little_tools/dimacs2txt.cpp

// =====================================================================================
//
//       Filename:  dimcas2txt.cpp
//
//    Description:  converter
//
//        Version:  1.0
//        Created:  2017年 12月 26日 星期二 07:53:19 -00
//       Revision:  none
//       Compiler:  g++
//
//         Author:  Jinkun Lin, jkunlin@gmail.com
//   Organization:  School of EECS, Peking University
//
// =====================================================================================
#include <iostream>
#include <vector>
#include <fstream>
#include <algorithm>
using namespace std;

vector<vector<int>> adjacency_list;
vector<int> vertex_weight;

void fast_build(string file_name) {/*{{{*/
	ifstream in_file(file_name);
	if (!in_file) {
		cout << "in_file error" << endl;
		exit(1);
	}
	in_file.seekg (0, in_file.end);
	size_t file_len = in_file.tellg();
	in_file.seekg (0, in_file.beg);
	char *data = new char [file_len];

	in_file.read(data, file_len);
	in_file.close();


	//skip comments
	char *pos = data;
	while (*pos == '%') {
		while(*(pos++) != '\n');
	}

	//read vertex_count
	long vertex_count = 0, edge_count = 0;
	while (*pos < '0' || *pos > '9') {
		++pos;
	}
	while (*pos != ' ') {
		vertex_count = vertex_count * 10 +  *pos - '0';
		++pos;
	}
	//read edge_count
	while (*pos < '0' || *pos > '9') {
		++pos;
	}
	while (*pos >= '0' && *pos <= '9') {
		edge_count = edge_count * 10 +  *pos - '0';
		++pos;
	}


	//read vertex_weight
	vertex_weight.resize(vertex_count + 1);
	long v,w;
	for(vector<vector<long>>::size_type i=1; i<vertex_weight.size(); ++i){
		v = w = 0;
		//read vertex
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos != ' ') {
			v = v * 10 +  *pos - '0';
			++pos;
		}
		//read weight
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos >= '0' && *pos <= '9') {
			w = w * 10 +  *pos - '0';
			++pos;
		}
		vertex_weight[v]=w;
	}

	//read adjacency_list
	adjacency_list.resize(vertex_count + 1);

	vector<long> vertex_degree(vertex_count + 1, 0);
	char *stash_pos = pos;
	long v1, v2;
	for (long i = 0; i < edge_count; ++i) {
		v1 = v2 = 0;
		//read v1
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos != ' ') {
			v1 = v1 * 10 +  *pos - '0';
			++pos;
		}
		//read weight
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos >= '0' && *pos <= '9') {
			v2 = v2 * 10 +  *pos - '0';
			++pos;
		}
		vertex_degree[v1]++;
		vertex_degree[v2]++;
	}
	for (size_t v = 1; v < adjacency_list.size(); ++v) {
		adjacency_list[v].reserve(vertex_degree[v]);
	}
	pos = stash_pos;
	for (long i = 0; i < edge_count; ++i) {
		v1 = v2 = 0;
		//read v1
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos != ' ') {
			v1 = v1 * 10 +  *pos - '0';
			++pos;
		}
		//read weight
		while (*pos < '0' || *pos > '9') {
			++pos;
		}
		while (*pos >= '0' && *pos <= '9') {
			v2 = v2 * 10 +  *pos - '0';
			++pos;
		}
		adjacency_list[v1].push_back(v2);
		adjacency_list[v2].push_back(v1);
	}
	delete[] data;
}/*}}}*/

void convert(string file_name) {
	ofstream out_file(file_name);
	if (!out_file) {
		std::cout << "out_file error" << std::endl;
		abort();
	}
	out_file << adjacency_list.size()  - 1 << endl;
	for (size_t v = 1; v < adjacency_list.size(); ++v) {
		out_file << "   " << vertex_weight[v];
	}
	out_file << std::endl;

	for (size_t v = 1; v < adjacency_list.size(); ++v) {
		std::sort(adjacency_list[v].begin(), adjacency_list[v].end());
	}

	for (size_t v = 1; v < adjacency_list.size(); ++v) {
		int cur_i = 0;
		for (size_t u = 1; u < adjacency_list.size(); ++u) {
			if (cur_i == (int)adjacency_list[v].size()) {
				out_file << ' ' << 0;
				continue;
			}
			if ((int)u == adjacency_list[v][cur_i]) {
				out_file << ' ' << 1;
				cur_i++;
			}
			else {
				out_file << ' ' << 0;
			}
		}
		out_file << std::endl;
	}

	out_file.close();
}
int main(int argc, char const *argv[]) {
	if (argc != 2) {
		std::cout << "usage: " << std::endl;
		return 1;
	}
	string in_filename = argv[1];
	fast_build(in_filename);
	string out_filename (in_filename.begin(), in_filename.end() - 4);
	out_filename += "txt";
	convert(out_filename);
	return 0;
}