unihan_lang 0.3.0 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 07200fff605b22cae82d18c8b9a16041fb1fbcd53c9106a25582785a779631ad
-  data.tar.gz: 335f78c7b507b547bbad354856bb294655f1042943ed66d3899a607721c40efb
+  metadata.gz: 9edfe134026854b7c939b4aa8a941b4781a7cbb3b944531e322ad5ebb6560558
+  data.tar.gz: 5b0c33cefa96257e0e83cfdd5ca76f5d47ae75b437e2403a06d691cdadeb27b9
 SHA512:
-  metadata.gz: 7691fa9c50fc79921fa069e36a367cb37adc20a2be5fecdd4d768d037389b3fa434fa1078ab4bf05afc026f5a91d1b8215696f0eff0f2feb9987e10264a4bc32
-  data.tar.gz: acff2b9f98d296486e0a83f2b8909db4351042e7ec837c088d0ddf8377b99425ebda8eea4f99709861390fe141162f95717768196c7a022eb29d13030d9708f8
+  metadata.gz: 51672291df0ad4762c41d39265e6c141a9b22f57bcc899fd8cf7051d5ce46c9091e163a48d1e6e22b5c18d58345519f6aa9fbc850e2570fa1ede818a44699b99
+  data.tar.gz: 440fe3f331130bd56ce65b0a6ad3723e894521730941dafdaeaf7e897ba404405926b8803347593e349d2f08538d08850e43e59a158a58912cfcb07419163ad6

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    unihan_lang (0.3.0)
+    unihan_lang (0.3.1)
 
 GEM
   remote: https://rubygems.org/

data/HOW_TO_UPDATE.md

@@ -9,24 +9,5 @@ curl -u $RUBY_GEMS_ACCOUNT https://rubygems.org/api/v1/api_key.yaml > ~/.gem/cre
 # Publish
 
 ```bash
-# Add release branch
-VERSION=TODO
-git checkout master
-git checkout -b release/$VERSION
-vi lib/unihan_lang/version.rb # edit version
-git commit -am "Bump version $VERSION"
-
-# Publish to rubygems
-gem build unihan_lang.gemspec
-bundle install
-gem push unihan_lang-$VERSION.gem
-
-# GitHub release
-git tag $VERSION
-git push --tags
-open https://github.com/kyubey1228/unihan_lang/releases/new
-
-# merge to master
-git checkout master
-git merge release/$VERSION
+./bump_new_version.sh 1.2.3
 ```

data/README.ja.md

@@ -83,3 +83,4 @@ puts unihan.contains_zh_cn?("這個text不包含簡體字") # => false
 
 このライブラリは、テキストの言語を完全に正確に判定することを保証するものではありません。
 特に、短いテキストや複数の言語が混在するテキストの場合、判定が難しい場合があります。
+判定の際、句読点・数字・空白は無視されます。また、ひらがな・カタカナを含むテキストは中国語ではなく日本語として扱われます。

data/README.md

@@ -71,7 +71,7 @@ puts unihan.contains_zh_cn?("這個text不包含簡體字") # => false
 
 ## Features
 
-- `determine_language(text)`: Determines the language of the text ("ZH_TW", "ZH_CN", "JA", "Unknown").
+- `determine_language(text)`: Determines the language of the text ("zh_TW", "zh_CN", "JA", "Unknown").
 - `zh_tw?(text)`: Checks if the text is in Traditional Chinese.
 - `zh_cn?(text)`: Checks if the text is in Simplified Chinese.
 - `contains_chinese?(text)`: Checks if the text contains Chinese characters.
@@ -86,3 +86,4 @@ puts unihan.contains_zh_cn?("這個text不包含簡體字") # => false
 This library does not guarantee 100% accuracy in language identification.
 Particularly for short texts or texts containing multiple languages, determination may be challenging.
 The distinction between Traditional and Simplified Chinese is based on the Unihan database.
+Punctuation, digits, and whitespace are ignored when judging a sentence, and text containing Hiragana/Katakana is treated as Japanese rather than Chinese.

