thai_keyboard_corrector 0.1.1 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ade7acc9f2b07c72e55d0f89cfe1c3a23002f5c01ba19e5eff0b27b1c1d483ca
-  data.tar.gz: f010ec32ab5064c6346643162e029876d5c7450e4d584bf3d7cdc4e569ed1b59
+  metadata.gz: adf3f3e1587f08281375e2d9971af347e33f9d136ee36c9df2b94895174bdb80
+  data.tar.gz: c3585823ef59f640c80d80fa906a4ade237874c237ee898c7fb0da9235b9b875
 SHA512:
-  metadata.gz: 1005a6452fd77bd31271c0db49612ff39f8451a4be5de998524d2ccf5edbce384b753a90ec681b17ca9a8127327333e4847156ec502c02e850fdb370c7acedbb
-  data.tar.gz: b18cffcd017adb5f5d1dd64782f0a8b87c1bdccffa96028765c5a8ca48ddd284127092fa34d12160d72aaa83088bf7c50e5a0fc0feead296409e04554fbe2cc4
+  metadata.gz: 3edf85c588c050020ce4a087466067cd83576ba2de3922f8121529f215243cba63defe98845f0a1997c47187ab3e4cb504cd711692d70094de37aeb1c36df8bd
+  data.tar.gz: a6d8ce0cbc345143393ba0bbcb841d69fdd17eb376e2cf415815d1be7c76c635fa2ac532001e94d894f56921a968796d5ed07aff444f64564054b0d12a25522d

data/README.md

@@ -20,7 +20,7 @@
 
 ```bash
 gem install thai_keyboard_corrector
-````
+```
 
 (หรือเพิ่ม `gem "thai_keyboard_corrector"` ใน Gemfile)
 

data/lib/thai_keyboard_corrector/detector.rb

@@ -13,61 +13,27 @@ module ThaiKeyboardCorrector
     # Returns :thai_in_en, :en_in_th, :thai, :en, :mixed, :unknown
     def detect_layout(str)
       clean = str.strip
-      return :unknown if clean.empty?
+      clean_no_ws = clean.gsub(/\s+/, '') # strip ALL whitespace
 
-      thai_cnt, latin_cnt = char_stats(clean)
-      return :unknown if thai_cnt.zero? && latin_cnt.zero?
+      return :unknown if clean_no_ws.empty?
 
-      # ---------- pure-Latin ----------
-      if thai_cnt.zero?
-        # ▼ Treat 1-3-letter words as Thai-in-EN (they’re almost never real English)
-        return :thai_in_en if clean.length <= 3 &&
-                              hit_ratio(clean, Mapping::ENG_TO_THAI) >= FULL_HIT
+      thai_cnt, latin_cnt = char_stats(clean) # char_stats already ignores ws
 
-        return :thai_in_en if clean.match?(/[^A-Za-z]/) &&
-                              hit_ratio(clean, Mapping::ENG_TO_THAI) >= THRESHOLD
+      # Majority-vote rule ----------------------------------------------
+      return :en_in_th   if thai_cnt  > latin_cnt   # mostly Thai glyphs
+      return :thai_in_en if latin_cnt > thai_cnt    # mostly Latin letters
 
-        return :en
-      end
-
-      # ---------- pure-Thai ----------
-      if latin_cnt.zero?
-        return :thai if clean.length < 4 # ignore tiny words like “ดี”
-
-        eng = Mapping.map_thai_to_eng(clean)
-        vowelish = eng.count(VOWELS).positive?
-        if eng.match?(/\A[a-z]+\z/i) && vowelish &&
-           hit_ratio(clean, Mapping::THAI_TO_ENG) >= THRESHOLD
-          return :en_in_th
-        end
-
-        return :thai
-      end
-
-      # ---------- mixed ----------
-      return :mixed if thai_cnt.positive? && latin_cnt.positive?
