Recently I attended an interview where they asked me to write a C program to concatenate two strings without using strcat(), strlen() and strcmp() and that function should not exceed two (2) lines.
I know how to concatenate two strings without using strcat(). But my method has nearly 15 lines. I dont know how to write it in two lines.
I expect they wanted something like this:
void mystrcat(char * dest, const char * src)
{
//advance dest until we find the terminating null
while (*dest) ++dest;
//copy src to dest including terminating null, until we hit the end of src
//Edit: originally this:
//for (; *dest = *src, *src; ++dest, ++src);
//...which is the same as this
for (; *dest = *src; ++dest, ++src);
}
It doesn't return the end of the concatenated string like the real strcat, but that doesn't seem to be required.
I don't necessarily know if this sort of thing is a good interview question - it shows that you can code tersely, and that you know what strcat does, but that's about it.
Edit: as aib writes, the statement
while (*dest++ = *src++);
...is perhaps a more conventional way of writing the second loop (instead of using for).
Any function can be made to fit in a single line by simply removing all the \n.
However, I think you're looking for this answer:
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
int main(void)
{
char string1[32] = "Hello";
char string2[] = ", World!";
char *dst = string1 + strlen(string1);
char *src = string2;
while (*dst++ = *src++); //single statement
printf("\"%s\"\n", string1);
return EXIT_SUCCESS;
}
The explanation is rather simple:
src++ returns a pointer to the current character being copied before incrementing to point to the next one. * dereferences this pointer, and a similar expression on the LHS copies it to dst. Result of the whole = expression is the character that was copied, hence a simple while loops it until a \0 is encountered and copied.
However:
strcat() is easier to read and possibly much faster. Any other solution is feasible only when strcat() is not available. (Or when you're in an interview, apparently.)
And replace strcat() above with strncat() unless you're really really sure the destination string is big enough.
Edit: I missed the part about strlen() being disallowed. Here's the two-statement function:
void my_strcat(char * restrict dst, const char * restrict src)
{
while (*dst) ++dst; //move dst to the end of the string
while (*dst++ = *src++); //copy src to dst
}
Note that the standard strcat() function returns the original value of dst.
The proper answer to that question is that the question would demonstrate a skill that it is bad to have. They are wanting you to demonstrate the ability to write hacker code. They are wanting you to invent your own implementation of things provided already by every C compiler, which is waste of time. They are wanting you to write streamlined code which, by definition, is not readable. The 15 line implementation is probably better if it is more readable. Most projects do not fail because the developers wasted 150 clock cycles. Some do fail because someone wrote unmaintainable code. If you did have to write that, it would need a 15 line comment. So my answer to that would be, show me the performance metrics that defend needing to not use the standard libraries and requiring the most optimal solution. Time is much better spent on design and gathering those performance metrics.
Never forget - you are also interviewing them.
//assuming szA contains "first string" and szB contains "second string"
//and both are null terminated
// iterate over A until you get to null, then iterate over B and add to the end of A
// and then add null termination to A
// WARNING: memory corruption likely if either string is not NULL terminated
// WARNING: memory corruption likely if the storage buffer for A was not allocated large
// enough for A to store all of B's data
// Justification: Performance metric XXX has shown this optimization is needed
for(int i=0; szA[i]!='\0'; i++);
for(int j=0; (j==0)||(szB[j-1]!='\0'); j++) szA[i+j] = szB[j];
*edit, 9/27/2010
After reading some other solutions to this, I think the following is probably the best code answer:
//Posted by Doug in answer below this one
void my_strcat(char * dest, const char * src)
{
while (*dest) ++dest;
while (*dest++ = *src++);
}
But I would follow that up with a safe version of that:
void my_safe_strcat(char * dest, const unsigned int max_size, const char * src)
{
int characters_used=0;
while (*dest) { ++dest; characters_used++; }
while ( (characters_used < (max_size-1) ) && (*dest++ = *src++) ) characters_used++;
*dest = 0; //ensure we end with a null
}
And follow that up with (full answer, which compiler will optimize to be the same as above, along with application which was the real question):
void my_readable_safe_strcat(char * dest, const unsigned int max_size, const char * src)
{
unsigned int characters_used = 0;
while (*dest != '\0')
{
++dest;
characters_used++;
}
while ( (characters_used < (max_size-1) ) && (*dest = *src) )
{
dest++;
src++;
characters_used++;
}
*dest = 0; //ensure we end with a null
}
int _tmain(int argc, _TCHAR* argv[])
{
char szTooShort[15] = "First String";
char szLongEnough[50] = "First String";
char szClean[] = "Second String";
char szDirty[5] = {'f','g','h','i','j'};
my_readable_safe_strcat(szTooShort,15,szClean);
printf("This string should be cut off:\n%s\n\n",szTooShort);
my_readable_safe_strcat(szLongEnough,50,szClean);
printf("This string should be complete:\n%s\n\n",szLongEnough);
my_readable_safe_strcat(szLongEnough,50,szDirty);
printf("This string probably has junk data in it, but shouldn't crash the app:\n%s\n\n",szLongEnough);
}
Given that the task was to concatenate two strings, not to create a duplicate of strcat, I'd go with the simple option of creating a completely new string that is a combination of the two.
char buffer[REASONABLE_MAX] = {0};
snprintf(buffer, REASONABLE_MAX - 1, "%s%s", string1, string2);
Two lines? Bwah...
void another_strcat(char* str1, const char* str2)
{
strcpy(strchr(str1, '\0'), str2);
}
EDIT: I'm very upset that people are so against strcpy and strchr. Waah! So, I thought I'd play by the spirit of the rules:
char thing(char* p, const char* s)
{
return *p ? thing(&p[1], s) : *s ? (*p++ = *s++, thing(p, s)) : *p = '\0';
}
I still can't understand how anyone would take 2 whole lines ;-P.
void StringCatenation(char *str1,char *str2)
{
int len1,i=0;
for(len1=0;*(str1+len1);len1++);
do{
str1[len1+i]=str2[i];
i++;
}
while(*(str2+i);
}
I have a feeling such questions are meant to be elimination questions rather than selection. It is easier to eliminate candidates for them based on such convoluted questions rather than select candidates by asking them more real world questions. Just a rant from me, since I am also looking for a job and facing such questions and answered quite a few of them thanks to SO!
I tested this bit in VS2008, and it worked fine.
void NewStrCat(char* dest, const char* src)
{
while (*dest) ++dest;
while (*dest++ = *src++);
}
One line:
sprintf(string1, "%s%s", string1, string2);
(Note that this might possibly invoke undefined behavior.)
Addendum
The ISO C99 standard states that:
If copying takes place between objects that overlap, the behavior is undefined.
That being said, the code above will still probably work correctly. It works with MS VC 2010.