#! /bin/bash

texte=$(cat $1 | sed "s/[^a-Z|0-9| |-|]/ /g; y/àâéèêëîïôöùûü/aaeeeeiioouuu/; y/ÇÀÂÉÈÊËÎÏÔÖÙÛÜ/çAAEEEEIIOOUUU/; s/$/ /g" | tr -d '\n' | tr -s ' ' | tr '[:upper:]' '[:lower:]')

nombremot=$(wc -w <<< "$texte")
#nombreligne=$(wc -l <<< "$texte")

if [ "$(tail -c 2 <<< "$texte")" = ' ' ]
    then
        texte="$(rev <<< "$texte" | cut -c2- | rev)"
fi

if [ "$(head -c 1 <<< "$texte")" = ' ' ]
    then
        texte=$(cut -c2- <<< "$texte")
        #echo "$texte"
fi

tableau_texte=( $texte )

texte=$(tr ' ' '\n' <<< $texte)

#set -x

OK=0
occurence_max=0
until [ $OK = $nombremot ]
    do
        occurence=0


        mot=$(sed -n $((OK + 1))p <<< "$texte")
        lignepresencemot=$(grep -nx "${tableau_texte[$OK]}" <<< "$texte" | cut -d : -f 1 | tr '\n' ', ')


#set -x
#echo $occurence_max
        for test1 in $(grep -nx "${tableau_texte[$OK]}" <<< "$texte" | cut -d : -f 1 | sort -r | tr '\n' ' ')
            do
                occurence=$((++occurence))
            done

            typeset tableau_${occurence}+="$mot ${lignepresencemot:0:-1}##"

    if [ $occurence -gt $occurence_max ]
        then occurence_max=$occurence
    fi

#set +x
#set -x

        OK=$((++OK))
done


OK=1
until [ $OK -gt $occurence_max ]
    do
        arraytruc="tableau_$OK[*]"
        echo "${!arraytruc}" | tr '##' '\n' | awk ' !x[$0]++'
        # literal variable http://stackoverflow.com/questions/7142965/bash-array-parameter-expansion-within-parameter-name-bad-substitution
        OK=$((++OK))
#echo $OK
done
#set +x
unset tableau_texte

exit 0