imc.sh

#!/bin/bash
# imc.sh - image copy
# 2015 ed <irc.rizon.net>
# something something mit license
#
set -e
[ "x$2" == "x" ] &&
{
    echo
    echo 'requires 2 arguments'
    echo
    echo '1) op mode:'
    echo '   d, check for dupes'
    echo '   n, check for continuity'
    echo '   c, copy all images in folder'
    echo '   r, copy all images recursively'
    echo
    echo '2) target directory'
    echo
    exit 1
}

imagemask='jpg|jpeg|png|gif'
rm imc.todo.list 2>/dev/null || true
op="$1"
to="$2"
cpy=''

function getlatest()
{
    {
        # list target directory with fallback '0.jpg'
        echo '0.jpg'
        find "$to" -maxdepth 1 2>/dev/null || true
    } |

    # remove until final /
    sed 's_.*/__' |

    # numerical sort
    sort -n |

    # filter by number.whatever
    grep -E '^[0-9]+\.' |

    # return the number of the last entry
    tail -n 1 |
    sed 's/\./\n/' |
    head -n 1
}

# copy single level
[ "x$op" == "xc" ] &&
    cpy='. -maxdepth 1'

# copy recursively
[ "x$op" == "xr" ] &&
    cpy='.'

# and then
[ $cpy ] &&
{
    find $cpy -type f |
    grep -E "\.($imagemask)$" > imc.todo.list

    echo -n 'Starting after '
    fid=$(getlatest)
    echo $fid

    # get number of files to copy
    remains=$(
        cat imc.todo.list |
        wc -l
    )

    # make sure target exists and has md5 list
    mkdir -p "$to"
    touch "$to/.md5"

    cat imc.todo.list |
    while read file
    do
        echo -n "$remains  $file "
        remains=$(( remains - 1 ))
        hash=$(
            md5sum -- "$file" |
            head -c 32
        )

        # check if exists
        grep -qE "$hash" "$to/.md5" &&
        {
            echo -e "\033[1;33m no\033[0m"
            continue
        }

        fid=$((fid+1))
        ext="$(
            echo "$file" |
            sed 's/\./\n/g' |
            tail -n 1
        )"

        # copy image + add md5sum
        cp -- "$file" "$to/$fid.$ext"
        echo "$hash  $fid.$ext" >> "$to/.md5"
        echo -e "\033[1;32m Copied\033[0m"
    done
    rm imc.todo.list
    echo 'done'
    exit 0
}

# check for dupes
[ "x$op" == "xd" ] &&
{
    echo -e '\033[1;33mRunning dupe check\033[0m'

    md5sum "$to/"* |
    sed 's_  .*/_  _' > "$to/.md5"

    cat "$to/.md5" |
    sort |
    uniq -d -w 32 |
    while read hash
    do
        echo "Removing $hash"
        fn="$(
            echo "$hash" |
            cut -c 35-
        )"
        rm "$to/$fn"
    done
    echo -e '\033[1;32mFinished checking for dupes.\033[0m'
    echo 'Run a continuity check if changes were made'
    exit 0
}

# check continuity
[ "x$op" == "xn" ] &&
{
    echo -e '\033[1;33mRunning continuity check\033[0m'
    tn=0
    find "$to" -maxdepth 1 |
    grep -E "\.($imagemask)$" |
    sort -n > imc.cont.list

    # iterate over all existing files
    cat imc.cont.list |

    # remove everything until final slash
    sed 's_.*/__' |

    # filter to number.whatever
    grep -E '^[0-9]+\.' |

    # only keep number
    sed 's/\..*//' |

    while read fn
    do
        tn=$(( tn + 1 ))
        while [ $tn -lt $fn ]
        do
            # if tn and fn mismatch here,
            # pick the first file that still exists
            # off the end of the file list
            # and swap it up to tn
            tac imc.cont.list |
            while read repl
            do
                [ -e "$repl" ] &&
                {
                    rnum=$(
                        echo "$repl" |
                        sed 's_.*/__;s/[^0-9].*//'
                    )
                    [ $rnum -le $tn ] &&
                    {
                        echo -e "\033[1;31mNO MORE SUBSTITUTE FILES WTF\033[0m"
                        break
                    }
                    ext=$(
                        echo "$repl" |
                        sed 's/\./\n/g' |
                        tail -n 1
                    )
                    echo "Moving $rnum.$ext to $tn.$ext"
                    mv "$to/$rnum.$ext" "$to/$tn.$ext"
                    break
                }
            done
            tn=$(( tn + 1 ))
        done
    done
    rm imc.cont.list
    echo -e '\033[1;32mFinished checking continuity.\033[0m'
    echo 'Run a dupe check to refresh md5 cache if changes were made'
    exit 0
}

echo 'wat'