SDP PDFs by category

extract.sh

#!/bin/bash

ls -1 pdfs_by_category | sed 's/^\(.*\)\.txt/mkdir -p "extracts\/\1"/' | bash

function download_category {
    cd "extracts/$1"
    cat "../../pdfs_by_category/$1.txt" | xargs -L 1 wget
    cd -
}

find . -name '*.pdf' -exec textutil -convert txt {} \;

cat "pdfs_by_category/Other, please specify.txt" | xargs -L 1 curl -sI | grep 'x-ms-meta-typeofbusinessother' | sed 's/^.*: //' | atob

while read url; do
    type=$(curl -sI "$url" | grep 'x-ms-meta-typeofbusinessother' | sed 's/^.*: //' | atob)
    echo -e "$url\t$type" >> othertypes.tsv
done < "pdfs_by_category/Other, please specify.txt"