Ngram in bash with awk

ngram.sh

cat << 'EOF' > ngram.sh

#Helper function to count ngrams
function soniq(){
    sort | uniq -c | sort -nr 

}

#Usage: cat file.txt | ngram 3 | soniq | head -20 
function ngram(){
    local NGRAM_COUNT=${1:-1}
    sed -r 's/[^[:alnum:]]/ /g' | awk '{print tolower($0)}' \
    | sed -e 's/ /\n/g' | sed '/^$/d' \
    | awk -v nn=$NGRAM_COUNT '{line="";for(i=nn;i>0;i--){line = line" "arr[i]}print line; for(i=nn;i>0;i--){arr[i]=arr[i-1]};arr[0]=$0;}'

}
EOF

source ngram.sh