[SOLVED] C++ reads in ASCII 255 at the end of files?

casualfred · 03-19-2014, 05:23 PM

I am basically just trying to read in a file and output it to the terminal in C++, but I have noticed a strange character at the end of the output every time. I am using Slackware 14.1, gcc v4.8.2, and this occurs on files with both unix and dos line endings.

Code:

#include <iostream>
#include <fstream>
#include <string>
#include <cctype>

using namespace std;

int main()
{
   ifstream thefile;
   char fileName[50], ch;

   cout << endl << "Enter filename: ";
   cin.getline(fileName, 49);
   thefile.open(fileName);
   if (thefile.fail())
   {
      cout << "file not opened" << endl;
      return 0;
   }

   cout << endl;

   while (! thefile.eof())
   {
      ch = thefile.get();
      cout << ch;
      if (ch == -1) // ASCII 255 or it wraps around to -1
         cout << endl << "EOF";
   }

   return 0;
}

And it outputs:

Code:

Enter filename: expr.d1

45 + 60
45 - 60
(23 + 45)
12
(12)
((12))
((12 + 12 + 12))
ÿ
EOF

where my file "expr.d1" is:

Code:

45 + 60
45 - 60
(23 + 45)
12
(12)
((12))
((12 + 12 + 12))

So, I'm trying to figure out why it is outputting "ÿ", ASCII 255 at the end of the file. I have done some research, but I feel like I just don't know what's going on.

Thanks in advance for your help!

gengisdave · 03-19-2014, 06:15 PM

ascii 255 is the non breaking space, the html  , does your file have an empty ending line or some other character before eof (first tought is about a copy-paste from a web page)?

michaelk · 03-19-2014, 06:27 PM

Code:

  while (! thefile.eof())
   {
      ch = thefile.get();
      cout << ch;
      if (ch == -1) // ASCII 255 or it wraps around to -1
         cout << endl << "EOF";
   }

Lets look at the loop. If the file pointer is at the last character in the file the while is still true, get() will read the next character which is EOF so cout<<ch will print the character representation which is that strange thing you see. (ch==-1) evaluates true so EOF is printed.

rknichols · 03-19-2014, 06:29 PM

EOF does not get set when you read the last character in the file. It is the next read that sets EOF, but you are printing the returned value (-1) from that unsuccessful read.

Why does EOF work that way? Because it is entirely possible for the file to grow between the time you read the last character and your next read.

Oops, echo in here.

casualfred · 03-19-2014, 06:55 PM

Thanks for the replies! So would a do-while loop work better in my case do you think?

Edit: I went ahead and tried it, but it didn't seem to help. I don't recall ever having this problem before when trying to read in files... I'm not sure what I am doing wrong.

rknichols · 03-19-2014, 10:35 PM

Anything that prints ch as a character prior to testing for EOF is going to exhibit that problem. Of the myriad ways to code that loop, here is one:

Code:

while (! thefile.eof())
   {
      ch = thefile.get();
      if (ch == -1) // ASCII 255 or it wraps around to -1
         cout << endl << "EOF";
      else
         cout << ch; 
   }

And no, I'm not claiming that is the best way to code that.

casualfred · 03-19-2014, 11:25 PM

Alright, thanks rknichols. So, I could just filter the output like that. I think this is an answer to my question, so I'll mark it as solved.

But, let me be sure I understand what is happening here. There is an EOF flag C++ associates with an opened file. C++ sets that flag when it attempts a read, but finds there is nothing else to read. The failed read will always return this strange character.

Thanks for the great information and for taking the time to explain this stuff!

NevemTeve · 03-20-2014, 04:02 AM

(It's quite common mistake, almost everyone has it once... getchar/getc/fgetc/... return unsigned char or -1, so the value should be stored in 'int' not 'char')

sundialsvcs · 03-20-2014, 03:22 PM

Do it this way:

Code:

while not ( eof ) {
  get()
  if not ( eof ) {
    ... do something ...
  end-if
end-while

Or this way:

Code:

get()
while not ( eof ) {
  ... do something ...
  get()
end-while

Usually, eof indications are produced by attempts to "get" something. They don't look-ahead to see if eof is about to be reached. (The only language that I recall which attempted to do otherwise was Standard Pascal, and it was ugly.)

Your code is basically producing garbage at the last because it has just read past the end of the file. Although the eof indicator will have been tripped thereby, you're not checking for that after the get which ran-dry.

casualfred · 03-20-2014, 06:23 PM

Thanks sundialsvcs - yeah that's pretty much what I came up with in the end:

Code:

char ch;
ch = thefile.get();
while (! thefile.eof())
{
   cout << ch;
   ch = thefile.get;
}

But I've found a situation where your first example would work better also.

I think I understand what's going on now - thanks again everyone for the helpful guidance!

sundialsvcs · 03-20-2014, 11:43 PM

Heh... I like the second one better, myself, and normally do it that way.

rknichols · 03-21-2014, 10:12 AM

My own preference is for

Code:

while (get(), ! eof) {
    ... do something ...
}

dwhitney67 · 03-22-2014, 11:09 AM

You can always evaluate the file stream object after you have performed a call to get(). C++ ios stream objects (including fstream) implement an operator bool() method that returns the state of the stream.

Something like:

Code:

char ch;

while ((ch = file.get()) && file)
{
    std::cout << ch;
}
std::cout << "EOF" << std::endl;

You could also consider reading one line at a time from the data file.

Code:

char line[80];
while (file.getline(line, sizeof(line)) && file)
{
    std::cout << line << std::endl;
}
std::cout << "EOF" << std::endl

Edit... another option:

Code:

std::string line;
while (std::getline(file, line) && file)
{
    std::cout << line << std::endl;
}
std::cout << "EOF" << std::endl;

casualfred · 03-25-2014, 06:38 PM

Interesting. Thanks dwhitney67!