#!/bin/sh
#
# l'autre babelfish: translate
# the other babelfish: translate
#
# Depends on "curl", by Daniel Stenberg
#
# usage: translate "Where's the form? Don't ask. Don't think. [...]"
#        translate fr_en "Si une horloge était conscinete d'elle même [...]"
#        translate UTF8 fr_de "Les petits oiseaux gazouillent dans les arbres"
#
# Copypouet þ Benoît Rouits <brouits@free.fr>
# Inspired from 'babelfish', by Thomas Nemeth.

PROG=`basename $0`
VERSION="0.4" #2008/01/20

DEBUG=0
if [ "$1" = "--debug" ]; then
  DEBUG=1
  shift
fi

MY=`echo $LANG | cut -d"_" -f 1`
#MY is taken as your expected natural language
#i hope this works for all countries...
if [ $MY = "C" -o $MY = "" ]
then
 MY="en"
#as you can see, your language defaults to English if not guessed or given.
fi
[ "$DEBUG" = "1" ] && echo "DEBUG => MY = ${MY}"

#############beginning of FROM/TO default configuration##########
FROM="en" #set your prefered FROM to be more convenient for you
TO="$MY"  #but you can obviously switch the defaults...
#############end of default FROM/TO configuration################

FROMTO="${FROM}_$TO"
[ "$DEBUG" = "1" ] && echo "DEBUG => FROMTO = $FROMTO"

#############beginning of TEXT ENCODING default configuration##########
TEXT="If a clock was conscious of itself, it could never keep time. --K.Werner"
ENCODING="`echo $LANG | sed -e 's/.._...\(.*\)/\1/'`" #set your defaults here
#############end of default TEXT ENCODING configuration################

ENCODSTYPES="us-ascii iso8859-1 iso8859-2 iso8859-3 iso8859-4 iso8859-5 iso8859-iso8859-6 iso8859-7 iso8859-8 iso8859-9 iso8859-10 iso8859-11 iso8859-12 iso8859-13 ido8859-14 iso8859-15 UTF8"
TRADUCSTYPES="en_de en_es en_fr en_it en_pt en_ja en_ko en_zh de_en de_fr es_en fr_en fr_de it_en pt_en ja_en ko_en zh_en"

if [ "$1" = "--help" -o "$1" = "-h" ]
then
  echo "$PROG v$VERSION"
  echo "Usage : $PROG [encodage] [direction] mots"
  echo "Exemple d'encodage : iso8859-1, UTF8 ... (defaut : $ENCODING)"
  echo "Direction de traduction (défaut : $FROMTO) :"
  echo "en_de            Anglais   vers Allemand"
  echo "en_es            Anglais   vers Espagnol"
  echo "en_fr            Anglais   vers Français"
  echo "en_it            Anglais   vers Italien"
  echo "en_pt            Anglais   vers Portugais"
  echo "en_ja            Anglais   vers Japonais"
  echo "en_ko            Anglais   vers Koréen"
  echo "en_zh            Anglais   vers Chinois"
  echo "de_en            Allemand  vers Anglais"
  echo "de_fr            Allemand  vers Français"
  echo "es_en            Espagnol  vers Anglais"
  echo "fr_en            Français  vers Anglais"
  echo "fr_de            Français  vers Allemand"
  echo "it_en            Italien   vers Anglais"
  echo "pt_en            Portugais vers Anglais"
  echo "ja_en            Japonais  vers Anglais"
  echo "ko_en            Koréen    vers Anglais"
  echo "zh_en            Chinois   vers Anglais"
  exit 0
fi

for enc in $ENCODSTYPES
do
 if [ "$enc" = "$1" ]
 then
  ENCODING="$enc"
  shift
 fi
done

[ "$DEBUG" = "1" ] && echo "DEBUG => ENCODING = $ENCODING"

for ttyp in $TRADUCSTYPES
do
  if [ "$ttyp" = "$1" ]
  then
    FROMTO="$ttyp"
    [ "$DEBUG" = "1" ] && echo "DEBUG => FROMTO = $FROMTO"
    shift
  fi
done

[ $# -gt 0 ] && TEXT="$*"

[ "$DEBUG" = "1" ] && echo "DEBUG => TEXT = $TEXT"

DIRECTION=`echo "$FROMTO" | sed -e 's/\_/ -> /'`
[ "$DEBUG" = "1" ] && echo "DEBUG => DIRECTION = $DIRECTION"

TEXTURL=`echo "$TEXT" | sed -e 's/ /\+/g'`
[ "$DEBUG" = "1" ] && echo "DEBUG => TEXTURL = \"${TEXTURL}\""

#google specific:
URL="http://translate.google.com/translate_t"
[ "$DEBUG" = "1" ] && echo "DEBUG => URL = \"${URL}\""
FROMTO=`echo "$FROMTO" | sed -e 's/\_/|/'`
[ "$DEBUG" = "1" ] && echo "DEBUG => FROMTO = \"${FROMTO}\""

#-d "oe=$ENCODING" \ no more "oe" found in Jan 2007 version
[ "$DEBUG" = "1" ] && echo "DEBUG => curl -s -A \"Mozilla/1.0 (compatible; translate/$VERSION)\" -d \"hl=$MY\" -d \"ie=$ENCODING\" -d \"langpair=$FROMTO\" -d \"text=$TEXTURL\" \"$URL\" > $HOME/.translate.tmp"

curl -s -A "Mozilla/1.0 (compatible; translate/$VERSION)" \
-d "hl=$MY" -d "ie=$ENCODING" \
-d "langpair=$FROMTO" \
-d "text=$TEXTURL" \
"$URL" | grep result_box | sed -e '1d' | cut -b 1700-3000 | sed -e "s/&#39;/'/g" > "$HOME"/.translate.tmp
LANG=C sed -e 's?^.*id=result_box dir="ltr">\([^<]*\)</div>.*?\1?' < "$HOME"/.translate.tmp | iconv --from latin1 --to $ENCODING
rm $HOME/.translate.tmp