Bash cannot process git output one-by-one, items stuck together

I'm trying to run a script like below:

Changed_Files=$(git diff --diff-filter=dr --name-only origin/main..."$My_branch" | grep '^\s*foo/') 
  
for file in "${Changed_Files[@]}"; do 
  subdir=$(echo "$file" | cut -d / -f 2)
  if [[ "$subdir" == "bar1" ]]; then
    func1 "$file"
  else 
    echo ("Error")
    exit(1)
  fi
done

Sample Changed_Files:

foo/bar1/file1.json
foo/bar1/file2.json

Expected Result:

func1 gets run on both files.

Actual Result:

The script exits because subdir evaluates to bar1 bar1. Instead of processing the files one-by-one, the variable file will always contain the entire list of results from git diff.
I also tried

for i in "${!Changed_Files[@]}"; do
  file="${Changed_Files[$i]}"
  ...

But this gives the same result. How can I process items returned from git diff one-by-one instead of the entire output stuck together?

The problem that would cause the symptom described is that you're storing output in a string, not an array.

The right way to fix that in bash 4.x and later is:

readarray -t Changed_Files < <(git diff --diff-filter=dr --name-only origin/main..."$My_branch" | grep '^\s*foo/')

But don't use that exactly as given: \s is a PCRE extension, and POSIX-compliant grep implementations aren't guaranteed to support it. Use [[:space:]] instead.

If you need to support bash 3.x (like the version included with MacOS as well), you might consider a rewrite entirely in native shell, not relying on any external tools other than git diff itself:

# feature needed for +( ) to work; in older bash releases, even needed in PEs
shopt -s extglob

parentDirTgt=foo # the thing you were grepping for in original code

errors_seen=0
while IFS=/ read -r parentDir subDir rest <&3; do # fd 3 has "git diff" output
  parentDir=${parentDir##+([[:space:]])}          # strip leading whitespace
  [[ $parentDir = $parentDirTgt ]] || continue    # skip files not under parent
  [[ $rest ]] || continue       # skip files not under two directory layers
  file=$parentDir/$subDir/$rest # reconstruct full file name
  case $subDir in               # branch on subdirectory layer
    bar1) func1 "$file";;       # for directory foo/bar1, call function func1
    *)    echo "Ignoring change in unrecognized subdirectory $subDir" >&2
          errors_seen=1;;
  esac
done 3< <(git diff --diff-filter=dr --name-only origin/main..."$My_branch")

exit "$errors_seen"

This probably calls for some explanation.

• The while read looping practice is covered in BashFAQ #1. Instead of reading each line into just one variable, we split on /s and read into three variables: parentDir, subDir, and rest (containing everything after the second /).
• The use of <( ) instead of | to connect git diff to the loop is for reasons described in BashFAQ #24.
• The use of +([[:space:]]) is an extglob that matches as many spaces as possible, used in the context of the ${var##pattern} parameter expansion to strip the longest possible matching pattern from the beginning of a variable's contents.
• File descriptor 3 is used so that anything in func1 that tries to read from stdin will actually consumer original stdin instead of the git diff output.

See this running an an online sandbox at https://ideone.com/ZA1Cgn