tc211-termbase 0.2.1 → 0.2.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2a232b33a8293be27f9fac20557b959345339a8835faa56cf31cbf75df849ac3
-  data.tar.gz: b8baf017786cd822b711816e2e015c7b5b5e4e63bd01b6b34d81eb4806974932
+  metadata.gz: 423950f471ef5d9c689a3542150d988fc997489a12b4ffe3668520d84bf6561f
+  data.tar.gz: f32b9f74dd1bb3310338785e58eb3488ba88656fa859fc3816e4615d9bad65a1
 SHA512:
-  metadata.gz: 432adaba8dea3489f330e15d1e58a28c1843b936a944d6d86d5602401ccaf82f840f9646e64e6eecd88eb8ab289f23b30df68b244406da5227eff7aaa9fa1afe
-  data.tar.gz: 7579e6eecf194ec07063a6e6e13b79edec56a9baaba0c066d1e3aa84abb58126560cc48a861401acb4ae8f518bf442f60bfe484c00e6fc4fbbc5fa8931a82a1a
+  metadata.gz: d4bb9c3b774042e3fe3674d30f1eb8211be2e0ac26f4834ee53646056d0b2fae276f9011776df5e7c36eae2bc450dda178ec32d03d0db9336eff50a55d16dd02
+  data.tar.gz: 2dd00314060b0984f353112988ecf931310ab41f8b9a2737a5e9e55fcc8756f321a24085a7823fb3dd3a6e00c1c7357b84b4abe26bcc2741b2f38f78bb22d063

data/.github/workflows/rake.yml

@@ -0,0 +1,13 @@
+# Auto-generated by Cimas: Do not edit it manually!
+# See https://github.com/metanorma/cimas
+name: rake
+
+on:
+  push:
+    branches: [ master, main ]
+    tags: [ v* ]
+  pull_request:
+
+jobs:
+  rake:
+    uses: geolexica/ci/.github/workflows/rake.yml@main

data/.github/workflows/release.yml

@@ -0,0 +1,23 @@
+# Auto-generated by Cimas: Do not edit it manually!
+# See https://github.com/metanorma/cimas
+name: release
+
+on:
+  workflow_dispatch:
+    inputs:
+      next_version:
+        description: |
+          Next release version. Possible values: x.y.z, major, minor, patch or pre|rc|etc
+        required: true
+        default: 'skip'
+  repository_dispatch:
+    types: [ do-release ]
+
+
+jobs:
+  release:
+    uses: geolexica/ci/.github/workflows/release.yml@main
+    with:
+      next_version: ${{ github.event.inputs.next_version }}
+    secrets:
+      rubygems-api-key: ${{ secrets.GEOLEXICA_RUBYGEMS_API_KEY }}

data/.gitignore

@@ -12,4 +12,6 @@
 .rspec_status
 .vscode/
 concepts/
-termbase.yaml
+/termbase.yaml
+/termbase.meta.yaml
+/Gemfile.lock

data/exe/tc211-termbase-xlsx2yaml

@@ -63,15 +63,12 @@ end
 
 # collection[1206].inspect
 
-collection.to_file(File.join(output_dir, Pathname.new(filepath).basename.sub_ext(".yaml")))
-
 collection_output_dir = File.join(output_dir, "concepts")
 
 FileUtils.mkdir_p(collection_output_dir)
 
-collection.keys.each do |id|
-  collection[id].to_file(File.join(collection_output_dir, "concept-#{id}.yaml"))
-end
+concept_collection = collection.to_concept_collection
+concept_collection.save_to_files(collection_output_dir)
 
 # french = workbook.language_sheet("French")
 # french.sections[3].structure

data/lib/tc211/termbase/concept.rb

@@ -1,56 +1,105 @@
 module Tc211::Termbase
+  class Concept < Hash
+    attr_reader :id
 
-class Concept < Hash
-  attr_accessor :id
-  attr_accessor :terms
-  DEFAULT_LANGUAGE = "eng"
+    DEFAULT_LANGUAGE = "eng".freeze
 
-  def initialize(options={})
-    terms = options.delete(:terms) || []
-    terms.each do |term|
-      add_term(term)
+    def initialize(options = {})
+      super
+
+      terms = options.delete(:terms) || []
+      terms.each do |term|
+        add_term(term)
+      end
+
+      options.each_pair do |k, v|
+        send("#{k}=", v)
+      end
     end
 
-    options.each_pair do |k,v|
-      self.send("#{k}=", v)
+    # The concept id should ALWAYS be an integer.
+    # https://github.com/riboseinc/tc211-termbase/issues/1
+    def id=(newid)
+      @id = Integer(newid)
     end
-  end
 
