to_root.cpp

// A basic parser, it might get stuff wrong, but good enought for EDA
Int_t parse_year(std::string year)
{
	if(year == "")
		throw std::runtime_error("year is empty");
	if(year[0] == '[' || year[0] == 'c' || year[0] == 'p') // Handle format of [2000], c2000 and p2000
		return std::stoi(std::string(year.begin()+1, year.begin()+5));
	else if(isdigit(year[0]) == true)
		return std::stoi(std::string(year.begin(), year.begin()+4));
	throw std::runtime_error("cannot parse format");
}


void to_root()
{
	// Open the raw archive
	using string = std::string;
	auto rdf = ROOT::RDataFrame("library_raw", "library_raw.root");

	//Create a new archive that we'll be coping to
	auto f = new TFile("library.root", "recreate");
	auto t = new TTree("library", "checkout_data");

	// All the fields
	TDatime date;
	string usage, checkout_type, creator, material, title, publisher;
	std::vector<std::string> subjects;
	Int_t num_checkout, publication_year;

	// Assign column/branchs to the fields
	t->Branch("usage", &usage);
	t->Branch("checkout_type", &checkout_type);
	t->Branch("material", &material);
	t->Branch("checkout_month", &date);
	t->Branch("num_checkout", &num_checkout);
	t->Branch("title", &title);
	t->Branch("creator", &creator);
	t->Branch("subjects", &subjects);
	t->Branch("publisher", &publisher);
	t->Branch("publication_year", &publication_year);

	ProgressDisplay disp(*rdf.Count());
	size_t i=0;

	rdf.Foreach([&](string usage_, string checkout_type_, string material_, string checkout_year_, string checkout_month_, string checkouts_, string title_
		, string creator_, string subjects_, string publisher_, string publication_year_) {

		// Copy data to field
		usage = usage_;
		checkout_type = checkout_type_;
		material = material_;
		num_checkout = std::stoi(checkouts_);
		title = title_;
		creator = creator_;
		subjects = split(subjects_);
		publisher = publisher_;

		try {publication_year = parse_year(publication_year_);}
		catch(...) {publication_year = 0x7fffffff;}

		date.Set(std::stoi(checkout_year_), std::stoi(checkout_month_),0, 0, 0, 0);

		t->Fill();

		i++;
		if(i%100000 == 0)
			disp.update(i);


	}, {"UsageClass", "CheckoutType", "MaterialType", "CheckoutYear", "CheckoutMonth", "Checkouts", "Title", "Creator", "Subjects", "Publisher", "PublicationYear"});

	f->Write();
}