abide_dev_utils 0.9.5 → 0.10.1

data/lib/abide_dev_utils/xccdf/diff/benchmark/number_title.rb

@@ -0,0 +1,270 @@
+# frozen_string_literal: true
+
+require 'abide_dev_utils/xccdf/diff/benchmark/property_existence'
+
+module AbideDevUtils
+  module XCCDF
+    module Diff
+      # Diffs two XCCDF benchmarks using the title / number of the items as the primary
+      # diff properties.
+      class NumberTitleDiff
+        SKIP_DIFF_TYPES = %i[equal both].freeze
+
+        def initialize(numbered_children, other_numbered_children)
+          new_number_title_objs(numbered_children, other_numbered_children)
+        end
+
+        def diff
+          @diff ||= find_diffs(@number_title_objs, @other_number_title_objs)
+        end
+
+        def to_s
+          parts = []
+          @diff.each do |_, diffs|
+            diffs.each do |dh|
+              parts << dh[:diff_text]
+            end
+          end
+          parts.join("\n")
+        end
+
+        private
+
+        attr_writer :diff
+
+        def added_number_title_objs
+          added_titles = @self_prop_checker.added_titles
+          @other_number_title_objs.select do |nto|
+            added_titles.include?(nto.title)
+          end
+        end
+
+        def removed_number_title_objs
+          removed_titles = @self_prop_checker.removed_titles
+          @number_title_objs.select do |nto|
+            removed_titles.include?(nto.title)
+          end
+        end
+
+        def find_diffs(objs, other_objs)
+          diffs = []
+          added_number_title_objs.each do |nto|
+            change_type = %i[both added]
+            stand_in = NumberTitleContainerStandIn.new(change_type)
+            diffs << process_diffs([diff_hash(change_type, stand_in, nto)])
+          end
+          removed_number_title_objs.each do |nto|
+            change_type = %i[both removed]
+            stand_in = NumberTitleContainerStandIn.new(change_type)
+            diffs << process_diffs([diff_hash(change_type, nto, stand_in)])
+          end
+          objs.each do |obj|
+            obj_diffs = other_objs.each_with_object([]) do |other_obj, o_ary|
+              obj_diff = obj.diff(other_obj)
+              next if SKIP_DIFF_TYPES.include?(obj_diff[0])
+
+              o_ary << diff_hash(obj_diff, obj, other_obj)
+            end
+
+            processed_obj_diffs = process_diffs(obj_diffs)
+            diffs << processed_obj_diffs unless obj_diffs.empty?
+          end
+          diffs
+        end
+
+        def process_diffs(diffs)
+          return {} if diffs.empty?
+
+          raise "Unexpected diffs: #{diffs}" if diffs.length > 2
+
+          return diffs[0] if diffs.length == 1
+
+          if diffs[0][:type][0] == PropChecker.inverse_existence_state[diffs[1][:type][0]]
+            diffs[0]
+          else
+            diffs[1]
+          end
+        end
+
+        def diff_hash(diff_type, obj, other_obj)
+          {
+            self: obj.child,
+            other: other_obj.child,
+            type: diff_type,
+            text: diff_type_text(diff_type, obj, other_obj),
+            number: obj.number,
+            other_number: other_obj.number,
+            title: obj.title,
+            other_title: other_obj.title,
+          }
+        end
+
+        def diff_type_text(diff_type, obj, other_obj)
+          DiffTypeText.text(diff_type, obj, other_obj)
+        end
+
+        def new_number_title_objs(children, other_children)
+          number_title_objs = children.map { |c| NumberTitleContainer.new(c) }.sort
+          other_number_title_objs = other_children.map { |c| NumberTitleContainer.new(c) }.sort
+          @self_prop_checker = PropChecker.new(number_title_objs, other_number_title_objs)
+          @other_prop_checker = PropChecker.new(other_number_title_objs, number_title_objs)
+          number_title_objs.map { |n| n.prop_checker = @self_prop_checker }
+          other_number_title_objs.map { |n| n.prop_checker = @other_prop_checker }
+          @number_title_objs = number_title_objs
+          @other_number_title_objs = other_number_title_objs
+        end
+      end
+
+      # Creates string representations of diff types
+      class DiffTypeText
+        def self.text(diff_type, obj, other_obj)
+          case diff_type[0]
+          when :equal
+            'The objects are equal'
+          when :title
+            "Title changed: Number \"#{obj.number}\": #{obj.title} -> #{other_obj.title}"
+          when :number
+            number_diff_type_text(diff_type, obj, other_obj)
+          when :both
+            both_diff_type_text(diff_type, obj, other_obj)
+          when :add
+            "Add object with number \"#{other_obj.number}\" and title \"#{other_obj.title}\""
+          when :remove
