locabulary 0.2.0 → 0.3.0

data/lib/locabulary/exceptions.rb

@@ -0,0 +1,25 @@
+module Locabulary
+  # Container for all exceptions in the Locabulary ecosystem
+  module Exceptions
+    class RuntimeError < ::RuntimeError
+    end
+
+    # There is a problem with the hierarchy; A child is being defined without a defined parent.
+    class MissingHierarchicalParentError < RuntimeError
+      attr_reader :predicate_name, :error
+      def initialize(predicate_name, error)
+        @predicate_name = predicate_name
+        @error = error
+        super("Expected #{predicate_name.inspect} to have a welformed tree. Error: #{error}")
+      end
+    end
+
+    # There is a problem with the hierarchy; Instead of a tree we have a multitude of trees
+    class TooManyHierarchicalRootsError < RuntimeError
+      attr_reader :predicate_name, :roots
+      def initialize(predicate_name, roots)
+        super("Expected fewer root slugs for #{predicate_name.inspect}. Roots encountered: #{roots.inspect}")
+      end
+    end
+  end
+end

data/lib/locabulary/items/administrative_unit.rb

@@ -0,0 +1,65 @@
+require 'locabulary/exceptions'
+require 'locabulary/items/base'
+
+module Locabulary
+  module Items
+    # Responsible for exposing the data structure logic of the Administrative Units
+    #
+    # @see ./data/administrative_units.json
+    class AdministrativeUnit < Locabulary::Items::Base
+      configure do |config|
+        config.attribute_names = [
+          :predicate_name, :term_label, :term_uri, :description, :grouping, :classification, :affiliation, :default_presentation_sequence,
+          :homepage, :activated_on, :deactivated_on
+        ]
+      end
+
+      # [String] What is the URL of the homepage. Please note the term_uri is reserved for something that is more resolvable by machines.
+      #   And while the homepage may look resolvable, it is not as meaningful for longterm preservation.
+      attr_reader :homepage
+      attr_reader :classification
+      attr_reader :grouping
+      attr_reader :affiliation
+
+      private
+
+      attr_writer :homepage, :classification, :grouping, :affiliation
+
+      public
+
+      def initialize(*args)
+        super
+        @children = []
+      end
+
+      attr_reader :children
+
+      HIERARCHY_SEPARATOR = '::'.freeze
+      def slugs
+        term_label.split(HIERARCHY_SEPARATOR)
+      end
+
+      def parent_slugs
+        slugs[0..-2]
+      end
+
+      def parent_term_label
+        parent_slugs.join(HIERARCHY_SEPARATOR)
+      end
+
+      def root_slug
+        slugs[0]
+      end
+
+      def selectable?
+        children.count == 0
+      end
+
+      def selectable_label
+        slugs[1..-1].join(HIERARCHY_SEPARATOR)
+      end
+
+      alias selectable_id id
+    end
+  end
+end

data/lib/locabulary/items/base.rb

@@ -0,0 +1,105 @@
+require 'dry/configurable'
+module Locabulary
+  module Items
+    # A singular item in the controlled vocubulary.
+    # @see https://en.wikipedia.org/wiki/Resource_Description_Framework
+    class Base
+      extend Dry::Configurable
+
+      setting :attribute_names, [
+        :predicate_name, :term_label, :term_uri, :deposit_label, :description, :grouping, :affiliation, :default_presentation_sequence,
+        :activated_on, :deactivated_on
+      ].freeze
+
+      def attribute_names
+        self.class.config.attribute_names
+      end
+
+      # [String] the trait for a given subject that we are describing by way of the term_label/term_uri
+      attr_reader :predicate_name
+
+      # [String] the human friendly version of the meaning for this given trait
+      # @note For the time being, please regard the term_label as immutable; If you need a modification, deactivate this one and activate a
+      #    new one
+      attr_reader :term_label
+
+      # [String] the machine friendly version of the meaning for this given trait
+      # @note For the time being, please regard the term_uri as immutable; If you need a modification, deactivate this one and activate a
+      #    new one
+      attr_reader :term_uri
+
+      # [String] a side-car of more exhaustive information related to this particular term
+      attr_reader :description
+
+      # [Date] When was this particular item activated
+      attr_reader :activated_on
+
+      # [Date] When was this particular item deactivated
+      attr_reader :deactivated_on
+
+      # [Integer, nil] What is the order in which
+      # @see Locabulary::Item#presentation_sequence for details on how this is calculated
+      attr_reader :default_presentation_sequence
+
+      # @deprecated
+      attr_reader :deposit_label
+
+      # @deprecated
+      attr_reader :grouping
+
+      # @deprecated
+      attr_reader :affiliation
+
+      def initialize(attributes = {})
+        attribute_names.each do |key|
+          value = attributes.fetch(key) { attributes.fetch(key.to_s, nil) }
+          send("#{key}=", value)
+        end
+      end
+
+      def to_h
+        attribute_names.each_with_object({}) do |key, mem|
+          mem[key.to_s] = send(key) unless send(key).to_s.strip == ''
+          mem
+        end
+      end
+      alias as_json to_h
+
+      def to_persistence_format_for_fedora
+        return term_uri unless term_uri.to_s.strip == ''
+        term_label
+      end
+      alias id to_persistence_format_for_fedora
+
+      private
+
+      attr_writer(*config.attribute_names)
+
+      def predicate_name=(input)
+        @predicate_name = input.to_s
+      end
+
+      def default_presentation_sequence=(input)
+        @default_presentation_sequence = input.to_s.strip == '' ? nil : input.to_i
+      end
+
+      public
+
+      include Comparable
+
+      def <=>(other)
+        predicate_name_sort = predicate_name <=> other.predicate_name
+        return predicate_name_sort unless predicate_name_sort == 0
+        presentation_sequence_sort = presentation_sequence <=> other.presentation_sequence
+        return presentation_sequence_sort unless presentation_sequence_sort == 0
+        term_label <=> other.term_label
+      end
+
+      SORT_SEQUENCE_FOR_NIL = 100_000_000
+      private_constant :SORT_SEQUENCE_FOR_NIL
+      def presentation_sequence
+        default_presentation_sequence || SORT_SEQUENCE_FOR_NIL
+      end
+    end
+  end
+end

data/lib/locabulary/items.rb

@@ -0,0 +1,32 @@
+require 'locabulary/items/base'
+require 'hanami/utils/string'
+module Locabulary
+  # A container for the various types of Locabulary Items
+  module Items
+    module_function
+
+    # @api public
+    # @since 0.2.1
+    def build(options = {})
+      predicate_name = options.fetch(:predicate_name) { options.fetch('predicate_name') }
+      builder_for(predicate_name: predicate_name).call(options)
+    end
+
+    # @api public
+    # @since 0.2.1
+    #
+    # @param options [Hash]
+    # @option predicate_name [String] Used for lookup of the correct Locabulary::Item type
+    def builder_for(options = {})
+      predicate_name = options.fetch(:predicate_name)
+      possible_class_name_for_predicate_name = Hanami::Utils::String.new(predicate_name).singularize.classify
+      klass = begin
+        Items.const_get(possible_class_name_for_predicate_name)
+      rescue NameError
+        Items::Base
+      end
+      klass.method(:new)
+    end
+  end
+end
+require 'locabulary/items/administrative_unit'

