Merge pull request #409 from henry0312/support_cjk_symbols_and_punctuation

Add support for CJK Symbols and Punctuation

extension/cjk_test.go

@@ -197,4 +197,15 @@ func TestEastAsianLineBreaks(t *testing.T) {
 		},
 		t,
 	)
+	no = 8
+	testutil.DoTestCase(
+		markdown,
+		testutil.MarkdownTestCase{
+			No:          no,
+			Description: "Soft line breaks between east asian wide characters or punctuations are ignored",
+			Markdown:    "太郎は\\ **「こんにちわ」**\\ と、\r\n言った\r\nんです",
+			Expected:    "<p>太郎は\\ <strong>「こんにちわ」</strong>\\ と、言ったんです</p>",
+		},
+		t,
+	)
 }

util/util.go

@@ -836,11 +836,18 @@ func IsAlphaNumeric(c byte) bool {
 
 // IsEastAsianWideRune returns trhe if the given rune is an east asian wide character, otherwise false.
 func IsEastAsianWideRune(r rune) bool {
+	// https://en.wikipedia.org/wiki/CJK_Symbols_and_Punctuation
+	var CJKSymbolsAndPunctuation = &unicode.RangeTable{
+		R16: []unicode.Range16{
+			{0x3000, 0x303F, 1},
+		},
+	}
 	return unicode.Is(unicode.Hiragana, r) ||
 		unicode.Is(unicode.Katakana, r) ||
 		unicode.Is(unicode.Han, r) ||
 		unicode.Is(unicode.Lm, r) ||
-		unicode.Is(unicode.Hangul, r)
+		unicode.Is(unicode.Hangul, r) ||
+		unicode.Is(CJKSymbolsAndPunctuation, r)
 }
 
 // A BufWriter is a subset of the bufio.Writer .