Using regex for checking a .dat file

Question

I am reading a file using fgets. I need to check each line of the file against a regex. If there is a non alpha numeric character, it needs to exit the program with the line number and "bad" character displayed. What is happening is it is kicking out before the "bad" character. Here is my .dat file:

howard jim dave 
joe
(
Maggie

My output of the program is:

file opened
Digit: howard jim dave 
is not alphanumeric on line: 1
Exiting program!
File closed

What should happen is it should kick out on line 3, as you can see that is not happening.

Here is my regex which is in my main.h file:

#ifndef MAIN_H
#define MAIN_H

#ifdef  __cplusplus
extern "C" {
#endif

#define BUFF 1024
#define to_find "^[a-zA-Z0-9]+$"

Here is my fileCheck.c

#include "main.h"

int fileCheck(FILE *fp)
{

    int ret_val;
    int line_count = 0;
    char file[BUFF];
    regex_t regex;

    if (regcomp(&regex, to_find, REG_EXTENDED) != 0)
    {
        fprintf(stderr, "Failed to compile regex '%s'\n", to_find);
        return EXIT_FAILURE;
    }

    if (fp != NULL)
    {
        while (fgets(file, BUFF, fp))
        {
            line_count++;

            if ((ret_val = regexec(&regex, file, 0, NULL, 0)) != 0)
            {
                printf("Digit: %s is not alphanumeric on line: %d\n", file, line_count);
                printf("Exiting program!\n");
                return EXIT_FAILURE;
            }
        }
    }

}

I am not sure if the "\n" character is the problem or not. I do not think it is. I am well aware if isalnum() but I am tasked with a regex. What would be a possible solution for this problem? Thank you for your suggestions.

EDIT: I wanted to mention that when I used fscanf instead of fgets, the above regex worked just fine. The reason for the change is I need to count each line. If I am correct, fscanf ignores newline character. I need some way to count a newline. Is it possible to count a new using fscanf? My original file read loop was:

while (fscanf(fp, "%11023s", file) != EOF
{
    line_count++;
    if (regexec(&regex, file, 0, NULL, 0) != 0)
    {
        printf("%s%d wrong:\n, file, line_count);
        return EXIT_FAILURE;
    }
}

I had something in my mind but I wasn't sure, and it seems more likely now that the first regex seems to be failing. I'm not versed in C, but does `if ((ret_val = regexec(&regex, file, 0, NULL, 0)) != 0)` really work? Shouldn't that be `if (regexec(&regex, file, 0, NULL, 0) != 0)`? — Jerry, Sep 15 '13 at 18:18
@Jerry you are correct, I had ret_val in an older version but even that change did not make a difference. I still get my original error. — Clint, Sep 15 '13 at 18:42
You are using the regex `^[a-zA-Z0-9 ]+$` as well? (I think it's better to have " " than `\\s` in this case) Otherwise, I think it's more about reading line by line than the regex, and I can't help much here :( If maybe that [answer](http://stackoverflow.com/questions/3501338/c-read-file-line-by-line) could help? — Jerry, Sep 15 '13 at 19:54

score 1 · Accepted Answer · 2013-09-16T01:15:18.163

howard jim dave contains whitespaces.

Edit3:
The reason I focused on a match that looks only for valid lines was that you seemed to
be using a simple test scenario that later will be more comples.
However, if this is just what you need it for, the real solution is to just look for
a non-alphanumeric non-whitespace character.
If the regex flavor you are using require a match from beginning to end,
this won't work.

  #define to_find "[^a-zA-Z0-9\\s]" 
     or, 
  #define to_find "[^a-zA-Z0-9\\ \\t\\f\\r\\n]"

   . . .
     Then down here if the regex matches, it found non alpha numeric

  if ( regexec(&regex, file, 0, NULL, 0)) == 0 )
  {
      printf("Digit: %s is not alphanumeric on line: %d\n", file, line_count);
      printf("Exiting program!\n");
      return EXIT_FAILURE;
  }

Edit2:
Is this a Posix engine? What error code does regcomp() return? You should set REG_EXTENDED as one of the cflag parameters.
Unfortunately the (?: pattern ) construct is an extended specification.

Might as well throw the kitchen sink at it
REG_EXTENDED | REG_NEWLINE

Try those flaqs and plop
"^\\s*[a-zA-Z0-9]+(?:\\s+[a-zA-Z0-9]+)*\\s*$" directly into regcomp()

This can help with the error code:

 int res_compile = 0;
 if ( (res_compile=regcomp(&regex, to_find, REG_EXTENDED) ) != 0)
 {
   fprintf(stderr, "Failed to compile regex '%s'\nError code:  %d\n", to_find, res_compile);
 }

Original: Maybe you need

 # ^\s*[a-zA-Z0-9]+(?:\s+[a-zA-Z0-9]+)*\s*$

 ^ 
 \s* 
 [a-zA-Z0-9]+ 
 (?: \s+ [a-zA-Z0-9]+ )*
 \s* 
 $

Or

 # \A[^\S\r\n]*[a-zA-Z0-9]+(?:[^\S\r\n]+[a-zA-Z0-9]+)*\s*\z

 \A 
 [^\S\r\n]* 
 [a-zA-Z0-9]+ 
 (?: [^\S\r\n]+ [a-zA-Z0-9]+ )*
 \s*
 \z

I appreciate your answer, however I am getting a `warning: unknown escape sequence '\s'` and when I run the program the regex fails to compile. — Clint, Sep 15 '13 at 17:58
@Clint Try escaping the backslashes. If there is \, put \\ instead. — Jerry, Sep 15 '13 at 18:02
@Jerry, I just did that and still get a `Failed to compile regex '^\s*[a-zA-Z0-9]+(?:\s*[a-zA-Z0-9]+)*\s*$'` — Clint, Sep 15 '13 at 18:05
@Jerry No, I do appreciate everyone's help. That one gives me the original error I stated in my post. — Clint, Sep 15 '13 at 18:13
@Clint - is this a Posix engine? What error code does regcomp() return? You should set REG_EXTENDED as one of the cflag parameters. Unfortunately the `(?: pattern )` construct is an extended specification. Try that flaq and plop `"^\\s*[a-zA-Z0-9]+(?:\\s+[a-zA-Z0-9]+)*\\s*$"` directly into regcomp(). — , Sep 15 '13 at 20:39

Using regex for checking a .dat file

1 Answers1