From d1e8558dc57210fcb902d359a8df9208365ca954 Mon Sep 17 00:00:00 2001
From: carolyncole <1599081+carolyncole@users.noreply.github.com>
Date: Wed, 5 Mar 2025 15:12:52 -0500
Subject: [PATCH] Remove Dataspace Communities (#762)

refs #685
---
 app/lib/research_data_collection.rb       |  20 ----
 app/models/dataspace_communities.rb       | 135 ----------------------
 app/models/dataspace_community.rb         |  57 ---------
 spec/lib/research_data_collection_spec.rb |  26 -----
 spec/models/dataspace_communities_spec.rb |  44 -------
 5 files changed, 282 deletions(-)
 delete mode 100644 app/lib/research_data_collection.rb
 delete mode 100644 app/models/dataspace_communities.rb
 delete mode 100644 app/models/dataspace_community.rb
 delete mode 100644 spec/lib/research_data_collection_spec.rb
 delete mode 100644 spec/models/dataspace_communities_spec.rb

diff --git a/app/lib/research_data_collection.rb b/app/lib/research_data_collection.rb
deleted file mode 100644
index 0309ff33..00000000
--- a/app/lib/research_data_collection.rb
+++ /dev/null
@@ -1,20 +0,0 @@
-# frozen_string_literal: true
-
-require 'csv'
-
-##
-# A DataSpace collection that contains research data and should be harvested.
-class ResearchDataCollection
-  attr_reader :parent_community, :community, :collection_name, :handle, :collection_id, :item_count
-
-  ##
-  # Take a CSV::Row and parse it out into the values we'll need at indexing time
-  def initialize(csv_row)
-    @parent_community = csv_row["ParentCommunity"]
-    @community = csv_row["Community"]
-    @collection_name = csv_row["CollectionName"]
-    @handle = csv_row["Handle"]
-    @collection_id = csv_row["CollectionID"]
-    @item_count = csv_row["ItemCount"].to_i
-  end
-end
diff --git a/app/models/dataspace_communities.rb b/app/models/dataspace_communities.rb
deleted file mode 100644
index ef48edbc..00000000
--- a/app/models/dataspace_communities.rb
+++ /dev/null
@@ -1,135 +0,0 @@
-# frozen_string_literal: true
-
-require "httparty"
-require "json"
-
-# Fetches information about DataSpace Communities (and collections)
-# and handles the nested structure of Communities.
-#
-# rubocop:disable Style/Next
-class DataspaceCommunities
-  attr_reader :tree
-
-  # @param filename [<String>] File name with communities information cached from DataSpace (used for testing).
-  # Fetches data from DataSpace directly when no filename is provided.
-  def initialize(filename = nil)
-    @tree = []
-    @flat_list = nil
-    if filename.blank?
-      load_from_dataspace
-    else
-      load_from_file(filename)
-    end
-  end
-
-  # Returns community information for a given ID
-  # @param id [<Int>] ID of the community.
-  def find_by_id(id)
-    flat_list.find { |community| community.id == id }
-  end
-
-  # Returns the name of the root community for given community ID.
-  # @param id [<Int>] ID of the community.
-  def find_root_name(id)
-    root_id = find_path_ids(id, []).last
-    found = find_by_id(root_id)
-    return if found.nil?
-
-    found.name
-  end
-
-  # Returns an array with the names (from root to sub-community) to the given community ID.
-  # @param id [<Int>] ID of the community.
-  def find_path_name(id)
-    ids = find_path(id)
-    ids.map { |path_id| find_by_id(path_id).name }
-  end
-
-  private
-
-  # Loads community information straight from Dataspace API
-  # See DSpace API reference: https://dataspace.princeton.edu/rest/
-  def load_from_dataspace
-    @tree = []
-    communities_url = "#{Rails.configuration.pdc_discovery.dataspace_url}/rest/communities?expand=all"
-    Rails.logger.info "Fetching communities information from #{communities_url}"
-    response = HTTParty.get(communities_url)
-    d_communities = JSON.parse(response.body)
-    d_communities.each do |d_community|
-      root_community = d_community['parentCommunity'].nil?
