Reading a file containing a blank line

Hey everyone,

i want to read in a file that has two parts. The first part is letters and the second part is numbers. They are separated by a blank line.

I'm having trouble actually finding the blank line that separates the two parts of the text file.

So the file format is:
a b c d
e f g h
i j k l

1 2 3
4 5 6
7 8 9

and so forth.

Here's what I'm trying to do

Code:

---

        string t1, t2, t3, t4;
        int n1, n2, n3;
        int blankFound = 0;
        while (!in1.eof()&&(blankFound==0)) {            // read first part of file
                string s;
                getline(in1, s);
                if(s.compare("")==0){
                      blankFound=1;
                }
                in1 >> t1 >> t2 >> t3 >> t4;
                cout << t1 << t2 << t3 << t4 << endl;
        }
        while(!in1.eof()) {                                        //read second part
                in1 >> n1 >> n2 >> n3;
                cout << n1 << n2 << n3 << endl;
        }

---

I get the output (I'm missing the first line of the file somewhere!)

efgh
ijlk
123
456
789

My mixing of getline() and the >> operator is probably not good and is not working out.

Anyway I can test for a blank line with >> operator? It just seems to skip ahead and read the next text, and if i do noskipws it just reads every white space which isnt whta i want either.

Any help wud be great,

Cheers!

Why wouldn't you just test your input for length = 0 ?
Blank the blank line should be the only one with length < 1.

Finding the blank line is not at all important, since operator>> skips whitespace anyway, but you do need to know where you are in the file; if only to extract the right type variables from it. You can also do this in a single loop.

To find out where you are, you need to execute a statement like

in1 >> c;

where c is a char of any name. Then figure out if that c represents a text digit '0' through '9'. If c is a digit, then do

in1.putback(c);

and read the group of numbers from in1. Else, do in1.putback(c); and read the group of letters as strings.

One more thing that I would remove in the existing code is that !in1.eof() expression. For info on what happens when you do that, read this: Cprogramming.com FAQ > Why it's bad to use feof() to control a loop. Despite the title, everything that page says, applies here. You could actually make the in1 >> c expression your loop condition to remedy the mistake, and the loop will end on time; just make sure to always execute the putback(c) call before you actually read the file.

Quote:

---

Originally Posted by CommonTater

Why wouldn't you just test your input for length = 0 ?
Blank the blank line should be the only one with length < 1.

---

Hey Tater, do you mean if i change

Code:

---

if(s.compare("")==0){

---

to

Code:

---

if(s.length()==0){

---

that still skips the first line in the file.

Also hey whiteflags,

I need to find the blank line because the file changes from having four variables per line to having 3 variables per line. The four variables can actually be a mix of numbers and letters so I can't rely on a test for variable type.


Ideally I'd like to do this without getline and just use the >> operator to get the data and also discover the blank line.

Thanx guys any ideas wud be great!

You're doomed to be disappointed then, since ">> skips whitespace" means that it will skip the blank line without comment, if you let it.

And once you determine that s has positive length you need to get your t1 t2 t3 t4 variables out of s, not out of the file.

Hmmmm... I figured that since a blank line would still have a newline at the end (beginning?) the function would return with length = 0 .... And you are right, in that case he'd have to get his variables out of strings read from the file.

And of course, the way to get variables out of strings is to use a stringstream. So far as I can see, there's no way to actually use a stringstream in I/O (whether oversight or deliberate decision I don't know), so you have to use your string s in the getline call, then use that s to initialize a stringstream object, then use >> to get things out of the stringstream.

Hmmm... Ok.... I suppose I should let this one go, since I'm so new to C++ anything but the most barebones simple advice is bound to be wrong. It's just that is how I would have done it in C...

Never have eof() as the loop condition. Bad idea.


Of course, as an alternative to iostream-based parsing, you could use Boost.Spirit for parsing. In spirit, a parser for your file format looks like this:

Code:

---

((char_ - char_(" \n")) % char_(" \n")) >> "\n\n" >> (int_ % char_(" \n"))

---

This reads: parse a list of single characters (not space or newline) separated by space or newline, followed by two consecutive newlines, followed by a list of integers separated by space or newline.

This parser is rather unflexible in that it requires exactly one whitespace character to separate any two characters. This makes it way easier to write, because a blank line is hard to recognize otherwise. There are other ways to cope with this too, but this is the easiest.

Once you have this parser, you can parse.

Code:

---

#include <boost/spirit/include/qi_match_attr.hpp>
#include <boost/spirit/include/qi_char.hpp>
#include <boost/spirit/include/qi_int.hpp>
#include <boost/spirit/include/qi_string.hpp>
#include <boost/spirit/include/qi_operator.hpp>
using namespace boost::spirit::qi;

void foo() {
  std::vector<char> characters;
  std::vector<int> numbers;
  in1 >> match(
    ((char_ - char_(" \n")) % char_(" \n")) >> "\n\n" >> (int_ % char_(" \n")),
    characters, numbers);
  // characters and numbers now contain the data from the file
}

---