Commit 10ce930c authored by unknown's avatar unknown
Browse files

UTF8 UCA based collations.

parent 25e262a0

mysys/charset-def.c

+33 −1
Original line number Diff line number Diff line
@@ -40,6 +40,24 @@ extern CHARSET_INFO my_charset_ucs2_slovak_uca_ci; 
extern CHARSET_INFO my_charset_ucs2_spanish2_uca_ci;

#endif



#ifdef HAVE_CHARSET_utf8

extern CHARSET_INFO my_charset_utf8_general_uca_ci;

extern CHARSET_INFO my_charset_utf8_icelandic_uca_ci;

extern CHARSET_INFO my_charset_utf8_latvian_uca_ci;

extern CHARSET_INFO my_charset_utf8_romanian_uca_ci;

extern CHARSET_INFO my_charset_utf8_slovenian_uca_ci;

extern CHARSET_INFO my_charset_utf8_polish_uca_ci;

extern CHARSET_INFO my_charset_utf8_estonian_uca_ci;

extern CHARSET_INFO my_charset_utf8_spanish_uca_ci;

extern CHARSET_INFO my_charset_utf8_swedish_uca_ci;

extern CHARSET_INFO my_charset_utf8_turkish_uca_ci;

extern CHARSET_INFO my_charset_utf8_czech_uca_ci;

extern CHARSET_INFO my_charset_utf8_danish_uca_ci;

extern CHARSET_INFO my_charset_utf8_lithuanian_uca_ci;

extern CHARSET_INFO my_charset_utf8_slovak_uca_ci;

extern CHARSET_INFO my_charset_utf8_spanish2_uca_ci;

#endif



my_bool init_compiled_charsets(myf flags __attribute__((unused)))

{

  CHARSET_INFO *cs;
@@ -92,7 +110,6 @@ my_bool init_compiled_charsets(myf flags __attribute__((unused))) 
  add_compiled_collation(&my_charset_ucs2_general_ci);

  add_compiled_collation(&my_charset_ucs2_bin);

  add_compiled_collation(&my_charset_ucs2_general_uca);

  add_compiled_collation(&my_charset_ucs2_general_uca);

  add_compiled_collation(&my_charset_ucs2_icelandic_uca_ci);

  add_compiled_collation(&my_charset_ucs2_latvian_uca_ci);

  add_compiled_collation(&my_charset_ucs2_romanian_uca_ci);
@@ -117,6 +134,21 @@ my_bool init_compiled_charsets(myf flags __attribute__((unused))) 
#ifdef HAVE_CHARSET_utf8

  add_compiled_collation(&my_charset_utf8_general_ci);

  add_compiled_collation(&my_charset_utf8_bin);

  add_compiled_collation(&my_charset_utf8_general_uca_ci);

  add_compiled_collation(&my_charset_utf8_icelandic_uca_ci);

  add_compiled_collation(&my_charset_utf8_latvian_uca_ci);

  add_compiled_collation(&my_charset_utf8_romanian_uca_ci);

  add_compiled_collation(&my_charset_utf8_slovenian_uca_ci);

  add_compiled_collation(&my_charset_utf8_polish_uca_ci);

  add_compiled_collation(&my_charset_utf8_estonian_uca_ci);

  add_compiled_collation(&my_charset_utf8_spanish_uca_ci);

  add_compiled_collation(&my_charset_utf8_swedish_uca_ci);

  add_compiled_collation(&my_charset_utf8_turkish_uca_ci);

  add_compiled_collation(&my_charset_utf8_czech_uca_ci);

  add_compiled_collation(&my_charset_utf8_danish_uca_ci);

  add_compiled_collation(&my_charset_utf8_lithuanian_uca_ci);

  add_compiled_collation(&my_charset_utf8_slovak_uca_ci);

  add_compiled_collation(&my_charset_utf8_spanish2_uca_ci);

#endif



  /* Copy compiled charsets */

strings/ctype-uca.c

+701 −51

File changed.

Preview size limit exceeded, changes collapsed.

strings/ctype-utf8.c

+3 −4
Original line number Diff line number Diff line
@@ -2057,7 +2057,7 @@ static MY_COLLATION_HANDLER my_collation_ci_handler = 
    my_hash_sort_utf8

};



static MY_CHARSET_HANDLER my_charset_handler=

MY_CHARSET_HANDLER my_charset_utf8_handler=

{

    NULL,		/* init */

    my_ismbchar_utf8,
@@ -2109,7 +2109,7 @@ CHARSET_INFO my_charset_utf8_general_ci= 
    3,			/* mbmaxlen     */

    0,			/* min_sort_char */

    255,		/* max_sort_char */

    &my_charset_handler,

    &my_charset_utf8_handler,

    &my_collation_ci_handler

};
@@ -2137,13 +2137,12 @@ CHARSET_INFO my_charset_utf8_bin= 
    3,			/* mbmaxlen     */

    0,			/* min_sort_char */

    255,		/* max_sort_char */

    &my_charset_handler,

    &my_charset_utf8_handler,

    &my_collation_mb_bin_handler

};





#ifdef MY_TEST_UTF8



#include <stdio.h>



static void test_mb(CHARSET_INFO *cs, uchar *s)