Finding unique elements in an string array in C

问题:

C bothers me with its handling of strings. I have a pseudocode like this in my mind:

char *data[20]; 

char *tmp; int i,j;

for(i=0;i<20;i++) {
  tmp = data[i]; 
  for(j=1;j<20;j++) 
  {
    if(strcmp(tmp,data[j]))
      //then except the uniqueness, store them in elsewhere
  }
}

But when i coded this the results were bad.(I handled all the memory stuff,little things etc.) The problem is in the second loop obviously :D. But i cannot think any solution. How do i find unique strings in an array.

Example input : abc def abe abc def deg entered unique ones : abc def abe deg should be found.

回答1:

You could use qsort to force the duplicates next to each other. Once sorted, you only need to compare adjacent entries to find duplicates. The result is O(N log N) rather than (I think) O(N^2).

Here is the 15 minute lunchtime version with no error checking:

  typedef struct {
     int origpos;
     char *value;
  } SORT;

  int qcmp(const void *x, const void *y) {
     int res = strcmp( ((SORT*)x)->value, ((SORT*)y)->value );
     if ( res != 0 )
        return res;
     else
        // they are equal - use original position as tie breaker
        return ( ((SORT*)x)->origpos - ((SORT*)y)->origpos );
  }

  int main( int argc, char* argv[] )
  {
     SORT *sorted;
     char **orig;
     int i;
     int num = argc - 1;

     orig = malloc( sizeof( char* ) * ( num ));
     sorted = malloc( sizeof( SORT ) * ( num ));

     for ( i = 0; i < num; i++ ) {
        orig[i] = argv[i + 1];
        sorted[i].value = argv[i + 1];
        sorted[i].origpos = i;
        }

     qsort( sorted, num, sizeof( SORT ), qcmp );

     // remove the dups (sorting left relative position same for dups)
     for ( i = 0; i < num - 1; i++ ) {
        if ( !strcmp( sorted[i].value, sorted[i+1].value ))
           // clear the duplicate entry however you see fit
           orig[sorted[i+1].origpos] = NULL;  // or free it if dynamic mem
        }

     // print them without dups in original order
     for ( i = 0; i < num; i++ )
        if ( orig[i] )
           printf( "%s ", orig[i] );

     free( orig );
     free( sorted );
  }

回答2:

char *data[20];
int i, j, n, unique[20];

n = 0;
for (i = 0; i < 20; ++i)
{
    for (j = 0; j < n; ++j)
    {
        if (!strcmp(data[i], data[unique[j]]))
           break;
    }

    if (j == n)
        unique[n++] = i;
}

The indexes of the first occurrence of each unique string should be in unique[0..n-1] if I did that right.

回答3:

Why are you starting second loop from 1?

You should start it from i+1. i.e.

for(j=i+1;j<20;j++) 

Like if the list is

abc
def
abc
abc
lop

then

when i==4

tmp="lop"

but then the second loop starts which is from 1 to 19. This means it will get a value of 4 too at one stage, and then

data[4], which is "lop", will be same as tmp. So although "lop" is unique but it will be flagged as repeated.

Hope it was helpful.

回答4:

Think a bit more about your problem -- what you really want to do is look at the PREVIOUS strings to see if you've already seen it. So, for each string n, compare it to strings 0 through n-1.

print element 0 (it is unique)
for i = 1 to n
  unique = 1
  for j = 0 to i-1 (compare this element to the ones preceding it)
    if element[i] == element[j]
       unique = 0
       break from loop
  if unique, print element i

回答5:

Might it be that your test is if (strcmp (this, that)) which will succeed if the two are different? !strcmp is probably what you want there.