CSV Parsing

Hi,

I am trying to get information from a CSV file, and import it into my program. The code I have so far is listed below:

Code:

#include <iostream>
#include <fstream>
#include <string>
using namespace std;

int main () {
  string line; //this will contain the data read from the file
  ifstream myfile ("example.csv"); //opening the file.
  if (myfile.is_open()) //if the file is open
  {
    while (! myfile.eof() ) //while the end of file is NOT reached
    {
      getline (myfile,line); //get one line from the file
      //DO THE CSV PARSING HERE.
     //FOR NOW I WILL JUST OUTPUT THE LINE:
      cout << line << endl; //and output it
    }
    myfile.close(); //closing the file
  }
  else cout << "Unable to open file"; //if the file is not open output <--

  return 0;
}

How does this look? Im very new to coding, so I was wondering if anyone might be able to give some input on how I can do the CSV parsing. Are there tutorials online for this? Would anyone be able to give some insight?

Regards,

James

Using .eof to control a loop is a Bad Idea, inasmuch as it doesn't actually work.

Given a generic CSV file, all you do is look for commas (and new lines, of course). If you're allowed something like 1234,"Lastname, Firstname", 1 with a comma inside quotes that doesn't count then you'll have to be a little more careful.

It is C++? I had some help on another forum, im really new to this but I assumed it was ok.

Is it easy to convert this to C?

if this isn't part of some larger project that actually is in C++ i would suggest this:

http://us3.php.net/manual/en/function.fgetcsv.php

using C++ to do a job like this is like calling in a nuclear strike to smoosh a bug.

Hi,

Thank you for the reply. I have to use C unfortunatly, but I really dont want to branch into C# or C++ as I want to take it one step at a time.

Grr... you have to use C? Right, you made me move this for nothing... but why did you start by writing it in C++ if you have to use C?

Sorry. Ive been getting help on this project to try and help me learn how to start making a simple program. This is what I have so far except I didnt realise it had branched off from C. I would like this to be in the C forum please as that is what I want it to be coded in.

Originally Posted by tabstop

Using .eof to control a loop is a Bad Idea, inasmuch as it doesn't actually work.

Please elaborate on that. What's wrong with

Code:

filename.getline( buffer, n );
while( !filename.eof() ) {
    ...[do stuff]...
    filename.getline( buffer, n );
}

?

Originally Posted by R.Stiltskin

Please elaborate on that. What's wrong with

Code:

filename.getline( buffer, n );
while( !filename.eof() ) {
    ...[do stuff]...
    filename.getline( buffer, n );
}

?

I have two objections to that:

1. Code duplication of the getline() call leads to brittleness.
2. The use of eof() is simply unnecessary, period.

Code:

while(filename.getline(buffer, n))
{
    ...[do stuff]...
}

Originally Posted by brewbuck

I have two objections to that:

1. Code duplication of the getline() call leads to brittleness.
2. The use of eof() is simply unnecessary, period.

Code:

while(filename.getline(buffer, n))
{
    ...[do stuff]...
}

I don't doubt you, and I've definitely used that construct in the past, having simply copied it from texts without really understanding what it was doing. Now that I'm actually thinking about what these functions return, I can't find anything that says that [ifstream].getline() or .read() ever return anything other than the stream *this. Presumably they return 0 or NULL when they fail to extract any characters?

Originally Posted by R.Stiltskin

Presumably they return 0 or NULL when they fail to extract any characters?

No, they still return the stream by reference. However, the stream is convertible to void*, and through that conversion it is convertible to bool, and thus would evaluate to false when the eof bit (and other status bits) is set.

So apparently, first there is
(1) an implicit conversion from member reference to void*, and then
(2) a second conversion from void* to bool?

For some reason I find this interesting enough to have spent over an hour googling around for further clarification. So far, I haven't found anything to explain conversion (1), and for conversion (2) the best that I've come up with is from the C++ draft standard:

4.12 Boolean conversions
An rvalue of arithmetic, unscoped enumeration, pointer, or pointer to member type can be converted to an rvalue of type bool. A zero value, null pointer value, or null member pointer value is converted to false; any other value is converted to true. An rvalue of type std::nullptr_t can be converted to an rvalue of type bool; the resulting value is false.

Taking 4.12 at face value, in order for conversion (2) to result in false, that has to be a conversion from a null pointer value...

... so conversion (1) must have resulted in a pointer with a null value ...

... so therefore should I conclude that a reference to a class member can be implicitly converted to a void* with a NULL value if least one of the fail, bad, or eof flags is set, and a non-NULL value otherwise?

(Why is it so difficult to find this written somewhere?)

Originally Posted by R.Stiltskin

So far, I haven't found anything to explain conversion (1)

Check section 27.4.4, in particular subsection 27.4.4.3, of the 2003 edition of the C++ Standard:

operator void*() const;
Returns: If fail() then a null pointer; otherwise some non-null pointer to indicate success.

You're my favorite witch! Thanks, laserlight.

Still, it irks me that an everyday usage that "everyone" is supposed to know "if(bla.read(...)" should require so much effort to find out why it works -- as opposed to "if(!bla.eof())" which is well-documented and easy to understand.