cosing 0.2.0 → 0.3.0

data/lib/cosing/annex/base.rb

@@ -12,11 +12,10 @@ module Cosing
       end
 
       def add_rule(params)
-        return unless params[:reference_number]
+        return unless reference = params[:reference_number]
 
-        @rules[params[:reference_number]] = self.class::Rule.new(
-          **params.merge(annex: self.class.name.gsub("::", " "))
-        )
+        annex = self.class.name.gsub("::", " ")
+        @rules[reference] = self.class::Rule.new(**params.merge(annex:))
       end
 
       def lookup(reference_number)
@@ -24,8 +23,8 @@ module Cosing
           reference_number.to_s,
           fuzzy_find(reference_number.to_s)
         )
-      rescue KeyError => e
-        return
+      rescue KeyError
+        nil
       end
 
       private
@@ -33,8 +32,7 @@ module Cosing
       def fuzzy_find(reference_number)
         return @rules[reference_number] if @rules.key?(reference_number)
 
-        candidates = @rules.keys.grep(/#{reference_number}[abcd]/)
-        candidates
+        @rules.keys.grep(/#{reference_number}[abcd]/)
           .map { |candidate| @rules[candidate] }
           .tap do |candidates|
             if candidates.empty?

data/lib/cosing/annex/ii.rb

@@ -6,6 +6,24 @@ module Cosing
       class Rule < Rule
         attribute :inn, Types::String
       end
+
+      def self.load
+        new.tap do |annex|
+          Annex.parse("data/annex.II.csv") do |row|
+            ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :identified_ingredients,
+              split: ";"
+            )
+
+            annex.add_rule(
+              row.merge(
+                identified_ingredients: ingredients.compact
+              )
+            )
+          end
+        end
+      end
     end
   end
 end

data/lib/cosing/annex/iii.rb

@@ -10,6 +10,31 @@ module Cosing
         attribute :other_restrictions, Types::String
         attribute :product_type, Types::String
       end
+
+      def self.load
+        new.tap do |annex|
+          Annex.parse("data/annex.III.csv") do |row|
+            common_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :common_ingredients,
+              split: ";"
+            )
+            identified_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :identified_ingredients,
+              split: ";"
+            )
+
+            annex.add_rule(
+              row.merge(
+                common_ingredients: common_ingredients.compact,
+                identified_ingredients: identified_ingredients.compact,
+                other_restrictions: row[:other]
+              )
+            )
+          end
+        end
+      end
     end
   end
 end

data/lib/cosing/annex/iv.rb

@@ -11,6 +11,25 @@ module Cosing
         attribute :product_type, Types::String
         attribute :wording_of_conditions, Types::String
       end
+
+      def self.load
+        new.tap do |annex|
+          Annex.parse("data/annex.IV.csv") do |row|
+            ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :identified_ingredients,
+              split: ";"
+            )
+
+            annex.add_rule(
+              row.merge(
+                identified_ingredients: ingredients.compact,
+                other_restrictions: row[:other]
+              )
+            )
+          end
+        end
+      end
     end
   end
 end

data/lib/cosing/annex/rule.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Cosing
   module Annex
     class Rule < Dry::Struct

data/lib/cosing/annex/sccs_opinion.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Cosing
   module Annex
     class SccsOpinion < Dry::Struct

data/lib/cosing/annex/v.rb

@@ -10,6 +10,31 @@ module Cosing
         attribute :product_type, Types::String
         attribute :wording_of_conditions, Types::String
       end
+
+      def self.load
+        new.tap do |annex|
+          Annex.parse("data/annex.V.csv") do |row|
+            common_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :common_ingredients,
+              split: ";"
+            )
+            identified_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :identified_ingredients,
+              split: ";"
+            )
+
+            annex.add_rule(
+              row.merge(
+                common_ingredients: common_ingredients.compact,
+                identified_ingredients: identified_ingredients.compact,
+                other_restrictions: row[:other]
+              )
+            )
+          end
+        end
+      end
     end
   end
 end

