obscenity2 0.0.1

data/lib/obscenity/active_model.rb

@@ -0,0 +1,18 @@
+if defined?(ActiveModel)
+  module ActiveModel
+    module Validations
+      class ObscenityValidator < ActiveModel::EachValidator
+        
+        def validate_each(record, attribute, value)
+          if options.present? && options.has_key?(:sanitize)
+            object = record.respond_to?(:[]) ? record[attribute] : record.send(attribute)
+            object = Obscenity.replacement(options[:replacement]).sanitize(object)
+          else
+            record.errors.add(attribute, options[:message] || 'cannot be profane') if Obscenity.profane?(value)
+          end
+        end
+        
+      end
+    end
+  end
+end

data/lib/obscenity/base.rb

@@ -0,0 +1,71 @@
+module Obscenity
+  class Base
+    class << self
+
+      def blacklist
+        @blacklist ||= set_list_content(Obscenity.config.blacklist)
+      end
+
+      def blacklist=(value)
+        @blacklist = value == :default ? set_list_content(Obscenity::Config.new.blacklist) : value
+      end
+
+      def whitelist
+        @whitelist ||= set_list_content(Obscenity.config.whitelist)
+      end
+
+      def whitelist=(value)
+        @whitelist = value == :default ? set_list_content(Obscenity::Config.new.whitelist) : value
+      end
+
+      def profane?(text)
+        blacklist.any? do |foul|
+          text =~ /\b#{foul}\b/i && !whitelist.include?(foul)
+        end
+      end
+
+      def sanitize(text)
+        blacklist.each do |foul|
+          text.gsub!(/\b#{foul}\b/i, replace(foul)) unless whitelist.include?(foul)
+        end
+        @scoped_replacement = nil
+        text
+      end
+
+      def replacement(chars)
+        @scoped_replacement = chars
+        self
+      end
+
+      def offensive(text)
+        words = []
+        blacklist.each do |foul|
+          words << foul if text =~ /\b#{foul}\b/i && !whitelist.include?(foul)
+        end
+        words.uniq
+      end
+
+      def replace(word)
+        content = @scoped_replacement || Obscenity.config.replacement
+        case content
+        when :vowels then word.gsub(/[aeiou]/i, '*')
+        when :stars  then '*' * word.size
+        when :nonconsonants then word.gsub(/[^bcdfghjklmnpqrstvwxyz]/i, '*')
+        when :default, :garbled then '$@!#%'
+        else content
+        end
+      end
+
+      private
+
+      def set_list_content(list)
+        case list
+        when Array then list
+        when String, Pathname then ::YAML.load_file( list.to_s )
+        else []
+        end
+      end
+
+    end
+  end
+end

data/lib/obscenity/config.rb

@@ -0,0 +1,52 @@
+module Obscenity
+  class Config
+
+    attr_accessor :replacement
+
+    DEFAULT_WHITELIST = []
+    DEFAULT_BLACKLIST = File.dirname(__FILE__) + "/../../config/blacklist.yml"
+
+    def initialize
+      yield(self) if block_given?
+      validate_config_options
+    end
+
+    def replacement
+      @replacement ||= :garbled
+    end
+
+    def blacklist
+      @blacklist ||= DEFAULT_BLACKLIST
+    end
+
+    def blacklist=(value)
+      @blacklist = value == :default ? DEFAULT_BLACKLIST : value
+    end
+
+    def whitelist
+      @whitelist ||= DEFAULT_WHITELIST
+    end
+
+    def whitelist=(value)
+      @whitelist = value == :default ? DEFAULT_WHITELIST : value
+    end
+
+    private
+
+    def validate_config_options
+      [@blacklist, @whitelist].each{ |content| validate_list_content(content) if content }
+    end
+
+    def validate_list_content(content)
+      case content
+      when Array    then !content.empty?       || raise(Obscenity::EmptyContentList.new('Content array is empty.'))
+      when String   then File.exists?(content) || raise(Obscenity::UnkownContentFile.new("Content file can't be found."))
+      when Pathname then content.exist?        || raise(Obscenity::UnkownContentFile.new("Content file can't be found."))
+      when Symbol   then content == :default   || raise(Obscenity::UnkownContent.new("The only accepted symbol is :default."))
+      else
+        raise Obscenity::UnkownContent.new("The content can be either an Array, Pathname, or String path to a .yml file.")
+      end
+    end
+
+  end
+end

