hashdeep

#! /bin/bash -

# Emulates the output of sha256deep for the given directories
# from the previously hashed table file_hashes.out
if [[ "$1" == "-h" ]] || [[ "$1" == "--help" ]]; then
    >&2 printf "Usage: $0 <dir>\n"
    exit 0
fi

dir="$(dirname $(readlink -f ${BASH_SOURCE[0]}))"

path="$(readlink -f "$1")"
path="${path//\\/\\\\}"
path="${path//$'\n'/\\n}"
escaped=$(printf '%s' "$path" | \
    sed -r 's/([\$\.\*\/\[\\^])/\\\1/g' | \
    sed 's/[]]/\[]]/g')
replace=$(printf '%s' "$1" | \
    sed -r 's/(["&\/\\])/\\\1/g')

printf -v pattern '^[0-9]*,[^,]*,%s' "$escaped"
printf -v rep 's/\(^[0-9]*,[^,]*,\)%s/\\1%s/' "$escaped" "$replace"
export pattern rep
export LC_ALL=C # byte-wise sorting

printf '%%%% hashdeep imitate by '
printf 'https://github.com/fraunhofer-izi/duplicates\n'
printf '%%%% size,sha256,filename\n'
printf '## Invoked from: %s\n' "$(pwd -P)"

parallel --will-cite -k --pipepart -a "$dir/file_hashes.out" \
    --block 100M 'grep -e "$pattern" | sed -e "$rep"'