inci_score 4.5.1 → 4.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c8d0453d792d9a6f5e628d13720a927c15ee14ba8441faf588736b22ab8bf618
-  data.tar.gz: d354ebb76fa432acc592da8e152e6a9712022c394cce7e8f951e6ebc9f94b93a
+  metadata.gz: 7c1ec290c92d1fb34798dbd04ec8f93a6ad1db12fd124ff2f7d0c4a50c108aa7
+  data.tar.gz: dad169f8a2486aeeac3a22df47097235d965b9435350d4800d662c9022f0a36c
 SHA512:
-  metadata.gz: dcfc4fcafe3b51ed9917937c36055c4043591b499250eae36be97cf4b58d101d0440bc12a57a2fe8e22c5fb4189ed20969cd6f0ab7f58cfa9028fa2ea681fda6
-  data.tar.gz: a34557698ffc1449c4ae4d4bb031465566506e92186c7a6e482ff655453566581d3645f8e78894d5afec6a3feb6066d887d30063356b6d3a91b408204a1a6ed4
+  metadata.gz: 34394e8b825952f6ff98696ba899592f23a0acd8e3b95331c1d645903ac9795569dc6e115ff1dca5df0ae4aee551b54aae0e32b91faa302b4338b4e37769d797
+  data.tar.gz: 1b932c8c80adfc5b0989383a7c31c7b78432890dd3b86c54c887b1d3146f6629332cb7abd16e50b70343113bf3beb76233f4cd0d77fd684e26e7377ca0a8ee7f

data/README.md

@@ -80,7 +80,7 @@ You can collect INCI data by using the available CLI interface:
 inci_score --src="ingredients: aqua, dimethicone, pej-10, noent"
 
 TOTAL SCORE:
-      	47.18
+      	53.22
 PRECISION:
       	75.0
 COMPONENTS:

data/config/catalog.yml

@@ -1,6 +1,7 @@
 ---
 aqua: 0
 water: 0
+aroma: 2
 parfum: 3
 fragrance: 3
 phosphatidylcholine: 1
@@ -1580,6 +1581,7 @@ fagus sylvatica: 0
 farnesol: 2
 farnesyl acetate: 0
 fast green fcf: 4
+fava tonka: 1
 fermented vegetable extract: 0
 ferric chloride: 0
 ferula assa foetida: 0
@@ -3370,6 +3372,7 @@ pentasodium aminotrimethylene phosphonate: 3
 pentasodium pentetate: 3
 pentasodium triphosphate: 2
 pentetic acid: 3
+pentylene glycol: 2
 perfluorodecalin: 3
 perfluoropolymethylisopropyl ether: 3
 perfluorotetralin: 3
@@ -4019,6 +4022,7 @@ prunus avium: 0
 prunus cerasus: 0
 prunus domestica: 0
 prunus dulcis: 0
+prunus amygdalus dulcis (sweet almond) oil: 0
 prunus persica: 0
 prunus serotina: 0
 prunus spinosa: 0
@@ -4831,6 +4835,8 @@ thymus extract: 4
 thymus hydrolysate: 4
 thymus serpyllum: 0
 thymus vulgaris: 0
+thymus vulgaris leaf oil: 0
+thymus vulgaris leaf water: 0
 tilia americana: 0
 tilia cordata: 0
 tilia vulgaris: 0
@@ -4908,6 +4914,7 @@ tripeptide-29: 1
 triphenyl trimethicone: 4
 tripotassium edta: 4
 trisodium edta: 4
+trisodium ethylenediamine disuccinate: 2
 trisodium hedta: 4
 trisodium lauroampho pg-acetate phosphate chloride: 3
 trisodium methylglycinediacetate: 1

data/lib/inci_score/config.rb

@@ -5,7 +5,6 @@ require 'yaml'
 module InciScore
   module Config
     CATALOG = YAML::load_file(File::expand_path('../../../config/catalog.yml', __FILE__)).freeze
-    CIR = File.readlines(File::expand_path('../../../config/cir', __FILE__)).freeze
     HAZARDS = YAML::load_file(File::expand_path('../../../config/hazards.yml', __FILE__)).freeze
   end
 end

data/lib/inci_score/ingredient.rb

@@ -2,7 +2,6 @@
 
 module InciScore
   class Ingredient
-    SLASH = '/'
     SLASH_RULE = /(?<!ate)\//.freeze
     PARENTHESIS = %w[( ) [ ]].freeze
     PARENTHESIS_RULE = /(\(.+\)|\[.+\])/.freeze
@@ -11,24 +10,23 @@ module InciScore
 
     def initialize(raw)
       @raw = raw.to_s
-      @values = fetch_values
+      @values = fetch_values.uniq
       freeze
     end
 
