more efficient utf32_iterator_set()

[disorder] / lib / unidata.h

diff --git a/lib/unidata.h b/lib/unidata.h
index a21ec51..c2daccb 100644 (file)
--- a/lib/unidata.h
+++ b/lib/unidata.h
@@ -89,8 +89,12 @@ struct unidata {
   char sentence_break;
 };
 extern const struct unidata *const unidata[];
+extern const struct unicode_utf8_row {
+  uint8_t count;
+  uint8_t min2, max2;
+} unicode_utf8_valid[];
 #define UNICODE_NCHARS 1114112
-#define UNICODE_MODULUS 128
+#define UNICODE_MODULUS 16
 #define UNICODE_BREAK_START 196608
 #define UNICODE_BREAK_END 917504
 #define UNICODE_BREAK_TOP 918016