Segmentation Fault?

I can do the problem in java, but I was wondering if I could write the code this way in C. Can I store a series of pointers to strings in an array? So that array[0] could contain a pointer to an string in memory? I get a segmentation fault when i run my code, and a bunch of formatting errors. The point is to create a simple mail merge type program, where each line of the file being read looks like |string|string|string| etc.. up to 9 fields.

Code:

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

#define MAXSIZE 25

int main( int argc, char *argv[] )
{
    char line[ MAXSIZE ]; //hold lines of file
    char fields[ MAXSIZE ]; //hold tokenized strings (need to be a pointer?)
    char *token = NULL;
    FILE *read = NULL; //file to be read
    int i;

    read = fopen( "data.txt", "r" );

    while( !feof( read ) )//read each record
    {
        fgets( line, MAXSIZE, read );
        token = strtok( line, "|" ); //tokenize the line
        
        i = 0;
        while( token )
        {
            fields[ i ] = token; //store pointers in array (if possible?)
            token = strtok( NULL, "|" );
            i++;
        }

        //formatting...
        printf( "Welcome back, %s %c\n", fields[ 1 ], '!' ); 
        printf( "We hope that you and all the members\n"); 
        printf( "of the %s %s", fields[ 0 ], " family are constantly\n" ); 
        printf( "reminding your neighbors there\n"); 
        printf( "on %s %s", fields[ 5 ], " to shop with us\n" ); 
        printf( "As usual, we will ship your order to\n"); 
        printf( "%4s %2s %2s %2s\n", fields[ 3 ], fields[ 1 ], fields[ 2 ], fields[ 0 ] ); 
        printf( "%4s %2s\n", fields[ 4 ], fields[ 5 ] );
        printf( "%4s %s %s %s\n", fields[ 6 ], ", ", fields[ 7 ], fields[ 8 ] );    
    }

    return EXIT_SUCCESS;
}

Code:

Errors:

merge.c: In function âmainâ:
merge.c:25: warning: assignment makes integer from pointer without a cast
merge.c:30: warning: format â%sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:32: warning: format â%sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:34: warning: format â%sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:36: warning: format â%4sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:36: warning: format â%2sâ expects type âchar *â, but argument 3 has type âintâ
merge.c:36: warning: format â%2sâ expects type âchar *â, but argument 4 has type âintâ
merge.c:36: warning: format â%2sâ expects type âchar *â, but argument 5 has type âintâ
merge.c:37: warning: format â%4sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:37: warning: format â%2sâ expects type âchar *â, but argument 3 has type âintâ
merge.c:38: warning: format â%4sâ expects type âchar *â, but argument 2 has type âintâ
merge.c:38: warning: format â%sâ expects type âchar *â, but argument 4 has type âintâ
merge.c:38: warning: format â%sâ expects type âchar *â, but argument 5 has type âintâ

fields should be char *, not char.

--
Mats

> char fields[ MAXSIZE ];
char *fields[ MAXSIZE ];
would be more consistent with all the usage of the array you have.

> while( !feof( read ) )
Read the FAQ as to why feof() to control a loop is bad.

Do this instead

Code:

    while( fgets( line, MAXSIZE, read ) != NULL )//read each record
    {
        fgets( line, MAXSIZE, read );
        token = strtok( line, "|" ); //tokenize the line
        // etc
    }

> while( token )
while( token && i < MAXSIZE )
would be much safer.

A final
fclose(read);
would be good as well.

Originally Posted by Salem

> char fields[ MAXSIZE ];
while( token && i < MAXSIZE )
would be much safer.

Yes, but assuming that the string has LESS tokens than it's own length should be safe, so I don't think that is likely to happen (actually, I fail to see how strtok could give you a number of tokens less than HALF of the size of the original string, because the token will consist of one non-separator character and one separator character, so two characters are needed to make a single token). However, if the size of the fields array is different from (smaller than half + 1 of) MAXSIZE, then this is definitely necessary to prevent overflow of the token pointers.

--
Mats

True, but I like to put the array subscript into the conditional as a matter of habit, even when there seems little chance of it being critical to the code.

Originally Posted by Salem

True, but I like to put the array subscript into the conditional as a matter of habit, even when there seems little chance of it being critical to the code.

Yes, I agree, better safe than sorry.

--
Mats

Thanks guys, i made the necessary changes and it's almost working. I get a segmentation fault again, but it's after a couple of the fields get printed. So at least I know its partially working. It seems that everything is working, I am storing the tokens inside the array and able to print the proper fields. But I don't get past the first record, in fact, I don't even finish the first record. I get no errors upon compiling, but this is what my program produces.

Code:

Welcome back, "Peter" ! <-----Tokenizing works, this was the second entry stored
We hope that you and all the members
of the "Johnson"  family are constantly <-----Storage of tokens in array seems to work
reminding your neighbors there
Segmentation fault <-------Is this a memory issue?  Can someone clarify...

Are you sure you get the right number of tokens back. You are explicitly accessing fields[n] up to 8, which means that i should be 9 to allow for that - or you need to check the field to see it it's been filled in [best done by assiging

Code:

char *fields[MAXSIZE] = { 0 };

so that you have a zero field if it's not been assigned by the code].

--
Mats

Post your latest version of this program with all the changes you've made.

when i was sitting down for breakfast, i thought the buffer size might have been too small because I forgot to change the constant. After pushing it to 100, instead of 25 the output is properly generated, and no segmentation fault occurs. However, formatting in C is beginning to annoy me. The first block of output is properly formatted, but later blocks deviate by a space or two although I am using &#37;4s for each. Any formatting tips?

p.s. Just so i'm clear as to whats going on in my program, I was storing a pointer to each string(token) inside the array. So when field[0] was called, inside was a pointer, which pointed to a string, and that's why I could store a string inside each index of the array?

Yes, fields will hold a pointer to the somewhere in "line". But my point was more that if you don't set the fields[x] to a valid string, then you can NOT print that.

Not sure what you mean by formatting not working. If you give a width for a string field, however, it will still print WIDER than that if I remember correctly.

--
Mats