From c35e37a304d4cb83ac30d84633cff404b32961c8 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 14:59:55 +0900
Subject: [PATCH 1/6] Add: ContinuousLanguageLocale module

---
 lib/ingreedy.rb                            |  1 +
 lib/ingreedy/amount_parser.rb              | 12 ++++++-
 lib/ingreedy/continuous_language_locale.rb |  9 ++++++
 lib/ingreedy/dictionary_collection.rb      | 14 ++++++--
 lib/ingreedy/root_parser.rb                | 11 ++++++-
 spec/ingreedy_spec.rb                      | 37 ++++++++++++++++++++++
 6 files changed, 79 insertions(+), 5 deletions(-)
 create mode 100644 lib/ingreedy/continuous_language_locale.rb

diff --git a/lib/ingreedy.rb b/lib/ingreedy.rb
index ef355ad..d33f6f2 100644
--- a/lib/ingreedy.rb
+++ b/lib/ingreedy.rb
@@ -1,6 +1,7 @@
 path = File.expand_path(File.join(File.dirname(__FILE__), "ingreedy"))
 
 require File.join(path, "case_insensitive_parser")
+require File.join(path, "continuous_language_locale")
 require File.join(path, "ingreedy_parser")
 require File.join(path, "dictionary_collection")
 
diff --git a/lib/ingreedy/amount_parser.rb b/lib/ingreedy/amount_parser.rb
index 0250533..c657db5 100644
--- a/lib/ingreedy/amount_parser.rb
+++ b/lib/ingreedy/amount_parser.rb
@@ -3,11 +3,17 @@
 module Ingreedy
   class AmountParser < Parslet::Parser
     include CaseInsensitiveParser
+    include ContinuousLanguageLocale
 
     rule(:whitespace) do
-      match("\s")
+      if use_whitespace?(current_locale)
+        match("\s")
+      else
+        match("\s").maybe
+      end
     end
 
+
     rule(:integer) do
       match("[0-9]").repeat(1)
     end
@@ -58,6 +64,10 @@ class AmountParser < Parslet::Parser
 
     private
 
+    def current_locale
+      Ingreedy.dictionaries.current_locale
+    end
+
     def word_digits
       Ingreedy.dictionaries.current.numbers.keys
     end
diff --git a/lib/ingreedy/continuous_language_locale.rb b/lib/ingreedy/continuous_language_locale.rb
new file mode 100644
index 0000000..ed0c811
--- /dev/null
+++ b/lib/ingreedy/continuous_language_locale.rb
@@ -0,0 +1,9 @@
+module Ingreedy
+  module ContinuousLanguageLocale
+    CONTINUOUS_LANGUAGES_LOCALES = [:ja].freeze
+
+    def use_whitespace?(locale)
+      !CONTINUOUS_LANGUAGES_LOCALES.include?(locale)
+    end
+  end
+end
diff --git a/lib/ingreedy/dictionary_collection.rb b/lib/ingreedy/dictionary_collection.rb
index 9af269a..167ca45 100644
--- a/lib/ingreedy/dictionary_collection.rb
+++ b/lib/ingreedy/dictionary_collection.rb
@@ -12,17 +12,25 @@ def []=(locale, attributes)
     end
 
     def current
+      fetch_dictionary(current_locale)
+    end
+
+    def current_locale
+      find_locale
+    end
+
+    private
+
+    def find_locale
       candidate_locales.each do |locale|
         if dictionary = fetch_dictionary(locale)
-          return dictionary
+          return locale
         end
       end
 
       raise "No dictionary found for locales: #{candidate_locales}"
     end
 
-    private
-
     def candidate_locales
       Array(Ingreedy.locale || i18n_gem_locales || :en)
     end
diff --git a/lib/ingreedy/root_parser.rb b/lib/ingreedy/root_parser.rb
index 4d1b0d9..82949a0 100644
--- a/lib/ingreedy/root_parser.rb
+++ b/lib/ingreedy/root_parser.rb
@@ -1,6 +1,7 @@
 module Ingreedy
   class RootParser < Parslet::Parser
     include CaseInsensitiveParser
+    include ContinuousLanguageLocale
 
     rule(:range) do
       AmountParser.new.as(:amount) >>
@@ -19,7 +20,11 @@ class RootParser < Parslet::Parser
     end
 
     rule(:whitespace) do
-      match("\s")
+      if use_whitespace?(current_locale)
+        match("\s")
+      else
+        match("\s").maybe
+      end
     end
 
     rule(:container_amount) do
@@ -119,6 +124,10 @@ def parse
     private
 
     attr_reader :original_query
+  
+    def current_locale
+      Ingreedy.dictionaries.current_locale
+    end
 
     def imprecise_amounts
       Ingreedy.dictionaries.current.imprecise_amounts
diff --git a/spec/ingreedy_spec.rb b/spec/ingreedy_spec.rb
index 4821a6e..46b2285 100644
--- a/spec/ingreedy_spec.rb
+++ b/spec/ingreedy_spec.rb
@@ -447,6 +447,43 @@
   end
 end
 
