Normalizing a string

[Krusty]

Good time of day, everyone!

In my program, I have encountered the necessity of normalizing strings so I could pass them to the execvp() method.

I want my routine to proceed the string in the following way: it must remove all the spaces, tabulations and newline characters that precede and follow the string; if the substring is put in the quotation marks, the method must remove them and leave their content as is. The last character is set to null.

That means that I want the function to return the same string for all of the following input cases:

ls
ls		                
   	   ls  
"ls"
	"ls"

If I simply print out the results and their lengths by calling printf(), everything seems all right (they coincide). But the issue is, if I invoke execvp(), it works only for the two first strings (in the second one ls is followed by tabs and spaces) and throws "No such file or directory" for other cases.

This is just a sketch but nonetheless:

#define TRUE 1
#define FALSE 0

void set_format(char *str) {
  int i, flag; // flag means we encountered quatation marks

  printf("\nBefore formatting\n|%s|\n", str);

  for (i = 0, flag = TRUE;
       strchr(" \t\n\"", str[i]) != NULL;
       i++ ) {
    if (str[i]=='\"') {
      flag = FALSE;
      break;
    }    
  }

  if (flag) str+=i;
  else str+=i+1;

  for (i = strlen(str)-1;
       strchr(" \t\n\"", str[i])!=NULL;
       i-- ) {
    if (str[i]=='\"') {
      if (!flag) break;
      else puts("Invalid input");
    }    
  }

  str[i]=(char)NULL;

  printf("\nAfter formatting\n|%s|\nSize: %d\n", str, strlen(str));

}

I'm probably missing something obvious...
Thanks in advance!

[MK27]

I think my approach with this would be to go thru the string once, beginning to end, using switch/case on the value of each character:

int len = strlen(string);
for (i=0;i<len;i++) {
     switch (string[i]) {
             case(' '):  remove_char(string,i);
             case('"'):  remove_char(string,i);
             default: continue;

If all you ever want to do is remove the character, so the cases are all the same, then you just need a single long if ( || || ) instead of the switch/case, I guess.

[Krusty]

Thank you for a quick response!

The problem is that the following input is possible

  "  command  "

And then the routine is supposed to save the substring in between the quitation marks, returning

  command

And catch the case when we have unclosed quatation mark.

But it's not even the case. What I don't understand is that the function DOES normalize the string, at least this is how printf() makes it seem. I don't understand what's wrong with the strings execvp() doesn't accept.

[MTK]

Here is a function that trims whitespace off ends of a string. Not the it requires #include <string.h>

void trim_string(char* str)
{
	int i = 0;
	while((str[i] == ' ' || str[i] == '\n' || str[i] == '\r' || str[i] == '\t') && str[i] != '\0') i++;
	if(str[i] != '\0') {
		memmove(str, str+i, (strlen(str)+1-i)*sizeof(char));
		i = strlen(str)-1;
		while(str[i] == ' ' || str[i] == '\n' || str[i] == '\r' || str[i] == '\t') i--;
		str[i+1] = '\0';
	} else {
		str[0] = '\0';
	}
}

This function keeps whitespace enclosed in non-whitespace characters intact.

P.S. I thought that "Trim" means do like the function above and "Normalize" means trim and then turn any enclosed whitespace into single spaces, right?

[dwks]

You can always just pass all this to a shell (/bin/sh, for example) and let it remove the quotes etc properly. For example, you could use system() or exec*() a command like

/bin/sh -c /bin/ls "where"

I think the shell will do the proper thing in that case.

[itCbitC]

How are you calling execvp()? What arguments are you passing it? Can you post that here?

[Krusty]

MTK, loads of thanks to you for that code! It worked just fine for me, I believe it won't be any problem to add quatations.
No, you're right, trimming is what I needed, I'm not really good at terminology.

You can always just pass all this to a shell (/bin/sh, for example) and let it remove the quotes etc properly. For example, you could use system() or exec*() a command like

/bin/sh -c /bin/ls "where"

I think the shell will do the proper thing in that case.

Mini-shell is what I'm attempting to write, this is a part of a coveyor processing

How are you calling execvp()? What arguments are you passing it? Can you post that here?

Seems like the issue was with my routine...

Thank you, everyone!

[shiryu3]

Have you tried strtok() ?

[shiryu3]

You might use strtok for this

#include <stdio.h>
#include <string.h>
#define LENGTH 64

int main()
{
        char input[LENGTH] ;
        char *ptr ;

        printf("Input string: ") ;
        fgets(input,LENGTH,stdin) ;

        ptr = strtok(input,"\040\t\n") ;

        printf("[%s]\n", ptr) ;
        return 0;
}


/* OUTPUT

[voltron@shiryu3 forum01]$ ./a.out
Input string: stringOne
[stringOne]
[voltron@shiryu3 forum01]$ ./a.out
Input string:    strongTwo
[strongTwo]
[voltron@shiryu3 forum01]$ ./a.out
Input string:                   stringThree
[stringThree]

*/