#!/bin/sh

if [ $# -eq 0 ] || [ "$1" = "-h" ]; then
    cat<<EOF
Usage: ${1##*/} FOLDERS

Convert all 'bad' encoding to UTF-8/LF.

WARNING: It will fail for encodings other the one explicitly supported below.

EOF
fi

if [ -z "$(command -v recode)" ]; then
    echo "recode needed."
    exit
fi

for i ; do
    while IFS= read -r j; do
        CODING=$(file "$j")

        if [ -n "$(echo $CODING | grep 'ISO-8859')" ]; then
            echo "ISO-8859:   [$j]"
            recode latin1..utf-8 "$j"

        elif [ -n "$(echo $CODING | grep 'Non-ISO extended-ASCII')" ]; then
            echo "cp1252:     [$j]"
            recode cp1252..utf-8 "$j"

        elif [ -n "$(echo $CODING | grep 'UTF-16 Unicode text')" ]; then
            echo "UTF-16:     [$j]"
            recode utf-16..utf-8 "$j"

        elif [ -n "$(echo $CODING | grep 'UTF-8 Unicode (with BOM)')" ]; then
            echo "UTF-8 BOM:  [$j]"
            sed -i '1s/^.//' "$j"
            ## sed -i is not the fastest depending on the implementations. The
            ## following commands work, but may be overkill.
            # dd iflag=skip_bytes skip=3 if=file.srt of=temp.srt
            # dd bs=1 skip=3 if=file.srt of=temp.srt
            # tail -c +32 file.srt > temp.srt
        fi

        if [ -n "$(echo $CODING | grep 'CRLF')" ]; then
            echo "CRLF:       [$j]"
            sed -i 's/\r//g' "$j"
        fi

    done <<EOF
$(find "$i" -type f -size -50M -print)
EOF
done