use WWW::Mechanize::TreeBuilder;
use vars qw($DEBUG);
use Data::Printer;
+use Text::Wrap;
tie my $uuid, 'OSSP::uuid::tie';
$uuid= ["v1"];
my %sites =
(herc => {url => 'http://main.hercjobs.org/jobs/search?keywords=professor+AND+%28genomics+OR+bioinformatics+OR+biology+OR+informatics%29&discipline=academic-faculty&category=academic-faculty&category=allied-health&category=biological-biomedical-sciences&category=computer-information-sciences&category=education&category=interdisciplinary&category=mathematics-statistics&category=medical-research&category=physical-sciences&sort=DATE_POSTED+DESC',
- next_selector => [class => "bti-pagination-previous-link bti-pagination-prev-next"],
+ next_selector => [class => "bti-pagination-previous-link bti-pagination-prev-next",text => '>'],
job_selector => [url_regex => qr{^\/jobs\/\d+/.+}],
university => [itemprop=>"hiringOrganization",itemtype=>"http://schema.org/Organization"],
description => [class=>"bti-jd-description",itemprop=>"description"],
print format_job($university,$position,$j_u->URI->abs(),$description,$date);
$m->back();
}
- $m->follow_link(@{$s->{next_selector}});
+ $m->follow_link(@{$s->{next_selector}}) or die "Unable to find next link";
}
}
sub format_job {
my ($university,$position,$url,$text,$date) = @_;
- $text =~ s/(\n)(\s*)/$1 /m;
+ $text = wrap(' ',' ',$text);
my $ret = <<"EOF";
** TODO $university -- $position
:PROPERTIES: