ossert 0.1.0

data/config/stats.yml

@@ -0,0 +1,198 @@
+stats:
+  agility:
+    total:
+      metrics:
+        - issues_closed_percent
+        - issues_non_owner_percent
+        - issues_with_contrib_comments_percent
+        - issues_all_count
+        - issues_processed_in_avg
+        - issues_processed_in_median
+        - pr_closed_percent
+        - pr_non_owner_percent
+        - pr_with_contrib_comments_percent
+        - pr_all_count
+        - pr_processed_in_avg
+        - pr_processed_in_median
+        - first_pr_date_int
+        - last_pr_date_int
+        - first_issue_date_int
+        - last_issue_date_int
+        - last_release_date_int
+        - releases_count
+        - commits_count_since_last_release_count
+        - stale_branches_count
+        - life_period
+        - last_changed
+      uniq_attributes:
+        - dependencies
+        - issues_open
+        - issues_closed
+        - issues_owner
+        - issues_non_owner
+        - issues_with_contrib_comments
+        - issues_total
+        - pr_open
+        - pr_merged
+        - pr_closed
+        - pr_owner
+        - pr_non_owner
+        - pr_with_contrib_comments
+        - pr_total
+        - releases_total_gh
+        - releases_total_rg
+        - stale_branches
+        - branches
+      attributes:
+        dependencies: Array
+        issues_open: Array
+        issues_closed: Array
+        issues_owner: Array
+        issues_non_owner: Array
+        issues_with_contrib_comments: Array
+        issues_total: Array
+        issues_processed_in_avg: Ossert::TooLong
+        issues_processed_in_median: Ossert::TooLong
+        pr_open: Array
+        pr_merged: Array
+        pr_closed: Array
+        pr_owner: Array
+        pr_non_owner: Array
+        pr_with_contrib_comments: Array
+        pr_total: Array
+        pr_processed_in_avg: Ossert::TooLong
+        pr_processed_in_median: Ossert::TooLong
+        releases_total_gh: Array
+        releases_total_rg: Array
+        stale_branches: Array
+        branches: Array
+        first_pr_date: Ossert::Zero
+        last_pr_date: Ossert::Zero
+        first_issue_date: Ossert::Zero
+        last_issue_date: Ossert::Zero
+        last_release_date: Ossert::Zero
+        commits_count_since_last_release: Ossert::Zero
+        last_year_commits: Ossert::Zero
+    quarter:
+      metrics:
+        - issues_processed_in_avg
+        - issues_processed_in_median
+        - issues_closed_percent
+        - issues_actual_count
+        - issues_all_count
+        - pr_processed_in_avg
+        - pr_processed_in_median
+        - pr_closed_percent
+        - pr_actual_count
+        - pr_all_count
+        - releases_count
+        - commits
+      uniq_attributes:
+        - issues_open
+        - issues_closed
+        - issues_total
+        - issues_actual
+        - pr_open
+        - pr_closed
+        - pr_total
+        - pr_actual
+        - branches
+        - releases
+        - releases_total_gh
+        - releases_total_rg
+      attributes:
+        issues_open: Array
+        issues_closed: Array
+        issues_total: Array
+        issues_actual: Array
+        pr_open: Array
+        pr_merged: Array
+        pr_closed: Array
+        pr_total: Array
+        pr_actual: Array
+        branches: Array
+        releases: Array
+        releases_total_gh: Array
+        releases_total_rg: Array
+        commits: Ossert::Zero
+        issues_processed_in_days: Array
+        pr_processed_in_days: Array
+  community:
+    total:
+      metrics:
+        - dependants_count
+        - users_creating_issues_count
+        - contributors_count
+        - users_commenting_issues_count
+        - users_creating_pr_count
+        - users_commenting_pr_count
+        - watchers_count
+        - stargazers_count
+        - forks_count
+        - users_involved_count
+        - users_involved_no_stars_count
+        - total_downloads_count
+      uniq_attributes:
+        - dependants
+        - users_creating_issues
+        - users_commenting_issues
+        - users_creating_pr
+        - users_commenting_pr
+        - contributors
+        - watchers
+        - stargazers
+        - forks
+        - users_involved
+        - owners_github
+        - owners_rubygems
+      attributes:
+        dependants: Array
+        users_creating_issues: Array
+        users_commenting_issues: Array
+        users_creating_pr: Array
+        users_commenting_pr: Array
+        contributors: Array
+        watchers: Array
+        stargazers: Array
+        forks: Array
+        users_involved: Array
+        owners_github: Array
+        owners_rubygems: Array
+        total_downloads: Ossert::Zero
+        delta_downloads: Ossert::Zero
+    quarter:
+      metrics:
+        - users_creating_issues_count
+        - users_commenting_issues_count
+        - users_creating_pr_count
+        - users_commenting_pr_count
+        - stargazers_count
+        - forks_count
+        - users_involved_count
+        - users_involved_no_stars_count
+        - total_downloads_count
+      uniq_attributes:
+        - users_creating_issues
+        - users_commenting_issues
+        - users_creating_pr
+        - users_commenting_pr
+        - contributors
+        - stargazers
+        - forks
+        - users_involved
+        - owners_github
+        - owners_rubygems
+      absolute_attributes:
+        - total_downloads
+      attributes:
+        users_creating_issues: Array
+        users_commenting_issues: Array
+        users_creating_pr: Array
+        users_commenting_pr: Array
+        contributors: Array
+        stargazers: Array
+        forks: Array
+        users_involved: Array
+        download_divergence: Ossert::Zero
+        total_downloads: Ossert::Zero
+        delta_downloads: Ossert::Zero

