#/bin/bash


wget https://docs.google.com/spreadsheets/d/e/2PACX-1vTyDVQFiC-qN4Ryp8H4GlGkKYbMOdN0KL7ygCu2cbiYLMwqmPcLDvEp-wOeVCg0s0AOl20rUjY3p5XK/pubhtml -O indice.html

cat indice.html | grep -Po '>\Khttps://archive.org.*?(?=</a>)' > listaEnlaces.txt

cat listaEnlaces.txt | while read url; do
    echo "Descargando $url";

    wget $url -O indiceAnio.html

    cat indiceAnio.html | grep -Po '/download/.*?pdf(?=")' > pdfs.txt

    cat pdfs.txt | while read urlPDF; do
        echo "Descargados PDF: $urlPDF";

        wget https://archive.org$urlPDF

        sleep 1
    done

    rm indiceAnio.html
    rm pdfs.txt

    sleep 2

done

rm listaEnlaces.txt
rm indice.html