Commit 729f819d authored by sasha@asksasha.com's avatar sasha@asksasha.com

Merge spachev@bk-internal.mysql.com:/home/bk/mysql-5.1-new

into  asksasha.com:/reiser-data/mysql-dev/mysql-5.1-new
parents ed2a86c1 9a769048
......@@ -1611,3 +1611,4 @@ vio/viotest-sslconnect.cpp
vio/viotest.cpp
zlib/*.ds?
zlib/*.vcproj
BitKeeper/etc/RESYNC_TREE
......@@ -187,6 +187,35 @@ static test_file* cur_file;
static test_file* file_stack_end;
uint start_lineno; /* Start line of query */
/* Stores regex substitutions */
struct st_regex
{
char* pattern; /* Pattern to be replaced */
char* replace; /* String or expression to replace the pattern with */
int icase; /* true if the match is case insensitive */
};
struct st_replace_regex
{
DYNAMIC_ARRAY regex_arr; /* stores a list of st_regex subsitutions */
/*
Temporary storage areas for substitutions. To reduce unnessary copying
and memory freeing/allocation, we pre-allocate two buffers, and alternate
their use, one for input/one for output, the roles changing on the next
st_regex substition. At the end of substitutions buf points to the
one containing the final result.
*/
char* buf;
char* even_buf;
uint even_buf_len;
char* odd_buf;
uint odd_buf_len;
};
struct st_replace_regex *glob_replace_regex= 0;
static char TMPDIR[FN_REFLEN];
static char delimiter[MAX_DELIMITER]= DEFAULT_DELIMITER;
static uint delimiter_length= 1;
......@@ -226,6 +255,9 @@ static void init_re(void);
static int match_re(my_regex_t *, char *);
static void free_re(void);
static int reg_replace(char** buf_p, int* buf_len_p, char *pattern, char *replace,
char *string, int icase);
static const char *embedded_server_groups[]=
{
"server",
......@@ -336,6 +368,7 @@ Q_EXIT,
Q_DISABLE_RECONNECT, Q_ENABLE_RECONNECT,
Q_IF,
Q_DISABLE_PARSING, Q_ENABLE_PARSING,
Q_REPLACE_REGEX,
Q_UNKNOWN, /* Unknown command. */
Q_COMMENT, /* Comments, ignored. */
......@@ -429,6 +462,7 @@ const char *command_names[]=
"if",
"disable_parsing",
"enable_parsing",
"replace_regex",
0
};
......@@ -469,6 +503,7 @@ struct st_replace *init_replace(my_string *from, my_string *to, uint count,
uint replace_strings(struct st_replace *rep, my_string *start,
uint *max_length, const char *from);
void free_replace();
static void free_replace_regex();
static int insert_pointer_name(reg1 POINTER_ARRAY *pa,my_string name);
void free_pointer_array(POINTER_ARRAY *pa);
static int initialize_replace_buffer(void);
......@@ -1927,6 +1962,204 @@ static char *get_string(char **to_ptr, char **from_ptr,
DBUG_RETURN(start);
}
/*
Finds the next (non-escaped) '/' in the expression.
(If the character '/' is needed, it can be escaped using '\'.)
*/
#define PARSE_REGEX_ARG \
while (p < expr_end) \
{\
char c= *p;\
if (c == '/')\
{\
if (last_c == '\\')\
{\
buf_p[-1]= '/';\
}\
else\
{\
*buf_p++ = 0;\
break;\
} \
} \
else\
*buf_p++ = c;\
\
last_c= c;\
p++;\
} \
/*
Initializes the regular substitution expression to be used in the
result output of test.