-  # The concept id should ALWAYS be an integer.
-  # https://github.com/riboseinc/tc211-termbase/issues/1
-  def id=(newid)
-    @id = Integer(newid)
-  end
+    def add_term(term)
+      self[term.language_code] = term
+    end
 
-  def add_term(term)
-    self[term.language_code] = term
-  end
+    def terms
+      values
+    end
 
-  def default_term
-    if self[DEFAULT_LANGUAGE]
-      self[DEFAULT_LANGUAGE]
-    else
-      puts "[tc211-termbase] term (lang: #{keys.first}, ID: #{id}) is missing a corresponding English term, probably needs updating."
-      self[keys.first]
+    def default_term
+      if self[DEFAULT_LANGUAGE]
+        self[DEFAULT_LANGUAGE]
+      else
+        puts "[tc211-termbase] term (lang: #{keys.first}, ID: #{id}) is \
+          missing a corresponding English term, probably needs updating."
+        self[keys.first]
+      end
     end
-  end
 
-  def to_hash
-    default_hash = {
-      "term" => default_term.term,
-      "termid" => id
-    }
+    def to_hash
+      default_hash = {
+        "term" => default_term.term,
+        "termid" => id,
+      }
 
-    self.inject(default_hash) do |acc, (lang, term)|
-      acc.merge!(lang => term.to_hash)
+      inject(default_hash) do |acc, (lang, term)|
+        acc.merge!(lang => term.to_hash)
+      end
     end
-  end
 
-  def to_file(filename)
-    File.open(filename,"w") do |file|
-      file.write(to_hash.to_yaml)
+    def to_file(filename)
+      File.open(filename, "w") do |file|
+        file.write(to_hash.to_yaml)
+      end
     end
-  end
 
+    def to_glossarist_concept
+      concept = Glossarist::ManagedConcept.new(data: { id: id.to_s })
+
+      localized_concepts = []
+
+      terms.map do |term|
+        next if term.nil?
+
+        localized_concepts << term.to_localized_concept_hash
+      end
+
+      concept.localized_concepts = localized_concepts
+
+      concept
+    end
+  end
 end
-end
+
+# term: abbreviation
+# termid: 2
+# eng:
+#   id: 2
+#   term: abbreviation
+#   definition: designation formed by omitting words or letters from a longer
+#     form and designating the same concept
+#   language_code: eng
+#   notes: []
+#   examples: []
+#   entry_status: valid
+#   classification: preferred
+#   authoritative_source:
+#     ref: ISO 1087-1:2000
+#     clause: 3.4.9
+#     link: https://www.iso.org/standard/20057.html
+#   lineage_source: ISO/TS 19104:2008
+#   lineage_source_similarity: 1
+#   date_accepted: 2008-11-15 00:00:00.000000000 +08:00
+#   review_date: 2013-01-29 00:00:00.000000000 +08:00
+#   review_status: final
+#   review_decision: accepted
+#   review_decision_date: 2016-10-01 00:00:00.000000000 +08:00
+#   review_decision_event: Publication of ISO 19104:2016
+#   review_decision_notes: Authoritative reference changed from ISO 1087-1:2000
+#     to ISO 1087-1:2000, 3.4.9. Lineage source added as ISO/TS 19104:2008
+#   release: '2'

data/lib/tc211/termbase/concept_collection.rb

@@ -1,32 +1,40 @@
 require_relative "concept"
 
 module Tc211::Termbase
+  class ConceptCollection < Hash
+    def add_term(term)
+      if self[term.id]
+        self[term.id].add_term(term)
+      else
+        self[term.id] = Concept.new(
+          id: term.id,
+          terms: [term],
+        )
+      end
+    end
 
-class ConceptCollection < Hash
-
-  def add_term(term)
-    if self[term.id]
-      self[term.id].add_term(term)
-    else
-      self[term.id] = Concept.new(
-        id: term.id,
-        terms: [term]
-      )
+    def to_hash
+      inject({}) do |acc, (id, concept)|
+        acc.merge!(id => concept.to_hash)
+      end
     end
-  end
 
-  def to_hash
-    self.inject({}) do |acc, (id, concept)|
-      acc.merge!(id => concept.to_hash)
+    def to_file(filename)
+      File.open(filename, "w") do |file|
+        file.write(to_hash.to_yaml)
+      end
     end
-  end
 