data/config/translations.yml

@@ -0,0 +1,44 @@
+translations:
+  # Agility
+  dependencies_count: 'Number of Runtime Dependencies'
+  issues_closed_percent: 'Issues Closed %'
+  issues_non_owner_percent: 'Issues Created Not By Repo Owner'
+  issues_with_contrib_comments_percent: 'Issues With Contibutors Comments'
+  issues_all_count: 'Number of Issues'
+  issues_processed_in_avg: 'Average Issue Processing Time'
+  issues_processed_in_median: 'Median Issue Processing Time'
+  first_issue_date_int: 'First Issue Created At'
+  last_issue_date_int: 'Last Issue Created At'
+  pr_closed_percent: 'Pull Requests Closed %'
+  pr_non_owner_percent: 'Pull Requests Created Not By Repo Owner'
+  pr_with_contrib_comments_percent: 'Pull Requests With Contibutors Comments'
+  pr_all_count: 'Number of Pull Requests'
+  pr_processed_in_avg: 'Average Pull Request Processing Time'
+  pr_processed_in_median: 'Median Pull Request Processing Time'
+  first_pr_date_int: 'First Pull Request Created At'
+  last_pr_date_int: 'Last Pull Request Created At'
+  pr_actual_count: 'Number of Legacy Pull Requests'
+  last_release_date_int: 'Last Release Date'
+  releases_count: 'Number of Releases'
+  commits_count_since_last_release_count: 'Commits Since Last Release'
+  stale_branches_count: 'Number of Stale Branches'
+  life_period: 'Active Years Age'
+  last_changed: 'Last Changed At'
+  issues_actual_count: 'Number of Legacy Issues'
+  commits: 'Number of Commits Made'
+
+  # Community
+  dependants_count: 'Number of Dependants'
+  users_creating_issues_count: 'Number of Users Creating Issues'
+  users_commenting_issues_count: 'Number of Users Commenting Issues'
+  users_creating_pr_count: 'Number of Users Creating Pull Requests'
+  users_commenting_pr_count: 'Number of Users Commenting Pull Requests'
+  watchers_count: 'Number of Watchers'
+  stargazers_count: 'Number of Stargazers'
+  contributors_count: 'Number of Contributors'
+  forks_count: 'Number of Forks'
+  users_involved_count: 'Number of Total Users Involved'
+  users_involved_no_stars_count: 'Number of Users Involved Without Stargazers'
+  total_downloads_count: 'Number of Downloads'
+  delta_downloads: 'Number of Downloads Increased By'
+  download_divergence: 'Downloads Growth From Previous Period %'

data/db/migrate/001_create_projects.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+Sequel.migration do
+  up do
+    extension :pg_json
+
+    create_table(:projects) do
+      primary_key :id
+      String :name, null: false
+      String :github_name, null: false
+      String :rubygems_name, null: false
+      String :reference, null: false
+      json :agility_total_data
+      json :agility_quarters_data
+      json :community_total_data
+      json :community_quarters_data
+    end
+  end
+
+  down do
+    drop_table(:projects)
+  end
+end

data/db/migrate/002_create_exceptions.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+Sequel.migration do
+  up do
+    create_table(:exceptions) do
+      primary_key :id
+      String :name, null: false
+      String :github_name, null: false
+    end
+  end
+
+  down do
+    drop_table(:exceptions)
+  end
+end

