MDEV-25904 New collation functions to compare InnoDB style trimmed NO…

… PAD strings

include/m_ctype.h

@@ -330,6 +330,60 @@ struct my_collation_handler_st
  const uchar *, size_t, const uchar *, size_t, my_bool);
  int (*strnncollsp)(CHARSET_INFO *,
  const uchar *, size_t, const uchar *, size_t);
+ /*
+ strnncollsp_nchars() - similar to strnncollsp() but assumes that both
+ strings were originally CHAR(N) values with the
+ same N, then were optionally space-padded,
+ or optionally space-trimmed.
+
+ In other words, this function compares in the way
+ if we insert both values into a CHAR(N) column
+ and then compare the two column values.
+
+ It compares the same amount of characters from the two strings.
+ This is especially important for NOPAD collations.
+
+ If CHAR_LENGTH of the two strings are different,
+ the shorter string is virtually padded with trailing spaces
+ up to CHAR_LENGTH of the longer string, to guarantee that the
+ same amount of characters are compared.
+ This is important if the two CHAR(N) strings are space-trimmed 
+ (e.g. like in InnoDB compact format for CHAR).
+
+ The function compares not more than "nchars" characters only.
+ This can be useful to compare CHAR(N) space-padded strings
+ (when the exact N is known) without having to truncate them before
+ the comparison.
+
+ For example, Field_string stores a "CHAR(3) CHARACTER SET utf8mb4" value
+ of "aaa" as 12 bytes in a record buffer:
+ - 3 bytes of the actual data, followed by
+ - 9 bytes of spaces (just fillers, not real data)
+ The caller can pass nchars=3 to compare CHAR(3) record values.
+ In such case, the comparator won't go inside the 9 bytes of the fillers.
+
+ If N is not known, the caller can pass max(len1,len2) as the "nchars" value
+ (i.e. the maximum of the OCTET_LENGTH of the two strings).
+
+ Notes on complex collations.
+
+ This function counts contraction parts as individual characters.
+ For example, the Czech letter 'ch' (in Czech collations)
+ is ordinarily counted by the "nchars" limit as TWO characters
+ (although it is only one letter).
+ This corresponds to what CHAR(N) does in INSERT.
+
+ If the "nchars" limit tears apart a contraction, only the part fitting
+ into "nchars" characters is used. For example, in case of a Czech collation,
+ the string "ach" with nchars=2 is compared as 'ac': the contraction
+ 'ch' is torn apart and the letter 'c' acts as an individual character.
+ This emulates the same comparison result with the scenario when we insert
+ 'ach' into a CHAR(2) column and then compare it.
+ */
+ int (*strnncollsp_nchars)(CHARSET_INFO *,
+ const uchar *str1, size_t len1,
+ const uchar *str2, size_t len2,
+ size_t nchars);
  size_t (*strnxfrm)(CHARSET_INFO *,
  uchar *dst, size_t dstlen, uint nweights,
  const uchar *src, size_t srclen, uint flags);

sql/field.cc

@@ -7433,23 +7433,10 @@ Field_string::compatible_field_size(uint field_metadata,
 
 int Field_string::cmp(const uchar *a_ptr, const uchar *b_ptr)
 {
- size_t a_len, b_len;
-
- if (field_charset->mbmaxlen != 1)
- {
- size_t char_len= field_length/field_charset->mbmaxlen;
- a_len= my_charpos(field_charset, a_ptr, a_ptr + field_length, char_len);
- b_len= my_charpos(field_charset, b_ptr, b_ptr + field_length, char_len);
- }
- else
- a_len= b_len= field_length;
- /*
- We have to remove end space to be able to compare multi-byte-characters
- like in latin_de 'ae' and 0xe4
- */
- return field_charset->coll->strnncollsp(field_charset,
- a_ptr, a_len,
- b_ptr, b_len);
+ return field_charset->coll->strnncollsp_nchars(field_charset,
+ a_ptr, field_length,
+ b_ptr, field_length,
+ Field_string::char_length());
 }
 
 
@@ -7848,19 +7835,6 @@ int Field_varstring::cmp(const uchar *a_ptr, const uchar *b_ptr)
 }
 
