For acceleration of spell check I needed in UDF function which be able to replace substrings based on regular expressions.

Such PHP functions are ereg_replace and preg_replace.

Trying to write UDF for MySQL I has confronted with a strange problem.

At compilation in a separate executed file my function works correctly and gives out correct results.

But when the same function I compile as UDF - I recive interesting effects.

MySQL at performance UDF through times breaks off connection. And when the connection does not break off - than MySQL gives out wrong results.

comple the UDF:

this is source:

#ifdef STANDARD #include <stdio.h> #include <string.h> #else #include <global.h> #include <my_sys.h> #endif #include <mysql.h> #include <m_ctype.h> #include <m_string.h> #include <regex.h>

#define NS 10

#ifdef HAVE_DLOPEN

char* reg_replace(char* string, char* pattern, char* replace, int icase, int extended);

extern "C" { my_bool ereg_replace_init(UDF_INIT *initid, UDF_ARGS *args, char *message); char* ereg_replace(UDF_INIT *initid, UDF_ARGS *args, char *result, unsigned long *length, char *is_null, char *error); my_bool eregi_replace_init(UDF_INIT *initid, UDF_ARGS *args, char *message); char* eregi_replace(UDF_INIT *initid, UDF_ARGS *args, char *result, unsigned long *length, char *is_null, char *error); }

my_bool ereg_replace_init(UDF_INIT *initid, UDF_ARGS *args, char *message) { if (args->arg_count != 3 || args->arg_type[0] != STRING_RESULT || args->arg_type[1] != STRING_RESULT || args->arg_type[2] != STRING_RESULT ) { strcpy(message,"ereg_replace require 3 string arguments"); return 1; } return 0; }

char *ereg_replace(UDF_INIT *initid, UDF_ARGS *args, char *result, unsigned long *length, char *is_null, char *error) { if (result = reg_replace(args->args[0], args->args[1], args->args[2], 0, 1)) { *length = strlen(result); return result; } else { *is_null = 1; return result; } }

my_bool eregi_replace_init(UDF_INIT *initid, UDF_ARGS *args, char *message) { if (args->arg_count != 3 || args->arg_type[0] != STRING_RESULT || args->arg_type[1] != STRING_RESULT || args->arg_type[2] != STRING_RESULT ) { strcpy(message,"eregi_replace require 3 string arguments"); return 1; } return 0; }

char *eregi_replace(UDF_INIT *initid, UDF_ARGS *args, char *result, unsigned long *length, char *is_null, char *error) { if (result = reg_replace(args->args[0], args->args[1], args->args[2], 1, 1)) { *length = strlen(result); return result; } else { *is_null = 1; return result; } }

char* reg_replace(char* string, char* pattern, char* replace, int icase, int extended) { char *result; regex_t* re; regmatch_t* subs; char *buf, *nbuf, *walkbuf; const char *walk; int buf_len; int pos, tmp, string_len, new_l; int err, copts = 0;

string_len = strlen(string);

if (icase) copts = REG_ICASE; if (extended) copts |= REG_EXTENDED;

re = (regex_t*) malloc(sizeof(regex_t)); subs = (regmatch_t*) malloc(sizeof(regmatch_t)*NS);

err = regcomp(re, pattern, copts);

if (err) { result = (char*) malloc(strlen(string) + 1); memcpy(result, string, strlen(string)+1); free(re); free(subs); return result; }

buf_len = 2 * string_len + 1; buf = (char*) malloc(buf_len * sizeof(char)); if (!buf) { regfree(re); free(re); free(subs); return 0; }

err = pos = 0; buf[0] = '\0';

while (!err) { err = regexec(re, &string[pos], (size_t) NS, subs, (pos ? REG_NOTBOL : 0)); if (!err) { new_l = strlen(buf) + subs[0].rm_so; walk = replace; while (*walk) if ('\\' == *walk && '0' <= walk[1] && '9' >= walk[1] && subs[walk[1] - '0'].rm_so > -1 && subs[walk[1] - '0'].rm_eo > -1) { new_l += subs[walk[1] - '0'].rm_eo - subs[walk[1] - '0'].rm_so; walk += 2; } else { new_l++; walk++; }

