X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~mdw/git/disorder/blobdiff_plain/c2e01e0ac91c61de69158113d2ab59f6831b2941..11f598d144c1e98322090063faa796ec2a5fb470:/lib/unidata.h diff --git a/lib/unidata.h b/lib/unidata.h index dbd15e6..aa8814d 100644 --- a/lib/unidata.h +++ b/lib/unidata.h @@ -74,13 +74,14 @@ enum unicode_Sentence_Break { }; extern const char *const unicode_Sentence_Break_names[]; enum unicode_flags { - unicode_normalize_before_casefold = 1 + unicode_normalize_before_casefold = 1, + unicode_compatibility_decomposition = 2 }; struct unidata { - const uint32_t *compat; - const uint32_t *canon; + const uint32_t *decomp; const uint32_t *casefold; + const uint32_t *composed; unsigned char ccc; char general_category; uint8_t flags; @@ -89,6 +90,10 @@ 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 16 #define UNICODE_BREAK_START 196608