Avoid calling iswalnum for ascii characters

iswalnum can be pretty expensive as its a shared library call.

1 files changed, 12 insertions, 11 deletions

diff --git a/src/unicode.hh b/src/unicode.hh
index a385b6eb..fd35ecf0 100644
--- a/src/unicode.hh
+++ b/src/unicode.hh
@@ -60,12 +60,23 @@ inline bool is_blank(Codepoint c) noexcept
            is_horizontal_blank(c) ;
 }
 
+inline bool is_basic_alpha(Codepoint c) noexcept
+{
+    return (c >= 'a' and c <= 'z') or (c >= 'A' and c <= 'Z');
+}
+
+inline bool is_basic_digit(Codepoint c) noexcept
+{
+    return c >= '0' and c <= '9';
+}
+
 enum WordType { Word, WORD };
 
 template<WordType word_type = Word>
 inline bool is_word(Codepoint c, ConstArrayView<Codepoint> extra_word_chars = {'_'}) noexcept
 {
-    return iswalnum((wchar_t)c) or contains(extra_word_chars, c);
+    return (c < 128 ? is_basic_alpha(c) or is_basic_digit(c) : iswalnum((wchar_t)c)) or
+           contains(extra_word_chars, c);
 }
 
 template<>
@@ -79,16 +90,6 @@ inline bool is_punctuation(Codepoint c, ConstArrayView<Codepoint> extra_word_cha
     return not (is_word(c, extra_word_chars) or is_blank(c));
 }
 
-inline bool is_basic_alpha(Codepoint c) noexcept
-{
-    return (c >= 'a' and c <= 'z') or (c >= 'A' and c <= 'Z');
-}
-
-inline bool is_basic_digit(Codepoint c) noexcept
-{
-    return c >= '0' and c <= '9';
-}
-
 inline bool is_identifier(Codepoint c) noexcept
 {
     return is_basic_alpha(c) or is_basic_digit(c) or