-      return :thai_in_en if hit_ratio(clean, Mapping::ENG_TO_THAI) >= THRESHOLD
-      return :en_in_th   if hit_ratio(clean, Mapping::THAI_TO_ENG) >= THRESHOLD
-
-      :mixed
+      # If counts are equal (or zero) we can’t be sure
+      thai_cnt.zero? && latin_cnt.zero? ? :unknown : :mixed
     end
 
     # helpers ----------------------------------------------------------------
     def char_stats(str)
-      clean = str.gsub(/\s+/, '') # remove ALL Unicode whitespace
+      clean = str.gsub(/\s+/, '')
       thai  = clean.each_char.count { |c| THAI_RANGE.include?(c.ord) }
-      latin = clean.each_char.count { |c| c =~ /[A-Za-z]/ }
+      latin = clean.each_char.count { |c| c.match?(/[A-Za-z]/) }
       [thai, latin]
     end
     private_class_method :char_stats
-
-    def hit_ratio(str, table)
-      chars = str.gsub(/\s+/, '').chars # whitespace-free array
-      return 0.0 if chars.empty?
-
-      hits = chars.count { |c| table.key?(c) }
-      hits.to_f / chars.length
-    end
-    private_class_method :hit_ratio
   end
 end

data/lib/thai_keyboard_corrector/mapping.rb

@@ -2,8 +2,8 @@
 
 module ThaiKeyboardCorrector
   # Mapping module provides methods to convert between English and Thai characters
-  module Mapping
-    # Base (lower-case) map
+  module mapping
+    # 1. Base map – lower-case EN → Thai Kedmanee
     BASE = {
       'q' => 'ๆ', 'w' => 'ไ', 'e' => 'ำ', 'r' => 'พ', 't' => 'ะ',
       'y' => 'ั', 'u' => 'ี', 'i' => 'ร', 'o' => 'น', 'p' => 'ย',
@@ -14,8 +14,32 @@ module ThaiKeyboardCorrector
       'n' => 'ื', 'm' => 'ท', ',' => 'ม', '.' => 'ใ', '/' => 'ฝ'
     }.freeze
 
+    # 2. Shift-layer Thai glyphs → underlying EN key
+    SHIFT = {
+      # ── Number row (Shift+1 … Shift+0) – Thai digits
+      '๑' => '1', '๒' => '2', '๓' => '3', '๔' => '4', '๕' => '5',
+      '๖' => '6', '๗' => '7', '๘' => '8', '๙' => '9', '๐' => '0',
+
+      # ── Top-letter row (Q–P)
+      'ฃ' => 'w', 'ฅ' => 'e', 'ฆ' => 'r', 'ฑ' => 't', 'ํ' => 'y',
+      'ฐ' => 'u', 'ณ' => 'i', 'ญ' => 'o', 'ธ' => 't', # (Shift+T duplicate)
+
+      # ── Home row (A–L)
+      'ฤ' => 'a', 'ฦ' => 's', 'ฌ' => 'h', 'ศ' => 'l', 'ษ' => ';', 'ฮ' => "'",
+
+      # ── Bottom row (Z–/)
+      'ฒ' => 'z', 'ฬ' => 'x',
+      'ฯ' => 'm',  # Thai paiyannoi
+      '฿' => '.',  # Baht sign (Shift+.)
+      '๏' => '/',  # Thai “head mark”
+
+      # Already mapped earlier (duplicates kept for clarity—harmless):
+      '๛' => ',' # end-paragraph mark
+    }.freeze
+
+    # 3. Final maps (frozen once, never mutated)
     ENG_TO_THAI = BASE.merge(BASE.transform_keys(&:upcase)).freeze
-    THAI_TO_ENG = BASE.invert.freeze # ⇒ always lower-case
+    THAI_TO_ENG = BASE.invert.merge(SHIFT).freeze
 
     module_function
 

data/lib/thai_keyboard_corrector/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module ThaiKeyboardCorrector
-  VERSION = '0.1.1'
+  VERSION = '0.2.1'
 end

data/lib/thai_keyboard_corrector.rb

@@ -13,11 +13,11 @@ module ThaiKeyboardCorrector
   # @return [String] corrected or original string
   def correct(str)
     case detect_layout(str)
-    when :thai_in_en
+    when :thai_in_en  # mostly Latin → convert EN→TH
       Mapping.map_eng_to_thai(str)
-    when :en_in_th
+    when :en_in_th    # mostly Thai  → convert TH→EN
       Mapping.map_thai_to_eng(str)
-    else
+    else              # :thai, :en, :mixed, :unknown
       str
     end
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: thai_keyboard_corrector
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.1
 platform: ruby
 authors:
 - Chayut Orapinpatipat
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-06-20 00:00:00.000000000 Z
+date: 2025-06-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec