// UpdateUniprot.cpp


#if defined(_WIN32) || defined(_WIN64)

// C4290 - the compiler ignores exception specifications
#pragma warning(disable: 4290)

#endif


#include <fstream>
#include <iostream>
#include <stdexcept>
#include <string>
#include <cctype>
#include <cerrno>
#include <cstring>
#include <boost/scoped_ptr.hpp>
#include "GzipReader.hpp"
#include "SystemError.hpp"


using namespace NGBW;


int main(int argc, char **argv)
{
	try {
		std::ios_base::sync_with_stdio(false);

		if(argc != 4)
			throw std::invalid_argument("usage: pdb-to-fasta dataset outputname inputname");

		boost::scoped_ptr<std::ofstream> output_file;

		if(std::memcmp(argv[2], "-", 2) != 0) {
			output_file.reset(new std::ofstream(argv[2]));

			if(!output_file->good()) {
				std::string message("can't open ");

				message.append(argv[2]);

				throw SystemError("main", message, errno);
			}
		}

		std::ostream &output = output_file ? *output_file : std::cout;
		GzipReader reader(argv[3]);

		output.exceptions(std::ios_base::badbit);

		while (1) {
			const char *line = reader.GetLine();

			if (reader.Eof())
				break;

			if (line[0] == '>') {
				output.write(">ngbw|", 6);

				output << argv[1];

				// protein id

				size_t last = 1;

				while (line[last] != '_') {
					if (line[last] == '\0')
						throw std::runtime_error("Unexpected comment format");

					last += 1;
				}

				output.put('|');

				for (size_t i = 1 ; i < last ; i += 1)
					output.put(std::toupper(line[i]));

				// chain id

				last += 1;

				size_t first = last;

				while (!std::isspace(line[last])) {
					if (line[last] == '\0')
						throw std::runtime_error("Unexpected comment format");

					last += 1;
				}

				output.put('|');

				for (size_t i = first ; i < last ; i += 1)
					output.put(std::toupper(line[i]));

				// description

				const char *description = std::strstr(line + last, "length:");

				if (description != NULL) {
					while (!std::isspace(*description) && *description != '\0')
						description += 1;

					while (std::isspace(*description) && *description != '\0')
						description += 1;

					output.put('|');

					output << description;
				}

				output.put('\n');
			}
			else {
				output << line;

				output.put('\n');
			}
		}

		return 0;
	}
	catch(const std::exception &err) {
		std::cerr << err.what() << std::endl;

		return 1;
	}
	catch(...) {
		std::cerr << "An unknown exception was thrown" << std::endl;

		return 1;
	}
}