ddiff

#!/usr/bin/env bash
# Directory/file-wise diff. Compares file/folder names. Does not compare file contents.
#
# author: andreasl

script_name="${0##*/}"
IFS= read -r -d '' script_description <<HELP_EOF
${script_name}
Compare file trees in a readable format using find, sort, comm and sed.
Only compares file/folder names, not contents!

Usage:
  ${script_name} [OPTIONS] <dir1> <dir2>
  ${script_name} <dir1> <dir2> [OPTIONS]

Options:
  -1, --in-1:                          Only show entries that are in dir1.
  -2, --in-2:                          Only show entries that are in dir2.
  -C, --no-color:                      Disable colored output.
  -d, --only-different:                Show only directories/files that exist only in dir1 or only in dir2.
  -e <pattern>, --exclude <pattern>:   Exclude glob pattern <pattern>, e.g. "*.git".
  -h, --help:                          Print the help message.
  -s --only-same:                      Show only directories/files that exist in both dir1 and dir2.
                                       Same as -1 -2.

Examples:
  ${script_name} -e "*.git" -C -d dev dev_backup
  ${script_name} ~/Photos ~/Media -s
  ${script_name} --help
HELP_EOF

# parse command line arguments
excludes=()
must_be_in_1=false
must_be_in_2=false
only_different=false
only_same=false
source_dir=.
target_dir=.
use_colors=true
while [ "$#" -gt 0 ]; do
    case "$1" in
    -1 | --in-1)
        must_be_in_1=true
        ;;
    -2 | --in-2)
        must_be_in_2=true
        ;;
    -C | --no-color)
        use_colors=false
        ;;
    -d | --only-different)
        only_different=true
        ;;
    -e | --exclude)
        excludes+=(-path "$2" -prune -o)
        shift
        ;;
    -h | --help)
        printf -- "$script_description"
        exit 0
        ;;
    -s | --only-same)
        only_same=true
        ;;
    *) # unknown option
        source_dir="$1"
        shift
        target_dir="$1"
        ;;
    esac
    shift
done

# get the file trees and do the comparisons
# `cd` into the dir and strip the leading `./` instead of using `-printf "%P\n"`, which is GNU-only
tree1="$(cd "$source_dir" && find . "${excludes[@]}" -print | sed -e 's|^\./||' -e 's|^\.$||' | sort)"
tree2="$(cd "$target_dir" && find . "${excludes[@]}" -print | sed -e 's|^\./||' -e 's|^\.$||' | sort)"
output="$(comm <(printf '%s\n' "$tree1") <(printf '%s\n' "$tree2"))"

# postprocess the output from comm, which marks its three columns by leading tabs
# `$'\t'` is a real tab character; writing `\t` in the regex instead would only work with GNU
# sed, as BSD/macOS sed reads it as a literal `t`.
tab=$'\t'
output="$(sed -E "s:(^[^${tab}].*):1 \1:g" <<<"$output")"
output="$(sed -E "s:^${tab}{2}(.*):3 \1:g" <<<"$output")"
output="$(sed -E "s:^${tab}{1}(.*):2 \1:g" <<<"$output")"

# filter the output
if [ "$must_be_in_1" == true ] || [ "$only_same" == true ]; then
    output="$(sed '/^2 /d' <<<"$output")"
fi
if [ "$must_be_in_2" == true ] || [ "$only_same" == true ]; then
    output="$(sed '/^1 /d' <<<"$output")"
fi
if [ "$only_different" == true ]; then
    output="$(sed '/^3 /d' <<<"$output")"
fi

# colorize output
if [ "$use_colors" == true ]; then
    red='\\e[31m'
    green='\\e[32m'
    orange='\\e[33;1m'
    nc='\\e[0m'

    output="$(sed -E "s:^1 (.*):${red}1 \1${nc}:g" <<<"$output")"
    output="$(sed -E "s:^2 (.*):${green}2 \1${nc}:g" <<<"$output")"
    output="$(sed -E "s:^3 (.*):${orange}3 \1${nc}:g" <<<"$output")"
fi

printf '%b\n' "$output"