data/bump_new_version.sh

@@ -0,0 +1,38 @@
+#!/bin/bash
+
+if [ -z "$1" ]; then
+  echo "Usage: $0 <version>"
+  exit 1
+fi
+
+# ~/.gem/credentials
+if [ ! -f ~/.gem/credentials ]; then
+  echo "Error: ~/.gem/credentials not found. Please set up your RubyGems credentials."
+  exit 1
+fi
+
+PROJECT_NAME="unihan_lang"
+GITHUB_REPO="kyubey1228/unihan_lang"
+VERSION=$1
+echo "Start bumping version: $VERSION"
+
+# Publish
+# Add release branch
+git checkout master
+git checkout -b release/$VERSION
+sed -i '' "s/VERSION = \".*\"/VERSION = \"$VERSION\"/" lib/$PROJECT_NAME/version.rb
+git commit -am "Bump version $VERSION"
+
+# Publish to rubygems
+gem build $PROJECT_NAME.gemspec
+bundle install
+gem push $PROJECT_NAME-$VERSION.gem
+
+# GitHub release
+git tag $VERSION
+git push --tags
+open https://github.com/$GITHUB_REPO/releases/new
+
+# merge to master
+git checkout master
+git merge release/$VERSION

data/lib/unihan_lang/chinese_score_analyzer.rb

@@ -10,10 +10,12 @@ module UnihanLang
       @variant_mapping = variant_mapping
       @traditional_score = 0
       @simplified_score = 0
+      @contains_kana = false
       analyze
     end
 
     def dominant_language
+      return "Unknown" if contains_kana?
       return "Unknown" if total_chinese.zero?
       return "ZH_TW" if traditional_score > simplified_score
       return "ZH_CN" if simplified_score > traditional_score
@@ -22,18 +24,37 @@ module UnihanLang
     end
 
     def language_ratio
-      return :unknown if total_chinese != @text.length
+      return :unknown if contains_kana?
+      return :unknown if total_chinese.zero? || total_chinese != meaningful_length
       return :tw if traditional_score > simplified_score
-      return :cn if simplified_score >= traditional_score
+      return :cn if simplified_score > traditional_score
 
       :unknown
     end
 
     private
 
+    def contains_kana?
+      @contains_kana
+    end
+
+    def kana_char?(char)
+      ord = char.ord
+      # 0x3040-0x309F: ひらがな, 0x30A0-0x30FF: カタカナ, 0x31F0-0x31FF: カタカナ拡張
+      (ord >= 0x3040 && ord <= 0x309F) || (ord >= 0x30A0 && ord <= 0x30FF) || (ord >= 0x31F0 && ord <= 0x31FF)
+    end
+
+    # Punctuation, digits, and whitespace are ignored so they don't
+    # invalidate an otherwise all-Chinese sentence (e.g. "這是中文。").
+    def meaningful_length
+      @text.chars.count { |char| char.match?(/\p{L}/) }
+    end
+
     def analyze
       @total_chinese = 0
       @text.chars.each do |char|
+        @contains_kana ||= kana_char?(char)
+
         next unless @chinese_processor.chinese_character?(char)
 
         @total_chinese += 1

data/lib/unihan_lang/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module UnihanLang
-  VERSION = "0.3.0"
+  VERSION = "0.3.1"
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: unihan_lang
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.3.1
 platform: ruby
 authors:
 - kyubey1228
 bindir: exe
 cert_chain: []
-date: 2025-02-20 00:00:00.000000000 Z
+date: 2026-08-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -70,6 +70,7 @@ files:
 - README.ja.md
 - README.md
 - Rakefile
+- bump_new_version.sh
 - data/Unihan_Variants.txt
 - lib/unihan_lang.rb
 - lib/unihan_lang/chinese_processor.rb