Formats and prints recent StackOverflow careers postings to console

help-wanted.pl

#!/usr/bin/perl
use URI::Escape;

my $url = 'http://careers.stackoverflow.com/jobs/feed?searchTerm=';

# Either use search terms given as command-line arguments, or use default terms
my $q   = join('+', map { uri_escape($_); } (@ARGV > 0) ?
                @ARGV : ('-.net', '-css', '-c#', '-android', '-php', '-ios', '-vb.net', '-excel' ));

# Read XML from child process
open(INPUT, "-|") || exec "curl -# '$url$q' | xmllint --format -";

# Reopen STDOUT so output is sent to pager
unless (-t STDOUT) {
  my $pager = $ENV{PAGER} || '/usr/bin/less';
  open(STDOUT, "| $pager");
}

while (<INPUT>) {
  s/&lt;/</g;
  s/&gt;/>/g;
  s/&amp;/&/g;
  s/&amp;/&/g;
  s/&nbsp;/ /g;
  s/&copy;/(C)/g;
  s/&trade;/(TM)/g;
  s/&pound;/£/g;
  s/&euro;/€/g;
  s/&rdquo;/"/g;
  s/&ldquo;/"/g;
  s/&bdquo;/"/g;
  s/&rsquo;/'/g;
  s/&lsquo;/'/g;
  s/&ndash;/-/g;
  s/&mdash;/-/g;
  s/\xc2\x92/'/g;
  s/&hellip;/.../g;
  s/&szlig;/ß/g;
  s/&(.)uml;/$1/g;
  s/&(.)circ;/$1/g;
  s/&(.)ring;/$1/g;
  s/&(.)slash;/$1/g;
  s/&(.)tilde;/$1/g;
  s/&(.)acute;/$1/g;
  s/&(.)cedil;/$1/g;
  s/&(.)grave;/$1/g;

  s/<p>/\n/g;
  s/<\/p>/\n/g;
  s/<br>/\n/g;
  s/<br \/>/\n/g;
  s/<li>\s*/* /g;
  s/<(a|b|em|strong|span|ul|ol)( [^>]*)?>//g;
  s/<\/(a|b|em|strong|span|li|ul|ol)>//g;
  s/\n\n+/\n\n/g;
  s/^<\/description/      <\/description/mg;

  print $_;
}