Search code examples
cgdb

Why "%79[^\n]\n" is causing a segmentation fault?

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

int main()
{
        char line[80];
        FILE *in = fopen("spooky.csv", "r");
        FILE *file1 = fopen("ufos.csv", "w");
        FILE *file2 = fopen("disappearances.csv", "w");
        FILE *file3 = fopen("others.csv", "w");
        while (fscanf(in, "%79[^\n]\n", line) == 1) {
                if (strstr(line, "UFO"))
                        fprintf(file1, "%s\n", line);
                else if (strstr(line, "Disappearance"))
                        fprintf(file2, "%s\n", line);
                else
                        fprintf(file3, "%s\n", line);
        }
        fclose(file1);
        fclose(file2);
        fclose(file3);
        return 0;
}

This code is returning a runtime error of segmentation fault.

enter image description here

Does it matter on what hardware I'm compiling this program? I'm using a Fedora 38 (Workstation) with Intel Core i7 (7th Gen).

Solution

  • The problem has nothing to do with the fscanf format: as indicated in the error message, the argument s to the internal function __vfscanf_internal has the value 0x0 indicating a null pointer passed for the FILE * argument.

    You should check for fopen failure to open the files and report this condition with a meaningful error message.

    Also note these remarks:

    • you should close in.
    • you could use fgets() instead of the fscanf() format.
    • the fscanf format will fail converting an initial empty line in spooky.csv, which should not occur if it is an actual CSV file.
    • the trailing \n in the conversion specification will cause the newline to be consumed along with any initial white space in the next line, which may not be intended. Use %79[^\n]%*[\n] to avoid this, yet consume subsequent empty lines.
    • the cde will silently break long lines into chunks of 79 bytes. Modifying the code for arbitrary long lines is non trivial, but a larger buffer might be a good option.
    • you might want to special case the first line if it is expected to be a header line with the field names.

    Here is a modified version:

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

FILE *fopen_check(const char *filename, const char *mode) {
    FILE *fp = fopen(filename, mode);
    if (fp == NULL) {
        fprintf(stderr, "cannot open %s: %s\n", filename, strerror(errno));
        exit(1);
    }
    return fp;
}

int main(void) {
    char line[80];

    FILE *in = fopen_check("spooky.csv", "r");
    FILE *file1 = fopen_check("ufos.csv", "w");
    FILE *file2 = fopen_check("disappearances.csv", "w");
    FILE *file3 = fopen_check("others.csv", "w");

    fscanf(in, "%*[\n]");  // skip initial empty lines

    while (fscanf(in, "%79[^\n]%*[\n]", line) == 1) {
        if (strstr(line, "UFO"))
            fprintf(file1, "%s\n", line);
        else
        if (strstr(line, "Disappearance"))
            fprintf(file2, "%s\n", line);
        else
            fprintf(file3, "%s\n", line);
    }
    fclose(in);
    fclose(file1);
    fclose(file2);
    fclose(file3);
    return 0;
}