Returns: st_replace_regex struct with pairs of substitutions
*/
static struct st_replace_regex* init_replace_regex(char* expr)
{
struct st_replace_regex* res;
char* buf,*expr_end;
char* p;
char* buf_p;
uint expr_len= strlen(expr);
char last_c = 0;
struct st_regex reg;
/* my_malloc() will die on fail with MY_FAE */
res=(struct st_replace_regex*)my_malloc(
sizeof(*res)+expr_len ,MYF(MY_FAE+MY_WME));
my_init_dynamic_array(&res->regex_arr,sizeof(struct st_regex),128,128);
buf= (char*)res + sizeof(*res);
expr_end= expr + expr_len;
p= expr;
buf_p= buf;
/* for each regexp substitution statement */
while (p < expr_end)
{
bzero(&reg,sizeof(reg));
/* find the start of the statement */
while (p < expr_end)
{
if (*p == '/')
break;
p++;
}
if (p == expr_end || ++p == expr_end)
{
if (res->regex_arr.elements)
break;
else
goto err;
}
/* we found the start */
reg.pattern= buf_p;
/* Find first argument -- pattern string to be removed */
PARSE_REGEX_ARG
if (p == expr_end || ++p == expr_end)
goto err;
/* buf_p now points to the replacement pattern terminated with \0 */
reg.replace= buf_p;
/* Find second argument -- replace string to replace pattern */
PARSE_REGEX_ARG
if (p == expr_end)
goto err;
/* skip the ending '/' in the statement */
p++;
/* Check if we should do matching case insensitive */
if (p < expr_end && *p == 'i')
reg.icase= 1;
/* done parsing the statement, now place it in regex_arr */
if (insert_dynamic(&res->regex_arr,(gptr) &reg))
die("Out of memory");
}
res->odd_buf_len= res->even_buf_len= 8192;
res->even_buf= (char*)my_malloc(res->even_buf_len,MYF(MY_WME+MY_FAE));
res->odd_buf= (char*)my_malloc(res->odd_buf_len,MYF(MY_WME+MY_FAE));
res->buf= res->even_buf;
return res;
err:
my_free((gptr)res,0);
die("Error parsing replace_regex \"%s\"", expr);
return 0;
}
/*
Execute all substitutions on val.
Returns: true if substituition was made, false otherwise
Side-effect: Sets r->buf to be the buffer with all substitutions done.
IN:
struct st_replace_regex* r
char* val
Out:
struct st_replace_regex* r
r->buf points at the resulting buffer
r->even_buf and r->odd_buf might have been reallocated
r->even_buf_len and r->odd_buf_len might have been changed
TODO: at some point figure out if there is a way to do everything
in one pass
*/
static int multi_reg_replace(struct st_replace_regex* r,char* val)
{
uint i;
char* in_buf, *out_buf;
int* buf_len_p;
in_buf= val;
out_buf= r->even_buf;
buf_len_p= &r->even_buf_len;
r->buf= 0;
/* For each substitution, do the replace */
for (i= 0; i < r->regex_arr.elements; i++)
{
struct st_regex re;
char* save_out_buf= out_buf;
get_dynamic(&r->regex_arr,(gptr)&re,i);
if (!reg_replace(&out_buf, buf_len_p, re.pattern, re.replace,
in_buf, re.icase))
{
/* if the buffer has been reallocated, make adjustements */
if (save_out_buf != out_buf)
{
if (save_out_buf == r->even_buf)
r->even_buf= out_buf;
else
r->odd_buf= out_buf;
}
r->buf= out_buf;
if (in_buf == val)
in_buf= r->odd_buf;
swap_variables(char*,in_buf,out_buf);
buf_len_p= (out_buf == r->even_buf) ? &r->even_buf_len :
&r->odd_buf_len;
}
}
return (r->buf == 0);
}
/*
Parse the regular expression to be used in all result files
from now on.
The syntax is --replace_regex /from/to/i /from/to/i ...
i means case-insensitive match. If omitted, the match is
case-sensitive
*/
static void get_replace_regex(struct st_query *q)
{
char *expr= q->first_argument;
free_replace_regex();
if (!(glob_replace_regex=init_replace_regex(expr)))
die("Could not init replace_regex");
q->last_argument= q->end;
}
/*
Get arguments for replace. The syntax is:
......@@ -1979,6 +2212,18 @@ static void get_replace(struct st_query *q)
DBUG_VOID_RETURN;
}
static void free_replace_regex()
{
if (glob_replace_regex)
{
my_free(glob_replace_regex->even_buf,MYF(MY_ALLOW_ZERO_PTR));
