lib/ting.rb in ting-0.11.0 vs lib/ting.rb in ting-0.12.0

@@ -48,18 +48,33 @@
       hsh[syll] = Ting.writer(:hanyu, :accents).(
         Ting.reader(:hanyu, :numbers).(syll.downcase)
       )
     end
 
+    # The longest syllables are six letters long (chuang, shuang, zhuang).
+    SYLLABLE_REGEXP = /[A-Za-züÜ]{1,6}\d?/
 
     def pretty_tones(string)
-      string.gsub('u:','ü').gsub(/[A-Za-züÜ]{1,7}\d?/) do |syll|
-        SYLLABLE_CACHE[syll]
+      string = string.gsub('u:', 'ü') # (note that this implicitly dups the string)
+      # Scan through the string, replacing syllable by syllable.
+      pos = 0
+      while match = string.match(SYLLABLE_REGEXP, pos)
+        syllable = match[0]
+        replacement = SYLLABLE_CACHE[syllable]
+        match_pos = match.begin(0)
+        # If this syllable starts with a vowel and is preceded by a letter (not whitespace or
+        # control characters), insert an apostrophe before it.
+        if match_pos > 0 && string[match_pos - 1] =~ /[[:alpha:]]/ && syllable =~ /^[AEOaoe]/
+          replacement = "'" + replacement
+        end
+        string[match_pos, syllable.length] = replacement
+        pos = match_pos + replacement.length
       end
+      string
     end
 
     def bpmf(string)
-      string.gsub('u:','ü').scan(/[A-Za-züÜ]{1,7}\d?/).map do |m|
+      string.gsub('u:', 'ü').scan(SYLLABLE_REGEXP).map do |m|
         Ting.writer(:zhuyin, :marks).(
           Ting.reader(:hanyu, :numbers).(m.downcase)
         )
       end.join(' ')
     end