diff options
Diffstat (limited to 'strings/strcoll.ic')
| -rw-r--r-- | strings/strcoll.ic | 267 |
1 files changed, 266 insertions, 1 deletions
diff --git a/strings/strcoll.ic b/strings/strcoll.ic index c647a5ef57e..9dfccb9018c 100644 --- a/strings/strcoll.ic +++ b/strings/strcoll.ic @@ -15,11 +15,18 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ - #ifndef MY_FUNCTION_NAME #error MY_FUNCTION_NAME is not defined #endif +/* + Define strnncoll() and strnncollsp() by default, + unless "#define DEFINE_STRNNCOLL 0" is specified. +*/ +#ifndef DEFINE_STRNNCOLL +#define DEFINE_STRNNCOLL 1 +#endif + /* The weight for automatically padded spaces when comparing strings with @@ -54,6 +61,8 @@ #endif +#if DEFINE_STRNNCOLL + /** Scan a valid character, or a bad byte, or an auto-padded space from a string and calculate the weight of the scanned sequence. @@ -278,6 +287,8 @@ MY_FUNCTION_NAME(strnncollsp)(CHARSET_INFO *cs __attribute__((unused)), } #endif +#endif /* DEFINE_STRNNCOLL */ + #ifdef DEFINE_STRNXFRM #ifndef WEIGHT_MB2_FRM @@ -322,11 +333,261 @@ MY_FUNCTION_NAME(strnxfrm)(CHARSET_INFO *cs, #endif /* DEFINE_STRNXFRM */ +#if defined(DEFINE_STRNXFRM_UNICODE) || defined(DEFINE_STRNXFRM_UNICODE_NOPAD) + +/* + Store sorting weights using 2 bytes per character. + + This function is shared between + - utf8mb3_general_ci, utf8_bin, ucs2_general_ci, ucs2_bin + which support BMP only (U+0000..U+FFFF). + - utf8mb4_general_ci, utf16_general_ci, utf32_general_ci, + which map all supplementary characters to weight 0xFFFD. +*/ + +#ifndef MY_MB_WC +#error MY_MB_WC must be defined for DEFINE_STRNXFRM_UNICODE +#endif + +#ifndef OPTIMIZE_ASCII +#error OPTIMIZE_ASCII must be defined for DEFINE_STRNXFRM_UNICODE +#endif + +#ifndef UNICASE_MAXCHAR +#error UNICASE_MAXCHAR must be defined for DEFINE_STRNXFRM_UNICODE +#endif + +#ifndef UNICASE_PAGE0 +#error UNICASE_PAGE0 must be defined for DEFINE_STRNXFRM_UNICODE +#endif + +#ifndef UNICASE_PAGES +#error UNICASE_PAGES must be defined for DEFINE_STRNXFRM_UNICODE +#endif + + +static size_t +MY_FUNCTION_NAME(strnxfrm_internal)(CHARSET_INFO *cs, + uchar *dst, uchar *de, + uint *nweights, + const uchar *src, const uchar *se) +{ + my_wc_t UNINIT_VAR(wc); + uchar *dst0= dst; + + DBUG_ASSERT(src || !se); + DBUG_ASSERT((cs->state & MY_CS_LOWER_SORT) == 0); + DBUG_ASSERT(0x7F <= UNICASE_MAXCHAR); + + for (; dst < de && *nweights; (*nweights)--) + { + int res; +#if OPTIMIZE_ASCII + if (src >= se) + break; + if (src[0] <= 0x7F) + { + wc= UNICASE_PAGE0[*src++].sort; + PUT_WC_BE2_HAVE_1BYTE(dst, de, wc); + continue; + } +#endif + if ((res= MY_MB_WC(cs, &wc, src, se)) <= 0) + break; + src+= res; + if (wc <= UNICASE_MAXCHAR) + { + MY_UNICASE_CHARACTER *page; + if ((page= UNICASE_PAGES[wc >> 8])) + wc= page[wc & 0xFF].sort; + } + else + wc= MY_CS_REPLACEMENT_CHARACTER; + PUT_WC_BE2_HAVE_1BYTE(dst, de, wc); + } + return dst - dst0; +} + + +static size_t +MY_FUNCTION_NAME(strnxfrm)(CHARSET_INFO *cs, + uchar *dst, size_t dstlen, uint nweights, + const uchar *src, size_t srclen, uint flags) +{ + uchar *dst0= dst; + uchar *de= dst + dstlen; + dst+= MY_FUNCTION_NAME(strnxfrm_internal)(cs, dst, de, &nweights, + src, src + srclen); + DBUG_ASSERT(dst <= de); /* Safety */ + + if (dst < de && nweights && (flags & MY_STRXFRM_PAD_WITH_SPACE)) + dst+= my_strxfrm_pad_nweights_unicode(dst, de, nweights); + + my_strxfrm_desc_and_reverse(dst0, dst, flags, 0); + + if ((flags & MY_STRXFRM_PAD_TO_MAXLEN) && dst < de) + dst+= my_strxfrm_pad_unicode(dst, de); + return dst - dst0; +} + + +#ifdef DEFINE_STRNXFRM_UNICODE_NOPAD +static size_t +MY_FUNCTION_NAME(strnxfrm_nopad)(CHARSET_INFO *cs, + uchar *dst, size_t dstlen, + uint nweights, + const uchar *src, size_t srclen, uint flags) +{ + uchar *dst0= dst; + uchar *de= dst + dstlen; + dst+= MY_FUNCTION_NAME(strnxfrm_internal)(cs, dst, de, &nweights, + src, src + srclen); + DBUG_ASSERT(dst <= de); /* Safety */ + + if (dst < de && nweights && (flags & MY_STRXFRM_PAD_WITH_SPACE)) + { + size_t len= de - dst; + set_if_smaller(len, nweights * 2); + memset(dst, 0x00, len); + dst+= len; + } + + my_strxfrm_desc_and_reverse(dst0, dst, flags, 0); + + if ((flags & MY_STRXFRM_PAD_TO_MAXLEN) && dst < de) + { + memset(dst, 0x00, de - dst); + dst= de; + } + return dst - dst0; +} +#endif + +#endif /* DEFINE_STRNXFRM_UNICODE || DEFINE_STRNXFRM_UNICODE_NOPAD */ + + + +#ifdef DEFINE_STRNXFRM_UNICODE_BIN2 + +/* + Store sorting weights using 2 bytes per character. + + These functions are shared between + - utf8mb3_general_ci, utf8_bin, ucs2_general_ci, ucs2_bin + which support BMP only (U+0000..U+FFFF). + - utf8mb4_general_ci, utf16_general_ci, utf32_general_ci, + which map all supplementary characters to weight 0xFFFD. +*/ + +#ifndef MY_MB_WC +#error MY_MB_WC must be defined for DEFINE_STRNXFRM_UNICODE_BIN2 +#endif + +#ifndef OPTIMIZE_ASCII +#error OPTIMIZE_ASCII must be defined for DEFINE_STRNXFRM_UNICODE_BIN2 +#endif + + +static size_t +MY_FUNCTION_NAME(strnxfrm_internal)(CHARSET_INFO *cs, + uchar *dst, uchar *de, + uint *nweights, + const uchar *src, + const uchar *se) +{ + my_wc_t UNINIT_VAR(wc); + uchar *dst0= dst; + + DBUG_ASSERT(src || !se); + + for (; dst < de && *nweights; (*nweights)--) + { + int res; +#if OPTIMIZE_ASCII + if (src >= se) + break; + if (src[0] <= 0x7F) + { + wc= *src++; + PUT_WC_BE2_HAVE_1BYTE(dst, de, wc); + continue; + } +#endif + if ((res= MY_MB_WC(cs, &wc, src, se)) <= 0) + break; + src+= res; + if (wc > 0xFFFF) + wc= MY_CS_REPLACEMENT_CHARACTER; + PUT_WC_BE2_HAVE_1BYTE(dst, de, wc); + } + return dst - dst0; +} + + +static size_t +MY_FUNCTION_NAME(strnxfrm)(CHARSET_INFO *cs, + uchar *dst, size_t dstlen, uint nweights, + const uchar *src, size_t srclen, uint flags) +{ + uchar *dst0= dst; + uchar *de= dst + dstlen; + dst+= MY_FUNCTION_NAME(strnxfrm_internal)(cs, dst, de, &nweights, + src, src + srclen); + DBUG_ASSERT(dst <= de); /* Safety */ + + if (dst < de && nweights && (flags & MY_STRXFRM_PAD_WITH_SPACE)) + dst+= my_strxfrm_pad_nweights_unicode(dst, de, nweights); + + my_strxfrm_desc_and_reverse(dst0, dst, flags, 0); + + if ((flags & MY_STRXFRM_PAD_TO_MAXLEN) && dst < de) + dst+= my_strxfrm_pad_unicode(dst, de); + return dst - dst0; +} + + +static size_t +MY_FUNCTION_NAME(strnxfrm_nopad)(CHARSET_INFO *cs, + uchar *dst, size_t dstlen, uint nweights, + const uchar *src, size_t srclen, uint flags) +{ + uchar *dst0= dst; + uchar *de= dst + dstlen; + dst+= MY_FUNCTION_NAME(strnxfrm_internal)(cs, dst, de, &nweights, + src, src + srclen); + DBUG_ASSERT(dst <= de); /* Safety */ + + if (dst < de && nweights && (flags & MY_STRXFRM_PAD_WITH_SPACE)) + { + size_t len= de - dst; + set_if_smaller(len, nweights * 2); + memset(dst, 0x00, len); + dst+= len; + } + + my_strxfrm_desc_and_reverse(dst0, dst, flags, 0); + + if ((flags & MY_STRXFRM_PAD_TO_MAXLEN) && dst < de) + { + memset(dst, 0x00, de - dst); + dst= de; + } + return dst - dst0; +} + +#endif /* DEFINE_STRNXFRM_UNICODE_BIN2 */ + + /* We usually include this file at least two times from the same source file, for the _ci and the _bin collations. Prepare for the second inclusion. */ #undef MY_FUNCTION_NAME +#undef MY_MB_WC +#undef OPTIMIZE_ASCII +#undef UNICASE_MAXCHAR +#undef UNICASE_PAGE0 +#undef UNICASE_PAGES #undef WEIGHT_ILSEQ #undef WEIGHT_MB1 #undef WEIGHT_MB2 @@ -335,4 +596,8 @@ MY_FUNCTION_NAME(strnxfrm)(CHARSET_INFO *cs, #undef WEIGHT_PAD_SPACE #undef WEIGHT_MB2_FRM #undef DEFINE_STRNXFRM +#undef DEFINE_STRNXFRM_UNICODE +#undef DEFINE_STRNXFRM_UNICODE_NOPAD +#undef DEFINE_STRNXFRM_UNICODE_BIN2 +#undef DEFINE_STRNNCOLL #undef DEFINE_STRNNCOLLSP_NOPAD |