Adding adapter for reading glossarist model V1 concepts

lib/glossarist.rb

@@ -26,6 +26,7 @@
 require_relative "glossarist/concept_manager"
 require_relative "glossarist/managed_concept"
 require_relative "glossarist/non_verb_rep"
+require_relative "glossarist/v1_reader"
 
 require_relative "glossarist/collections"
 

lib/glossarist/concept_manager.rb

@@ -17,11 +17,11 @@ def load_from_files(collection: nil)
       collection ||= ManagedConceptCollection.new
 
       Dir.glob(concepts_glob) do |filename|
-        concept = load_concept_from_file(filename)
-        concept.localized_concepts.each do |_lang, id|
-          localized_concept = load_localized_concept(id)
-          concept.add_l10n(localized_concept)
-        end
+        concept = if v1_collection?
+                    Glossarist::V1Reader.load_concept_from_file(filename)
+                  else
+                    load_concept_from_file(filename)
+                  end
 
         collection.store(concept)
       end
@@ -35,7 +35,14 @@ def save_to_files(managed_concepts)
     def load_concept_from_file(filename)
       concept_hash = Psych.safe_load(File.read(filename), permitted_classes: [Date])
       concept_hash["uuid"] = concept_hash["id"] || File.basename(filename, ".*")
-      ManagedConcept.new(concept_hash)
+
+      concept = ManagedConcept.new(concept_hash)
+      concept.localized_concepts.each do |_lang, id|
+        localized_concept = load_localized_concept(id)
+        concept.add_l10n(localized_concept)
+      end
+
+      concept
     rescue Psych::SyntaxError => e
       raise Glossarist::ParseError.new(filename: filename, line: e.line)
     end
@@ -71,11 +78,19 @@ def save_concept_to_file(concept)
     private
 
     def concepts_glob
-      File.join(path, "concept", "*.{yaml,yml}")
+      if v1_collection?
+        File.join(path, "concept-*.{yaml,yml}")
+      else
+        File.join(path, "concept", "*.{yaml,yml}")
+      end
     end
 
     def localized_concept_path(id)
       Dir.glob(File.join(path, "localized_concept", "#{id}.{yaml,yml}"))&.first
     end
+
+    def v1_collection?
+      @v1_collection ||= !Dir.glob(File.join(path, "concept-*.{yaml,yml}")).empty?
+    end
   end
 end

lib/glossarist/v1_reader.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+module Glossarist
+  # An adapter to read concepts in V1 format, converts them to v2 format and
+  # load into glossarist concept model.
+  class V1Reader
+    def self.load_concept_from_file(filename)
+      new.load_concept_from_file(filename)
+    end
+
+    def load_concept_from_file(filename)
+      concept_hash = Psych.safe_load(File.read(filename), permitted_classes: [Date])
+      ManagedConcept.new(generate_v2_concept_hash(concept_hash))
+    end
+
+    private
+
+    def generate_v2_concept_hash(concept_hash)
+      v2_concept = { "groups" => concept_hash["groups"] }
+      v2_concept["data"] = {
+        "identifier" => concept_hash["termid"],
+        "localized_concepts" => concept_hash.values.grep(Hash),
+      }
+
+      v2_concept
+    end
+  end
+end

spec/features/v1_serialization_spec.rb

@@ -0,0 +1,44 @@
+RSpec.describe "Serialization and deserialization" do
+  let(:concept_folder) { "concept_collection_v1" }
+  let(:output_folder) { File.join("concept_collection_v1", "output") }
+  let(:concept_files) { Dir.glob(File.join(fixtures_path(output_folder), "concept-*.{yaml,yml}")) }
+
+  it "correctly loads concepts from files" do
+    collection = Glossarist::ManagedConceptCollection.new
+    collection.load_from_files(fixtures_path(concept_folder))
+
+    concept_files.each do |filename|
+      concept_from_file = load_yaml_file(filename)
+      concept = collection[concept_from_file["data"]["identifier"]]
+
+      expect(concept.to_h["data"]).to eq(concept_from_file["data"])
+
+      concept.localized_concepts.each do |lang, id|
+        localized_concept_path = File.join(localized_concepts_folder, "#{id}.yaml")
+        localized_concept = load_yaml_file(localized_concept_path)
+
+        expect(localized_concept["data"]).to eq(concept.localizations[lang].to_h["data"])
+      end
+    end
+
+    Dir.mktmpdir do |tmp_path|
+      collection.save_to_files(tmp_path)
+
+      # check if concept and localized_concept folder exist
+      system "diff", fixtures_path(output_folder), tmp_path
+      expect($?.exitstatus).to eq(0) # no difference
+
+      # check content of conecept folder
+      system "diff", File.join(fixtures_path(output_folder), "concept"), File.join(tmp_path, "concept")
+      expect($?.exitstatus).to eq(0) # no difference
+
+      # check content of localized_conecept folder
+      system "diff", File.join(fixtures_path(output_folder), "localized_concept"), File.join(tmp_path, "localized_concept")
+      expect($?.exitstatus).to eq(0) # no difference
+    end
+  end
+
+  def load_yaml_file(filename)
+    Psych.safe_load(File.read(filename), permitted_classes: [Date])
+  end
+end

spec/fixtures/concept_collection_v1/concept-3.1.1.1.yaml

@@ -0,0 +1,26 @@
+---
+termid: 3.1.1.1
+term: entity
+groups:
+- foo
+- bar
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: entity
+  definition:
+  - content: concrete or abstract thing that exists, did exist, or can possibly exist,
+      including associations among these things
+  notes: []
+  examples:
+  - content: "{{person,Person}}, object, event, idea,
+      process, etc."
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.1
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/concept-3.1.1.2.yaml

@@ -0,0 +1,22 @@
+---
+termid: 3.1.1.2
+term: immaterial_entity
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: immaterial_entity
+  definition:
+  - content: "{{entity}} that does not occupy three-dimensional
+      space"
+  notes: []
+  examples:
+  - content: Idea, process, organization, etc.
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.2
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/concept-3.1.1.3.yaml

@@ -0,0 +1,23 @@
+---
+termid: 3.1.1.3
+term: material entity
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: material entity
+  definition:
+  - content: "{{entity}} that occupies three-dimensional
+      space"
+  notes:
+  - content: All material entities have certain characteristics that can be described
+      and therefore this concept is important for ontology purposes.
+  examples: []
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.3
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/concept-3.1.1.4.yaml

@@ -0,0 +1,21 @@
+---
+termid: 3.1.1.4
+term: non-biological entity
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: non-biological entity
+  definition:
+  - content: "{{material entity}} that is not and has
+      never been a living organism"
+  notes: []
+  examples: []
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.4
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/concept-3.1.1.5.yaml

@@ -0,0 +1,21 @@
+---
+termid: 3.1.1.5
+term: biological_entity
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: biological_entity
+  definition:
+  - content: "{{material entity}} that was or is a living
+      organism"
+  notes: []
+  examples: []
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.5
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/concept-3.1.1.6.yaml

@@ -0,0 +1,21 @@
+---
+termid: 3.1.1.6
+term: person
+eng:
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: person
+  definition:
+  - content: "{{biological_entity}} that is a human
+      being"
+  notes: []
+  examples: []
+  language_code: eng
+  entry_status: valid
+  sources:
+  - type: authoritative
+    origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.6
+      link: https://www.iso.org/standard/79779.html

spec/fixtures/concept_collection_v1/output/concept/2b1a0687-8c98-5780-98fe-211effd00338.yaml

@@ -0,0 +1,8 @@
+---
+data:
+  identifier: 3.1.1.1
+  localized_concepts:
+    eng: ae27e36e-858c-5026-9dc7-fa366c464f01
+  groups:
+  - foo
+  - bar

spec/fixtures/concept_collection_v1/output/concept/3c910f05-0d12-5a34-8cc0-10e0fe0a133e.yaml

@@ -0,0 +1,5 @@
+---
+data:
+  identifier: 3.1.1.4
+  localized_concepts:
+    eng: 5bf347a5-f1e5-5c33-b6d3-9cc74a8f6e64

spec/fixtures/concept_collection_v1/output/concept/56e21ba2-3f0f-545e-8fb6-ae1bbf3dd3f0.yaml

@@ -0,0 +1,5 @@
+---
+data:
+  identifier: 3.1.1.6
+  localized_concepts:
+    eng: cb1e4bff-262c-5a1c-98b1-10e9e9b21ba5

spec/fixtures/concept_collection_v1/output/concept/668152d7-206d-51a1-8a06-dfd25da33264.yaml

@@ -0,0 +1,5 @@
+---
+data:
+  identifier: 3.1.1.5
+  localized_concepts:
+    eng: 87d81109-7a89-53a4-af7c-92f99bd07a71

spec/fixtures/concept_collection_v1/output/concept/73fb6812-b8b4-57d4-bb77-985e3a2d1a26.yaml

@@ -0,0 +1,5 @@
+---
+data:
+  identifier: 3.1.1.3
+  localized_concepts:
+    eng: d2bd44d8-ef32-5695-9eab-ef82053e0d1d

spec/fixtures/concept_collection_v1/output/concept/a67f6427-43a4-57d9-9b62-87716634a4b0.yaml

@@ -0,0 +1,5 @@
+---
+data:
+  identifier: 3.1.1.2
+  localized_concepts:
+    eng: 48f1340a-517a-5533-ae55-99b304f0f8c9

spec/fixtures/concept_collection_v1/output/localized_concept/48f1340a-517a-5533-ae55-99b304f0f8c9.yaml

@@ -0,0 +1,20 @@
+---
+data:
+  dates: []
+  definition:
+  - content: "{{entity}} that does not occupy three-dimensional space"
+  examples:
+  - content: Idea, process, organization, etc.
+  notes: []
+  sources:
+  - origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.2
+      link: https://www.iso.org/standard/79779.html
+    type: authoritative
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: immaterial_entity
+  language_code: eng
+  entry_status: valid

spec/fixtures/concept_collection_v1/output/localized_concept/5bf347a5-f1e5-5c33-b6d3-9cc74a8f6e64.yaml

@@ -0,0 +1,19 @@
+---
+data:
+  dates: []
+  definition:
+  - content: "{{material entity}} that is not and has never been a living organism"
+  examples: []
+  notes: []
+  sources:
+  - origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.4
+      link: https://www.iso.org/standard/79779.html
+    type: authoritative
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: non-biological entity
+  language_code: eng
+  entry_status: valid

spec/fixtures/concept_collection_v1/output/localized_concept/87d81109-7a89-53a4-af7c-92f99bd07a71.yaml

@@ -0,0 +1,19 @@
+---
+data:
+  dates: []
+  definition:
+  - content: "{{material entity}} that was or is a living organism"
+  examples: []
+  notes: []
+  sources:
+  - origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.5
+      link: https://www.iso.org/standard/79779.html
+    type: authoritative
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: biological_entity
+  language_code: eng
+  entry_status: valid

spec/fixtures/concept_collection_v1/output/localized_concept/ae27e36e-858c-5026-9dc7-fa366c464f01.yaml

@@ -0,0 +1,21 @@
+---
+data:
+  dates: []
+  definition:
+  - content: concrete or abstract thing that exists, did exist, or can possibly exist,
+      including associations among these things
+  examples:
+  - content: "{{person,Person}}, object, event, idea, process, etc."
+  notes: []
+  sources:
+  - origin:
+      ref: ISO/TS 14812:2022
+      clause: 3.1.1.1
+      link: https://www.iso.org/standard/79779.html
+    type: authoritative
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: entity
+  language_code: eng
+  entry_status: valid