data/lib/cosing/annex/vi.rb

@@ -11,6 +11,31 @@ module Cosing
         attribute :product_type, Types::String
         attribute :wording_of_conditions, Types::String
       end
+
+      def self.load
+        new.tap do |annex|
+          Annex.parse("data/annex.VI.csv") do |row|
+            common_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :common_ingredients,
+              split: ";"
+            )
+            identified_ingredients = Cosing::Parser.transform_array!(
+              row,
+              key: :identified_ingredients,
+              split: ";"
+            )
+
+            annex.add_rule(
+              row.merge(
+                common_ingredients: common_ingredients.compact,
+                identified_ingredients: identified_ingredients.compact,
+                other_restrictions: row[:other]
+              )
+            )
+          end
+        end
+      end
     end
   end
 end

data/lib/cosing/annex.rb

@@ -33,133 +33,25 @@ module Cosing
         yield row.merge(
           cas_numbers: cas_numbers.compact,
           ec_numbers: ec_numbers.compact,
-          sccs_opinions: sccs_opinions.compact,
+          sccs_opinions: sccs_opinions.compact
         )
       end
     end
 
     def load
-      annex_ii = Annex::II.new.tap do |annex|
-        parse("data/annex.II.csv") do |row|
-          identified_ingredients = transform_array!(
-            row,
-            key: :identified_ingredients,
-            split: ";"
-          )
-
-          annex.add_rule(
-            row.merge(
-              identified_ingredients: identified_ingredients.compact
-            )
-          )
-        end
-      end
-
-      annex_iii = Annex::III.new.tap do |annex|
-        parse("data/annex.III.csv") do |row|
-          common_ingredients = transform_array!(
-            row,
-            key: :common_ingredients,
-            split: ";"
-          )
-          identified_ingredients = transform_array!(
-            row,
-            key: :identified_ingredients,
-            split: ";"
-          )
-
-          annex.add_rule(
-            row.merge(
-              common_ingredients: common_ingredients.compact,
-              identified_ingredients: identified_ingredients.compact,
-              other_restrictions: row[:other]
-            )
-          )
-        end
-      end
-
-      annex_iv = Annex::IV.new.tap do |annex|
-        parse("data/annex.IV.csv") do |row|
-          identified_ingredients = transform_array!(
-            row,
-            key: :identified_ingredients,
-            split: ";"
-          )
-
-          annex.add_rule(
-            row.merge(
-              identified_ingredients: identified_ingredients.compact,
-              other_restrictions: row[:other]
-            )
-          )
-        end
-      end
-
-      annex_v = Annex::V.new.tap do |annex|
-        parse("data/annex.V.csv") do |row|
-          common_ingredients = transform_array!(
-            row,
-            key: :common_ingredients,
-            split: ";"
-          )
-          identified_ingredients = transform_array!(
-            row,
-            key: :identified_ingredients,
-            split: ";"
-          )
-
-          annex.add_rule(
-            row.merge(
-              common_ingredients: common_ingredients.compact,
-              identified_ingredients: identified_ingredients.compact,
-              other_restrictions: row[:other]
-            )
-          )
-        end
-      end
-
-      annex_vi = Annex::VI.new.tap do |annex|
-        parse("data/annex.VI.csv") do |row|
-          common_ingredients = transform_array!(
-            row,
-            key: :common_ingredients,
-            split: ";"
-          )
-          identified_ingredients = transform_array!(
-            row,
-            key: :identified_ingredients,
-            split: ";"
-          )
-
-          annex.add_rule(
-            row.merge(
-              common_ingredients: common_ingredients.compact,
-              identified_ingredients: identified_ingredients.compact,
-              other_restrictions: row[:other]
-            )
-          )
-        end
-      end
-
-      {
-        ii: annex_ii,
-        iii: annex_iii,
-        iv: annex_iv,
-        v: annex_v,
-        vi: annex_vi
-      }
-    end
-
-    def transform_array!(params, key:, split:)
-      params
-        .delete(key)
-        .split(split)
-        .map(&:strip)
-        .reject { |n| n == "-" }
+      ii, iii, iv, v, vi = [
+        Annex::II,
+        Annex::III,
+        Annex::IV,
+        Annex::V,
+        Annex::VI
+      ].map(&:load)
+
+      { ii:, iii:, iv:, v:, vi: }
     end
 
     def build_sccs_opinions(row)
-      transform_array!(
+      Cosing::Parser.transform_array!(
         row,
         key: :sccs_opinions,
         split: ";"
@@ -174,23 +66,23 @@ module Cosing
     end
 