data/lib/obscenity/error.rb

@@ -0,0 +1,7 @@
+module Obscenity
+  class Error < RuntimeError; end
+  
+  class UnkownContent     < Error; end
+  class UnkownContentFile < Error; end
+  class EmptyContentList  < Error; end
+end

data/lib/obscenity/rack.rb

@@ -0,0 +1,91 @@
+module Rack
+  class Obscenity
+
+    def initialize(app, options = {})
+      @app, @options = app, options
+    end
+
+    def call(env)
+      rejectable  = false
+      post_params = Rack::Utils.parse_query(env['rack.input'].read, "&")
+      get_params  = Rack::Utils.parse_query(env['QUERY_STRING'],    "&")
+
+      if @options.has_key?(:reject)
+        rejactable = validate_rejectability_of( select_params(:reject, get_params.update(post_params)) )
+
+      elsif @options.has_key?(:sanitize)
+        get_params  = sanitize_contents_of(get_params)
+        post_params = sanitize_contents_of(post_params)
+
+        env['QUERY_STRING'] = Rack::Utils.build_query(get_params)
+        env['rack.input']   = StringIO.new(Rack::Utils.build_query(post_params))
+      end
+
+      rejactable ? reject : continue(env)
+    end
+
+    private
+
+    def continue(env)
+      @app.call(env)
+    end
+
+    def reject
+      length, content = 0, ''
+      if @options[:reject].is_a?(Hash)
+        if (message = @options[:reject][:message]).present?
+          content = message
+          length  = message.size
+        elsif (path = @options[:reject][:path]).present?
+          if (path = ::File.expand_path(path)) && ::File.exists?(path)
+            content = ::File.read(path)
+            length  = content.size
+          end
+        end
+      end
+
+      [422, {'Content-Type' => 'text/html', 'Content-Length' => length.to_s}, [content]]
+    end
+
+    def validate_rejectability_of(params = {})
+      should_reject_request = false
+      params.each_pair do |param, value|
+        if value.is_a?(Hash)
+          validates_rejectability_of(value)
+        elsif value.is_a?(String)
+          next unless value.size >= 3
+          if ::Obscenity.profane?(value)
+            should_reject_request = true
+            break
+          end
+        else
+          next
+        end
+      end
+      should_reject_request
+    end
+
+    def sanitize_contents_of(params)
+      sanitized_params   = {}
+      replacement_method = @options[:sanitize].is_a?(Hash) && @options[:sanitize][:replacement]
+      select_params(:sanitize, params).each{|param, value|
+        if value.is_a?(String)
+          next unless value.size >= 3
+          sanitized_params[param] = ::Obscenity.replacement(replacement_method).sanitize(value)
+        else
+          next
+        end
+      }
+      params.update(sanitized_params)
+    end
+
+    def select_params(key, params = {})
+      if @options[key].is_a?(Hash) && @options[key][:params].is_a?(Array)
+        params.select{ |param, vvalue| @options[key][:params].include?(param.to_sym) }
+      else
+        params
+      end
+    end
+
+  end
+end

data/lib/obscenity/rspec_matcher.rb

@@ -0,0 +1,9 @@
+if defined?(RSpec::Matchers)
+  require 'rspec/expectations'
+
+  RSpec::Matchers.define :be_profane do |expected|
+    match do |actual|
+      Obscenity.profane?(actual) == expected
+    end
+  end
+end