-    def to_s
-      values.join(SLASH)
-    end
-
     private
 
     def fetch_values
-      if parenthesis?
-        parenthesis = PARENTHESIS.join
-        parenthesis_values = raw.match(PARENTHESIS_RULE).captures.map { |c| c.delete(parenthesis) }
-        deparenthesized = raw.sub(PARENTHESIS_RULE, '').sub(/\s+/, ' ').strip
-        [deparenthesized].concat(parenthesis_values)
-      else
-        raw.split(SLASH_RULE).map(&:strip)
+      [raw].tap do |vals|
+        if parenthesis?
+          parenthesis = PARENTHESIS.join
+          parenthesis_values = raw.match(PARENTHESIS_RULE).captures.map { |c| c.delete(parenthesis) }
+          deparenthesized = raw.sub(PARENTHESIS_RULE, '').sub(/\s{2,}/, ' ').strip
+          vals << deparenthesized
+          vals.concat(parenthesis_values)
+        else
+          vals.concat(raw.split(SLASH_RULE).map(&:strip))
+        end
       end
     end
 

data/lib/inci_score/recognizer_rules.rb

@@ -7,7 +7,11 @@ module InciScore
     module Rules
       TOLERANCE = 3
 
-      Component = Struct.new(:name, :hazard)
+      Component = Struct.new(:name, :hazard) do
+        def to_s
+          "#{name} (#{hazard})"
+        end
+      end
 
       Key = ->(src) do
         score = Config::CATALOG[src]
@@ -24,24 +28,31 @@ module InciScore
         extend self
 
         Result = Struct.new(:name, :distance, :score) do
-          def tolerable?(size)
-            distance < TOLERANCE && distance <= (size-1)
+          def tolerable?(size, tolerance)
+            distance < tolerance && distance <= (size-1)
           end
         end
 
         def call(src)
           return if src.empty?
           size = src.size
+          t = tolerance(size)
           farthest = Result.new(nil, size)
           initial = src[0]
           result = Config::CATALOG.reduce(farthest) do |nearest, (name, score)|
             next nearest unless name.start_with?(initial)
-            next nearest if name.size > (size + TOLERANCE)
+            next nearest if name.size > (size + t)
             d = src.distance(name)
             nearest = Result.new(name, d, score) if d < nearest.distance
             nearest
           end
-          Component.new(result.name, result.score) if result.tolerable?(size)
+          Component.new(result.name, result.score) if result.tolerable?(size, t)
+        end
+
+        private
+
+        def tolerance(size)
+          Math.log(size, 1.8).round
         end
       end
 
@@ -53,7 +64,7 @@ module InciScore
         def call(src)
           return if src.size < TOLERANCE
           digits = src[0, MIN_MEANINGFUL]
-          pairs = Config::CATALOG.detect { |name, _| name.start_with?(digits) }.to_a.first
+          pairs = Config::CATALOG.detect { |name, _| name.start_with?(digits) }
           Component.new(*pairs) if pairs
         end
       end

data/lib/inci_score/response.rb

@@ -41,7 +41,7 @@ PRECISION:
       return '' if components.empty?
       %Q{
 COMPONENTS:
-      \t#{components.map { |c| "#{c.name} (#{c.hazard})" }.join(', ')}}
+      \t#{components.map(&:to_s).join(', ')}}
     end
 
     def unrecognized_str

data/lib/inci_score/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module InciScore
-  VERSION = '4.5.1'
+  VERSION = '4.6.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: inci_score
 version: !ruby/object:Gem::Version
-  version: 4.5.1
+  version: 4.6.0
 platform: ruby
 authors:
 - costajob
@@ -106,7 +106,6 @@ files:
 - README.md
 - bin/inci_score
 - config/catalog.yml
-- config/cir
 - config/hazards.yml
 - ext/levenshtein.c
 - lib/inci_score.rb