Commit 23d7544efc8bcf2bf48dec32d00b755678a1830c

Authored by Marcin Woliński
1 parent 610e6695

NARROW NO-BREAK SPACE and ZERO WIDTH SPACE

morfeusz/charset/charset_utils.hpp
... ... @@ -43,9 +43,12 @@ static inline std::vector<char> initializeWhitespaces() {
43 43 res[0x2008] = true; // PUNCTUATION SPACE
44 44 res[0x2009] = true; // THIN SPACE
45 45 res[0x200A] = true; // HAIR SPACE
  46 + res[0x200B] = true; // ZERO WIDTH SPACE (for marking words in CJK lanugages)
46 47 res[0x2028] = true; // LINE SEPARATOR
47 48 res[0x2029] = true; // PARAGRAPH SEPARATOR
  49 + res[0x202F] = true; // NARROW NO-BREAK SPACE
48 50 res[0x205F] = true; // MEDIUM MATHEMATICAL SPACE
  51 + res[0x2060] = true; // WORD JOINER (no-break version of 0x200B)
49 52 res[0x3000] = true; // IDEOGRAPHIC SPACE
50 53 return res;
51 54 }
... ...