midaWeb/website-size.sh

#!/bin/bash

resum=0
while getopts 'r:' OPTION; do
    case "$OPTION" in 
        r)
            resum=1
            ;;
        ?)
            echo "utilització: ./website-size.sh [-r (crear un fitxer resum)] example.org" >&2
            exit 1
            ;;
    esac
done
# shift "$(($OPTIND -1))"

output="./${!#}.log"
sites_output="./${!#}_sites.log"
mides_output="./${!#}_mides.log"
resum_output="./${!#}_resum.log"

year=$(date +"%Y")

echo "Analitzant ${!#}..."
sleep 1s
echo "Pot tardar una estona..."
echo ""

wget \
  --recursive --level=1 \
  --spider --server-response \
  --no-directories --no-parent \
  -e robots=off \
  --output-file="$output" "$@"

sleep 1s
if [ -f "$output" ]
then
    # Calculate and print estimated website size
    echo "Mida estimada de ${!#}: $(\
        grep -e "Content-Length" "$output" | \
        awk '{sum+=$2} END {printf("%.0f", sum / 1024)}'\
    ) kb"
    echo ""
    sleep 1s
    grep -e "--$year" $output | awk '{printf("%s\n", $3)}' | sort | uniq >> $sites_output
    sed -i '$ d' $sites_output
    sed -i '1d' $sites_output
    if [ $resum = 1 ]
    then
        echo "## Resum dels fitxers baixats en el fitxer $resum_output ##"
        sleep 1s
        while read line; do
            wget --server-response --no-directories --spider -e robots=off --output-file="$mides_output" "$line"
            echo "$line $(grep -e "Content-Length" $mides_output | awk '{printf("%.2fKb\n", $2 / 1024)}')" 
            rm $mides_output
        done < $sites_output >> $resum_output
    fi
    #grep -e "Content-Length" $output | awk '{printf("%.2fKb\n", $2 / 1024)}' >> $mides_output
    # echo "## Resum dels fitxers baixats ##"
    # sleep 1s
    #echo "$(paste $sites_output $mides_output | uniq)" 
    # Delete wget output file
    rm "$output"
    rm "$sites_output"
else
    echo "Unable to calculate estimated size."
    exit 1
fi

exit 0