-  def to_file(filename)
-    File.open(filename,"w") do |file|
-      file.write(to_hash.to_yaml)
+    def to_concept_collection
+      collection = Glossarist::ManagedConceptCollection.new
+
+      values.each do |term_concept|
+        next if term_concept.nil?
+
+        collection.store(term_concept.to_glossarist_concept)
+      end
+
+      collection
     end
   end
-
 end
-
-end

data/lib/tc211/termbase/information_sheet.rb

@@ -1,22 +1,18 @@
 require_relative "terminology_sheet"
 
 module Tc211::Termbase
+  class InformationSheet < TerminologySheet
+    def metadata_section
+      sheet_array = @sheet.simple_rows.to_a
+      MetadataSection.new(sheet_array)
+    end
 
-class InformationSheet < TerminologySheet
+    def to_hash
+      { "glossary" => metadata_section.to_hash }
+    end
 
-  def metadata_section
-    sheet_array = @sheet.simple_rows.to_a
-    section = MetadataSection.new(sheet_array)
+    def to_yaml
+      to_hash.to_yaml
+    end
   end
-
-  def to_hash
-    { "glossary" => metadata_section.to_hash }
-  end
-
-  def to_yaml
-    to_hash.to_yaml
-  end
-
 end
-
-end

data/lib/tc211/termbase/metadata_section.rb

@@ -1,136 +1,132 @@
 require_relative "sheet_section"
 
 module Tc211::Termbase
+  class MetadataSection < SheetSection
+    attr_accessor :header_row
+    attr_writer :attributes
+
+    GLOSSARY_HEADER_ROW_MATCH = {
+      # "English" uses "".
+      # "Arabic" uses "A".
+      # This is fixed in the MLGT as of 2018 Aug 6.
+      "A" => [nil, "Item", "A"],
+
+      "C" => ["Data Type"],
+      "D" => ["Special Instruction"],
+
+      # "Malay" has it empty ("")
+      # This is fixed in the MLGT as of 2018 Aug 6.
+      "E" => ["ISO 19135 Class.attribute", nil],
+
+      "F" => ["Domain"],
+    }.freeze
+
+    GLOSSARY_ROW_KEY_MAP = {
+      "A" => "name",
+      "B" => "value",
+      "C" => "datatype",
+      "D" => "special-instruction",
+      "E" => "19135-class-attribute",
+      "F" => "value-domain",
+    }.freeze
+
+    def initialize(rows, options = {})
+      super
+
+      self.class.match_header(@rows[0])
+      @header_row = @rows[0]
+      @body_rows = @rows[1..-1]
+      attributes
+      self
+    end
 
-class MetadataSection < SheetSection
-  attr_accessor :header_row
-  attr_accessor :attributes
-
-  GLOSSARY_HEADER_ROW_MATCH = {
-    # "English" uses "".
-    # "Arabic" uses "A".
-    # This is fixed in the MLGT as of 2018 Aug 6.
-    "A" => [nil, "Item", "A"],
-
-    "C" => ["Data Type"],
-    "D" => ["Special Instruction"],
-
-    # "Malay" has it empty ("")
-    # This is fixed in the MLGT as of 2018 Aug 6.
-    "E" => ["ISO 19135 Class.attribute", nil],
-
-    "F" => ["Domain"]
-  }
-
-  GLOSSARY_ROW_KEY_MAP = {
-    "A" => "name",
-    "B" => "value",
-    "C" => "datatype",
-    "D" => "special-instruction",
-    "E" => "19135-class-attribute",
-    "F" => "value-domain"
-  }
-
-  def initialize(rows, options={})
-    super
-
-    self.class.match_header(@rows[0])
-    @header_row = @rows[0]
-    @body_rows = @rows[1..-1]
-    attributes
-    self
-  end
-
-  def self.match_header(columns)
-    # puts "row #{row}"
-    columns.each do |key, value|
-      # puts "#{key}, #{value}"
-      if GLOSSARY_HEADER_ROW_MATCH[key]
-        unless GLOSSARY_HEADER_ROW_MATCH[key].include?(value)
-          raise RowHeaderMatchError.new("Metadata section header for column `#{key}` does not match expected value `#{value}`")
+    def self.match_header(columns)
+      # puts "row #{row}"
+      columns.each do |key, value|
+        # puts "#{key}, #{value}"
+        header_row_match = GLOSSARY_HEADER_ROW_MATCH[key]
+        if header_row_match && !header_row_match.include?(value)
+          raise RowHeaderMatchError.new(
+            "Metadata section header for column `#{key}` does not match \
+            expected value `#{value}`",
+          )
         end
       end
     end
-  end
-
 
-  def structure
-    GLOSSARY_ROW_KEY_MAP
-  end
+    def structure
+      GLOSSARY_ROW_KEY_MAP
+    end
 
-  def clean_key(k)
-    k.strip.
-      downcase.
-      gsub(/[()]/,"").
-      gsub(' ', '-')
-  end
+    def clean_key(key)
+      key.strip
+        .downcase
+        .gsub(/[()]/, "")
+        .gsub(" ", "-")
+    end
 
-  def clean_value(v)
-    return nil if v.nil?
+    def clean_value(value)
+      return nil if value.nil?
 
-    case v
-    when String
-      v.strip
-    else
-      v
+      case value
+      when String
+        value.strip
+      else
+        value
+      end
     end
-  end
 
-  def parse_row(row)
-    return nil if row.empty?
-    attribute = {}
+    def parse_row(row)
+      return nil if row.empty?
 
-    structure.each_pair do |key, value|
-      # puts"#{key}, #{value}, #{row[key]}"
-      attribute_key = clean_key(value)
-      attribute_value = clean_value(row[key])
-      next if attribute_value.nil?
-      attribute[attribute_key] = attribute_value
-    end
+      attribute = {}
 
-    # TODO: "Chinese" name is empty!
-    key = clean_key(attribute["name"] || "(empty)")
+      structure.each_pair do |key, value|
+        attribute_value = clean_value(row[key])
+        next if attribute_value.nil?
 
-    { key => attribute }
-  end
+        attribute[clean_key(value)] = attribute_value
+      end
 
-  def attributes
-    return @attributes if @attributes
+      # TODO: "Chinese" name is empty!
+      key = clean_key(attribute["name"] || "(empty)")
 
-    @attributes = {}
-    @body_rows.each do |row|
-      result = parse_row(row)
-      @attributes.merge!(result) if result
+      { key => attribute }
     end
-    @attributes
-  end
 
-  def fields
-
-    # "operating-language-country"=>
-    #  {"name"=>"Operating Language Country",
-    #   "value"=>"410",
-    #   "datatype"=>"Country Code",
-    #   "special-instruction"=>
-    #    "ftp.ics.uci.edu/pub/ietf/http/related/iso3166.txt",
-    #   "19135-class-attribute"=>"RE_Register.operatingLanguage",
-    #   "value-domain"=>
-    #    "<<Data Type>>RE_Locale.country \n" +
-    #    "[ISO 3166-1 3-character numeric country code]"},
-    #
-
-    attributes.inject({}) do |acc, (k, v)|
-      acc.merge({
-        k => v["value"]
-      })
+    def attributes
+      return @attributes if @attributes
+
+      @attributes = {}
+      @body_rows.each do |row|
+        result = parse_row(row)
+        @attributes.merge!(result) if result
+      end
+      @attributes
     end
-  end
 
-  def to_hash
-    {
-      "metadata" => fields
-    }
-  end
+    def fields
+      # "operating-language-country"=>
+      #  {"name"=>"Operating Language Country",
+      #   "value"=>"410",
+      #   "datatype"=>"Country Code",
+      #   "special-instruction"=>
+      #    "ftp.ics.uci.edu/pub/ietf/http/related/iso3166.txt",
+      #   "19135-class-attribute"=>"RE_Register.operatingLanguage",
+      #   "value-domain"=>
+      #    "<<Data Type>>RE_Locale.country \n" +
+      #    "[ISO 3166-1 3-character numeric country code]"},
+      #
+
+      attributes.inject({}) do |acc, (k, v)|
+        acc.merge({ k => v["value"] })
+      end
+    end
 
-end
+    def to_hash
+      {
+        "metadata" => fields,
+      }
+    end
+  end
 end

data/lib/tc211/termbase/sheet_section.rb

@@ -1,30 +1,29 @@
 
 module Tc211::Termbase
+  class SheetSection
 
-class SheetSection
+    class RowHeaderMatchError < StandardError; end
 
-  class RowHeaderMatchError < StandardError; end
-  class UnknownHeaderError < StandardError; end
+    class UnknownHeaderError < StandardError; end
 
-  attr_accessor :sheet_content
+    attr_accessor :sheet_content
 
-  def initialize(rows, options={})
-    # rows is an array of rows!
-    raise unless rows.is_a?(Array)
-    @rows = rows
-    # @has_header = options[:has_header].nil? ? true : options[:has_header]
-    self
-  end
+    def initialize(rows, _options = {})
+      # rows is an array of rows!
+      raise unless rows.is_a?(Array)
 
-  # Abstract method
-  def self.match_header(row)
-    false
-  end
+      @rows = rows
+      # @has_header = options[:has_header].nil? ? true : options[:has_header]
+      self
+    end
 
-  def self.identify_type(row)
+    # Abstract method
+    def self.match_header(_row)
+      false
+    end
 
-  end
+    def self.identify_type(_row); end
 
-  # TODO
+    # TODO
+  end
 end
-end