data/db/migrate/003_add_meta_to_projects.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+Sequel.migration do
+  up do
+    extension :pg_json
+
+    alter_table(:projects) do
+      add_column :meta_data, JSON
+    end
+  end
+
+  down do
+    drop_column :projects, :meta_data
+  end
+end

data/db/migrate/004_add_timestamps_to_projects.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+Sequel.migration do
+  up do
+    run 'ALTER TABLE projects ADD COLUMN created_at TIMESTAMP DEFAULT NOW() NOT NULL;'
+    run 'ALTER TABLE projects ADD COLUMN updated_at TIMESTAMP DEFAULT NOW() NOT NULL;'
+  end
+
+  down do
+    drop_column :projects, :created_at
+    drop_column :projects, :updated_at
+  end
+end

data/db/migrate/005_create_classifiers.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+Sequel.migration do
+  up do
+    extension :pg_json
+
+    create_table(:classifiers) do
+      primary_key :id
+      String :section, null: false
+      json :reference_values
+    end
+
+    run 'ALTER TABLE classifiers ADD COLUMN created_at TIMESTAMP DEFAULT NOW() NOT NULL;'
+    run 'ALTER TABLE classifiers ADD COLUMN updated_at TIMESTAMP DEFAULT NOW() NOT NULL;'
+  end
+
+  down do
+    drop_table(:classifiers)
+  end
+end

data/lib/ossert/classifiers/decision_tree.rb

@@ -0,0 +1,112 @@
+# frozen_string_literal: true
+require 'graphr'
+require 'decisiontree'
+
+module Ossert
+  module Classifiers
+    class DecisionTree
+      GRADES = %w(
+        ClassA
+        ClassB
+        ClassC
+        ClassD
+        ClassE
+      ).freeze
+
+      class << self
+        attr_accessor :all
+
+        def for_current_projects
+          new(Project.projects_by_reference)
+        end
+
+        def current
+          all.last
+        end
+      end
+
+      def initialize(train_group)
+        @train_group = train_group
+
+        (self.class.all ||= []) << self
+      end
+
+      attr_reader :train_group
+      attr_reader :agility_total_dec_tree,
+                  :agility_last_year_dec_tree,
+                  :community_total_dec_tree,
+                  :community_total_dec_tree,
+                  :community_last_year_dec_tree
+
+      def check(project)
+        {
+          agility: {
+            total: agility_total_check(project),
+            last_year: agility_last_year_check(project)
+          },
+          community: {
+            total: community_total_check(project),
+            last_year: community_last_year_check(project)
+          }
+        }
+      end
+
+      SECTION_DATA = {
+        agility_total: ->(project) { project.agility.total.metric_values },
+        agility_last_year: ->(project) { project.agility.quarters.last_year_data },
+        community_total: ->(project) { project.community.total.metric_values },
+        community_last_year: ->(project) { project.community.quarters.last_year_data }
+      }.freeze
+
+      SECTION_METRICS = {
+        agility_total: Stats::AgilityTotal.metrics,
+        agility_last_year: Stats::AgilityQuarter.metrics,
+        community_total: Stats::CommunityTotal.metrics,
+        community_last_year: Stats::CommunityQuarter.metrics
+      }.freeze
+
+      [:community, :agility].each do |section|
+        [:total, :last_year].each do |type|
+          name = "#{section}_#{type}".to_sym
+          define_method("#{name}_check") do |project|
+            public_send("#{name}_dec_tree").predict(SECTION_DATA[name].call(project))
+          end
+        end
+      end
+
+      def ready?
+        agility_total_dec_tree.presence &&
+          agility_last_year_dec_tree.presence &&
+          community_total_dec_tree.presence &&
+          community_last_year_dec_tree.presence
+      end
+
+      def train
+        data = initialize_data
+
+        trees = SECTION_METRICS.map do |section, metrics|
+          ::DecisionTree::ID3Tree.new(metrics, data[section], 'ClassE', :continuous)
+        end.tap(&:train)
+
+        @agility_total_dec_tree,
+        @agility_last_year_dec_tree,
+        @community_total_dec_tree,
+        @community_last_year_dec_tree = trees
+      end
+
+      def initialize_data
+        result = { agility_total: [], agility_last_year: [], community_total: [], community_last_year: [] }
+
+        GRADES.each_with_object(train_group) do |grade, grouped_projects|
+          grouped_projects[grade].each do |project|
+            SECTION_DATA.each do |section, data_collector|
+              result[section] << (data_collector.call(project) << grade)
+            end
+          end
+        end
+
+        result
+      end
+    end
+  end
+end

