CSV++ is C++ Interface for libcsv

Dependencies

• A C++11 compiler
• libcsv http://sourceforge.net/projects/libcsv/
• boost::lexical_cast -- available on ubuntu as libboost-dev

Usage

The reader is all fancy lambda type detection, so one just defines a lamba matching the CSV file and off you go.

Let's assume you have a two-column CSV like the following:

6,joe
3,louise
2,mary
1,louise

You can parse it with following code:

std::map<std::string, int> groups;
auto parser = csv::make_parser(
  [&groups](const int count, const std::string name) {
    groups[name] += count;
  });
parser.ParseFile("/path/to/file");

Or, with a functor

class Grouper {
  std::map<std::string, int> groups;
  void operator()(const int count, const std::string name) {
    groups[name] += count;
  }
};
Grouper grouper;
auto parser = csv::make_parser(grouper);
parser.ParseFile("/path/to/file");

How it Works

It inspects the provided lamba and builds a std::tuple with the same types. Each field is converted from a string to the final type using boost::lexical_cast, and the provided lamda is called once per row.

Performance

Not really a goal of the library. I don't think it does anything crazy and should be fast enough.

Limitations

1. No real support for files with a variable number of columns, reading a subset of columns, etc.