Dynamic C String Question

[intercom5]

I'm writing a program in C, and thus have to use C strings. The
problem that I am having is I don't know how to reallocate the space
for a C string outside the scope of that string. For example:
int main(void)
{
char *string1;
string1 = malloc(6);
sprintf(string1 , "Hello");
foo(string1);

}

void foo(char *string)
{
string = realloc(string, 12);
strcat(string, " World");
}
In this example, I reallocate the space for a string declacred in
another function in the function foo. But, after it returns to main
from the function foo, the C string will only contain "Hello". The
changes from the function foo have disappered. I suspect this is
because the reallocation has gone out of scope - and I have no idea how
to keep the changes after the function returns.
Any help would greatly be appreciated.

--Sachin

[infobahn]

Charlie Gordon wrote:

"Karthik Kumar" <ka************ *******@yahoo.c om> wrote in message
news:41c90b47$1 @darkstar...

realloc is a multi-edged sword, indeed :) . Use it with care.

Or even better : realloc() is too complex to use for newbies, it has an error
prone API, it is *strongly* recommended to not use it at all !

I used to think the same thing, but the reality is that realloc is
just too useful to discard.

It does take a modicum of care to use realloc correctly but, once
you have learned how to do it, it's actually not at all complex. Anyone
who cannot master realloc is going to struggle in their programming
career, for the simple reason that they are required to master lots
of stuff that is much, much more complex than realloc.

[Flash Gordon]

On Wed, 22 Dec 2004 06:50:35 +0000 (UTC)
infobahn <in******@btint ernet.com> wrote:

Karthik Kumar wrote:

<snip>

pointer to pointer could be the answer.

int main(void)
{
char *string1;
string1 = malloc(6);
You need to test to see if malloc has failed.
sprintf(string1 , "Hello");
foo(&string1);

}

void foo(char **string)
{
*string = realloc(string, 12);

This won't work. You meant:

*string = realloc(*string , 12);

Had you tested your code before posting, you'd have discovered this.

Also, if the realloc fails you have lost your pointer to the memory you
still have allocated.

void foo(char **my_string)
{
char *tmp = realloc(*my_str ing, 12);
if (tmp == NULL) {
/* handle error */
}
else {
*my_string = tmp
}
}

I've also changed the name because identifiers starting with str
followed by another letter, such as string, are reserved.
--
Flash Gordon
Living in interesting times.
Although my email address says spam, it is real and I read it.

[Al Bowers]

infobahn wrote:

The following code is based heavily on your own code; I have changed
the indentation to make it more readable to others, and added error
checking, but I haven't "fixed" the code to my own style and preference,
tempting though the idea was.

I realize that you are using the routines provided by the op, but
without too much trouble you can provide protection should the function
be called with no previous allocations, i.e. should *string == NULL.
If you do this then you need not worry with doing the initial allocation
in function main. Instead, use the function for the initial string and
subsequent appends.

int foo(char **string) /* 9 */
{
char *p = realloc(*string , 12); /* 10 */
if(p != NULL) /* 11 */
{
*string = p; /* 12 */
strcat(*string, " World"); /* 13 */
}
return p == NULL; /* 14 */
}

Should *string be NULL and the allocation succeed, you would
need to assign string[0] the value of '\0' for the strcat
function to work.

Example:

#include <stdlib.h>
#include <string.h>
#include <stdio.h>

int dCatStr(char **s, const char *catstr)
{
char *tmp;
size_t curlen = *s?strlen(*s):0 ;

if((tmp = realloc(*s,curl en+strlen(catst r)+1)) != NULL)
{
if(curlen == 0) *tmp = '\0';
*s = tmp;
strcat(*s,catst r);
}
return tmp?1:0;
}

int main(void)
{
char *string1 = NULL;

if(dCatStr(&str ing1,"Hello"))
if(dCatStr(&str ing1," World!"))
printf("string1 = \"%s\"\n",strin g1);
else puts("FAILURE: to append");
else puts("FAILURE: to append");
free(string1);
return 0;
}

--
Al Bowers
Tampa, Fl USA
mailto: xa******@myrapi dsys.com (remove the x to send email)
http://www.geocities.com/abowers822/

[infobahn]

Flash Gordon wrote:

On Wed, 22 Dec 2004 06:50:35 +0000 (UTC)
infobahn <in******@btint ernet.com> wrote:

Karthik Kumar wrote:

<snip>

pointer to pointer could be the answer.

int main(void)
{
char *string1;
string1 = malloc(6);

You need to test to see if malloc has failed.

Well, Earthik Kumar does. If you glance over my
reply to the OP, you'll find that I addressed
this issue, and all the others you mentioned, except ...

<snip>
void foo(char **my_string)
{
char *tmp = realloc(*my_str ing, 12);
if (tmp == NULL) {
/* handle error */
}
else {
*my_string = tmp
}
}

I've also changed the name because identifiers starting with str
followed by another letter, such as string, are reserved.

Not local identifiers.

[infobahn]

Al Bowers wrote:

infobahn wrote:

The following code is based heavily on your own code; I have changed
the indentation to make it more readable to others, and added error
checking, but I haven't "fixed" the code to my own style and preference,
tempting though the idea was.

I realize that you are using the routines provided by the op,

Right.
but
without too much trouble you can provide protection should the function
be called with no previous allocations, i.e. should *string == NULL.
If you do this then you need not worry with doing the initial allocation
in function main. Instead, use the function for the initial string and
subsequent appends.
I thought I'd squeezed all the juice out of it, but you're right - I
should have suggested this myself. Apologies for my omission.

<snip>
Example:

#include <stdlib.h>
#include <string.h>
#include <stdio.h>

int dCatStr(char **s, const char *catstr)
{
char *tmp;

It would be profitable to add an assertion here:

assert(s != NULL);

before dereferencing s. This would of course involve including
<assert.h> and, in C90, enclosing the remainder of the code in
a block { }, or separating the definition of curlen from the
test on *s (for the obvious reason).

<snip>

[Emmanuel Delahaye]

infobahn wrote on 22/12/04 :

I've also changed the name because identifiers starting with str
followed by another letter, such as string, are reserved.

Not local identifiers.

Saiz who ?

--
Emmanuel
The C-FAQ: http://www.eskimo.com/~scs/C-faq/faq.html
The C-library: http://www.dinkumware.com/refxc.html

"C is a sharp tool"

[infobahn]

Emmanuel Delahaye wrote:

infobahn wrote on 22/12/04 :

I've also changed the name because identifiers starting with str
followed by another letter, such as string, are reserved.

Not local identifiers.

Saiz who ?

ANSI, if I'm reading the Standard correctly.

****
4.13 FUTURE LIBRARY DIRECTIONS

The following names are grouped under individual headers for
convenience. All external names described below are reserved no
matter what headers are included by the program.
****

The possibility remains that I am not reading the Standard correctly.

[Emmanuel Delahaye]

infobahn wrote on 23/12/04 :

Emmanuel Delahaye wrote:

infobahn wrote on 22/12/04 :

I've also changed the name because identifiers starting with str
followed by another letter, such as string, are reserved.

Not local identifiers.

Saiz who ?

ANSI, if I'm reading the Standard correctly.

****
4.13 FUTURE LIBRARY DIRECTIONS

The following names are grouped under individual headers for
convenience. All external names described below are reserved no
matter what headers are included by the program.
****

The possibility remains that I am not reading the Standard correctly.

A static (hence 'not external') function with str* (say strcmp()) will
shadow the external strcmp() from the standard library. According to
the standard, technically correct, but bad practice IMO.

--
Emmanuel
The C-FAQ: http://www.eskimo.com/~scs/C-faq/faq.html
The C-library: http://www.dinkumware.com/refxc.html

"C is a sharp tool"

[infobahn]

Emmanuel Delahaye wrote:

<snip>

A static (hence 'not external') function with str* (say strcmp()) will
shadow the external strcmp() from the standard library. According to the
standard, technically correct, but bad practice IMO.

Shadowing names that are already known to exist in the standard library
is, of course, not just bad practice but terrible practice.

In this case, however, no such shadowing was done. In my own code, I
avoid starting names with str*, but I saw no reason to amend the OP's
code in my own article, since I figured I'd already given him enough
to think about. :-)