untested utf32_is_word_boundary() and associated table changes

[disorder] / lib / unidata.h

diff --git a/lib/unidata.h b/lib/unidata.h
index 859575479c6401124aba03226f1ce731fafe14ce..5e6df0db7a3a40ee447b78328fd35dab3d1de6e2 100644 (file)
--- a/lib/unidata.h
+++ b/lib/unidata.h
@@ -33,6 +33,17 @@ enum unicode_gc_cat {
   unicode_gc_Zp,
   unicode_gc_Zs
 };
+enum unicode_Word_Break {
+  unicode_Word_Break_ALetter,
+  unicode_Word_Break_Extend,
+  unicode_Word_Break_ExtendNumLet,
+  unicode_Word_Break_Format,
+  unicode_Word_Break_Katakana,
+  unicode_Word_Break_MidLetter,
+  unicode_Word_Break_MidNum,
+  unicode_Word_Break_Numeric,
+  unicode_Word_Break_Other
+};
 enum unicode_flags {
   unicode_normalize_before_casefold = 1
 };
@@ -46,6 +57,7 @@ struct unidata {
   unsigned char ccc;
   char gc;
   uint8_t flags;
+  char word_break;
 };
 extern const struct unidata *const unidata[];
 #define UNICODE_NCHARS 195200