#!/bin/bash
#Author: LingYi
#Date: 2016.4.12
#Func: Delete duplicate files
#EG  : $0 [ DIR1 DIR2 ... DIRn ]
 
#Define the mnt file, confirming the write authority by yourself.  
md5sum_result_log="/tmp/$(date +%Y%m%d%H%M%S)"
 
echo -e "\033[1;31mMd5suming ...\033[0m"
  
find $@ -type f -print0 | xargs -0 -I {} md5sum {} | tee -a $md5sum_result_log
files_sum=$(cat $md5sum_result_log | wc -l)
 
# Define array, using the value of md5 as index, filename as element. 
# Firstly, you must do advance declaration to make sure the it‘s supported by bash. 
declare -A md5sum_value_arry
 
while read md5sum_value md5sum_filename
do 
    #Space in a file name, in order to support this case ，using the ‘+’ as the segmentation charater. 
    #So, if ‘+‘ appears in a file name, there will be problems. The use should choose the manual mode to delete redundant files.
    md5sum_value_arry[$md5sum_value]="${md5sum_value_arry[$md5sum_value]}+$md5sum_filename"
    (( _${md5sum_value}+=1 ))
done <$md5sum_result_log
 
 
# counting the duplicate file groups and the size of redundant files in this loop.
groups_sum=0
repfiles_size=0
for md5sum_value_index in ${!md5sum_value_arry[@]}
do
    if eval [[ \${_${md5sum_value_index}} -gt 1 ]]; then
        let groups_sum++
        need_print_indexes="$need_print_indexes $md5sum_value_index"
        eval repfile_sum=\$\(\( \$_$md5sum_value_index - 1 \)\)
        repfile_size=$( ls -lS "`echo ${md5sum_value_arry[$md5sum_value_index]}|awk -F‘+‘ ‘{print $2}‘`" | awk ‘{print $5}‘)
        repfiles_size=$(( repfiles_size + repfile_sum*repfile_size ))
    fi
done 
 
#Outputing the statistical information.
echo -e "\033[1;31mFiles: $files_sum    Groups: $groups_sum    \
Size: ${repfiles_size}B $((repfiles_size/1024))K $((repfiles_size/1024/1024))M\033[0m"
[[ $groups_sum -eq 0 ]] && exit
 
#The use chooses whether to check the file grouping or not. 
read -n 1 -s -t 300 -p ‘Show detailed information ?‘ user_ch
[[ $user_ch == ‘n‘ ]] && echo || {
    [[ $user_ch == ‘q‘ ]] && exit
    for print_value_index in $need_print_indexes
    do
        echo -ne "\n\033[1;35m$((++i)) \033[0m"
        eval echo -ne "\\\033[1\;34m$print_value_index [ \$_${print_value_index} ]:\\\033[0m"
        echo ${md5sum_value_arry[$print_value_index]} | tr ‘+‘ ‘\n‘
    done | more
}
 
#The user can choose the way of deleting file here. 
echo -e "\n\nManual Selection by default !"
echo -e " 1 Manual selection\n 2 Random selection" 
echo -ne "\033[1;31m"
read -t 300 USER_CH
echo -ne "\033[0m"
[[ $USER_CH == ‘q‘ ]] && exit
[[ $USER_CH -ne 2 ]] && USER_CH=1 || {
    echo -ne "\033[31mWARNING: you have choiced the Random Selection mode, files will be deleted at random !\nAre you sure ?\033[0m"
    read -t 300  yn
    [[ $yn == ‘q‘ ]] && exit
    [[ $yn != ‘y‘ ]] && USER_CH=1
}
 
#Handle files according to the user‘s selection 
echo -e "\033[31m\nWarn: keep the first file by default.\033[0m"
for exec_value_index in $need_print_indexes
do
    #This loop contains an array of files that are about to be deleted.
    for((i=0,j=2;i<$(echo ${md5sum_value_arry[$exec_value_index]} | grep -o ‘+‘ | wc -l); i++,j++))
    do 
        file_choices_arry[i]="$(echo ${md5sum_value_arry[$exec_value_index]}|awk -F‘+‘ ‘{print $J}‘ J=$j)"
    done
     
    eval file_sum=\$_$exec_value_index
    if [[ $USER_CH -eq 1 ]]; then
        #If the user selects a manual mode, handle the duplicate file group one by one in a loop.
        echo -e "\033[1;34m$exec_value_index\033[0m"
        for((j=0; j<${#file_choices_arry[@]}; j++))
        do
            echo "[ $j ]  ${file_choices_arry[j]}"
        done
        read -p "Number of the file you want to keep: " num_ch
        [[ $num_ch == ‘q‘ ]] && exit
        seq 0 $((${#file_choices_arry[@]}-1)) | grep -w $num_ch &>/dev/null || num_ch=0
    else
        num_ch=0
    fi
    #If the user selects the automatic deletion mode, then delete the redundant files
    for((n=0; n<${#file_choices_arry[@]}; n++))
    do
        [[ $n -ne $num_ch ]] && {
            echo -ne "\033[1mDeleting file \" ${file_choices_arry[n]} \" ... \033[0m"
            rm -f "${file_choices_arry[n]}"  
            [[ $? -eq 0 ]] && echo -e "\033[1;32mOK" || echo -e "\033[1;31mFAIL"
            echo -ne "\033[0m"
        }
    done
done