partial md5sum

partial-md5.sh

#!/bin/bash

help_info() {
    echo "Usage : $0 file"; exit 0
}

if [ $# -lt 1 ]; then
    help_info
else
    if [ -f "$1" ]; then
        FILE=$1
    else
        help_info
    fi
fi

OS=$(uname -a | awk '{print $1}')

case $OS in
    Linux) MD5=md5sum;;
    Darwin) MD5=md5;;
    *) echo "Your device is not supported yet."; exit 0;;
esac

# BLOCK_SIZE 1M
BLOCK_SIZE=1048576

filesize=$(ls -al "$FILE" | awk '{print $5}')

# stage0 0 ~ 30M, All
if [ $filesize -lt $(expr 30 \* $BLOCK_SIZE) ]; then
    dd if="$FILE" bs=$BLOCK_SIZE 2> /dev/null | $MD5 | awk '{print $1}'; exit 0
fi

# stage1 30  ~ 200M, 1/3
# stage2 200 ~ 400M, 60M
# stage3 400 ~ 600M, 80M
# stage* 600M ~    ,100M
threshold_200M=$(expr 200 \* $BLOCK_SIZE)

stage_x=$(expr $filesize / $threshold_200M)

one_second() {
    half=$(expr $(expr $filesize / 2) / $BLOCK_SIZE)
}

one_third() {
    a_third=$(expr $(expr $filesize / 3) / $BLOCK_SIZE)
}

case $stage_x in
    0) one_third; dd if="$FILE" bs=$BLOCK_SIZE skip=$a_third count=$a_third 2> /dev/null | $MD5 | awk '{print $1}';;
    1) one_second; dd if="$FILE" bs=$BLOCK_SIZE skip=$(expr $half - 30) count=60 2> /dev/null | $MD5 | awk '{print $1}';;
    2) one_second; dd if="$FILE" bs=$BLOCK_SIZE skip=$(expr $half - 40) count=80 2> /dev/null | $MD5 | awk '{print $1}';;
    *) one_second; dd if="$FILE" bs=$BLOCK_SIZE skip=$(expr $half - 50) count=100 2> /dev/null | $MD5 | awk '{print $1}';;
esac