my_free(glob_replace_regex->odd_buf,MYF(MY_ALLOW_ZERO_PTR));
my_free((char*) glob_replace_regex,MYF(0));
glob_replace_regex=0;
}
}
void free_replace()
{
DBUG_ENTER("free_replace");
......@@ -2277,6 +2522,7 @@ int connect_n_handle_errors(struct st_query *q, MYSQL* con, const char* host,
err:
free_replace();
free_replace_regex();
return error;
}
......@@ -3126,6 +3372,214 @@ void dump_result_to_reject_file(const char *record_file, char *buf, int size)
str_to_file(fn_format(reject_file, record_file,"",".reject",2), buf, size);
}
static void check_regerr(my_regex_t* r, int err)
{
char err_buf[1024];
if (err)
{
my_regerror(err,r,err_buf,sizeof(err_buf));
die("Regex error: %s\n", err_buf);
}
}
/*
auxiluary macro used by reg_replace
makes sure the result buffer has sufficient length
*/
#define SECURE_REG_BUF if (buf_len < need_buf_len)\
{\
int off= res_p - buf;\
buf= (char*)my_realloc(buf,need_buf_len,MYF(MY_WME+MY_FAE));\
res_p= buf + off;\
buf_len= need_buf_len;\
}\
/*
Performs a regex substitution
IN:
buf_p - result buffer pointer. Will change if reallocated
buf_len_p - result buffer length. Will change if the buffer is reallocated
pattern - regexp pattern to match
replace - replacement expression
string - the string to perform substituions in
icase - flag, if set to 1 the match is case insensitive
*/
static int reg_replace(char** buf_p, int* buf_len_p, char *pattern,
char *replace, char *string, int icase)
{
my_regex_t r;
my_regmatch_t* subs;
char* buf_end, *replace_end;
char* buf= *buf_p;
int len;
int buf_len,need_buf_len;
int cflags= REG_EXTENDED;
int err_code;
char* res_p,*str_p,*str_end;
buf_len= *buf_len_p;
len= strlen(string);
str_end= string + len;
/* start with a buffer of a reasonable size that hopefully will not
need to be reallocated
*/
need_buf_len= len * 2 + 1;
res_p= buf;
SECURE_REG_BUF
buf_end = buf + buf_len;
if (icase)
cflags |= REG_ICASE;
if ((err_code=my_regcomp(&r,pattern,cflags,&my_charset_latin1)))
{
check_regerr(&r,err_code);
return 1;
}
subs= (my_regmatch_t*)my_malloc(sizeof(my_regmatch_t) * (r.re_nsub+1),
MYF(MY_WME+MY_FAE));
*res_p= 0;
str_p= string;
replace_end= replace + strlen(replace);
/* for each pattern match instance perform a replacement */
while (!err_code)
{
/* find the match */
err_code= my_regexec(&r,str_p, r.re_nsub+1, subs,
(str_p == string) ? REG_NOTBOL : 0);
/* if regular expression error (eg. bad syntax, or out of memory) */
if (err_code && err_code != REG_NOMATCH)
{
check_regerr(&r,err_code);
my_regfree(&r);
return 1;
}
/* if match found */
if (!err_code)
{
char* expr_p= replace;
int c;
/*
we need at least what we have so far in the buffer + the part
before this match
*/
need_buf_len= (res_p - buf) + subs[0].rm_so;
/* on this pass, calculate the memory for the result buffer */
while (expr_p < replace_end)
{
int back_ref_num= -1;
c= *expr_p;
if (c == '\\' && expr_p + 1 < replace_end)
{
back_ref_num= expr_p[1] - '0';
}
/* found a valid back_ref (eg. \1)*/
if (back_ref_num >= 0 && back_ref_num <= (int)r.re_nsub)
{
int start_off,end_off;
if ((start_off=subs[back_ref_num].rm_so) > -1 &&
(end_off=subs[back_ref_num].rm_eo) > -1)
{
need_buf_len += (end_off - start_off);
}
expr_p += 2;
}
else
{
expr_p++;
need_buf_len++;
}
}
need_buf_len++;
/*
now that we know the size of the buffer,
make sure it is big enough
*/
SECURE_REG_BUF
/* copy the pre-match part */
if (subs[0].rm_so)
{
memcpy(res_p,str_p,subs[0].rm_so);
res_p += subs[0].rm_so;
}
expr_p= replace;
/* copy the match and expand back_refs */
while (expr_p < replace_end)
{
int back_ref_num= -1;
c= *expr_p;
if (c == '\\' && expr_p + 1 < replace_end)
{
back_ref_num= expr_p[1] - '0';
}
if (back_ref_num >= 0 && back_ref_num <= (int)r.re_nsub)
{
int start_off,end_off;
if ((start_off=subs[back_ref_num].rm_so) > -1 &&