     def build_cas_numbers(row)
-      transform_array!(
+      Cosing::Parser.transform_array!(
         row,
         key: :cas_number,
         split: "/"
       ).map do |cas_number|
-        match = cas_number.match(/(?<cas_number>\d{2,7}-\d{2}-\d)/)
+        match = cas_number.match(Patterns::CAS_NUMBER)
         match[:cas_number] if match
       end
     end
 
     def build_ec_numbers(row)
-      transform_array!(
+      Cosing::Parser.transform_array!(
         row,
         key: :ec_number,
         split: "/"
       ).map do |ec_number|
-        match = ec_number.match(/(?<ec_number>\d{3}-\d{3}-\d)/)
+        match = ec_number.match(Patterns::EC_NUMBER)
         match[:ec_number] if match
       end
     end

data/lib/cosing/database.rb

@@ -4,80 +4,86 @@ module Cosing
   class Database
     attr_reader :annexes, :ingredients
 
+    ANNOTATION_PATTERN = %r{([IVX]+)/([IVX]*/)?([\dabcd,]+)}
+
     def initialize(annexes)
       @annexes = annexes
       @ingredients = {}
     end
 
     def add_ingredient(params)
-      restrictions = transform_array!(params, key: :restriction, split: "\n")
-      annotation_pattern = %r{([IVX]+)/([IVX]*/)?([\dabcd,]+)}
+      restrictions = Parser
+        .transform_array!(params, key: :restriction, split: "\n")
+      regulations = extract_regulations(restrictions).compact
+      cas_numbers = extract_cas_numbers(params).compact
+      einecs_numbers = extract_einecs_numbers(params).compact
+      functions = Parser
+        .transform_array!(params, key: :functions, split: ",")
+        .compact
 
-      regulations = restrictions
-        .flat_map do |restriction|
-          matches = restriction.scan(annotation_pattern)
-          next unless matches.any?
+      @ingredients[params[:reference_number]] = Ingredient.new(
+        functions:,
+        restrictions:,
+        regulations:,
+        cas_numbers:,
+        einecs_numbers:,
+        **params
+      )
+    end
 
-          hits = matches.flat_map do |match|
-            numeral, _, reference_number = match
+    def save(filepath, pretty: false)
+      output = if pretty
+        JSON.pretty_generate(@ingredients.to_h)
+      else
+        JSON.dump(@ingredients.to_h)
+      end
 
-            reference_number.split(",").flat_map do |number|
-              #debugger if number == "41" && numeral == "VI"
-              @annexes[numeral.downcase.to_sym].lookup(number)
-            end
-          end
+      File.write(filepath, output)
+    end
 
-          hits.compact
-        end
+    private
 
-      cas_numbers = transform_array!(
+    def extract_cas_numbers(params)
+      Parser.transform_array!(
         params,
         key: :cas_number,
         split: "/"
       ).map do |cas_number|
-        match = cas_number.match(/(?<cas_number>\d{2,7}-\d{2}-\d)/)
+        match = cas_number.match(Patterns::CAS_NUMBER)
         match[:cas_number] if match
       end
+    end
 
-      functions = transform_array!(params, key: :functions, split: ",")
-
-      einecs_numbers = transform_array!(
+    def extract_einecs_numbers(params)
+      Parser.transform_array!(
         params,
         key: :einecs_number,
         split: "/"
       ).map do |einecs_number|
-        match = einecs_number.match(/(?<einecs_number>\d{3}-\d{3}-\d)/)
+        match = einecs_number.match(Patterns::EINECS_NUMBER)
         match[:einecs_number] if match
       end
-
-      @ingredients[params[:reference_number]] = Ingredient.new(
-        functions:,
-        restrictions: restrictions.compact,
-        regulations: regulations.compact,
-        cas_numbers: cas_numbers.compact,
-        einecs_numbers: einecs_numbers.compact,
-        **params
-      )
     end
 
-    def save(filepath, pretty: false)
-      output = if pretty
-                 JSON.pretty_generate(@ingredients.to_h)
-               else
-                 JSON.dump(@ingredients.to_h)
-               end
+    def extract_regulations(restrictions)
+      restrictions
+        .flat_map do |restriction|
+          matches = restriction.scan(ANNOTATION_PATTERN)
+          next if matches.none?
 