data/lib/obscenity/version.rb

@@ -0,0 +1,5 @@
+module Obscenity
+
+  VERSION = '0.0.1'
+
+end

data/lib/obscenity.rb

@@ -0,0 +1,38 @@
+require 'obscenity/error'
+require 'obscenity/config'
+require 'obscenity/base'
+require 'obscenity/version'
+
+if defined?(::RSpec)
+  require 'obscenity/rspec_matcher'
+end
+
+module Obscenity extend self
+
+  attr_accessor :config
+
+  def configure(&block)
+    @config = Config.new(&block)
+  end
+
+  def config
+    @config ||= Config.new
+  end
+
+  def profane?(word)
+    Obscenity::Base.profane?(word)
+  end
+
+  def sanitize(text)
+    Obscenity::Base.sanitize(text)
+  end
+
+  def replacement(chars)
+    Obscenity::Base.replacement(chars)
+  end
+
+  def offensive(text)
+    Obscenity::Base.offensive(text)
+  end
+
+end

data/obscenity.gemspec

@@ -0,0 +1,20 @@
+# coding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'obscenity/version'
+
+Gem::Specification.new do |s|
+  s.name = "obscenity2"
+  s.version = Obscenity::VERSION
+
+  s.authors = ["Thiago Jackiw"]
+  s.description = " Obscenity is a profanity filter gem for Ruby/Rubinius, Rails (through ActiveModel), and Rack middleware "
+  s.email = "tjackiw@gmail.com"
+  s.files = `git ls-files`.split($/)
+  s.homepage = "http://github.com/tjackiw/obscenity"
+  s.licenses = ["MIT"]
+  s.require_paths = ["lib"]
+  s.summary = "Obscenity is a profanity filter gem for Ruby/Rubinius, Rails (through ActiveModel), and Rack middleware"
+  s.test_files = s.files.grep(%r{^(test|spec|features)/})
+end
+

data/test/helper.rb

@@ -0,0 +1,26 @@
+require 'rubygems'
+require 'yaml'
+require 'test/unit'
+require 'shoulda'
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+
+require 'active_model'
+require 'obscenity'
+require 'obscenity/active_model'
+
+module Dummy
+  class BaseModel
+    include ActiveModel::Validations
+
+    attr_accessor :title
+
+    def initialize(attr_names)
+      attr_names.each{ |k,v| send("#{k}=", v) }
+    end
+  end
+end
+
+class Test::Unit::TestCase
+end

data/test/static/422.html

@@ -0,0 +1 @@
+We don't accept profanity

data/test/test_active_model.rb

@@ -0,0 +1,69 @@
+require 'helper'
+
+class TestActiveModel < Test::Unit::TestCase
+  
+  def generate_new_class(name, options = {})
+    Dummy.send(:remove_const, name) if Dummy.const_defined?(name)
+    klass = Class.new(Dummy::BaseModel) do
+      validates :title, options
+    end
+    Dummy.const_set(name, klass)
+  end
+
+  should "be invalid when title is profane" do
+    klass = generate_new_class("Post", obscenity: true)
+    post  = klass.new(title: "He who poops, shits itself")
+    assert !post.valid?
+    assert post.errors.has_key?(:title)
+    assert_equal ['cannot be profane'], post.errors[:title]
+  end
+
+  should "be invalid when title is profane and should include a custom error message" do
+    klass = generate_new_class("Post", obscenity: { message: "can't be profane!" })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert !post.valid?
+    assert post.errors.has_key?(:title)
+    assert_equal ["can't be profane!"], post.errors[:title]
+  end
+
+  should "sanitize the title using the default replacement" do
+    klass = generate_new_class("Post", obscenity: { sanitize: true })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert post.valid?
+    assert !post.errors.has_key?(:title)
+    assert_equal "He who poops, $@!#% itself", post.title
+  end
+  
+  should "sanitize the title using the :garbled replacement" do
+    klass = generate_new_class("Post", obscenity: { sanitize: true, replacement: :garbled })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert post.valid?
+    assert !post.errors.has_key?(:title)
+    assert_equal "He who poops, $@!#% itself", post.title
+  end
+
+  should "sanitize the title using the :stars replacement" do
+    klass = generate_new_class("Post", obscenity: { sanitize: true, replacement: :stars })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert post.valid?
+    assert !post.errors.has_key?(:title)
+    assert_equal "He who poops, ***** itself", post.title
+  end
+
+  should "sanitize the title using the :vowels replacement" do
+    klass = generate_new_class("Post", obscenity: { sanitize: true, replacement: :vowels })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert post.valid?
+    assert !post.errors.has_key?(:title)
+    assert_equal "He who poops, sh*ts itself", post.title
+  end
+
+  should "sanitize the title using a custom replacement" do
+    klass = generate_new_class("Post", obscenity: { sanitize: true, replacement: '[censored]' })
+    post  = klass.new(title: "He who poops, shits itself")
+    assert post.valid?
+    assert !post.errors.has_key?(:title)
+    assert_equal "He who poops, [censored] itself", post.title
+  end
+
+end

