a problem about difference between GNU/Linux and Windows

[Whistler]

This is from my .ini parser:

PHP Code:

      // check if this is a session line (e.g., [SECTION])
      if (str[0] == '[' && str[length - 1] == ']') {
         strcpy(section, &str[1]);
         section[length - 2] = 0; // remove the ] 


This works in Windows, but in GNU/Linux I have to use:

PHP Code:

      // check if this is a session line (e.g., [SECTION])
      if (str[0] == '[' && str[length - 2] == ']') {
         strcpy(section, &str[1]);
         section[length - 3] = 0; // remove the ] 


The problem is, why does the GNU/Linux has to use a different way ? Also I'm using gcc 3.4.2 for both of the OS's

I have also pre-processed the string read from the .ini file with this function:

PHP Code:

void trim(char *str)
{
   int pos = 0;
   char *dest = str;

   // skip leading blanks
   while (str[pos] == ' ' || str[pos] == '\t' || str[pos] == '\n')
      pos++;

   while (str[pos]) {
      *dest++ = str[pos];
      pos++;
   }

   *dest = '\0'; // store the null

   int i = strlen(str) - 1;
   while (i > 0 && (str[i] == ' ' || str[i] == '\t' || str[i] == '\n')) { // remove trailing blanks
      str[i--] = '\0';
   }
} 


[koraX]

I think I know where is your problem
imagine you have text file made under windows, so end line is CR+LF. And now you will load this .ini file into your parser in windows and linux :

WINDOWS :
if you call fgets, it will detect newline as CR+LF and end character in your string is \n
You will use [length - 2] and everything is allright

LINUX :
if you call fgets, it reads until LF and string will have \r\n in the end. So you must use [length -3] to skip to section name

How to fix it :
I use something like this in my parser to skip trailing whitespaces

#include <string>

int main()
{
	std::string line = "    test string   \r\t\n  \r  \t\n  ";
	std::string result;

	std::string::size_type j;
	std::string::size_type begin = 0;
	std::string::size_type end = line.size();

	// skip all leading whitespaces
	for (j = begin; j < line.size(); ++j) {
		if (line[j] > ' ') 
			break; 
		++begin;
	}
	// skip all trailing whitespaces
	for (j = end; j > 0; --j) {
		if (line[j - 1] > ' ') 
			break; 
		--end;
	}

	if (end > begin)
		result = line.substr (begin, end - begin);
	else
		result = " ";
}

However this code is for std::string class, not evil char array

You only check for \n in your trim function. You should also check for \r, or better, check for all whitespace characters by comparing them with space, like I did in previous example

more info : http://en.wikipedia.org/wiki/Newline