Convert an array stored in a string variable back to an array

Question

This will create arr[0]=hy, arr[1]="hello man":

declare -a arr=(hy "hello man")

I have the 'hy "hello man"' this as whole string stored in variable such that:

echo "$var"
hy "hello man"

How can I split it back to an array? I tried IFS= IFS=" " readarray and all but array only results into arr[0]="hy "hello man"" and not arr[0]=hy, arr[1]="hello man"

Desired array:

arr[0]=hy arr[1]="hello man"

For testing purpose:

arg () { printf "%d args:" "$#"; [ "$#" -eq 0 ] || printf " <%s>" "$@"; echo; }

arg hy "hello man"
2 args: <hy> <hello man>

But

var='hy "hello man"'
arg $t
1 arg: <hy "hello man">

Is there a quoting trick that can help with my problem?

Answer 1

If you're bash is new enough that has loadable_builtins which include dsv, you could try something like this.

#!/usr/bin/env bash

arg(){
  local array
  enable dsv || return
  dsv -Sga array -d' ' "$1"
  declare -p array
}

var='hy "hello man"'

arg "$var"

---

Output

declare -a array=([0]="hy" [1]="hello man")

---

As per the OP's requested output, the modified function.

arg(){
  local array
  enable dsv || return
  dsv -Sga array -d' ' "$1"
  printf '%s args: ' "${#array[*]}"
  printf '<%s>' "${array[@]}"
}

var='hy "hello man"'

arg "$var"

---

Output

2 args: <hy><hello man>

---

enable -p should show the loadable_builtin dsv

---

• See help enable
• See help dsv

Answer 2

Some explanation about dsv and an alternative using sed

In addition to Jetchisel's correct answer, there are some details.

And as some installation don't have builtins installed, I'v posted a stronger alternative than eval using sed to parse correctly submitted string.

1. Basic usage:

As previously commented on other posts, eval is evil!! For showing this I will use:

var='my "Hello world." $(</etc/passwd)'

Then

enable dsv
dsv -d\  -a arry "$var"
declare  -p arry

declare -a arry=([0]="my" [1]="Hello world." [2]="\$(</etc/passwd)")

Yes, for this use case, switches -S or -g is not required.

2. Usage

After run help dsv I've read a lot of explanation the most usefull paragraph is:

Parse STRING, a line of delimiter-separated values, into individual
fields, and store them into the indexed array ARRAYNAME starting at
index 0. The parsing understands and skips over double-quoted strings. 
If ARRAYNAME is not supplied, "DSV" is the default array name.
If the delimiter is a comma, the default, this parses comma-
separated values as specified in RFC 4180.
....
The -d option specifies the delimiter. The delimiter is the first
character of the DELIMS argument. Specifying a DELIMS argument that
contains more than one character is not supported and will produce
unexpected results. The -S option enables shell-like quoting: double-
quoted strings can contain backslashes preceding special characters,
and the backslash will be removed; and single-quoted strings are
processed as the shell would process them. The -g option enables a
greedy split: sequences of the delimiter are skipped at the beginning
and end of STRING, and consecutive instances of the delimiter in STRING
do not generate empty fields. If the -p option is supplied, dsv leaves
quote characters as part of the generated field; otherwise they are
removed.

Shortened:

 -a ARRAYNAME Array name to populate (default: DSV)
 -d DELIMS    Separator character (one character only!)
 -S           Shell like (differences single or double quote)
 -g           Greedy (ignore leading and trailing delimiters)
 -p           Preserve quotes      

3. About eval is evil

If you try same string with one of other solution posted here:

Suggested by pmf:

unset arr; declare -a arr="($var)"
declare -p arr

Then command will dump my whole password file into array:

declare -a arr=([0]="my" [1]="Hello world." [2]="root:x:0:0:root:/root:/bin/bash
" [3]="daemon:x:1:1:daemon:/usr/sbin:/usr/sbin/nologin" [4]="bin:x:2:2:bin:/bin:
/usr/sbin/nologin" [5]="sys:x:3:3:sys:/dev:/usr/sbin/nologin" [6]="sync:x:4:6553
... 59 more lines ...
/usr/sbin/nologin" [122]="_lxd:x:107:117::/var/lib/lxd/:/bin/false")

Or other answer, suggested by Hamidreza Shafizadeh:

unset arr; eval "arr=($var)"
declare -p arr

Will render same array!

So you have to be confident with variable's content source!!!

4. Alternative using sed.

If you don't have bash-builtins installed, and still want some security, you could use sed from Replace every comma not enclosed in a pair of double quotes:

IFS=$'\t' read -a arr < <(
    sed -e ':a;s/^\(\("[^"]*"\|'\''[^'\'']*'\''\|[^" '\'']*\)*\) /\1\t/;ta' <<<"$var"
)
declare -p arr

declare -a arr=([0]="hy" [1]="\"hello man\"" [2]="\$(</etc/passwd)")

4.1. Using sed in a function.

string2arry() {
    IFS=$'\t' read -a "$2" < <(
        sed -e <<<"$1" \
            ':a;s/^\(\("[^"]*"\|'\''[^'\'']*'\''\|[^" '\'']*\)*\) /\1\t/;ta'
    )
}
string2arry "$var" arr
declare -p arr

declare -a arr=([0]="hy" [1]="\"hello man\"" [2]="\$(</etc/passwd)")

4.1b. Same, cleaning quotes

string2arry() {
    local -i _i
    local -n _res="$2"
    IFS=$'\t' read -a _res < <(
        sed -e <<<"$1" \
            ':a;s/^\(\("[^"]*"\|'\''[^'\'']*'\''\|[^" '\'']*\)*\) /\1\t/;ta'
    )
    for _i in ${!_res[@]}; do
        lc=${_res[_i]::1} rc=${_res[_i]: -1}
        [[ $lc == "$rc" ]] && [[ -z ${lc/[\'\"]} ]] &&
            _res[_i]=${_res[_i]:1: -1}
    done
}
string2arry "$var" arr
declare -p arr

declare -a arr=([0]="hy" [1]="hello man" [2]="\$(</etc/passwd)")