+describe Ingreedy, "continuous language" do
+  before(:all) do
+    Ingreedy.dictionaries[:ja] = {
+      units: { gram: ["g"] },
+      numbers: { "一" => 1 },
+    }
+    Ingreedy.locale = :ja
+  end
+
+  after(:all) do
+    Ingreedy.locale = nil
+  end
+
+  it "parses correctly" do
+    result = Ingreedy.parse "200g砂糖"
+
+    expect(result.amount).to eq(200)
+    expect(result.unit).to eq(:gram)
+    expect(result.ingredient).to eq("砂糖")
+  end
+
+  it "parses correctly with reverse format" do
+    result = Ingreedy.parse "砂糖200g"
+
+    expect(result.amount).to eq(200)
+    expect(result.unit).to eq(:gram)
+    expect(result.ingredient).to eq("砂糖")
+  end
+
+  it "parses correctly with numbers" do
+    result = Ingreedy.parse "卵一g"
+
+    expect(result.amount).to eq(1)
+    expect(result.ingredient).to eq("卵")
+  end
+end
+
 describe Ingreedy, "error handling" do
   it "wraps Parslet exceptions in a custom exception" do
     expect do

From a9df09bdb82cc79942e8db0e6c138c81f5af5533 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 15:01:21 +0900
Subject: [PATCH 2/6] Improve: reverse format

- avoid parsing sugger 100g -> sugge, 100, g when continuous language
---
 lib/ingreedy/root_parser.rb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/ingreedy/root_parser.rb b/lib/ingreedy/root_parser.rb
index 82949a0..1f30219 100644
--- a/lib/ingreedy/root_parser.rb
+++ b/lib/ingreedy/root_parser.rb
@@ -92,7 +92,7 @@ class RootParser < Parslet::Parser
 
     rule(:reverse_format) do
       # e.g. flour 200g
-      ((whitespace >> quantity).absent? >> any).repeat.as(:ingredient) >>
+      ((whitespace >> quantity >> any.absent?).absent? >> any).repeat.as(:ingredient) >>
         whitespace >>
         quantity
     end

From 15164287985c914e28a0e27405d42d20944a5b63 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 16:31:25 +0900
Subject: [PATCH 3/6] Remove: newline

---
 lib/ingreedy/amount_parser.rb | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/ingreedy/amount_parser.rb b/lib/ingreedy/amount_parser.rb
index c657db5..d6f291e 100644
--- a/lib/ingreedy/amount_parser.rb
+++ b/lib/ingreedy/amount_parser.rb
@@ -13,7 +13,6 @@ class AmountParser < Parslet::Parser
       end
     end
 
-
     rule(:integer) do
       match("[0-9]").repeat(1)
     end

From ce01093d9d020d0990773128fda98c443c518038 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 17:40:36 +0900
Subject: [PATCH 4/6] Add: Taiwan and Thailand to CONTINUOUS_LANGUAGES_LOCALES

---
 lib/ingreedy/continuous_language_locale.rb    |  2 +-
 .../continuouse_language_locale_spec.rb       | 23 +++++
 spec/ingreedy_spec.rb                         | 91 ++++++++++++++-----
 3 files changed, 90 insertions(+), 26 deletions(-)
 create mode 100644 spec/ingreedy/continuouse_language_locale_spec.rb

diff --git a/lib/ingreedy/continuous_language_locale.rb b/lib/ingreedy/continuous_language_locale.rb
index ed0c811..1773f2b 100644
--- a/lib/ingreedy/continuous_language_locale.rb
+++ b/lib/ingreedy/continuous_language_locale.rb
@@ -1,6 +1,6 @@
 module Ingreedy
   module ContinuousLanguageLocale
-    CONTINUOUS_LANGUAGES_LOCALES = [:ja].freeze
+    CONTINUOUS_LANGUAGES_LOCALES = %i(ja th zh-TW)
 
     def use_whitespace?(locale)
       !CONTINUOUS_LANGUAGES_LOCALES.include?(locale)
diff --git a/spec/ingreedy/continuouse_language_locale_spec.rb b/spec/ingreedy/continuouse_language_locale_spec.rb
new file mode 100644
index 0000000..3b62b92
--- /dev/null
+++ b/spec/ingreedy/continuouse_language_locale_spec.rb
@@ -0,0 +1,23 @@
+require 'ingreedy/continuous_language_locale'
+
+RSpec.describe Ingreedy::ContinuousLanguageLocale do
+  include Ingreedy::ContinuousLanguageLocale
+
+  describe '#use_whitespace?' do
+    context 'when the locale is a continuous language' do
+      it 'returns false' do
+        expect(use_whitespace?(:ja)).to be_falsey
+        expect(use_whitespace?(:th)).to be_falsey
+        expect(use_whitespace?(:'zh-TW')).to be_falsey
+      end
+    end
+
+    context 'when the locale is not a continuous language' do
+      it 'returns true' do
+        expect(use_whitespace?(:en)).to be_truthy
+        expect(use_whitespace?(:fr)).to be_truthy
+        expect(use_whitespace?(:'zh-CN')).to be_truthy
+      end
+    end
+  end
+end
diff --git a/spec/ingreedy_spec.rb b/spec/ingreedy_spec.rb
index 46b2285..66a6488 100644
--- a/spec/ingreedy_spec.rb
+++ b/spec/ingreedy_spec.rb
@@ -448,39 +448,80 @@
 end
 
 describe Ingreedy, "continuous language" do
-  before(:all) do
-    Ingreedy.dictionaries[:ja] = {
-      units: { gram: ["g"] },
-      numbers: { "一" => 1 },
-    }
-    Ingreedy.locale = :ja
-  end
+  context "Japanese" do
+    before do
+      Ingreedy.dictionaries[:ja] = {
+        units: { gram: ["g"], other: ["個"] },
+        numbers: { "一" => 1 },
+      }
+      Ingreedy.locale = :ja
+    end
 
-  after(:all) do
-    Ingreedy.locale = nil
-  end
+    after do
+      Ingreedy.locale = nil
+    end
 
-  it "parses correctly" do
-    result = Ingreedy.parse "200g砂糖"
+    it "parses correctly" do
+      result = Ingreedy.parse "200g砂糖"
 
-    expect(result.amount).to eq(200)
-    expect(result.unit).to eq(:gram)
-    expect(result.ingredient).to eq("砂糖")
-  end
+      expect(result.amount).to eq(200)
+      expect(result.unit).to eq(:gram)
+      expect(result.ingredient).to eq("砂糖")
+    end
 
-  it "parses correctly with reverse format" do
-    result = Ingreedy.parse "砂糖200g"
+    it "parses correctly with reverse format" do
+      result = Ingreedy.parse "砂糖200g"
 
-    expect(result.amount).to eq(200)
-    expect(result.unit).to eq(:gram)
-    expect(result.ingredient).to eq("砂糖")
+      expect(result.amount).to eq(200)
+      expect(result.unit).to eq(:gram)
+      expect(result.ingredient).to eq("砂糖")
+    end
+
+    it "parses correctly with numbers" do
+      result = Ingreedy.parse "卵一個"
+
+      expect(result.amount).to eq(1)
+      expect(result.unit).to eq(:other)
+      expect(result.ingredient).to eq("卵")
+    end
   end
 
-  it "parses correctly with numbers" do
-    result = Ingreedy.parse "卵一g"
+  context "Taiwanese" do
+    before do
+      Ingreedy.dictionaries[:"zh-TW"] = {
+        units: { gram: ["g"], other: ["个"] },
+        numbers: { "一" => 1 },
+      }
+      Ingreedy.locale = :"zh-TW"
+    end
 
-    expect(result.amount).to eq(1)
-    expect(result.ingredient).to eq("卵")
+    after do
+      Ingreedy.locale = nil
+    end
+
+    it "parses correctly" do
+      result = Ingreedy.parse "200g砂糖"
+
+      expect(result.amount).to eq(200)
+      expect(result.unit).to eq(:gram)
+      expect(result.ingredient).to eq("砂糖")
+    end
+
+    it "parses correctly with reverse format" do
+      result = Ingreedy.parse "砂糖200g"
+
+      expect(result.amount).to eq(200)
+      expect(result.unit).to eq(:gram)
+      expect(result.ingredient).to eq("砂糖")
+    end
+
+    it "parses correctly with numbers" do
+      result = Ingreedy.parse "一个鸡蛋"
+
+      expect(result.amount).to eq(1)
+      expect(result.unit).to eq(:other)
+      expect(result.ingredient).to eq("鸡蛋")
+    end
   end
 end
 

From 72046b1e8d45e2ae32c8a36cb050f7cbf831cb02 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 18:00:16 +0900
Subject: [PATCH 5/6] Refactore: method to get current locale