 
-static int cmp_str_prefix(const uchar *ua, size_t alen, const uchar *ub,
- size_t blen, size_t prefix, CHARSET_INFO *cs)
-{
- const char *a= (char*)ua, *b= (char*)ub;
- MY_STRCOPY_STATUS status;
- prefix/= cs->mbmaxlen;
- alen= cs->cset->well_formed_char_length(cs, a, a + alen, prefix, &status);
- blen= cs->cset->well_formed_char_length(cs, b, b + blen, prefix, &status);
- return cs->coll->strnncollsp(cs, ua, alen, ub, blen);
-}
-
-
-
 int Field_varstring::cmp_prefix(const uchar *a_ptr, const uchar *b_ptr,
  size_t prefix_len)
 {
@@ -7880,8 +7854,12 @@ int Field_varstring::cmp_prefix(const uchar *a_ptr, const uchar *b_ptr,
  a_length= uint2korr(a_ptr);
  b_length= uint2korr(b_ptr);
  }
- return cmp_str_prefix(a_ptr+length_bytes, a_length, b_ptr+length_bytes,
- b_length, prefix_len, field_charset);
+ return field_charset->coll->strnncollsp_nchars(field_charset,
+ a_ptr + length_bytes,
+ a_length,
+ b_ptr + length_bytes,
+ b_length,
+ prefix_len / field_charset->mbmaxlen);
 }
 
 
@@ -8659,7 +8637,10 @@ int Field_blob::cmp_prefix(const uchar *a_ptr, const uchar *b_ptr,
  memcpy(&blob1, a_ptr+packlength, sizeof(char*));
  memcpy(&blob2, b_ptr+packlength, sizeof(char*));
  size_t a_len= get_length(a_ptr), b_len= get_length(b_ptr);
- return cmp_str_prefix(blob1, a_len, blob2, b_len, prefix_len, field_charset);
+ return field_charset->coll->strnncollsp_nchars(field_charset,
+ blob1, a_len,
+ blob2, b_len,
+ prefix_len / field_charset->mbmaxlen);
 }
 
 

strings/ctype-big5.c

@@ -6711,6 +6711,7 @@ static MY_COLLATION_HANDLER my_collation_handler_big5_chinese_ci=
  NULL, /* init */
  my_strnncoll_big5_chinese_ci,
  my_strnncollsp_big5_chinese_ci,
+ my_strnncollsp_nchars_big5_chinese_ci,
  my_strnxfrm_big5_chinese_ci,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -6727,6 +6728,7 @@ static MY_COLLATION_HANDLER my_collation_handler_big5_bin=
  NULL, /* init */
  my_strnncoll_big5_bin,
  my_strnncollsp_big5_bin,
+ my_strnncollsp_nchars_big5_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -6743,6 +6745,7 @@ static MY_COLLATION_HANDLER my_collation_handler_big5_chinese_nopad_ci=
  NULL, /* init */
  my_strnncoll_big5_chinese_ci,
  my_strnncollsp_big5_chinese_nopad_ci,
+ my_strnncollsp_nchars_big5_chinese_nopad_ci,
  my_strnxfrm_big5_chinese_nopad_ci,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -6759,6 +6762,7 @@ static MY_COLLATION_HANDLER my_collation_handler_big5_nopad_bin=
  NULL, /* init */
  my_strnncoll_big5_bin,
  my_strnncollsp_big5_nopad_bin,
+ my_strnncollsp_nchars_big5_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,

strings/ctype-bin.c

@@ -125,6 +125,17 @@ static int my_strnncollsp_binary(CHARSET_INFO * cs __attribute__((unused)),
 }
 
 
+static int my_strnncollsp_nchars_binary(CHARSET_INFO * cs __attribute__((unused)),
+ const uchar *s, size_t slen,
+ const uchar *t, size_t tlen,
+ size_t nchars)
+{
+ set_if_smaller(slen, nchars);
+ set_if_smaller(tlen, nchars);
+ return my_strnncoll_binary(cs, s, slen, t, tlen, 0);
+}
+
+
 static int my_strnncoll_8bit_bin(CHARSET_INFO * cs __attribute__((unused)),
  const uchar *s, size_t slen,
  const uchar *t, size_t tlen,
@@ -199,6 +210,17 @@ static int my_strnncollsp_8bit_bin(CHARSET_INFO * cs __attribute__((unused)),
 }
 
 
+static int my_strnncollsp_nchars_8bit_bin(CHARSET_INFO * cs,
+ const uchar *a, size_t a_length,
+ const uchar *b, size_t b_length,
+ size_t nchars)
+{
+ set_if_smaller(a_length, nchars);
+ set_if_smaller(b_length, nchars);
+ return my_strnncollsp_8bit_bin(cs, a, a_length, b, b_length);
+}
+
+
 static int my_strnncollsp_8bit_nopad_bin(CHARSET_INFO * cs
  __attribute__((unused)),
  const uchar *a, size_t a_length,
@@ -487,6 +509,7 @@ MY_COLLATION_HANDLER my_collation_8bit_bin_handler =
  my_coll_init_8bit_bin,
  my_strnncoll_8bit_bin,
  my_strnncollsp_8bit_bin,
+ my_strnncollsp_nchars_8bit_bin,
  my_strnxfrm_8bit_bin,
  my_strnxfrmlen_simple,
  my_like_range_simple,
@@ -503,6 +526,7 @@ MY_COLLATION_HANDLER my_collation_8bit_nopad_bin_handler =
  my_coll_init_8bit_bin,
  my_strnncoll_8bit_bin,
  my_strnncollsp_8bit_nopad_bin,
+ my_strnncollsp_nchars_8bit_bin,
  my_strnxfrm_8bit_nopad_bin,
  my_strnxfrmlen_simple,
  my_like_range_simple,
@@ -519,6 +543,7 @@ static MY_COLLATION_HANDLER my_collation_binary_handler =
  NULL, /* init */
  my_strnncoll_binary,
  my_strnncollsp_binary,
+ my_strnncollsp_nchars_binary,
  my_strnxfrm_8bit_bin,
  my_strnxfrmlen_simple,
  my_like_range_simple,

strings/ctype-cp932.c

@@ -34667,6 +34667,7 @@ static MY_COLLATION_HANDLER my_collation_handler_cp932_japanese_ci=
  NULL, /* init */
  my_strnncoll_cp932_japanese_ci,
  my_strnncollsp_cp932_japanese_ci,
+ my_strnncollsp_nchars_cp932_japanese_ci,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -34683,6 +34684,7 @@ static MY_COLLATION_HANDLER my_collation_handler_cp932_bin=
  NULL, /* init */
  my_strnncoll_cp932_bin,
  my_strnncollsp_cp932_bin,
+ my_strnncollsp_nchars_cp932_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -34699,6 +34701,7 @@ static MY_COLLATION_HANDLER my_collation_handler_cp932_japanese_nopad_ci=
  NULL, /* init */
  my_strnncoll_cp932_japanese_ci,
  my_strnncollsp_cp932_japanese_nopad_ci,
+ my_strnncollsp_nchars_cp932_japanese_nopad_ci,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -34715,6 +34718,7 @@ static MY_COLLATION_HANDLER my_collation_handler_cp932_nopad_bin=
  NULL, /* init */
  my_strnncoll_cp932_bin,
  my_strnncollsp_cp932_nopad_bin,
+ my_strnncollsp_nchars_cp932_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,

strings/ctype-czech.c

@@ -610,6 +610,7 @@ static MY_COLLATION_HANDLER my_collation_latin2_czech_ci_handler =
  NULL, /* init */
  my_strnncoll_czech,
  my_strnncollsp_czech,
+ my_strnncollsp_nchars_generic_8bit,
  my_strnxfrm_czech,
  my_strnxfrmlen_czech,
  my_like_range_czech,

strings/ctype-euc_kr.c

@@ -9957,6 +9957,7 @@ static MY_COLLATION_HANDLER my_collation_handler_euckr_korean_ci=
  NULL, /* init */
  my_strnncoll_euckr_korean_ci,
  my_strnncollsp_euckr_korean_ci,
+ my_strnncollsp_nchars_euckr_korean_ci,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -9973,6 +9974,7 @@ static MY_COLLATION_HANDLER my_collation_handler_euckr_bin=
  NULL, /* init */
  my_strnncoll_euckr_bin,
  my_strnncollsp_euckr_bin,
+ my_strnncollsp_nchars_euckr_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -9989,6 +9991,7 @@ static MY_COLLATION_HANDLER my_collation_handler_euckr_korean_nopad_ci=
  NULL, /* init */
  my_strnncoll_euckr_korean_ci,
  my_strnncollsp_euckr_korean_nopad_ci,
+ my_strnncollsp_nchars_euckr_korean_nopad_ci,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -10005,6 +10008,7 @@ static MY_COLLATION_HANDLER my_collation_handler_euckr_nopad_bin=
  NULL, /* init */
  my_strnncoll_euckr_bin,
  my_strnncollsp_euckr_nopad_bin,
+ my_strnncollsp_nchars_euckr_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,

strings/ctype-eucjpms.c

@@ -67495,6 +67495,7 @@ static MY_COLLATION_HANDLER my_collation_eucjpms_japanese_ci_handler =
  NULL, /* init */
  my_strnncoll_eucjpms_japanese_ci,
  my_strnncollsp_eucjpms_japanese_ci,
+ my_strnncollsp_nchars_eucjpms_japanese_ci,
  my_strnxfrm_mb, /* strnxfrm */
  my_strnxfrmlen_simple,
  my_like_range_mb, /* like_range */
@@ -67511,6 +67512,7 @@ static MY_COLLATION_HANDLER my_collation_eucjpms_bin_handler =
  NULL, /* init */
  my_strnncoll_eucjpms_bin,
  my_strnncollsp_eucjpms_bin,
+ my_strnncollsp_nchars_eucjpms_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -67527,6 +67529,7 @@ static MY_COLLATION_HANDLER my_collation_eucjpms_japanese_nopad_ci_handler =
  NULL, /* init */
  my_strnncoll_eucjpms_japanese_ci,
  my_strnncollsp_eucjpms_japanese_nopad_ci,
+ my_strnncollsp_nchars_eucjpms_japanese_nopad_ci,
  my_strnxfrm_mb_nopad, /* strnxfrm */
  my_strnxfrmlen_simple,
  my_like_range_mb, /* like_range */
@@ -67543,6 +67546,7 @@ static MY_COLLATION_HANDLER my_collation_eucjpms_nopad_bin_handler =
  NULL, /* init */
  my_strnncoll_eucjpms_bin,
  my_strnncollsp_eucjpms_nopad_bin,
+ my_strnncollsp_nchars_eucjpms_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,

strings/ctype-gb2312.c

@@ -6362,6 +6362,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gb2312_chinese_ci=
  NULL, /* init */
  my_strnncoll_gb2312_chinese_ci,
  my_strnncollsp_gb2312_chinese_ci,
+ my_strnncollsp_nchars_gb2312_chinese_ci,
  my_strnxfrm_mb, /* strnxfrm */
  my_strnxfrmlen_simple,
  my_like_range_mb, /* like_range */
@@ -6378,6 +6379,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gb2312_bin=
  NULL, /* init */
  my_strnncoll_gb2312_bin,
  my_strnncollsp_gb2312_bin,
+ my_strnncollsp_nchars_gb2312_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -6394,6 +6396,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gb2312_chinese_nopad_ci=
  NULL, /* init */
  my_strnncoll_gb2312_chinese_ci,
  my_strnncollsp_gb2312_chinese_nopad_ci,
+ my_strnncollsp_nchars_gb2312_chinese_nopad_ci,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -6410,6 +6413,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gb2312_nopad_bin=
  NULL, /* init */
  my_strnncoll_gb2312_bin,
  my_strnncollsp_gb2312_nopad_bin,
+ my_strnncollsp_nchars_gb2312_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,

strings/ctype-gbk.c

@@ -10645,6 +10645,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gbk_chinese_ci=
  NULL, /* init */
  my_strnncoll_gbk_chinese_ci,
  my_strnncollsp_gbk_chinese_ci,
+ my_strnncollsp_nchars_gbk_chinese_ci,
  my_strnxfrm_gbk_chinese_ci,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -10661,6 +10662,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gbk_bin=
  NULL, /* init */
  my_strnncoll_gbk_bin,
  my_strnncollsp_gbk_bin,
+ my_strnncollsp_nchars_gbk_bin,
  my_strnxfrm_mb,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -10677,6 +10679,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gbk_chinese_nopad_ci=
  NULL, /* init */
  my_strnncoll_gbk_chinese_ci,
  my_strnncollsp_gbk_chinese_nopad_ci,
+ my_strnncollsp_nchars_gbk_chinese_nopad_ci,
  my_strnxfrm_gbk_chinese_nopad_ci,
  my_strnxfrmlen_simple,
  my_like_range_mb,
@@ -10693,6 +10696,7 @@ static MY_COLLATION_HANDLER my_collation_handler_gbk_nopad_bin=
  NULL, /* init */
  my_strnncoll_gbk_bin,
  my_strnncollsp_gbk_nopad_bin,
+ my_strnncollsp_nchars_gbk_nopad_bin,
  my_strnxfrm_mb_nopad,
  my_strnxfrmlen_simple,
  my_like_range_mb,