After reading from text file, results are displayed incorrectly - c

Question

I've written a program that reads four variables (three strings and one character) every line from a text file. But when I display the variables, an unexpected character pops up at the end of each line. (I've ensured that the lengths of the variables are large enough).
Why is this? (Overflowing buffers, again?) And how do I fix this?
Text file contents:
M0001 Cool Name F 123-456789
M0002 Name Cool M 987-654321
Code:
#include <stdio.h>
#include <stdlib.h>
int main() {
FILE *text;
char id[6], name[101], gender, contact[13];
text = fopen("test.txt", "r");
while (fscanf(text, "%s %[^\n]s %c %s\n", id, name, &gender, contact) != EOF)
printf("%s %s %c %s\n", id, name, gender, contact);
fclose(text);
return 0;
}
The output I expect:
M0001 Cool Name F 123-456789
M0002 Name Cool M 987-654321
What I get instead:
M0001 Cool Name F 123-456789 1⁄4
M0002 Name Cool M 987-654321 1⁄4

Answer 1

in the call to fscanf(), the format string: "%s %[^\n]s %c %s\n" is not correct.
the '[^\n]' will read to the end of the line (which will overflow the input buffer: `name'. Then the next char is NOT an 's' because the next character is the newline.
should compare the returned value to 4, not EOF
the input/format specifiers '%[...]' and '%s' have no problem overflowing the input buffer, so should ALWAYS have a MAX_CHARACTERS modifier that is one less than the length of the input buffer (those format specifiers always append a NUL byte to the input
The following proposed code:
cleanly compiles
documents why each header file is included
performs the desired functionality
splits the 'name' into 'firstname' and 'lastname' for easier handling and to match the format of the input data
properly checks the returned value from fscanf()
properly checks for any error from fopen() and if an error is returned, properly outputs the error message and the text indicating why the system thinks the function failed to stderr
uses an appropriate format string for the calls to fscanf() and printf()
replaces 'magic' numbers with meaningful names via a enum statement
And now the proposed code:
#include <stdio.h> // fopen(), fclose(), fscanf(), perror(), printf()
#include <stdlib.h> // exit(), EXIT_FAILURE
enum{
MAX_ID_LEN = 6,
MAX_NAME_LEN = 20,
MAX_CONTACT_LEN = 13
};
int main( void )
{
char id[ MAX_ID_LEN ];
char firstname[ MAX_NAME_LEN ];
char lastname[ MAX_NAME_LEN ];
char gender;
char contact[ MAX_CONTACT_LEN ];
FILE *text = fopen("test.txt", "r");
if( !text )
{
perror( "fopen to read 'test.txt' failed" );
exit( EXIT_FAILURE );
}
// implied else, fopen successful
while (5 == fscanf(text, "%5s %19s %19s %c %12s",
id, firstname, lastname, &gender, contact) )
{
printf("%s %s %s %c %s\n",
id, firstname, lastname, gender, contact);
}
fclose(text);
return 0;
}

Answer 2

%[^\n]s eats up everything from that point on and puts it in name. So only id and name are filled. gender and contact have 'random' contents coming from the program stack (as they are not initialized).
By accident the your stack had 1/4 in gender + contact.
On my machine, the program crashes.

Answer 3

As the number of space-delimited words in your name apparently is variable, you can only use %[^\n]s to grab "as much as possible" – but that will also eat up any and all following relevant data. A quick solution would be to re-design the input format and place the name at the very end; then, your fscanf argument would be:
"%s %c %s %s\n", id, &gender, contact, name
Alternatively, rewrite the code to use less fscanf and more 'manual' parsing:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>
int main (void)
{
FILE *text;
char id[6], name[101], gender, contact[13];
char *lookback;
int result;
unsigned int line_number = 0;
text = fopen ("test.txt", "r");
if (text == NULL)
{
printf ("file not found!\n");
return EXIT_FAILURE;
}
do
{
result = fscanf(text, "%s %[^\n]s\n", id, name);
line_number++;
if (result == EOF)
break;
if (result != 2)
{
printf ("error in data file on line %u (expected at least 2 items)\n", line_number);
break;
}
/* at this point, 'name' also contains 'gender' and 'contact' */
lookback = strrchr (name, ' ');
if (lookback == NULL || strlen(lookback+1) > 12)
{
printf ("error in data file on line %u (expected 'contact')\n", line_number);
break;
}
/* lookback+1 because lookback itself points to the space */
strcpy (contact, lookback+1);
/* cut off at lookback */
*lookback = 0;
lookback = strrchr (name, ' ');
if (lookback == NULL || strlen(lookback+1) != 1)
{
printf ("error in data file on line %u (expected 'gender')\n", line_number);
break;
}
/* lookback now points to the space before the gender */
gender = toupper(lookback[1]);
if (gender != 'F' && gender != 'M')
{
printf ("error in data file on line %u (expected 'M' or 'F')\n", line_number);
break;
}
/* cut off again at lookback; now name is complete */
*lookback = 0;
printf ("%s %s %c %s\n", id, name, gender, contact);
} while (1);
fclose(text);
return EXIT_SUCCESS;
}
This method does have a couple of associated drawbacks. One of the perks of scanf is that it normalizes whitespace; multiple spaces and tabs (or even returns) will silently be translated to a single space before scanning. This code, on the other hand, explicitly checks for a single space character. If there is variation in the whitespace in your data file, you must account for that as well.
With the last two items 'manually' processed, you can opt to not use fscanf at all. You can read an entire line of text at once with fgets (which also has a line length check built in) and look for spaces using strchr and strrchr. To counter possible whitespace problems, search the line for tabs and double spaces and change these to a single space.