data/test/test_base.rb

@@ -0,0 +1,258 @@
+require 'helper'
+
+class TestBase < Test::Unit::TestCase
+
+  context "#respond_to?" do
+    should "respond to methods and attributes" do
+      [:blacklist, :whitelist, :profane?, :sanitize, :replacement, :offensive, :replace].each do |field|
+        assert Obscenity::Base.respond_to?(field)
+      end
+    end
+  end
+
+  context "#blacklist" do
+    context "without custom config" do
+      setup { Obscenity::Base.blacklist = :default }
+      should "use the default content file when no config is found" do
+        assert Obscenity::Base.blacklist.is_a?(Array)
+        assert_equal 565, Obscenity::Base.blacklist.size
+      end
+    end
+    context "with custom config" do
+      setup { Obscenity::Base.blacklist = ['bad', 'word'] }
+      should "respect the config options" do
+        assert_equal ['bad', 'word'], Obscenity::Base.blacklist
+      end
+    end
+  end
+
+  context "#whitelist" do
+    context "without custom config" do
+      setup { Obscenity::Base.whitelist = :default }
+      should "use the default content file when no config is found" do
+        assert Obscenity::Base.whitelist.is_a?(Array)
+        assert Obscenity::Base.whitelist.empty?
+      end
+    end
+    context "with custom config" do
+      setup { Obscenity::Base.whitelist = ['safe', 'word'] }
+      should "respect the config options" do
+        assert_equal ['safe', 'word'], Obscenity::Base.whitelist
+      end
+    end
+  end
+
+  context "#profane?" do
+    context "without whitelist" do
+      context "without custom config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = :default
+        }
+        should "validate the profanity of a word based on the default list" do
+          assert Obscenity::Base.profane?('ass')
+          assert Obscenity::Base.profane?('biatch')
+          assert !Obscenity::Base.profane?('hello')
+        end
+      end
+      context "with custom blacklist config" do
+        setup { Obscenity::Base.blacklist = ['ass', 'word'] }
+        should "validate the profanity of a word based on the custom list" do
+          assert Obscenity::Base.profane?('ass')
+          assert Obscenity::Base.profane?('word')
+          assert !Obscenity::Base.profane?('biatch')
+        end
+      end
+    end
+    context "with whitelist" do
+      context "without custom blacklist config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "validate the profanity of a word based on the default list" do
+          assert Obscenity::Base.profane?('ass')
+          assert !Obscenity::Base.profane?('biatch')
+          assert !Obscenity::Base.profane?('hello')
+        end
+      end
+      context "with custom blacklist/whitelist config" do
+        setup {
+          Obscenity::Base.blacklist = ['ass', 'word']
+          Obscenity::Base.whitelist = ['word']
+        }
+        should "validate the profanity of a word based on the custom list" do
+          assert Obscenity::Base.profane?('ass')
+          assert !Obscenity::Base.profane?('word')
+          assert !Obscenity::Base.profane?('biatch')
+        end
+      end
+    end
+  end
+
+  context "#sanitize" do
+    context "without whitelist" do
+      context "without custom config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = :default
+        }
+        should "sanitize and return a clean text based on the default list" do
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.sanitize('Yo assclown, sup')
+          assert_equal "Hello world", Obscenity::Base.sanitize('Hello world')
+        end
+      end
+      context "with custom blacklist config" do
+        setup { Obscenity::Base.blacklist = ['ass', 'word'] }
+        should "sanitize and return a clean text based on a custom list" do
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.sanitize('Yo word, sup')
+          assert_equal "Hello world", Obscenity::Base.sanitize('Hello world')
+        end
+      end
+    end
+    context "with whitelist" do
+      context "without custom blacklist config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "sanitize and return a clean text based on the default blacklist and custom whitelist" do
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.sanitize('Yo assclown, sup')
+          assert_equal "Yo biatch, sup", Obscenity::Base.sanitize('Yo biatch, sup')
+        end
+      end
+      context "with custom blacklist/whitelist config" do
+        setup {
+          Obscenity::Base.blacklist = ['clown', 'biatch']
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "validate the profanity of a word based on the custom list" do
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.sanitize('Yo clown, sup')
+          assert_equal "Yo biatch, sup", Obscenity::Base.sanitize('Yo biatch, sup')
+        end
+      end
+    end
+  end
+
+  context "#replacement" do
+    context "without whitelist" do
+      context "without custom config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = :default
+        }
+        should "sanitize and return a clean text based on the default list" do
+          assert_equal "Yo ********, sup", Obscenity::Base.replacement(:stars).sanitize('Yo assclown, sup')
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.replacement(:garbled).sanitize('Yo assclown, sup')
+          assert_equal "Yo *sscl*wn, sup", Obscenity::Base.replacement(:vowels).sanitize('Yo assclown, sup')
+          assert_equal "Oh, *h*t!", Obscenity::Base.replacement(:nonconsonants).sanitize('Oh, 5hit!')
+          assert_equal "Yo [censored], sup", Obscenity::Base.replacement('[censored]').sanitize('Yo assclown, sup')
+          assert_equal "Hello World", Obscenity::Base.replacement(:default).sanitize('Hello World')
+        end
+      end
+      context "with custom blacklist config" do
+        setup { Obscenity::Base.blacklist = ['ass', 'word', 'w0rd'] }
+        should "sanitize and return a clean text based on a custom list" do
+          assert_equal "Yo ****, sup", Obscenity::Base.replacement(:stars).sanitize('Yo word, sup')
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.replacement(:garbled).sanitize('Yo word, sup')
+          assert_equal "Yo w*rd, sup", Obscenity::Base.replacement(:vowels).sanitize('Yo word, sup')
+          assert_equal "Yo w*rd, sup", Obscenity::Base.replacement(:nonconsonants).sanitize('Yo w0rd, sup')
+          assert_equal "Yo [censored], sup", Obscenity::Base.replacement('[censored]').sanitize('Yo word, sup')
+          assert_equal "Hello World", Obscenity::Base.replacement(:default).sanitize('Hello World')
+        end
+      end
+    end
+    context "with whitelist" do
+      context "without custom blacklist config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "sanitize and return a clean text based on the default blacklist and custom whitelist" do
+          assert_equal "Yo ********, sup", Obscenity::Base.replacement(:stars).sanitize('Yo assclown, sup')
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.replacement(:garbled).sanitize('Yo assclown, sup')
+          assert_equal "Yo *sscl*wn, sup", Obscenity::Base.replacement(:vowels).sanitize('Yo assclown, sup')
+          assert_equal "What an *r**", Obscenity::Base.replacement(:nonconsonants).sanitize('What an ar5e')
+          assert_equal "Yo [censored], sup", Obscenity::Base.replacement('[censored]').sanitize('Yo assclown, sup')
+          assert_equal "Yo biatch, sup", Obscenity::Base.replacement(:default).sanitize('Yo biatch, sup')