data/lib/locabulary/json_creator.rb

@@ -0,0 +1,105 @@
+require "google/api_client"
+require "google_drive"
+require 'highline/import'
+require 'locabulary'
+require 'locabulary/items'
+require 'json'
+
+module Locabulary
+  # Responsible for capturing predicate_name from a given source and writing it to a file
+  class JsonCreator
+    def initialize(document_key, predicate_name, data_fetcher = default_data_fetcher)
+      @document_key = document_key
+      @predicate_name = predicate_name
+      @output_filepath = Locabulary.filename_for_predicate_name(predicate_name: predicate_name)
+      @data_fetcher = data_fetcher
+    end
+
+    attr_reader :document_key, :predicate_name, :data_fetcher, :spreadsheet_data, :json_data
+    attr_accessor :output_filepath
+
+    def create_or_update
+      rows = data_fetcher.call(document_key)
+      data = extract_data_from(rows)
+      convert_to_json(data)
+    end
+
+    # :nocov:
+    def write_to_file
+      File.open(output_filepath, "w") do |f|
+        f.puts json_data
+      end
+    end
+    # :nocov:
+
+    private
+
+    def extract_data_from(rows)
+      spreadsheet_data = []
+      header = rows[0]
+      rows[1..-1].each do |row|
+        # The activated_on is a present hack reflecting a previous value
+        row_data = { "predicate_name" => predicate_name, "activated_on" => "2015-07-22", "default_presentation_sequence" => nil }
+        row.each_with_index do |cell, index|
+          row_data[header[index]] = cell unless cell.to_s.strip == ''
+        end
+        spreadsheet_data << row_data
+      end
+      spreadsheet_data
+    end
+
+    def default_data_fetcher
+      ->(document_key) { GoogleSpreadsheet.new(document_key).all_rows }
+    end
+
+    def convert_to_json(data)
+      json_array = data.map do |row|
+        Locabulary::Items.build(row).to_h
+      end
+      @json_data = JSON.pretty_generate("predicate_name" => predicate_name, "values" => json_array)
+    end
+
+    # :nocov:
+    # Responsible for interacting with Google Sheets and retrieiving relevant information
+    class GoogleSpreadsheet
+      attr_reader :access_token, :document_key, :session
+
+      private :session
+
+      def initialize(document_key)
+        @document_key = document_key
+        configure_oauth!
+        @session = GoogleDrive.login_with_oauth(access_token)
+      end
+
+      def configure_oauth!
+        client = Google::APIClient.new
+        auth = client.authorization
+        auth.client_id = client_secrets.fetch('client_id')
+        auth.client_secret = client_secrets.fetch('client_secret')
+        auth.scope = ["https://www.googleapis.com/auth/drive.readonly"]
+        auth.redirect_uri = "urn:ietf:wg:oauth:2.0:oob"
+        puts "\n Open the following URL, login with your credentials and get the authorization code \n\n #{auth.authorization_uri}\n\n"
+        auth.code = ask('Authorization Code: ')
+        auth.fetch_access_token!
+        @access_token = auth.access_token
+      end
+
+      def all_rows
+        session.spreadsheet_by_key(document_key).worksheets[0].rows
+      end
+
+      def client_secrets
+        @secrets ||= YAML.load(File.open(File.join(secrets_path)))
+      end
+
+      def secrets_path
+        if File.exist? File.join(File.dirname(__FILE__), '../../config/client_secrets.yml')
+          File.join(File.dirname(__FILE__), '../../config/client_secrets.yml')
+        else
+          File.join(File.dirname(__FILE__), '../../config/client_secrets.example.yml')
+        end
+      end
+    end
+  end
+end

data/lib/locabulary/schema.rb

@@ -2,26 +2,17 @@ require 'dry/validation'
 require 'dry/validation/schema'
 
 module Locabulary
-  class Schema < Dry::Validation::Schema
-    key(:predicate_name) { |predicate_name| predicate_name.format?(/\A[a-z_]+\Z/) & predicate_name.filled? }
-    key(:values) do |values|
-      values.array? do
-        values.each do |value|
-          value.hash? do
-            value.key(:term_label) { |term_label| term_label.filled? }
-            value.optional(:term_uri) { |term_label| term_label.none? | term_label.str?  }
-            value.optional(:deposit_label) { |deposit_label| deposit_label.none? | deposit_label.str?  }
-            value.optional(:description) { |description| description.none? | description.str?  }
-            value.optional(:grouping) { |grouping| grouping.none? | grouping.str?  }
-            value.optional(:affiliation) { |affiliation| affiliation.none? | affiliation.str?  }
-            value.optional(:default_presentation_sequence) do |default_presentation_sequence|
-              default_presentation_sequence.none? | default_presentation_sequence.int?
-            end
-            value.key(:activated_on) { |activated_on| activated_on.format?(/\A\d{4}-\d{2}-\d{2}\Z/) }
-            value.optional(:deactivated_on) { |deactivated_on| deactivated_on.none? | deactivated_on.format?(/\A\d{4}-\d{2}-\d{2}\Z/) }
-          end
-        end
-      end
+  # Responsible for providing a defined and clear schema for each of the locabulary items.
+  Schema = Dry::Validation.Schema do
+    key(:predicate_name).required(format?: /\A[a-z_]+\Z/)
+    key(:values).each do
+      key(:term_label).required(:str?)
+      optional(:description).maybe(:str?)
+      optional(:grouping).maybe(:str?)
+      optional(:affiliation).maybe(:str?)
+      optional(:default_presentation_sequence).maybe(:int?)
+      key(:activated_on).required(format?: /\A\d{4}-\d{2}-\d{2}\Z/)
+      optional(:deactivated_on).maybe(format?: /\A\d{4}-\d{2}-\d{2}\Z/)
     end
   end
 end

data/lib/locabulary/version.rb

@@ -0,0 +1,3 @@
+module Locabulary
+  VERSION = '0.3.0'.freeze
+end

data/lib/locabulary.rb

@@ -1,43 +1,12 @@
 require 'date'
 require 'json'
+require 'locabulary/exceptions'
+require 'locabulary/items'
 