+            "Remove object with number \"#{obj.number}\" and title \"#{obj.title}\""
+          else
+            raise ArgumentError, "Unknown diff type: #{diff_type}"
+          end
+        end
+
+        def self.number_diff_type_text(diff_type, obj, other_obj)
+          case diff_type[1]
+          when :added
+            "Number changed (New Number): Title \"#{obj.title}\": #{obj.number} -> #{other_obj.number}"
+          when :exists
+            "Number changed (Existing Number): Title \"#{obj.title}\": #{obj.number} -> #{other_obj.number}"
+          else
+            raise ArgumentError, "Unknown diff type for number change: #{diff_type[1]}"
+          end
+        end
+
+        def self.both_diff_type_text(diff_type, obj, other_obj)
+          case diff_type[1]
+          when :added
+            "Added object: Title \"#{other_obj.title}\": Number \"#{other_obj.number}\""
+          when :removed
+            "Removed object: Title \"#{obj.title}\": Number \"#{obj.number}\""
+          else
+            raise ArgumentError, "Unknown diff type for both change: #{diff_type[1]}"
+          end
+        end
+      end
+
+      # Checks properties for existence in both benchmarks.
+      class PropChecker < AbideDevUtils::XCCDF::Diff::PropertyExistenceChecker
+        attr_reader :all_numbers, :all_titles, :other_all_numbers, :other_all_titles
+
+        def initialize(number_title_objs, other_number_title_objs)
+          super
+          @all_numbers = number_title_objs.map(&:number)
+          @all_titles = number_title_objs.map(&:title)
+          @other_all_numbers = other_number_title_objs.map(&:number)
+          @other_all_titles = other_number_title_objs.map(&:title)
+        end
+
+        def title(title)
+          property_existence(title, @all_titles, @other_all_titles)
+        end
+
+        def number(number)
+          property_existence(number, @all_numbers, @other_all_numbers)
+        end
+
+        def added_numbers
+          added(@all_numbers, @other_all_numbers)
+        end
+
+        def removed_numbers
+          removed(@all_numbers, @other_all_numbers)
+        end
+
+        def added_titles
+          added(@all_titles, @other_all_titles)
+        end
+
+        def removed_titles
+          removed(@all_titles, @other_all_titles)
+        end
+      end
+
+      class NumberTitleDiffError < StandardError; end
+      class InconsistentDiffTypeError < StandardError; end
+
+      # Holds a number and title for a child of a benchmark
+      # and provides methods to compare it to another child.
+      class NumberTitleContainer
+        include ::Comparable
+        attr_accessor :prop_checker
+        attr_reader :child, :number, :title
+
+        def initialize(child, prop_checker = nil)
+          @child = child
+          @number = child.number.to_s
+          @title = child.title.to_s
+          @prop_checker = prop_checker
+        end
+
+        def diff(other)
+          return %i[equal exist] if number == other.number && title == other.title
+
+          if number == other.number && title != other.title
+            c_diff = correlate_prop_diff_types(@prop_checker.title(other.title),
+                                               other.prop_checker.title(other.title))
+            [:title, c_diff]
+          elsif title == other.title && number != other.number
+            c_diff = correlate_prop_diff_types(@prop_checker.number(other.number),
+                                               other.prop_checker.number(other.number))
+            [:number, c_diff]
+          else
+            %i[both exist]
+          end
+        rescue StandardError => e
+          err_str = [
+            'Error diffing number and title',
+            "Number: #{number}",
+            "Title: #{title}",
+            "Other number: #{other.number}",
+            "Other title: #{other.title}",
+            e.message,
+          ]
+          raise NumberTitleDiffError, err_str.join(', ')
+        end
+
+        def <=>(other)
+          child <=> other.child
+        end
+
+        private
+
+        def correlate_prop_diff_types(self_type, other_type)
+          inverse_diff_type = PropChecker.inverse_existence_state[self_type]
+          return other_type if inverse_diff_type.nil?
+
+          self_type
+        end
+      end
+
+      # Stand-in object for a NumberTitleContainer when the NumberTitleContainer
+      # would not exist. This is used when a change is an add or remove.
+      class NumberTitleContainerStandIn
+        attr_reader :child, :number, :title
+
+        def initialize(change_type)
+          @change_type = change_type
+          @child = nil
+          @number = ''
+          @title = ''
+        end
+      end
+    end
+  end
+end