-      if root_community
-        node = DataspaceCommunity.new(d_community, true)
-        @tree << node
-      end
-    end
-    @tree
-  end
-
-  # Loads community information from a pre-saved file with the information
-  def load_from_file(filename)
-    @tree = []
-    Rails.logger.info "Loading communities information from #{filename}"
-    content = File.read(filename)
-    d_communities = JSON.parse(content)
-    d_communities.each do |d_community|
-      root_community = d_community['parentCommunity'].nil?
-      if root_community
-        node = DataspaceCommunity.new(d_community, false)
-        @tree << node
-      end
-    end
-    @tree
-  end
-
-  # Returns an array with all the DataspaceCommunity as a flat array.
-  # This array is used inernally to perform searches by ID (it's faster to search a flat array
-  # than to search a nested structure.)
-  def flat_list
-    @flat_list ||= begin
-      nodes = []
-      @tree.each do |node|
-        nodes += flat_node(node)
-      end
-      nodes
-    end
-  end
-
-  # Returns an array of all the communities and sub-communities for a given community.
-  # @param community [<DataspaceCommunity>] a community object.
-  def flat_node(community)
-    list = [community]
-    community.subcommunities.each do |sub|
-      list += flat_node(sub)
-    end
-    list
-  end
-
-  # Returns the path (from root to sub-community) to the community as an array of IDs.
-  # @param id [<Int>] ID of the community.
-  def find_path(id)
-    find_path_ids(id, []).reverse
-  end
-
-  # Returns the ID of the root community for a given community.
-  # @param id [<Int>] ID of the community.
-  # @note No testing coverage but not called and private
-  def find_root(id)
-    find_path_ids(id, []).last
-  end
-
-  # Returns the path (community to parent to grandparent) to the community as an array of IDs.
-  # @param id [<Int>] ID of the community.
-  # @param path [Array<Int>] Array of IDs.
-  def find_path_ids(id, path)
-    community = find_by_id(id)
-    return [] if community.nil?
-
-    path << id
-    if community.parent_id.nil?
-      path
-    else
-      find_path_ids(community.parent_id, path)
-    end
-  end
-end
-# rubocop:enable Style/Next
diff --git a/app/models/dataspace_community.rb b/app/models/dataspace_community.rb
deleted file mode 100644
index d825eb57..00000000
--- a/app/models/dataspace_community.rb
+++ /dev/null
@@ -1,57 +0,0 @@
-# frozen_string_literal: true
-
-require "httparty"
-require "json"
-
-# rubocop:disable Rails/Delegate
-class DataspaceCommunity
-  attr_accessor :id, :name, :handle, :collections, :subcommunities, :parent_id
-
-  # d_community is a hash with DSpace community information
-  def initialize(d_community, fetch_from_dataspace = false)
-    @id = d_community['id']
-    @name = d_community['name']
-    @handle = d_community['handle']
-    @parent_id = fetch_from_dataspace ? d_community.dig('parentCommunity', 'id') : d_community['parent_id']
-
-    @collections = []
-    d_community['collections'].each do |d_collection|
-      @collections << { id: d_collection['id'], name: d_collection['name'] }
-    end
-
-    @subcommunities = []
-    d_community['subcommunities'].each do |d_sub_community|
-      if fetch_from_dataspace
-        # Fetch data from dataspace since the subcommunity information does not come on the community response
-        # by default.
-        sub_community_url = "#{Rails.configuration.pdc_discovery.dataspace_url}/rest/communities/#{d_sub_community['id']}?expand=all"
-        response = HTTParty.get(sub_community_url)
-        d_sub_community = JSON.parse(response.body)
-        @subcommunities << DataspaceCommunity.new(d_sub_community, true)
-      else
-        # Use the data that we got as-is.
-        # This is used when building the nodes from cache where we have all the information already available.
-        @subcommunities << DataspaceCommunity.new(d_sub_community, false)
-      end
-    end
-  end
-
-  # @note Not covered in testing but not called
-  def to_hash
-    hash = {
-      id: @id,
-      name: @name,
-      handle: @handle,
-      collections: @collections,
-      subcommunities: @subcommunities,
-      parent_id: @parent_id
-    }
-    hash
-  end
-
-  # @note Not covered in testing but not called
-  def to_json(opts)
-    to_hash.to_json(opts)
-  end
-end
-# rubocop:enable Rails/Delegate
diff --git a/spec/lib/research_data_collection_spec.rb b/spec/lib/research_data_collection_spec.rb
deleted file mode 100644
index bfadfbdb..00000000
--- a/spec/lib/research_data_collection_spec.rb
+++ /dev/null
@@ -1,26 +0,0 @@
-# frozen_string_literal: true
-
-RSpec.describe ResearchDataCollection do
-  let(:csv_data) do
-    {
-      "ParentCommunity" => "Princeton Plasma Physics Laboratory",
-      "Community" => "Advanced Projects",
-      "CollectionName" => "Socio Economic",
-      "Handle" => "88435/dsp01sf268746p",
-      "CollectionID" => "1305",
-      "ItemCount" => "5",
-      nil => nil
-    }
-  end
-  let(:csv_row) { CSV::Row.new(csv_data.keys, csv_data.values) }
-  let(:rdc) { described_class.new(csv_row) }
-
-  it "takes a CSV::Row as an argument" do
-    expect(rdc.parent_community).to eq "Princeton Plasma Physics Laboratory"
-    expect(rdc.community).to eq "Advanced Projects"
-    expect(rdc.collection_name).to eq "Socio Economic"
-    expect(rdc.handle).to eq "88435/dsp01sf268746p"
-    expect(rdc.collection_id).to eq "1305"
-    expect(rdc.item_count).to eq 5
-  end
-end
diff --git a/spec/models/dataspace_communities_spec.rb b/spec/models/dataspace_communities_spec.rb
deleted file mode 100644
index f495b2eb..00000000
--- a/spec/models/dataspace_communities_spec.rb
+++ /dev/null
@@ -1,44 +0,0 @@
-# frozen_string_literal: true
-
-require 'rails_helper'
-
-RSpec.describe DataspaceCommunities do
-  let(:communities) { described_class.new('./spec/fixtures/files/dataspace_communities.json') }
-  let(:communities_blank) { described_class.new('') }
-
-  describe "#find_by_id" do
-    it "finds a root community" do
-      astrophysical = communities.find_by_id(186)
-      expect(astrophysical.name).to eq "Astrophysical Sciences"
-      expect(astrophysical.parent_id).to be nil
-      expect(astrophysical.collections.count).to eq 1
-    end
-
-    it "finds a subcommunity" do
-      adv_projects = communities.find_by_id(347)
-      expect(adv_projects.name).to eq "Advanced Projects"
-      expect(adv_projects.parent_id).to eq 346
-    end
-  end
-
-  describe "#find_root_name" do
-    it "reports root name for a root community" do
-      # We expect root communities to report their name as root.
-      pppl_id = 346
-      expect(communities.find_root_name(pppl_id)).to eq "Princeton Plasma Physics Laboratory"
-    end
-    it "reports root name for a subcommunity" do
-      pppl_adv_projects_id = 347
-      expect(communities.find_root_name(pppl_adv_projects_id)).to eq "Princeton Plasma Physics Laboratory"
-    end
-  end
-
-  describe "#find_path_name" do
-    it "finds path for a root community" do
-      expect(communities.find_path_name(346)).to eq ["Princeton Plasma Physics Laboratory"]
-    end
-    it "finds path for a subcommunity" do
-      expect(communities.find_path_name(347)).to eq ["Princeton Plasma Physics Laboratory", "Advanced Projects"]
-    end
-  end
-end