 # @since 0.1.0
 module Locabulary
-  VERSION='0.2.0'.freeze
   DATA_DIRECTORY = File.expand_path("../../data/", __FILE__).freeze
 
-  class RuntimeError < ::RuntimeError
-  end
-
-  class Item
-    include Comparable
-    ATTRIBUTE_NAMES = [:predicate_name, :term_label, :term_uri, :deposit_label, :description, :grouping, :affiliation, :default_presentation_sequence, :activated_on, :deactivated_on].freeze
-
-    attr_reader(*ATTRIBUTE_NAMES)
-
-    def initialize(attributes={})
-      ATTRIBUTE_NAMES.each do |key|
-        value = attributes.fetch(key) { attributes.fetch(key.to_s, nil) }
-        send("#{key}=", value)
-      end
-    end
-
-    SORT_SEQUENCE_FOR_NIL = 100_000_000
-    def <=>(other)
-      value = presentation_sequence <=> other.presentation_sequence
-      return value unless value == 0
-      term_label <=> other.term_label
-    end
-
-    def presentation_sequence
-      default_presentation_sequence || SORT_SEQUENCE_FOR_NIL
-    end
-
-    private
-
-    attr_writer(*ATTRIBUTE_NAMES)
-  end
-
   module_function
 
   # @api public
@@ -50,25 +19,68 @@ module Locabulary
   def active_items_for(options = {})
     predicate_name = options.fetch(:predicate_name)
     as_of = options.fetch(:as_of) { Date.today }
+    builder = Items.builder_for(predicate_name: predicate_name)
     active_cache[predicate_name] ||= begin
-      filename = filename_for_predicate_name(predicate_name: predicate_name)
-      json = JSON.parse(File.read(filename))
-      predicate_name = json.fetch('predicate_name')
-      json.fetch('values').each_with_object([]) do |item_values, mem|
-        activated_on = Date.parse(item_values.fetch('activated_on'))
-        next unless activated_on < as_of
-        deactivated_on_value = item_values.fetch('deactivated_on', nil)
-        if deactivated_on_value.nil?
-          mem << Item.new(item_values.merge('predicate_name' => predicate_name))
-        else
-          deactivated_on = Date.parse(deactivated_on_value)
-          next unless deactivated_on >= as_of
-          mem << Item.new(item_values.merge('predicate_name' => predicate_name))
-        end
-        mem
-      end.sort
+      collector = []
+      with_active_extraction_for(predicate_name, as_of) do |data|
+        collector << builder.call(data.merge('predicate_name' => predicate_name))
+      end
+      collector.sort
+    end
+  end
+
+  # @api public
+  # @since 0.2.0
+  def active_hierarchical_root(options = {})
+    predicate_name = options.fetch(:predicate_name)
+    as_of = options.fetch(:as_of) { Date.today }
+    builder = Items.builder_for(predicate_name: predicate_name)
+    active_hierarchical_root_cache[predicate_name] ||= begin
+      items = []
+      hierarchy_graph_keys = {}
+      top_level_slugs = Set.new
+      with_active_extraction_for(predicate_name, as_of) do |data|
+        item = builder.call(data.merge('predicate_name' => predicate_name))
+        items << item
+        top_level_slugs << item.root_slug
+        hierarchy_graph_keys[item.term_label] = item
+      end
+      associate_parents_and_childrens_for(hierarchy_graph_keys, items, predicate_name)
+      raise Exceptions::TooManyHierarchicalRootsError.new(predicate_name, top_level_slugs.to_a) if top_level_slugs.size > 1
+      hierarchy_graph_keys.fetch(top_level_slugs.first)
+    end
+  end
+
+  def associate_parents_and_childrens_for(hierarchy_graph_keys, items, predicate_name)
+    items.each do |item|
+      begin
+        hierarchy_graph_keys.fetch(item.parent_term_label).children << item unless item.parent_slugs.empty?
+      rescue KeyError => error
+        raise Exceptions::MissingHierarchicalParentError.new(predicate_name, error)
+      end
     end
   end
+  private :associate_parents_and_childrens_for
+
+  def with_active_extraction_for(predicate_name, as_of)
+    filename = filename_for_predicate_name(predicate_name: predicate_name)
+    json = JSON.parse(File.read(filename))
+    json.fetch('values').each do |data|
+      yield(data) if data_is_active?(data, as_of)
+    end
+  end
+  private :with_active_extraction_for
+
+  def data_is_active?(data, as_of)
+    activated_on = Date.parse(data.fetch('activated_on'))
+    return false unless activated_on < as_of
+    deactivated_on_value = data.fetch('deactivated_on', nil)
+    return true if deactivated_on_value.nil?
+    deactivated_on = Date.parse(deactivated_on_value)
+    return false unless deactivated_on >= as_of
+    true
+  end
+  private :data_is_active?
 
