

I get this output:

String:  how , value:
String:  are , value:
String:  you? , value:  4
String:  fine , value:
String:  and , value:
String:  you? , value:  7
String:  fine, , value:
String:  thank , value:
String:  you , value:  5
String:  , value:

And my expected output is like this:

STRING: how are you?, VALUE: 4
STRING: fine and you?, VALUE: 7
STRING: fine, thank you, VALUE: 5

How to fix this? I'd like to preserve the structure of this script since is part of another bigger awk script with a similar output. Thanks in advance.

Solution

OP's current code stores each white-space delimited word as a separate entry in the array (aka word splitting - as Gordon mentions in comments). We see this with the following:

$ typeset -p A
declare -a A=([0]="how" [1]="are" [2]="you?|4" [3]="fine" [4]="and" [5]="you?|7" [6]="fine," [7]="thank" [8]="you|5")

A couple ideas for storing each line of awk output as a separate array entry:

# let bash map each line as a separate array entry

mapfile   -t A < <( awk '...' )
readarray -t A < <( awk '...' )       # readarray is a synonym for mapfile

# use a while/read loop to process each line separately

A=()                                  # init array

while read -r line
do
    A+=( "${line}" )                  # add ${line} as a new array entry
done < <( awk '...' )

All of these produce:

$ typeset -p A
declare -a A=([0]="how are you?|4" [1]="fine and you?|7" [2]="fine, thank you|5") 

---

As for OP's follow-on for/read/echo loop, bash array indices start with 0 so an array with n items will require a loop that goes from 0 to n-1 (ie, change i<=n to i<n - as Andrej mentions in comments). Rolling this into OP's current for loop:

n=${#A[*]}

for ((i=0; i<n; i++))
do
    IFS='|' read -ra B <<< "${A[$i]}"

    echo "String: " ${B[0]} ", value: " ${B[1]}
done

Another option is to feed the array directly to a while/read loop, eg:

while IFS='|' read -r string value
do
    echo "String: ${string} , value: ${value}"
done < <(printf "%s\n" "${A[@]}")

These both generate:

String:  how are you? , value:  4
String:  fine and you? , value:  7
String:  fine, thank you , value:  5

Answered By - markp-fuso

Answer Checked By - Pedro (WPSolving Volunteer)