data/lib/ossert/classifiers/growing/check.rb

@@ -0,0 +1,172 @@
+# frozen_string_literal: true
+module Ossert
+  module Classifiers
+    class Growing
+      class Check
+        GRADES = %w(
+          A
+          B
+          C
+          D
+          E
+        ).freeze
+        KLASS_2_GRADE = {
+          'ClassA' => 'A',
+          'ClassB' => 'B',
+          'ClassC' => 'C',
+          'ClassD' => 'D',
+          'ClassE' => 'E'
+        }.freeze
+
+        class << self
+          def process_using(action, config, project, classifiers, last_year_offset = 1)
+            checks_rates = config['checks'].map do |check_name|
+              [
+                check_name.to_sym,
+                check_class_by(check_name).new(
+                  config, project, classifiers, last_year_offset
+                ).send(action)
+              ]
+            end
+            checks_rates.to_h
+          end
+
+          def check(*args)
+            process_using(*args.unshift(:grade_as_hash))
+          end
+
+          def grade(*args)
+            process_using(*args.unshift(:grade))
+          end
+
+          def check_class_by(check_name)
+            Kernel.const_get(
+              "Ossert::Classifiers::Growing::Check::#{check_name.capitalize}"
+            )
+          end
+        end
+
+        class Base
+          def initialize(config, project, classifiers, last_year_offset = 1)
+            @config = config
+            @project = project
+            @classifiers = classifiers
+            @last_year_offset = last_year_offset
+          end
+
+          def metrics_type
+            self.class.name.split('::').last.downcase
+          end
+
+          def metrics
+            return @metrics if defined? @metrics
+
+            @metrics = @config['metrics'][metrics_type].map do |section, section_metrics|
+              [
+                section,
+                section_metrics.map do |metric, weight|
+                  [metric, weight.to_d]
+                end.to_h
+              ]
+            end.to_h
+          end
+
+          def max_gain
+            @max_gain ||= (metrics['last_year'].values.sum + metrics['total'].values.sum).to_d
+          end
+
+          def community_last_year_data
+            @community_last_year_data ||= @project.community.quarters.last_year_as_hash(@last_year_offset)
+          end
+
+          def community_total_data
+            @community_total_data ||= @project.community.total.metrics_to_hash
+          end
+
+          def agility_last_year_data
+            @agility_last_year_data ||= @project.agility.quarters.last_year_as_hash(@last_year_offset)
+          end
+
+          def agility_total_data
+            @agility_total_data ||= @project.agility.total.metrics_to_hash
+          end
+
+          def check
+            rates = GRADES.each_with_object({}) { |klass, res| res[klass] = 0.0.to_d }
+            strategy.each do |(section_type, data_types)|
+              Array(data_types).each do |data_type|
+                rate(
+                  rates, metrics[section_type.to_s], send("#{data_type}_#{section_type}_data"),
+                  @classifiers.fetch("#{data_type}_#{section_type}".to_sym)
+                )
+              end
+            end
+            rates
+          end
+
+          def grade
+            max = GRADES.count
+            sum = -0.6
+            check.sort.reverse.each do |(_, gain)|
+              sum += gain
+              sum += 0.1 if gain > trusted_probability
+            end
+            GRADES[(max - sum).to_i]
+          end
+
+          def grade_as_hash
+            max = GRADES.count
+            sum = -0.6
+            check.sort.reverse.each do |(_, gain)|
+              sum += gain
+              sum += 0.1 if gain > trusted_probability
+            end
+            { gain: sum, mark: GRADES[(max - sum).to_i] }
+          end
+
+          protected
+
+          def trusted_probability
+            @trusted_probability ||= @config['trusted_probability']
+          end
+
+          def rate(rates, metrics, data, classifier)
+            classifier.each_pair do |klass, qualified_metrics|
+              metrics.slice(*data.keys).each_pair do |metric, weight|
+                range = qualified_metrics[metric.to_s][:range]
+                rates[KLASS_2_GRADE[klass]] += weight / max_gain if range.cover? data[metric].to_f
+              end
+            end
+          end
+        end
+
+        class Popularity < Base
+          def strategy
+            {
+              last_year: :community,
+              total: :community
+            }
+          end
+        end
+
+        class Maintenance < Base
+          def strategy
+            {
+              last_year: :agility,
+              total: :agility
+            }
+          end
+        end
+
+        class Maturity < Base
+          def strategy
+            {
+              last_year: [:agility, :community],
+              total: [:agility, :community]
+            }
+          end
+        end
+      end
+    end
+  end
+end