   # @api public
   # @since 0.1.0
@@ -87,27 +99,12 @@ module Locabulary
     active_items_for(predicate_name: predicate_name).map(&:term_label)
   end
 
-
-  # @api public
-  def active_nested_labels_for(options = {})
-    format_active_items_for(active_labels_for(options))
-  end
-
-  # @api public
-  def properties_for_uri(options = {})
-    predicate_name = options.fetch(:predicate_name)
-    term_uri = options.fetch(:term_uri)
-    object = active_items_for(predicate_name: predicate_name).detect { |obj| obj.term_uri == term_uri }
-    return object. if object
-    term_uri
-  end
-
   # @api private
   def filename_for_predicate_name(options = {})
     predicate_name = options.fetch(:predicate_name)
     filename = File.join(DATA_DIRECTORY, "#{File.basename(predicate_name)}.json")
     return filename if File.exist?(filename)
-    raise Locabulary::RuntimeError, "Unable to find predicate_name: #{predicate_name}"
+    raise Locabulary::Exceptions::RuntimeError, "Unable to find predicate_name: #{predicate_name}"
   end
 
   # @api private
@@ -116,31 +113,13 @@ module Locabulary
   end
 
   # @api private
-  def reset_active_cache!
-    @active_cache = nil
+  def active_hierarchical_root_cache
+    @active_hierarchical_root_cache ||= {}
   end
 
   # @api private
-  def format_active_items_for(items)
-    root = {}
-    items.each do |item|
-      key, value = build_key_and_value(item)
-      root[key] ||= []
-      root[key] << value
-    end
-    root
-  end
-
-  # @api private
-  def build_key_and_value(text)
-    text_array = text.split(/(::)/)
-    return text, text if text_array.size == 1
-    return text, text_array.last if text_array.size == 3
-    key = ""
-    (0..(text_array.size-3)).each do |index|
-      key << text_array[index]
-    end
-    value = text_array.last
-    return key, value
+  def reset_active_cache!
+    @active_cache = nil
+    @active_hierarchical_root_cache = nil
   end
 end

data/locabulary.gemspec

@@ -1,7 +1,7 @@
 # coding: utf-8
 lib = File.expand_path('../lib', __FILE__)
 $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
-require 'locabulary'
+require 'locabulary/version'
 
 Gem::Specification.new do |spec|
   spec.name          = "locabulary"
@@ -20,12 +20,18 @@ Gem::Specification.new do |spec|
   spec.license = 'APACHE2'
 
   spec.add_dependency "json", "~> 1.8"
+  spec.add_dependency "dry-configurable"
+  spec.add_dependency "hanami-utils"
 
   spec.add_development_dependency "dry-validation"
   spec.add_development_dependency "bundler"
-  spec.add_development_dependency "minitest"
+  spec.add_development_dependency "rspec"
+  spec.add_development_dependency "rspec-its"
   spec.add_development_dependency "rake", "~> 10.0"
   spec.add_development_dependency 'google_drive'
   spec.add_development_dependency 'highline'
   spec.add_development_dependency "activesupport", "~>4.0"
+  spec.add_development_dependency "rubocop"
+  spec.add_development_dependency "simplecov"
+  spec.add_development_dependency "codeclimate-test-reporter"
 end

data/script/update_data_files.sh

@@ -1,8 +1,13 @@
 #!/usr/bin/env ruby -U
+if !ENV.key?('BUNDLE_GEMFILE')
+  $stderr.puts "You need to call the command via `bundle exec script/#{File.basename(__FILE__)}`"
+  $stderr.puts "You will also need to make sure you have valid information in ./config/client_secrets.example.yml."
+  abort(1)
+end
 
-require_relative 'json_creator'
+require 'locabulary/json_creator'
 
 puts "Updating Administrative Units"
-json_creator = JsonCreator.new "1oBW5FCTtYXsUi8roBiMRBLFY3dXamhTqy-kiG2rqu5Q", "administrative_units"
+json_creator = Locabulary::JsonCreator.new "1oBW5FCTtYXsUi8roBiMRBLFY3dXamhTqy-kiG2rqu5Q", "administrative_units"
 json_creator.create_or_update
 json_creator.write_to_file