Hello, the work is finished. Version 4 of the patch is attached to this message.
- Add rough description of the algorithm as comment to pg_utf8_increment() and pg_eucjp_increment(). - Fixed a bug of pg_utf8_increment() found while working. pg_(utf8|eucjp)_increment are retested on whole valid code points to be properly handled. - The comment previously pointed out as being wrong in grammar is left untouched. I'm sorry to bother you with my poor English. At Tue, 11 Oct 2011 16:55:00 +0900 (JST), Kyotaro HORIGUCHI <horiguchi.kyot...@oss.ntt.co.jp> wrote > > One thing I still think it would be useful to add, > > though, is some comments to pg_utf8_increment() and > > pg_eucjp_increment() describing the algorithm being used. Can you > > take a crack at that? > > Yes I'll do it in a day or two. Regards, -- Kyotaro Horiguchi NTT Open Source Software Center
diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 8ceea82..59f8c37 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5664,6 +5664,19 @@ pattern_selectivity(Const *patt, Pattern_Type ptype) /* + * This function is "character increment" function for bytea used in + * make_greater_string() that has same interface with pg_wchar_tbl.charinc. + */ +static bool +byte_increment(unsigned char *ptr, int len) +{ + if (*ptr >= 255) return false; + + (*ptr)++; + return true; +} + +/* * Try to generate a string greater than the given string or any * string it is a prefix of. If successful, return a palloc'd string * in the form of a Const node; else return NULL. @@ -5702,6 +5715,7 @@ make_greater_string(const Const *str_const, FmgrInfo *ltproc, Oid collation) int len; Datum cmpstr; text *cmptxt = NULL; + character_incrementer charincfunc; /* * Get a modifiable copy of the prefix string in C-string format, and set @@ -5763,29 +5777,33 @@ make_greater_string(const Const *str_const, FmgrInfo *ltproc, Oid collation) } } + if (datatype == BYTEAOID) + charincfunc = &byte_increment; + else + charincfunc = pg_database_encoding_character_incrementer(); + while (len > 0) { - unsigned char *lastchar = (unsigned char *) (workstr + len - 1); - unsigned char savelastchar = *lastchar; + int charlen = 1; + unsigned char *lastchar; + Const *workstr_const; + + if (datatype != BYTEAOID) + charlen = len - pg_mbcliplen(workstr, len, len - 1); + + lastchar = (unsigned char *) (workstr + len - charlen); /* - * Try to generate a larger string by incrementing the last byte. + * Try to generate a larger string by incrementing the last byte or + * character. */ - while (*lastchar < (unsigned char) 255) - { - Const *workstr_const; - - (*lastchar)++; - if (datatype != BYTEAOID) - { - /* do not generate invalid encoding sequences */ - if (!pg_verifymbstr(workstr, len, true)) - continue; - workstr_const = string_to_const(workstr, datatype); - } - else + if (charincfunc(lastchar, charlen)) + { + if (datatype == BYTEAOID) workstr_const = string_to_bytea_const(workstr, len); + else + workstr_const = string_to_const(workstr, datatype); if (DatumGetBool(FunctionCall2Coll(ltproc, collation, @@ -5804,20 +5822,10 @@ make_greater_string(const Const *str_const, FmgrInfo *ltproc, Oid collation) pfree(workstr_const); } - /* restore last byte so we don't confuse pg_mbcliplen */ - *lastchar = savelastchar; - /* - * Truncate off the last character, which might be more than 1 byte, - * depending on the character encoding. + * Truncate off the last character or byte. */ - if (datatype != BYTEAOID && pg_database_encoding_max_length() > 1) - len = pg_mbcliplen(workstr, len, len - 1); - else - len -= 1; - - if (datatype != BYTEAOID) - workstr[len] = '\0'; + len -= charlen; } /* Failed... */ diff --git a/src/backend/utils/mb/wchar.c b/src/backend/utils/mb/wchar.c index f23732f..a213636 100644 --- a/src/backend/utils/mb/wchar.c +++ b/src/backend/utils/mb/wchar.c @@ -1336,6 +1336,264 @@ pg_utf8_islegal(const unsigned char *source, int length) /* *------------------------------------------------------------------- + * character incrementer + * + * These functions accept "charptr", a pointer to the first byte of a + * maybe-multibyte character. Try `increment' the character and return + * true if successed. If these functions returns false, the character + * should be untouched. These functions must be implemented in + * correspondence with verifiers, in other words, the rewrited + * character by this function must pass the check by pg_*_verifier() + * if returns true. + * ------------------------------------------------------------------- + */ + +#ifndef FRONTEND +static bool +pg_generic_charinc(unsigned char *charptr, int len) +{ + unsigned char *lastchar = (unsigned char *) (charptr + len - 1); + unsigned char savelastchar = *lastchar; + const char *const_charptr = (const char *)charptr; + + while (*lastchar < (unsigned char) 255) + { + (*lastchar)++; + if (!pg_verifymbstr(const_charptr, len, true)) + continue; + return true; + } + + *lastchar = savelastchar; + return false; +} + +/* + * Calculate the valid UTF-8 byte sequence that is the smallest of the + * greater than the given valid UTF-8 byte sequence for single + * character. The word `valid' here means that pg_utf8_islegal() says + * that it is valid. pg_utf8_increment does this roughly with the + * following algorithm. + * + * - If the length of the sequence is 1, the byte is between 0x01 and + * 0x7f. It is simplly incremented if less than 0x7f. + * + * - Otherwise the bytes after the first one ranges between 0x80 and + * 0xbf. The sequence is incremented from the last byte and the + * carry is propagated up to the first byte. + * + * Only the UTF-8 byte sequences corresponding to the following + * unicode point which have no character allocated make this function + * to fail in spite of its validity. + * + * U+0007F, U+007FF, U+0FFFF, U+10FFFF + * + */ +static bool +pg_utf8_increment(unsigned char *charptr, int length) +{ + unsigned char a; + unsigned char bak[4]; + unsigned char limit; + + switch (length) + { + default: + /* reject lengths 5 and 6 for now */ + return false; + case 4: + bak[3] = charptr[3]; + a = charptr[3]; + if (a < 0xBF) + { + charptr[3]++; + break; + } + charptr[3] = 0x80; + /* FALL THRU */ + case 3: + bak[2] = charptr[2]; + a = charptr[2]; + if (a < 0xBF) + { + charptr[2]++; + break; + } + charptr[2] = 0x80; + /* FALL THRU */ + case 2: + bak[1] = charptr[1]; + a = charptr[1]; + switch (*charptr) + { + case 0xED: + limit = 0x9F; + break; + + case 0xF4: + limit = 0x8F; + break; + + default: + limit = 0xBF; + + } + + if (a < limit) + { + charptr[1]++; + break; + } + + charptr[1] = 0x80; + /* FALL THRU */ + case 1: + bak[0] = *charptr; + a = *charptr; + if (a == 0x7F || a == 0xDF || a == 0xEF || a == 0xF4) + { + /* Rewinding modified bytes and return fail. length is + * confirmed to be between 1 and 4 here. */ + memcpy(charptr, bak, length); + return false; + } + charptr[0]++; + break; + } + + return true; +} + +/* Calculate the valid EUC-JP byte sequence that is the smallest of + * the greater than the given valid EUC-JP byte sequence for single + * character. The word `valid' here means that pg_eucjp_verifier() + * syas that it is valid. This function does this roughly with the + * following algorithm. + * + * - If the sequence starts with SS2(0x8e), it must be a two-byte + * sequence representing JIS X 0201 characters with the second byte + * ranges between 0xa1 and 0xde. It is simplly incremented at the + * second byte if less than 0xde, and otherwise rewrite whole the + * sequence to 0xa1 0xa1 as the result. + * + * - If the sequence starts with SS3(0x8f), it must be a three-byte + * sequence which the last two bytes ranges between 0xa1 and 0xfe. + * The