-      File.write(filepath, output)
+          extract_hits(matches).compact
+        end
     end
 
-    private
+    def extract_hits(matches)
+      matches.flat_map do |match|
+        numeral, _, reference_number = match
 
-    def transform_array!(params, key:, split:)
-      params
-        .delete(key)
-        .split(split)
-        .map(&:strip)
-        .reject { |n| n == "-" }
+        reference_number.split(",").flat_map do |number|
+          # debugger if number == "41" && numeral == "VI"
+          @annexes[numeral.downcase.to_sym].lookup(number)
+        end
+      end
     end
   end
 end

data/lib/cosing/parser.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Cosing
+  module Parser
+    module_function
+
+    def transform_array!(params, key:, split:)
+      params
+        .delete(key)
+        .split(split)
+        .map!(&:strip)
+        .reject { |n| n == "-" }
+    end
+  end
+end

data/lib/cosing/patterns.rb

@@ -0,0 +1,7 @@
+module Cosing
+  module Patterns
+    CAS_NUMBER = /(?<cas_number>\d{2,7}-\d{2}-\d)/
+    EINECS_NUMBER = /(?<einecs_number>\d{3}-\d{3}-\d)/
+    EC_NUMBER = /(?<ec_number>\d{3}-\d{3}-\d)/
+  end
+end

data/lib/cosing/version.rb

@@ -1,6 +1,5 @@
 # frozen_string_literal: true
 
 module Cosing
-  VERSION = "0.2.0"
-  public_constant :VERSION
+  VERSION = "0.3.0"
 end

data/lib/cosing.rb

@@ -7,6 +7,8 @@ require "dry/types"
 require "dry/struct"
 require_relative "cosing/version"
 require_relative "cosing/types"
+require_relative "cosing/parser"
+require_relative "cosing/patterns"
 require_relative "cosing/ingredient"
 require_relative "cosing/annex"
 require_relative "cosing/database"
@@ -30,13 +32,10 @@ module Cosing
         liberal_parsing: true,
         header_converters: :symbol
       ) do |row|
-        row =
-          row
-            .to_h
-            .transform_values(&:to_s)
-            .transform_values(&:strip)
-
-        database.add_ingredient(row)
+        row
+          .to_h
+          .transform_values! { |value| value.to_s.strip }
+          .then { |row| database.add_ingredient(row) }
       end
     end
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: cosing
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.3.0
 platform: ruby
 authors:
 - Nolan J Tait
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-02-25 00:00:00.000000000 Z
+date: 2024-04-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: csv
@@ -59,12 +59,8 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
-- ".rspec"
-- ".rubocop.yml"
-- CHANGELOG.md
 - LICENSE.txt
 - README.md
-- Rakefile
 - cosing.gemspec
 - data/annex.II.csv
 - data/annex.III.csv
@@ -84,6 +80,8 @@ files:
 - lib/cosing/annex/vi.rb
 - lib/cosing/database.rb
 - lib/cosing/ingredient.rb
+- lib/cosing/parser.rb
+- lib/cosing/patterns.rb
 - lib/cosing/types.rb
 - lib/cosing/version.rb
 homepage: https://github.com/inhouse-work/cosing
@@ -109,7 +107,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.3
+rubygems_version: 3.5.7
 signing_key:
 specification_version: 4
 summary: COSING database

data/.rspec

@@ -1,3 +0,0 @@
---format documentation
---color
---require spec_helper

data/.rubocop.yml

@@ -1,9 +0,0 @@
-require:
-  - rubocop-inhouse
-
-inherit_gem:
-  rubocop-inhouse:
-    - config/default.yml
-
-AllCops:
-  TargetRubyVersion: 3.1

data/CHANGELOG.md

@@ -1,13 +0,0 @@
-## [Unreleased]
-
-## [0.1.2] - 2023-12-03
-
-- Fixes missing pathname require
-
-## [0.1.1] - 2023-12-03
-
-- Fixes gem paths
-
-## [0.1.0] - 2023-12-03
-
-- Initial release

data/Rakefile

@@ -1,14 +0,0 @@
-# frozen_string_literal: true
-
-require "bundler/gem_tasks"
-require "rspec/core/rake_task"
-
-RSpec::Core::RakeTask.new(:spec)
-
-task default: :spec
-
-desc "Output the cosing database into a json format"
-task :export do
-  database = Cosing.load
-  database.save
-end