pp_ucfirst: Avoid a conversion from UTF-8 to code point

This generates the UTF-8 for U+0345.  Then we can do a string compare
against that when parsing a UTF-8 string, instead of first converting to
numeric.

Author: khwilliamson

pp.c

@@ -4511,7 +4511,6 @@ PP_wrapped(pp_uc, 1, 0)
         U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
 
 #define GREEK_CAPITAL_LETTER_IOTA 0x0399
-#define COMBINING_GREEK_YPOGEGRAMMENI 0x0345
         /* All occurrences of these are to be moved to follow any other marks.
          * This is context-dependent.  We may not be passed enough context to
          * move the iota subscript beyond all of them, but we do the best we can
@@ -4552,9 +4551,8 @@ PP_wrapped(pp_uc, 1, 0)
 #else
             uv = _toUPPER_utf8_flags(s, send, tmpbuf, &upper_len, 0);
 #endif
-            if (   uv == GREEK_CAPITAL_LETTER_IOTA
-                && utf8_to_uv_or_die(s, send, 0) ==
-                                                COMBINING_GREEK_YPOGEGRAMMENI)
+            if (   UNLIKELY(uv == GREEK_CAPITAL_LETTER_IOTA)
+                && memBEGINs(s, this_len, COMBINING_GREEK_YPOGEGRAMMENI_UTF8))
             {
                 in_iota_subscript = TRUE;
             }

regen/unicode_constants.pl

@@ -1020,6 +1020,7 @@ END
 
 U+0300 string
 U+0307 string
+U+0345 string
 
 U+1E9E string_skip_if_undef
 

unicode_constants.h

@@ -57,6 +57,7 @@ bytes.
 
 #   define COMBINING_GRAVE_ACCENT_UTF8  "\xCC\x80"    /* U+0300 */
 #   define COMBINING_DOT_ABOVE_UTF8  "\xCC\x87"    /* U+0307 */
+#   define COMBINING_GREEK_YPOGEGRAMMENI_UTF8  "\xCD\x85"    /* U+0345 */
 
 #   define LATIN_CAPITAL_LETTER_SHARP_S_UTF8  "\xE1\xBA\x9E"    /* U+1E9E */
 
@@ -122,6 +123,7 @@ bytes.
 
 #   define COMBINING_GRAVE_ACCENT_UTF8  "\xAF\x41"    /* U+0300 */
 #   define COMBINING_DOT_ABOVE_UTF8  "\xAF\x48"    /* U+0307 */
+#   define COMBINING_GREEK_YPOGEGRAMMENI_UTF8  "\xB1\x46"    /* U+0345 */
 
 #   define LATIN_CAPITAL_LETTER_SHARP_S_UTF8  "\xBF\x63\x72"    /* U+1E9E */
 
@@ -187,6 +189,7 @@ bytes.
 
 #   define COMBINING_GRAVE_ACCENT_UTF8  "\xAD\x41"    /* U+0300 */
 #   define COMBINING_DOT_ABOVE_UTF8  "\xAD\x48"    /* U+0307 */
+#   define COMBINING_GREEK_YPOGEGRAMMENI_UTF8  "\xAF\x46"    /* U+0345 */
 
 #   define LATIN_CAPITAL_LETTER_SHARP_S_UTF8  "\xBF\x62\x71"    /* U+1E9E */