Download Thomson master journal list

thomson-download.pl

#!/usr/bin/perl
use warnings;
use strict;

use Mojo::UserAgent;
use Mojo::DOM;


my $ua = Mojo::UserAgent->new;
my @selection = qw( H Y B K D SS );
@selection = @ARGV if @ARGV;

foreach my $pc ( @selection ) {

my $page = 1;

while (1) {

	warn "# $pc $page\n";

	my $body = $ua->get('http://science.thomsonreuters.com/cgi-bin/jrnlst/jlresults.cgi?PC='.$pc.'&mode=print&Page='. $page )->res->body;

	last if $body !~ m/<DT>/;
	$page++;

	my $dom = Mojo::DOM->new( $body );


	foreach my $j ( split(/<DT>/, $body) ) {
		$j =~ s{<ol>.*</ol>}{}s;
		$j =~ s{<[^>]+>}{|}gs;
		$j =~ s{[\n\r]+}{}gs;
		$j =~ s{^\|(\d+)\.\s+}{$1|}gs;
		$j =~ s{\|\s+}{|}gs;
		$j =~ s{\|+}{|}gs;
		$j =~ s{\s+ISSN:\s+}{|};
		$j =~ s{\|$}{};
		print "$pc|$j\n";
	}

}

}