(end_off=subs[back_ref_num].rm_eo) > -1)
{
int block_len= end_off - start_off;
memcpy(res_p,str_p + start_off, block_len);
res_p += block_len;
}
expr_p += 2;
}
else
{
*res_p++ = *expr_p++;
}
}
/* handle the post-match part */
if (subs[0].rm_so == subs[0].rm_eo)
{
if (str_p + subs[0].rm_so >= str_end)
break;
str_p += subs[0].rm_eo ;
*res_p++ = *str_p++;
}
else
{
str_p += subs[0].rm_eo;
}
}
else /* no match this time, just copy the string as is */
{
int left_in_str= str_end-str_p;
need_buf_len= (res_p-buf) + left_in_str;
SECURE_REG_BUF
memcpy(res_p,str_p,left_in_str);
res_p += left_in_str;
str_p= str_end;
}
}
my_regfree(&r);
*res_p= 0;
*buf_p= buf;
*buf_len_p= buf_len;
return 0;
}
/* Append the string to ds, with optional replace */
......@@ -3139,6 +3593,16 @@ static void replace_dynstr_append_mem(DYNAMIC_STRING *ds, const char *val,
die("Out of memory in replace");
val=out_buff;
}
if (glob_replace_regex)
{
if (!multi_reg_replace(glob_replace_regex,(char*)val))
{
val= glob_replace_regex->buf;
len= strlen(val);
}
}
dynstr_append_mem(ds, val, len);
}
......@@ -3577,6 +4041,7 @@ static void run_query_normal(MYSQL *mysql, struct st_query *command,
end:
free_replace();
free_replace_regex();
/*
We save the return code (mysql_errno(mysql)) from the last call sent
......@@ -3887,7 +4352,8 @@ static void run_query_stmt(MYSQL *mysql, struct st_query *command,
end:
free_replace();
free_replace_regex();
if (!disable_warnings)
{
dynstr_free(&ds_prepare_warnings);
......@@ -3899,6 +4365,7 @@ static void run_query_stmt(MYSQL *mysql, struct st_query *command,
to the server into the mysqltest builtin variable $mysql_errno. This
variable then can be used from the test case itself.
*/
var_set_errno(mysql_stmt_errno(stmt));
#ifndef BUG15518_FIXED
mysql_stmt_close(stmt);
......@@ -4623,6 +5090,10 @@ int main(int argc, char **argv)
case Q_REPLACE:
get_replace(q);
break;
case Q_REPLACE_REGEX:
get_replace_regex(q);
break;
case Q_REPLACE_COLUMN:
get_replace_column(q);
break;
......
......@@ -419,3 +419,22 @@ select-me
ERROR 42000: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'insertz error query' at line 1
drop table t1;
drop table t1;
select "b" as col1, "c" as col2;
col1 col2
b c
select "b" as col1, "b" as col2, "c" as col3;
col1 col2 col3
b b c
seled "b" bs col1, "d" bs col2;
col1 col2
b d
select "raspberry and strawberry","blackberry","tomato";
raspberry and strawberry blackberry tomato
raspberry and strawberry blackberry tomato
mysqltest: At line 1: Error parsing replace_regex "a"
mysqltest: At line 1: Error parsing replace_regex "a;"
mysqltest: At line 1: Error parsing replace_regex "a"
mysqltest: At line 1: Error parsing replace_regex "a "
mysqltest: At line 1: Error parsing replace_regex "a b"
mysqltest: At line 1: Error parsing replace_regex "/a b c"
mysqltest: At line 1: Error parsing replace_regex "/a /b c "
......@@ -1061,3 +1061,30 @@ drop table t1;
drop table t1;
# test for replace_regex
--replace_regex /at/b/
select "at" as col1, "c" as col2;
--replace_regex /at/b/i
select "at" as col1, "AT" as col2, "c" as col3;
--replace_regex /a/b/ /ct/d/
select "a" as col1, "ct" as col2;
--replace_regex /(strawberry)/raspberry and \1/ /blueberry/blackberry/ /potato/tomato/;
select "strawberry","blueberry","potato";
--error 1
--exec echo "--replace_regex a" | $MYSQL_TEST 2>&1
--error 1
--exec echo "--replace_regex a;" | $MYSQL_TEST 2>&1
--error 1
--exec echo "replace_regex a;" | $MYSQL_TEST 2>&1
--error 1
--exec echo "replace_regex a ;" | $MYSQL_TEST 2>&1
--error 1
--exec echo "replace_regex a b; echo OK;" | $MYSQL_TEST 2>&1
--error 1
--exec echo "--replace_regex /a b c" | $MYSQL_TEST 2>&1
--error 1
--exec echo "replace_regex /a /b c ;" | $MYSQL_TEST 2>&1
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment