Scraping NJ Covid19 Municipal Data

scrape-nj.pl

#!/usr/bin/perl
# (c) Chris Ruvolo 2020.
# 2-clause BSD license:  https://opensource.org/licenses/BSD-2-Clause
use strict;
use utf8;
use feature 'unicode_strings';
binmode(STDOUT, ":utf8");

my $url = "https://www.nj.com/coronavirus/2020/04/where-is-the-coronavirus-in-nj-latest-map-update-on-county-by-county-cases-april-5-2020.html";

my $county = undef;

print "County,Municpality,Cases,Deaths\n";
open(DUMP, '-|', "elinks --dump '$url'");
binmode(DUMP, ":utf8");
while (<DUMP>) {
  if (/([A-Z][A-Z ]+) COUNTY/) {
    $county = lc $1;
    $county =~ s/^[a-z]/uc($&)/e;
    $county =~ s/ [a-z]/uc($&)/e;
  }
  my ($muni, $cases) = ($1, $2) if /• *([A-Z ]+): (\d+)/i;
  my $deaths = $2 if /(with|including) (\d+) (death|fatal)/ and defined $muni;
  print "$county,$muni,$cases,$deaths\n" if defined ($muni);
}
close(DUMP);