if (new_l + 1 > buf_len) { buf_len = 1 + buf_len + 2 * new_l; nbuf = (char*) malloc(buf_len); memcpy (nbuf, buf, strlen(buf)+1); free(buf); buf = nbuf; } tmp = strlen(buf);

buf[strlen(buf) + subs[0].rm_so + 1] = 0; memcpy(buf + strlen(buf), &string[pos], subs[0].rm_so);

walkbuf = &buf[tmp + subs[0].rm_so]; walk = replace; while (*walk) if ('\\' == *walk && '0' <= walk[1] && '9' >= walk[1] && subs[walk[1] - '0'].rm_so > -1 && subs[walk[1] - '0'].rm_eo > -1) { tmp = subs[walk[1] - '0'].rm_eo - subs[walk[1] - '0'].rm_so; memcpy (walkbuf, &string[pos + subs[walk[1] - '0'].rm_so], tmp); walkbuf += tmp; walk += 2; } else *walkbuf++ = *walk++; *walkbuf = '\0';

if (subs[0].rm_so == subs[0].rm_eo) { if (subs[0].rm_so + pos >= string_len) break; new_l = strlen (buf) + 1; if (new_l + 1 > buf_len) { buf_len = 1 + buf_len + 2 * new_l; nbuf = (char*) malloc(buf_len * sizeof(char)); memcpy (nbuf, buf, strlen(buf)+1); free(buf); buf = nbuf; } pos += subs[0].rm_eo + 1; buf [new_l-1] = string [pos-1]; buf [new_l] = '\0'; } else { pos += subs[0].rm_eo; } } else { new_l = strlen(buf) + strlen(&string[pos]); if (new_l + 1 > buf_len) { buf_len = new_l + 1; nbuf = (char*) malloc(buf_len * sizeof(char)); memcpy (nbuf, buf, strlen(buf)+1); free(buf); buf = nbuf; } memcpy(buf + strlen(buf), &string[pos], strlen(&string[pos]) + 1); } } regfree(re); free(re); free(subs);

if (strlen(buf) > 0) { result = (char*) malloc(strlen(buf) + 1); memcpy(result, buf, strlen(buf) + 1); } return result; }

#endif /* HAVE_DLOPEN */

<how to correct or work around the problem, if known (multiple lines)>

<organization of PR author (multiple lines)>

pc-linux-gnu on i686

Copyright (C) 2000 MySQL AB & MySQL Finland AB & TCX DataKonsult AB

This software comes with ABSOLUTELY NO WARRANTY. This is free software,

and you are welcome to modify and redistribute it under the GPL license

Server version 3.23.33-log

Protocol version 10

Connection Localhost via UNIX socket

UNIX socket /tmp/mysql.sock

Uptime: 1 day 1 hour 1 min 19 sec

Threads: 10 Questions: 961 Slow queries: 0 Opens: 57 Flush tables: 1 Open tables: 8 Queries per second avg: 0.011

<machine, os, target, libraries (multiple lines)>

System: Linux ukrbiz 2.4.1 #2 Thu Feb 15 16:16:51 EET 2001 i686 unknown

Architecture: i686

Some paths: /usr/bin/perl /usr/bin/make /usr/bin/gmake /usr/local/bin/gcc /usr/bin/cc

GCC: Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/2.95.2/specs

gcc version 2.95.2 19991024 (release)

Compilation info: CC='gcc' CFLAGS='' CXX='c++' CXXFLAGS='' LDFLAGS=''

LIBC:

lrwxrwxrwx 1 root root 13 Feb 13 15:16 /lib/libc.so.6 -> libc-2.2.1.so

-rwxr-xr-x 1 root root 1013224 Mar 22 2000 /lib/libc-2.1.3.so

-rwxr-xr-x 1 root root 4806279 Feb 13 15:16 /lib/libc-2.2.1.so

-rw-r--r-- 1 root root 24150394 Feb 13 15:14 /usr/lib/libc.a

-rw-r--r-- 1 root root 178 Feb 13 15:14 /usr/lib/libc.so

Configure command: ./configure --prefix=/usr/local/mysql --with-charset=win1251ukr --with-mysql d-ldflags=-rdynamic