public-suffix-list 0.0.1 → 0.0.2

data/README.rdoc

@@ -21,7 +21,25 @@ easier to use.
 
 == Synopsis
 
-  PublicSuffixList.new.tld("foobar.com") # downloads the latest data file and returns "com"
+  require "public_suffix_list"
+
+  # downloads and parses the latest data file and returns "com"
+  PublicSuffixList.new.tld("foobar.com")
+
+  # downloads and parses the latest data file and returns "foobar.com"
+  PublicSuffixList.new.cdn("foobar.com")
+
+  # downloads and parses the latest data file, caches it in /tmp, returns ["abc", "xyz", "co.uk"]
+  PublicSuffixList.new(:cache_dir => "/tmp").split("abc.xyz.co.uk")
+
+  # loads the cached data in /tmp and returns ["test", "nhs", "uk"]
+  PublicSuffixList.new(:cache_dir => "/tmp").split("test.nhs.uk")
+
+  # you don't have to instantiate PublicSuffixList every time you use it, of course...
+  p = PublicSuffixList.new
+  p.split("fee.fi.fo.com") # => ["fee.fi", "fo", "com"]
+  p.cdn("fee.fi.fo.com")   # => "fo.com", "cdn" is "canonical domain name"
+  p.tld("fee.fi.fo.com")   # => "com", "tld" is "top-level domain"
 
 == Requirements
 

data/lib/public_suffix_list.rb

@@ -5,7 +5,7 @@ require "public_suffix_list/parser.rb"
 
 class PublicSuffixList
 
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 
   def self.config
     @@config ||= Config.new
@@ -17,34 +17,73 @@ class PublicSuffixList
   
   class Config
 
+    attr_accessor :cache_dir
+    attr_accessor :cache_expiry_period
     attr_accessor :effective_tld_names_url
 
     def initialize
+      @cache_dir = nil
+      @cache_expiry_period = 30 * 24 * 60 * 60
       @effective_tld_names_url = "http://mxr.mozilla.org/mozilla-central/source/netwerk/dns/src/effective_tld_names.dat?raw=1"
     end
 
   end
 
-  def initialize(url = self.class.config.effective_tld_names_url)
-    @rules = Parser.parse(open(url))
+  def initialize(options = {})
+    @config = self.class.config.dup
+    options.each { |k, v| @config.send("#{k}=", v) }
+    if @config.cache_dir && File.directory?(@config.cache_dir) && File.exist?(File.join(@config.cache_dir, name))
+      uncache or (download and cache)
+    elsif @config.cache_dir && File.directory?(@config.cache_dir)
+      download and cache
+    else
+      download
+    end
   end
 
   def split(domain)
     domain = domain.split(".")
-    result = best(match(domain.dup, @rules))
-    [domain.dup.reverse.drop(result.size + 1).reverse.join("."), domain.dup.reverse.drop(result.size).first, result.reverse.join('.')]
+    result = best(match(domain, @rules))
+    [gimme!(domain, result.size), gimme!(domain), domain].reverse.map { |d| d ? d.join(".") : "" }
   end
 
   def tld(domain)
-    best(match(domain.split("."), @rules)).reverse.join(".")
+    domain = domain.split(".")
+    result = best(match(domain, @rules))
+    gimme!(domain, result.size).join(".")
+  end
+
+  def cdn(domain)
+    domain = domain.split(".")
+    result = best(match(domain, @rules))
+    gimme!(domain, result.size + 1).join(".")
   end
 
   private
 
+  def name
+    URI.parse(@config.effective_tld_names_url).path.split("/").last + ".cache"
+  end
+
+  def cache
+    @cache = {:rules => @rules, :created_at => Time.now}
+    open(File.join(@config.cache_dir, name), "w") { |f| Marshal.dump(@cache, f) }
+  end
+
+  def uncache
+    open(File.join(@config.cache_dir, name), "r") { |f| @cache = Marshal.load(f) }
+    @rules = @cache[:rules] if Time.now < @cache[:created_at] + @config.cache_expiry_period
+  end
+
+  def download
+    @rules = Parser.parse(open(@config.effective_tld_names_url))
+  end
+
   def match(domain, rules)
     return [] if domain.empty? or rules.empty?
-    set = []
+    domain = domain.dup
     first = domain.pop
+    set = []
     [[first, first], ["!#{first}", "!#{first}"], ["*", first]].each do |a, b|
       if rules[a]
         set << [b]
@@ -61,4 +100,8 @@ class PublicSuffixList
     result
   end
 
+  def gimme!(domain, n = 1)
+    domain.slice!(-n, n)
+  end
+
 end

data/spec/public_suffix_list.spec

@@ -2,12 +2,13 @@ require 'lib/public_suffix_list'
 
 def cookie_calculation(public_suffix_list, domain)
   s, d, t = public_suffix_list.split(domain)
-  s.empty? and d ? "Cookies may be set for #{domain}." : "Cookies may not be set for #{domain}."
+  (s.empty? and !d.empty?) ? "Cookies may be set for #{domain}." : "Cookies may not be set for #{domain}."
 end
 
 describe PublicSuffixList do
+
   it "should act as described at http://publicsuffix.org/format/" do
-    public_suffix_list = PublicSuffixList.new("spec/test.dat")
+    public_suffix_list = PublicSuffixList.new(:effective_tld_names_url => "spec/test.dat")
     cookie_calculation(public_suffix_list, "foo.com").should == "Cookies may be set for foo.com."
     cookie_calculation(public_suffix_list, "foo.bar.jp").should == "Cookies may be set for foo.bar.jp."
     cookie_calculation(public_suffix_list, "bar.jp").should == "Cookies may not be set for bar.jp."
@@ -18,4 +19,11 @@ describe PublicSuffixList do
     cookie_calculation(public_suffix_list, "pref.hokkaido.jp").should == "Cookies may be set for pref.hokkaido.jp."
     cookie_calculation(public_suffix_list, "metro.tokyo.jp").should == "Cookies may be set for metro.tokyo.jp."
   end
+
+  it "should calculate tld and cdn correctly" do
+    public_suffix_list = PublicSuffixList.new(:effective_tld_names_url => "spec/test.dat")
+    public_suffix_list.cdn("foo.bar.com").should == "bar.com"
+    public_suffix_list.tld("foo.bar.com").should == "com"
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: public-suffix-list
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors: 
 - Todd Sundsted