capture

#!/usr/bin/ruby

require "rubygems"
require "rbosa"

safari = OSA.app "Safari"

while(url = STDIN.gets) do
  url.chomp!
  STDERR.puts "Handling #{url}..."
  safari.make OSA::Safari::Document, :with_properties => { :url => url }
  sleep 5
  safari.print safari.documents[0], :print_dialog => false
  sleep 5
  safari.close safari.documents[0]
end

crawl

#!/usr/bin/ruby

require "rubygems"
require "open-uri"
require "nokogiri"

root_url = "http://mephisto.theomearas.net"
root_doc = Nokogiri::HTML open(root_url)

monthly_archive_paths = root_doc.css("a").inject([]) do |paths, a|
  paths << a["href"] if(a["href"] =~ %r(/archives))
  paths
end

STDERR.puts "Searching #{monthly_archive_paths.size} monthly archives..."

article_paths = []
monthly_archive_paths.each do |archive_path|
  monthly_archive_doc = Nokogiri::HTML open(root_url + archive_path)
  monthly_archive_doc.css("a").each do |a|
    article_paths << a["href"] if(a["href"] =~ %r(^/\d{4}/\d{1,2}/\d{1,2}/) && a["href"] !~ %r(#comments))
  end
end

STDERR.puts "Found #{article_paths.size} articles"

article_paths.each { |path| STDOUT.puts(root_url + path) }