bytes are incremented from the last within the range and + * propagate the carry up to the first byte. + * + * - If the sequence starts with the values other than the aboves and + * its MSB is set, it must be a two-byte sequence representing JIS + * X 0208 characters with both bytes ranges between 0xa1 and 0xfe. + * The bytes are incremented from the latter within the range and + * propagate the carry up to the first byte. + * + * - Otherwise the sequence is consists of single byte representing + * ASCII characters. It is incremented up to 0x7f. + * + * Only three EUC-JP byte sequences shown below which has no character + * allocated make this function to fail in spite of its validity. + * + * 0x7f, 0xfe 0xfe, 0x8f 0xfe 0xfe + * + */ +static bool +pg_eucjp_increment(unsigned char *charptr, int length) +{ + unsigned char bak[3]; + unsigned char c1, c2; + signed int i; + + c1 = *charptr; + + switch (c1) + { + case SS2: /* JIS X 0201 */ + if (length != 2) return false; + + c2 = charptr[1]; + + if (c2 > 0xde) + charptr[0] = charptr[1] = 0xa1; + else if (c2 < 0xa1) + charptr[1] = 0xa1; + else + charptr[1]++; + + break; + + case SS3: /* JIS X 0212 */ + if (length != 3) return false; + + for (i = 2 ; i > 0 ; i--) + { + bak[i] = charptr[i]; + c2 = charptr[i]; + if (c2 < 0xa1) + { + charptr[i] = 0xa1; + return true; + } + else if (c2 < 0xfe) + { + charptr[i]++; + break; + } + charptr[i] = 0xa1; + } + + + if (i == 0) /* Out of 3-byte code region */ + { + charptr[1] = bak[1]; + charptr[2] = bak[2]; + return false; + } + + break; + + default: + if (IS_HIGHBIT_SET(c1)) /* JIS X 0208? */ + { + if (length != 2) return false; + + for (i = 1 ; i >= 0 ; i--) /* i must be signed */ + { + bak[i] = charptr[i]; + c2 = charptr[i]; + if (c2 < 0xa1) + { + charptr[i] = 0xa1; + return true; + } + else if (c2 < 0xfe) + { + charptr[i]++; + break; + } + charptr[i] = 0xa1; + } + + if (i < 0) /* Out of 2 byte code region */ + { + charptr[0] = bak[0]; + charptr[1] = bak[1]; + return false; + } + } + else + { /* ASCII, single byte */ + if (c1 > 0x7e) + return false; + (*charptr)++; + } + } + + return true; +} +#endif + +/* + *------------------------------------------------------------------- * encoding info table * XXX must be sorted by the same order as enum pg_enc (in mb/pg_wchar.h) *------------------------------------------------------------------- @@ -1459,6 +1717,25 @@ pg_database_encoding_max_length(void) } /* + * give the character incrementer for the encoding for the current database + */ +character_incrementer +pg_database_encoding_character_incrementer(void) +{ + switch (GetDatabaseEncoding()) + { + case PG_UTF8: + return pg_utf8_increment; + + case PG_EUC_JP: + return pg_eucjp_increment; + + default: + return pg_generic_charinc; + } +} + +/* * Verify mbstr to make sure that it is validly encoded in the current * database encoding. Otherwise same as pg_verify_mbstr(). */ diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h index 826c7af..728175c 100644 --- a/src/include/mb/pg_wchar.h +++ b/src/include/mb/pg_wchar.h @@ -284,6 +284,8 @@ typedef int (*mblen_converter) (const unsigned char *mbstr); typedef int (*mbdisplaylen_converter) (const unsigned char *mbstr); +typedef bool (*character_incrementer) (unsigned char *mbstr, int len); + typedef int (*mbverifier) (const unsigned char *mbstr, int len); typedef struct @@ -389,6 +391,7 @@ extern int pg_encoding_mbcliplen(int encoding, const char *mbstr, extern int pg_mbcharcliplen(const char *mbstr, int len, int imit); extern int pg_encoding_max_length(int encoding); extern int pg_database_encoding_max_length(void); +extern character_incrementer pg_database_encoding_character_incrementer(void); extern int PrepareClientEncoding(int encoding); extern int SetClientEncoding(int encoding);
-- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers