2utf8

#!/bin/bash
#===============================================================================
#
#          FILE:  2utf8.sh
#
#         USAGE:  ./2utf8.sh
#
#   DESCRIPTION:
#
#       OPTIONS:  ---
#  REQUIREMENTS:  ---
#          BUGS:  ---
#         NOTES:  ---
#        AUTHOR:  adaptee (), adaptee@gmail.com
#       COMPANY:  Open Source Corporation
#       VERSION:  1.0
#       CREATED:  09/25/10 11:33:25 CST
#      REVISION:  ---
#===============================================================================

if  which tput 2>&1 >/dev/null ;then
    BRIGHTRED=$(tput bold;tput setaf 1)
    BRIGHTGREEN=$(tput bold;tput setaf 2)
    NOCOLOR=$(tput sgr0)
else
    BRIGHTRED='\e[01;31m'
    BRIGHTGREEN='\e[01;32m'
    NOCOLOR='\e[00m'
fi

green-echo ()
{
    echo -e "${BRIGHTGREEN}$@${NOCOLOR}"
}

red-echo ()
{
    echo -e "${BRIGHTRED}$@${NOCOLOR}"
}

function guess_encoding_by_enca()
{
    local target="$1"
    local encoding="unknown"

    # ask chardet to run in batchmode, only providing the encdding name
    # encoding name are forced upper case
    local guess=$(enca -i "${target}" | tr '[a-z]' '[A-Z]'  )
    #echo -e "[debug] >>${guess}<<"


    case "${guess}" in

        "ASCII"|"UTF-8" )
            encoding="done"
        ;;

        "UTF-16BE" )
            encoding="ucs-2be"
        ;;

        "UTF-16LE" )
            encoding="ucs-2le"
        ;;

        "UTF-32BE" )
            encoding="ucs-4be"
        ;;

        "UTF-32LE" )
            encoding="ucs-4le"
        ;;

        "GB2312"|"GBK"|"GB18030" )
            encoding="gb18030"
        ;;

        "BIG5" )
            encoding="big5"
        ;;

        "SHIFT_JIS" )
            encoding="sjis"
        ;;

        "EUC_JP" )
            encoding="ecu-jp"
        ;;

        * )
            encoding="unknown"
        ;;

    esac

    echo "${encoding}"

}


################################################################################

function guess_encoding_by_chardet()
{
    local target="$1"
    local encoding="unknown"

    # ask chardet to run in batchmode, only providing the encdding name
    # encoding name are forced upper case
    local guess=$(chardet -b "${target}" | tr '[a-z]' '[A-Z]'  )
    #echo -e "[debug] >>${guess}<<"


    case "${guess}" in

        "ASCII"|"UTF-8" )
            encoding="done"
        ;;

        "UTF-16BE" )
            encoding="ucs-2be"
        ;;

        "UTF-16LE" )
            encoding="ucs-2le"
        ;;

        "UTF-32BE" )
            encoding="ucs-4be"
        ;;

        "UTF-32LE" )
            encoding="ucs-4le"
        ;;

        "GB2312"|"GBK"|"GB18030" )
            encoding="gb18030"
        ;;

        "BIG5" )
            encoding="big5"
        ;;

        "SHIFT_JIS" )
            encoding="sjis"
        ;;

        "EUC_JP" )
            encoding="ecu-jp"
        ;;

        * )
            encoding="unknown"
        ;;

    esac

    echo "${encoding}"

}


function guess_encoding()
{
    local target="$1"
    local encoding="unknown"

    if  which chardet 2>&1 >/dev/null ; then
        encoding=$(guess_encoding_by_chardet "${target}")
    elif which enca 2>&1 >/dev/null ; then
        ehcoding=$(guess_encoding_by_enca "${target}")
    fi

    echo "${encoding}"
}


for item in "$@" ; do

    encoding=$(guess_encoding "${item}")

    case "${encoding}" in

        "done" )
            green-echo "${item} is already using UTF-8"
        ;;

        "unknown" )
            red-echo "${item} is using unknown encoding"
        ;;

        * )
            old="${item}.$$.old"
            temp="${item}.$$.temp"

            cp "${item}" "${old}"
            iconv -f "${encoding}" -t utf-8 -c "${item}"  > "${temp}"
            mv "${temp}" "${item}"

            echo "${item} is converted from ${encoding} to utf-8"

        ;;

    esac

done