---
 lib/ingreedy/amount_parser.rb         |  2 +-
 lib/ingreedy/dictionary.rb            |  5 +++--
 lib/ingreedy/dictionary_collection.rb | 18 +++++-------------
 lib/ingreedy/root_parser.rb           |  2 +-
 4 files changed, 10 insertions(+), 17 deletions(-)

diff --git a/lib/ingreedy/amount_parser.rb b/lib/ingreedy/amount_parser.rb
index d6f291e..20f5de1 100644
--- a/lib/ingreedy/amount_parser.rb
+++ b/lib/ingreedy/amount_parser.rb
@@ -64,7 +64,7 @@ class AmountParser < Parslet::Parser
     private
 
     def current_locale
-      Ingreedy.dictionaries.current_locale
+      Ingreedy.dictionaries.current.locale
     end
 
     def word_digits
diff --git a/lib/ingreedy/dictionary.rb b/lib/ingreedy/dictionary.rb
index 3b329fe..20b113d 100644
--- a/lib/ingreedy/dictionary.rb
+++ b/lib/ingreedy/dictionary.rb
@@ -1,14 +1,15 @@
 module Ingreedy
   class Dictionary
-    attr_reader :units, :numbers, :prepositions, :range_separators
+    attr_reader :units, :numbers, :prepositions, :range_separators, :locale
     attr_reader :imprecise_amounts
 
-    def initialize(units:, numbers: {}, prepositions: [], range_separators: %w{- ~}, imprecise_amounts: [])
+    def initialize(units:, numbers: {}, prepositions: [], range_separators: %w{- ~}, imprecise_amounts: [], locale: nil)
       @units = units
       @numbers = sort_by_length(numbers)
       @prepositions = prepositions
       @range_separators = range_separators
       @imprecise_amounts = imprecise_amounts
+      @locale = locale
     end
 
     # https://en.wikipedia.org/wiki/Number_Forms
diff --git a/lib/ingreedy/dictionary_collection.rb b/lib/ingreedy/dictionary_collection.rb
index 167ca45..ed47c51 100644
--- a/lib/ingreedy/dictionary_collection.rb
+++ b/lib/ingreedy/dictionary_collection.rb
@@ -8,29 +8,21 @@ def initialize
     end
 
     def []=(locale, attributes)
-      @collection[locale] = Dictionary.new(**attributes)
+      @collection[locale] = Dictionary.new(locale:, **attributes)
     end
 
     def current
-      fetch_dictionary(current_locale)
-    end
-
-    def current_locale
-      find_locale
-    end
-
-    private
-
-    def find_locale
       candidate_locales.each do |locale|
         if dictionary = fetch_dictionary(locale)
-          return locale
+          return dictionary
         end
       end
 
       raise "No dictionary found for locales: #{candidate_locales}"
     end
 
+    private
+
     def candidate_locales
       Array(Ingreedy.locale || i18n_gem_locales || :en)
     end
@@ -46,7 +38,7 @@ def i18n_gem_locales
     end
 
     def fetch_dictionary(locale)
-      @collection[locale] ||= Dictionary.new **load_yaml(locale)
+      @collection[locale] ||= Dictionary.new(locale:, **load_yaml(locale))
     rescue Errno::ENOENT
     end
 
diff --git a/lib/ingreedy/root_parser.rb b/lib/ingreedy/root_parser.rb
index 1f30219..ec4b884 100644
--- a/lib/ingreedy/root_parser.rb
+++ b/lib/ingreedy/root_parser.rb
@@ -126,7 +126,7 @@ def parse
     attr_reader :original_query
   
     def current_locale
-      Ingreedy.dictionaries.current_locale
+      Ingreedy.dictionaries.current.locale
     end
 
     def imprecise_amounts

From 75e164330b7eb3de6c8f1ab3cca1d2aeb32a8332 Mon Sep 17 00:00:00 2001
From: sota-horiuchi <sota-horiuchi@cookpad.com>
Date: Tue, 26 Mar 2024 18:44:34 +0900
Subject: [PATCH 6/6] Fix: Dictionary initialization arg order

---
 lib/ingreedy/dictionary_collection.rb | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/ingreedy/dictionary_collection.rb b/lib/ingreedy/dictionary_collection.rb
index ed47c51..ddcbb26 100644
--- a/lib/ingreedy/dictionary_collection.rb
+++ b/lib/ingreedy/dictionary_collection.rb
@@ -8,7 +8,7 @@ def initialize
     end
 
     def []=(locale, attributes)
-      @collection[locale] = Dictionary.new(locale:, **attributes)
+      @collection[locale] = Dictionary.new(**attributes, locale: locale)
     end
 
     def current
@@ -38,7 +38,7 @@ def i18n_gem_locales
     end
 
     def fetch_dictionary(locale)
-      @collection[locale] ||= Dictionary.new(locale:, **load_yaml(locale))
+      @collection[locale] ||= Dictionary.new(**load_yaml(locale), locale: locale)
     rescue Errno::ENOENT
     end