#!/usr/bin/perl -w

# print out the number of articles per publication year

# Example:
# 1995 	23
# 1996  34
# etc.

use strict;

my $fn = "urls_by_article.txt";

# articles are not indented.  All urls indented under article are the article's urls

my %years;

open(URLS, $fn) || die("Can't open $fn: $!\n");
my $line = <URLS>;
while ($line) {
	chomp($line);
	#print "line=$line\n";

	if ($line !~ /^\t/) {
		my $article = $line;
		#print "$article\n";

		my ($year) = $article =~ /dlib\/\w+(\d\d)\//;
		if ($year > 4) {
			$year = "19$year";
		}
		else {
			$year = "20$year";
		}

		$years{$year}++;

		$line = <URLS>;
		while ($line && $line =~ /^\t(.+)$/) {
			$line = <URLS>;
		}
	}
}

close URLS;

# Print out number of urls per year

print "year\tnum of articles\n";

foreach my $year (sort keys %years) {
	print "$year\t" . $years{$year} . "\n";
}