Merge pull request #406 from aycabta/support-halfwidth-kana-correctly

Character merging may increase the character width
ruby · Dec 23, 2021 · a5d8a6c · a5d8a6c
2 parents 50ed575 + 0895a0d
commit a5d8a6c
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 3 deletions.
diff --git a/lib/reline/line_editor.rb b/lib/reline/line_editor.rb
@@ -2029,8 +2029,16 @@ def finish
     last_byte_size = Reline::Unicode.get_prev_mbchar_size(@line, @byte_pointer)
     @byte_pointer += bytesize
     last_mbchar = @line.byteslice((@byte_pointer - bytesize - last_byte_size), last_byte_size)
-    if last_byte_size != 0 and (last_mbchar + str).grapheme_clusters.size == 1
-      width = 0
+    combined_char = last_mbchar + str
+    if last_byte_size != 0 and combined_char.grapheme_clusters.size == 1
+      # combined char
+      last_mbchar_width = Reline::Unicode.get_mbchar_width(last_mbchar)
+      combined_char_width = Reline::Unicode.get_mbchar_width(combined_char)
+      if combined_char_width > last_mbchar_width
+        width = combined_char_width - last_mbchar_width
+      else
+        width = 0
+      end
     end
     @cursor += width
     @cursor_max += width

diff --git a/lib/reline/unicode.rb b/lib/reline/unicode.rb
@@ -79,6 +79,8 @@ def self.escape_for_print(str)
 
   require 'reline/unicode/east_asian_width'
 
+  HalfwidthDakutenHandakuten = /[\u{FF9E}\u{FF9F}]/
+
   MBCharWidthRE = /
     (?<width_2_1>
       [#{ EscapedChars.map {|c| "\\x%02x" % c.ord }.join }] (?# ^ + char, such as ^M, ^H, ^[, ...)
@@ -93,6 +95,12 @@ def self.escape_for_print(str)
       #{ EastAsianWidth::TYPE_H }
     | #{ EastAsianWidth::TYPE_NA }
     | #{ EastAsianWidth::TYPE_N }
+    )(?!#{ HalfwidthDakutenHandakuten })
+  | (?<width_2_3>
+      (?: #{ EastAsianWidth::TYPE_H }
+        | #{ EastAsianWidth::TYPE_NA }
+        | #{ EastAsianWidth::TYPE_N })
+      #{ HalfwidthDakutenHandakuten }
     )
   | (?<ambiguous_width>
       #{EastAsianWidth::TYPE_A}
@@ -109,7 +117,7 @@ def self.get_mbchar_width(mbchar)
     m = mbchar.encode(Encoding::UTF_8).match(MBCharWidthRE)
     case
     when m.nil? then 1 # TODO should be U+FFFD � REPLACEMENT CHARACTER
-    when m[:width_2_1], m[:width_2_2] then 2
+    when m[:width_2_1], m[:width_2_2], m[:width_2_3] then 2
     when m[:width_3] then 3
     when m[:width_0] then 0
     when m[:width_1] then 1

diff --git a/test/reline/test_key_actor_emacs.rb b/test/reline/test_key_actor_emacs.rb
@@ -2306,6 +2306,22 @@ def test_ed_argument_digit_by_meta_num
     assert_line('abcd')
   end
 
+  def test_halfwidth_kana_width_dakuten
+    input_keys('ｶﾞｷﾞｹﾞｺﾞ')
+    assert_byte_pointer_size('ｶﾞｷﾞｹﾞｺﾞ')
+    assert_cursor(8)
+    assert_cursor_max(8)
+    input_keys("\C-b\C-b", false)
+    assert_byte_pointer_size('ｶﾞｷﾞ')
+    assert_cursor(4)
+    assert_cursor_max(8)
+    input_keys('ｸﾞ', false)
+    assert_byte_pointer_size('ｶﾞｷﾞｸﾞ')
+    assert_cursor(6)
+    assert_cursor_max(10)
+    assert_line('ｶﾞｷﾞｸﾞｹﾞｺﾞ')
+  end
+
   def test_input_unknown_char
     input_keys('͸') # U+0378 (unassigned)
     assert_line('͸')