data/lib/abide_dev_utils/xccdf/diff/benchmark/profile.rb

@@ -0,0 +1,104 @@
+# frozen_string_literal: true
+
+require 'abide_dev_utils/xccdf/diff/benchmark/property_existence'
+
+module AbideDevUtils
+  module XCCDF
+    module Diff
+      # Diffs two sets of XCCDF profiles.
+      class ProfileDiff
+        def initialize(profiles, other_profiles)
+          new_profile_rule_objs(profiles, other_profiles)
+        end
+
+        def diff_hash(diff_type, profile1, prof1_rules, profile2, prof2_rules)
+          {
+
+          }
+        end
+
+        private
+
+        def new_profile_rule_objs(profiles, other_profiles)
+          profile_objs = containers_from_profile_list(profiles)
+          other_profile_objs = containers_from_profile_list(other_profiles)
+          @self_prop_checker = PropertyExistenceChecker.new(profile_objs, other_profile_objs)
+          @other_prop_checker = PropertyExistenceChecker.new(other_profile_objs, profile_objs)
+          profile_objs.map { |p| p.prop_checker = @self_prop_checker }
+          other_profile_objs.map { |p| p.prop_checker = @other_prop_checker }
+          @profile_rule_objs = profile_objs
+          @other_profile_rule_objs = other_profile_objs
+        end
+
+        def containers_from_profile_list(profile_list)
+          profile_list.each_with_object([]) do |profile, ary|
+            ary << ProfileRuleContainer.new(profile)
+          end
+        end
+      end
+
+      # Checks property existence in both profiles.
+      class PropChecker < AbideDevUtils::XCCDF::Diff::Benchmark::PropertyExistence
+        def initialize(profile_rule_objs, other_profile_rule_objs)
+          super
+          @profile_rule_objs = profile_rule_objs
+          @other_profile_rule_objs = other_profile_rule_objs
+          @profiles = profile_rule_objs.map(&:profile)
+          @other_profiles = other_profile_rule_objs.map(&:profile)
+        end
+
+        def profile(profile)
+          profile_key = profile.respond_to?(:id) ? profile.id : profile
+          property_existence(profile_key, @profiles, @other_profiles)
+        end
+
+        def rule_in_profile(rule, profile, rule_key: :title)
+          rk = rule.respond_to?(rule_key) ? rule.send(rule_key) : rule
+          rules = @profiles.find { |p| p.id == profile }.linked_rule.map(&rk)
+          other_rules = @other_profiles.find { |p| p.id == profile }.linked_rule.map(&rk)
+          property_existence(rk, rules, other_rules)
+        end
+
+        def added_profiles
+          added(@other_profiles.map(&:id), @profiles.map(&:id))
+        end
+
+        def removed_profiles
+          removed(@profiles.map(&:id), @other_profiles.map(&:id))
+        end
+
+        def added_rules_by_profile
+          @rules_by_profile.each_with_object({}) do |(profile, rules), hsh|
+            next unless @other_rules_by_profile.key?(profile)
+
+            hsh[profile] = added(rules, @other_rules_by_profile[profile])
+          end
+        end
+
+        def removed_rules_by_profile
+          @rules_by_profile.each_with_object({}) do |(profile, rules), hsh|
+            next unless @other_rules_by_profile.key?(profile)
+
+            hsh[profile] = removed(rules, @other_rules_by_profile[profile])
+          end
+        end
+      end
+
+      class ProfileRuleContainer
+        include ::Comparable
+        attr_accessor :prop_checker
+        attr_reader :profile, :rules
+
+        def initialize(profile, prop_checker = nil)
+          @profile = profile
+          @rules = profile.linked_rule
+          @prop_checker = prop_checker
+        end
+
+        def <=>(other)
+          @profile.id <=> other.profile.id
+        end
+      end
+    end
+  end
+end

data/lib/abide_dev_utils/xccdf/diff/benchmark/property.rb

@@ -0,0 +1,127 @@
+# frozen_string_literal: true
+
+require 'amatch'
+
+module AbideDevUtils
+  module XCCDF
+    module Diff
+      # Diffs benchmark properties.
+      module BenchmarkPropertyDiff
+        DEFAULT_PROPERTY_DIFF_OPTS = {
+          rule_properties_for_similarity: %i[title description rationale fixtext],
+          rule_properties_for_confidence: %i[description rationale fixtext],
+          rule_confidence_property_threshold: 0.7,
+          rule_confidence_total_threshold: 0.5,
+          digest_similarity_threshold: 0.75,
+          digest_similarity_label_weights: {
+            'title' => 4.0,
+          },
+          digest_similarity_only_labels: %w[title description fixtext rationale],
+          digest_top_x_similarities: 10,
+        }.freeze
+
+        def safe_rule_prop(rule, prop)
+          rule.respond_to?(prop) ? rule.send(prop).to_s : :none
+        end
+
+        def self_rule_vals
+          @self_rule_vals ||= {}
+        end
+
+        def other_rule_vals
+          @other_rule_vals ||= {}
+        end
+
+        def add_rule_val(rule, prop, val, container: nil)
+          raise ArgumentError, 'container must not be nil' if container.nil?
+
+          return unless container.dig(rule, prop).nil?
+
+          container[rule] ||= {}
+          container[rule][prop] = val
+        end
+
+        def add_self_rule_val(rule, prop, val)
+          add_rule_val(rule, prop, val, container: self_rule_vals)
+        end
+
+        def add_other_rule_val(rule, prop, val)
+          add_rule_val(rule, prop, val, container: other_rule_vals)
+        end
+
+        def same_rule?(prop_similarities)
+          confidence_indicator = 0.0
+          opts[:rule_properties_for_confidence].each do |prop|
+            confidence_indicator += 1.0 if prop_similarities[prop] >= opts[:rule_confidence_property_threshold]
+          end
+          (confidence_indicator / opts[:rule_properties_for_confidence].length) >= opts[:rule_confidence_total_threshold]
+        end
+
+        def maxed_digest_similarities(child, other_children)
+          similarities = other_children.each_with_object([]) do |other_child, ary|
+            if other_child.digest_equal? child
+              ary << [1.0, other_child]
+              next
+            end
+
+            d_sim = child.digest_similarity(other_child,
+                                            only_labels: opts[:digest_similarity_only_labels],
+                                            label_weights: opts[:digest_similarity_label_weights])
+            ary << [d_sim, other_child]
+          end
+          max_digest_similarities(similarities)
+        end
+
+        def max_digest_similarities(digest_similarities)
+          digest_similarities.reject! { |s| s[0] < opts[:digest_similarity_threshold] }
+          return digest_similarities if digest_similarities.empty?
+
+          digest_similarities.max_by(opts[:digest_top_x_similarities]) { |s| s[0] }
+        end
+
+        def rule_property_similarity(rule1, rule2)
+          prop_similarities = {}
+          prop_diff = {}
+          opts[:rule_properties_for_similarity].each do |prop|
+            add_self_rule_val(rule1, prop, safe_rule_prop(rule1, prop).to_s)
+            add_other_rule_val(rule2, prop, safe_rule_prop(rule2, prop).to_s)
+            prop_similarities[prop] = self_rule_vals[rule1][prop].levenshtein_similar(other_rule_vals[rule2][prop])
+            if prop_similarities[prop] < 1.0
+              prop_diff[prop] = { self: self_rule_vals[rule1][prop], other: other_rule_vals[rule2][prop] }
+            end
+          end
+          total = prop_similarities.values.sum / opts[:rule_properties_for_similarity].length
+          {
+            total: total,
+            prop_similarities: prop_similarities,
+            prop_diff: prop_diff,
+            confident_same: same_rule?(prop_similarities),
+          }
+        end
+
+        def most_similar(child, maxed_digest_similarities)
+          most_similar_map = maxed_digest_similarities.each_with_object({}) do |similarity, h|
+            prop_similarities = rule_property_similarity(child, similarity[1])
+            if child.title.to_s == similarity[1].title.to_s
+              prop_similarities[:total] = 99.0 # magic number denoting a title match
+            end
+            h[prop_similarities[:total]] = { self: child, other: similarity[1] }.merge(prop_similarities)
+          end
+          most_similar_map[most_similar_map.keys.max]
+        end
+
+        def find_most_similar(children, other_children)
+          children.each_with_object({}) do |benchmark_child, h|
+            maxed_similarities = maxed_digest_similarities(benchmark_child, other_children)
+            next if maxed_similarities.empty?
+
+            best = most_similar(benchmark_child, maxed_similarities)
+            next if best.nil? || best.empty?
+
+            h[benchmark_child] = best
+          end
+        end
+      end
+    end
+  end
+end

data/lib/abide_dev_utils/xccdf/diff/benchmark/property_existence.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module AbideDevUtils
+  module XCCDF
+    module Diff
+      # PropertyExistenceChecker provides methods to check existence state of various properties
+      class PropertyExistenceChecker
+        def initialize(*_args); end
+
+        # Compares two arrays (or other iterables implementing `#to_a`)
+        # containing properies and returns an array of the properties
+        # that are added by other_props but not in self_props.
+        def added(self_props, other_props)
+          other_props.to_a - self_props.to_a
+        end
+
+        # Compares two arrays (or other iterables implementing `#to_a`)
+        # containing properies and returns an array of the properties
+        # that are removed by other_props but exist in self_props.
+        def removed(this, other)
+          this.to_a - other.to_a
+        end
+
+        # Returns a hash of existence states and their inverse.
+        def self.inverse_existence_state
+          {
+            removed: :added,
+            added: :removed,
+            exists: :exists,
+          }
+        end
+
+        private
+
+        def property_existence(property, self_props, other_props)
+          if self_props.include?(property) && !other_props.include?(property)
+            :removed
+          elsif !self_props.include?(property) && other_props.include?(property)
+            :added
+          else
+            :exists
+          end
+        end
+      end
+    end
+  end
+end