#!/usr/local/bin/bash

#
# CHANGELOG:
#
#       0.2.1 - pridano dekodovani 7bit transfer encoding (k base64)
#               2005-10-26
#
#	0.2.2 - minor changes to have server compatibility
#		2005-10-27
#
#	0.2.3 - minor corrections
#		2005-10-27
#

logger -p mail.info -t "$0" "executed"

# CONFIG
outdir="$1"                                     # vystupni adresar, zadany je prvnim parametrem
if [ -z "$outdir" ]
then
    outdir="/var/spool/mimExplode"
fi

if [ "$0" == "/usr/local/bin/mimExplode.spam" ]
then
outdir="$outdir"/spam
elif [ "$0" == "/usr/local/bin/mimExplode.nospam" ]
then
outdir="$outdir"/nospam
else
logger -p mail.warn -t "$0" "WARNING - default outdir used; $outdir"
exit
fi

# MAKE TEMP FILENAME
pid=$$.`date +%s`                               # todle je unikatni cislo, dle bash-e
tempdir="/tmp"                                  # adresar pro docasne soubory
input="$tempdir/$pid-input.mimexplode"          # vstup ulozime docasne sem, /tmp nebo /dev/shm (ramdisk)

# STDIN -> TEMP FILE
cat >"$input"                                   # ulozim stdin do souboru

## FILE -> BOUNDARY STRING
#
function get_boundary_from_file() { 

    # hledam prislusny content-type, boundary je na dalsim radku (-A 1)
    boundary=`grep "$1" -A 1 -e "multipart/"`   

    # nasli jsme neco?
    if [ -n "$boundary" ]
    then

        # nalezeni oddelovace a jeho vyparsovani z radku
        echo "$boundary" \
        | grep 'boundary' \
        | head -n 1 \
        | sed 's/^.*boundary[ ]*=[ ]*\"//' \
        | sed 's/".*$//'

    fi # -n "$boundary"
}

## OUTPUT
#
function deliver_part() {
    mv "$1" "$outdir/$pid.$2"
}

## PROCESSING THE PART OF MAIL
#
function process_part() {

    # overeni jestli jde o mail
    isMail=1

# my variant
#    for header in "return-path" "received" "from" "subject"
#    do
#        count=`grep "$part" -ie "^$header:" | wc -l`
#        if [ $count -eq 0 ]
#        then
#            isMail=0
#            break
#        fi
#    done

# better variant
    if [ `file "$1" | grep -e "smtp mail text" | wc -l` -eq 0 ]
    then
        isMail=0
    fi
#
    # pokud jde o mail, posleme ho na vystup
    if [ $isMail -eq 1 ]
    then
        deliver_part "$1" "$2"
    fi
}

## FRAGMENT PROCESSOR
#
function process_fragment() {

    # vstup
    part="$1"
    
    # zjistime hlavicku
    hdrlines=`grep "$part" -nEe '^$' | head -n 1 | cut -f1 -d:`
    header=`sed -n -e "1,$hdrlines p" "$part"`

    # TRANSPORT ENCODING
    transport=`echo "$header" | grep -Ee "^Content-Transfer-Encoding" | cut -f2 -d: | sed 's/ //'`
    newhead=1
    
#    echo "-$transport-"
    
    # decode
    case "$transport" in
    
        base64) # dekodovani base64
		logger -p mail.info -t "$0" "INFO - base64 attachment included"
                sed -e "1,$hdrlines s/.*//" "$part" | decode-base64 >"$part.decoded"
                # nahrazeni
                rm -f "$part"
                mv "$part.decoded" "$part"
                ;;

        7bit)   # dekodovani 7bit - jde o odstraneni hlavicek navic
		logger -p mail.info -t "$0" "INFO - 7bit attachment included"
                f=$(( $hdrlines + 1 ))
                t=`wc -l "$part" | sed 's/^ *//' | cut -f1 -d" "`
                sed -n -e "$f,$t p" "$part" >"$part.decoded"
                # nahrazeni
                rm -f "$part"
                mv "$part.decoded" "$part"
                ;;
                
        *)      # nezname kodovani - TODO: zalogovat $transport
                newhead=0
		logger -p mail.crit -t "$0" "CRITICAL - unknown transport encoding: $transport"
                ;;
                
    esac

    # prepocitani hlavicek
    if [ "$newhead" -eq 1 ]
    then
        hdrlines=`grep "$part" -nEe '^$' | head -n 1 | cut -f1 -d:`
        header=`sed -n -e "1,$hdrlines p" "$part"`
    fi
    
    # co s dekodovanym souborem?
    process_part $part "$2"
}

## FILE + LINE NUMBERS -> PARTS
#
function split_file_on_lines() {

    # s cim pracujeme?
    input=$1                                    # prvni parametr je vstupni soubor
    output="$tempdir/$pid-part.mimexplode"      # docasny soubor pro dekodovani
    shift
    
    # dalsi parametry jsou cisla radku, a prvni fragment nas nezajima
    n=1
    from=$(( $1 + 1 ))
    shift
    while [ -n "$1" ]
    do
        # vyriznuti casti
        to=$(( $1 - 1 ))
        sed -n -e "$from,$to p" "$input" >"$output"

        # zpracovani        
        process_fragment "$output" "$n"
    
        # dalsi prosim
        from=$(( $1 + 1 ))
        n=$(( $n + 1 ))
        shift
    done
    
    # uz to nepotrebujeme
    rm -f "$output" &>/dev/null
}

# GET BOUNDARY
boundary=`get_boundary_from_file "$input"`

# FOUND?
if [ -n "$boundary" ]
then

    # najdeme cisla radku (zajimave je ze bounary ted ma -- na zacatku)
    lines=`grep -n "$input" -Ee "^--$boundary" | cut -f1 -d:`
        
    # rozsekame vstupni soubor dle radku
    split_file_on_lines "$input" $lines

fi

# DELETE TEMP
rm -f "$input" &>/dev/null              # odstranim docasny soubor, bez zbytecnych kecu