padronpdfhtml.sh

#!/bin/bash

set -o nounset

case "${1:-}" in
--)
  pdf="$2"
  html="${pdf%.pdf}".html
  dest="${html}".gz
  [ -e "$dest" ] && exit
  pdftohtml -nodrm -s -i -noframes -stdout "$pdf"
  if grep <"$html" -E -v 'PADRÓN +AUDITADO +PLEBISCITO +NACIONAL +2020' | gzip -9 >"$dest"; then
    rm -vf "$html"
  fi
  ;;
*)
  tsp pdftohtml -S 8
  for x in *.pdf; do
    tsp pdftohtml "${BASH_SOURCE[0]}" -- "$x"
  done
  ;;
esac