parse-comodo.rb

#!/usr/bin/ruby
require 'nokogiri'
require 'curb'
require 'openssl'

doc = File.open("publiclyDisclosedSubCACerts"){|f| Nokogiri::HTML(f)}

certurls = {}

# The page as almost no metadata, so we are doing this by index
subtbl = doc.css("body table")[1]
subtbl.css("tr").each do |row|
	links = row.css("td")[2]
	if links.nil?
		next
	end
	hrefs = links.css("a")

	# Audit links suggest external subordinate
	if hrefs.any?{|a|a.children[0].to_s == "Audit"}
		next	
	end

	certurl = hrefs.select{|a|a.children[0].to_s == "Certificate"}.first["href"]
	if !certurl.nil?
		certurls[certurl] = true
	end
end

certurls.keys.each do |u|
	http = Curl.get(u)
	if http.response_code >= 200 &&  http.response_code < 300
		der = http.body_str
		begin
			crt = OpenSSL::X509::Certificate.new(der)
			subj = crt.subject.to_s(OpenSSL::X509::Name::RFC2253 & ~4).force_encoding("UTF-8")
			spki = OpenSSL::Digest::SHA256.hexdigest(crt.public_key.to_der)
			puts "#{subj}\t#{spki}"
		end
	end
end