+        end
+      end
+      context "with custom blacklist/whitelist config" do
+        setup {
+          Obscenity::Base.blacklist = ['clown', 'biatch']
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "validate the profanity of a word based on the custom list" do
+          assert_equal "Yo *****, sup", Obscenity::Base.replacement(:stars).sanitize('Yo clown, sup')
+          assert_equal "Yo $@!#%, sup", Obscenity::Base.replacement(:garbled).sanitize('Yo clown, sup')
+          assert_equal "Yo cl*wn, sup", Obscenity::Base.replacement(:vowels).sanitize('Yo clown, sup')
+          assert_equal "Yo cl*wn, sup", Obscenity::Base.replacement(:nonconsonants).sanitize('Yo clown, sup')
+          assert_equal "Yo [censored], sup", Obscenity::Base.replacement('[censored]').sanitize('Yo clown, sup')
+          assert_equal "Yo biatch, sup", Obscenity::Base.replacement(:default).sanitize('Yo biatch, sup')
+          assert_equal "Yo assclown, sup", Obscenity::Base.replacement(:default).sanitize('Yo assclown, sup')
+        end
+      end
+    end
+  end
+
+  context "#offensive" do
+    context "without whitelist" do
+      context "without custom config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = :default
+        }
+        should "return an array with the offensive words based on the default list" do
+          assert_equal ['assclown'], Obscenity::Base.offensive('Yo assclown, sup')
+          assert_equal [], Obscenity::Base.offensive('Hello world')
+        end
+      end
+      context "with custom blacklist config" do
+        setup { Obscenity::Base.blacklist = ['yo', 'word'] }
+        should "return an array with the offensive words based on a custom list" do
+          assert_equal ['yo', 'word'], Obscenity::Base.offensive('Yo word, sup')
+          assert_equal [], Obscenity::Base.offensive('Hello world')
+        end
+      end
+    end
+    context "with whitelist" do
+      context "without custom blacklist config" do
+        setup {
+          Obscenity::Base.blacklist = :default
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "return an array with the offensive words based on the default blacklist and custom whitelist" do
+          assert_equal ['assclown'], Obscenity::Base.offensive('Yo assclown, sup')
+          assert_equal [], Obscenity::Base.offensive('Yo biatch, sup')
+        end
+      end
+      context "with custom blacklist/whitelist config" do
+        setup {
+          Obscenity::Base.blacklist = ['clown', 'biatch']
+          Obscenity::Base.whitelist = ['biatch']
+        }
+        should "return an array with the offensive words based on the custom list" do
+          assert_equal ['clown'], Obscenity::Base.offensive('Yo clown, sup')
+          assert_equal [], Obscenity::Base.offensive('Yo biatch, sup')
+        end
+      end
+    end
+  end
+
+  context "#replace" do
+    should "replace the given word by the given replacement method" do
+      [
+        [:vowels,        {original: "Oh 5hit", clean: "Oh 5h*t"}],
+        [:nonconsonants, {original: "Oh 5hit", clean: "Oh *h*t"}],
+        [:stars,         {original: "Oh 5hit", clean: "Oh ****"}],
+        [:garbled,       {original: "Oh 5hit", clean: "Oh $@!#%"}],
+        [:default,       {original: "Oh 5hit", clean: "Oh $@!#%"}],
+        ["[censored]",   {original: "Oh 5hit", clean: "Oh [censored]"}],
+        [nil,            {original: "Oh 5hit", clean: "Oh $@!#%"}]
+      ].each do |replacement_method, content|
+        assert_equal content[:clean], Obscenity::Base.replacement(replacement_method).sanitize(content[:original]), "(replacement should match for #{replacement_method})"
+      end
+    end
+  end
+
+end