Example #1
0
csa_wt<t_wt, t_dens, t_inv_dens, t_sa_sample_strat, t_isa, t_alphabet_strat>::csa_wt(cache_config& config)
{
    if (!cache_file_exists(key_trait<alphabet_type::int_width>::KEY_BWT, config)) {
        return;
    }
    {
        auto event = memory_monitor::event("construct csa-alpbabet");
        int_vector_buffer<alphabet_type::int_width> bwt_buf(cache_file_name(key_trait<alphabet_type::int_width>::KEY_BWT,config));
        size_type n = bwt_buf.size();
        alphabet_type tmp_alphabet(bwt_buf, n);
        m_alphabet.swap(tmp_alphabet);
    }
    {
        auto event = memory_monitor::event("construct wavelet tree");
        int_vector_buffer<alphabet_type::int_width> bwt_buf(cache_file_name(key_trait<alphabet_type::int_width>::KEY_BWT,config));
        size_type n = bwt_buf.size();
        wavelet_tree_type tmp_wt(bwt_buf, n);
        m_wavelet_tree.swap(tmp_wt);
    }
    {
        auto event = memory_monitor::event("sample SA");
        sa_sample_type tmp_sa_sample(config);
        m_sa_sample.swap(tmp_sa_sample);
    }
    {
        auto event = memory_monitor::event("sample ISA");
        isa_sample_type isa_s(config, &m_sa_sample);
        util::swap_support(m_isa_sample, isa_s, &m_sa_sample, &m_sa_sample);
    }
}
Example #2
0
csa_sada<t_enc_vec, t_dens, t_inv_dens, t_sa_sample_strat, t_isa, t_alphabet_strat>::csa_sada(cache_config& config)
{
    create_buffer();
    if (!cache_file_exists(key_trait<alphabet_type::int_width>::KEY_BWT, config)) {
        return;
    }
    int_vector_buffer<alphabet_type::int_width> bwt_buf(cache_file_name(key_trait<alphabet_type::int_width>::KEY_BWT,config));
    size_type n = bwt_buf.size();
    {
        auto event = memory_monitor::event("construct csa-alpbabet");
        alphabet_type tmp_alphabet(bwt_buf, n);
        m_alphabet.swap(tmp_alphabet);
    }

    int_vector<> cnt_chr(sigma, 0, bits::hi(n)+1);
    for (typename alphabet_type::sigma_type i=0; i < sigma; ++i) {
        cnt_chr[i] = C[i];
    }
    // calculate psi
    {
        auto event = memory_monitor::event("construct PSI");
        // TODO: move PSI construct into construct_PSI.hpp
        int_vector<> psi(n, 0, bits::hi(n)+1);
        for (size_type i=0; i < n; ++i) {
            psi[ cnt_chr[ char2comp[bwt_buf[i]] ]++ ] = i;
        }
        std::string psi_file = cache_file_name(conf::KEY_PSI, config);
        if (!store_to_cache(psi, conf::KEY_PSI, config)) {
            return;
        }
    }
    {
        auto event = memory_monitor::event("encode PSI");
        int_vector_buffer<> psi_buf(cache_file_name(conf::KEY_PSI, config));
        t_enc_vec tmp_psi(psi_buf);
        m_psi.swap(tmp_psi);
    }
    {
        auto event = memory_monitor::event("sample SA");
        sa_sample_type tmp_sa_sample(config);
        m_sa_sample.swap(tmp_sa_sample);
    }
    {
        auto event = memory_monitor::event("sample ISA");
        isa_sample_type isa_s(config, &m_sa_sample);
        util::swap_support(m_isa_sample, isa_s, &m_sa_sample, (const sa_sample_type*)nullptr);
    }
}
Example #3
0
//bool construct_bwt(tMSS& file_map, const std::string& dir, const std::string& id) {
void construct_bwt(int_vector<8>& bwt, const cache_config& config){
    typedef int_vector<>::size_type size_type;
	tMSS::const_iterator key;
	key = config.file_map.find(constants::KEY_BWT);
    if ( config.file_map.end() == key) { // if bwt is not already registered in file_map
        std::string bwt_tmp_file_name = config.dir+"/tmp_bwt_"+util::to_string(util::get_pid())+"_"+util::to_string(util::get_id()); 
        write_R_output("csa", "construct BWT", "begin", 1, 0);
        const size_type buffer_size = 1000000;

		int_vector<8> text;
		util::load_from_file(text, config.file_map.find(constants::KEY_TEXT)->second.c_str() );
        size_type n = text.size();

        int_vector_file_buffer<> sa_buf(config.file_map.find(constants::KEY_SA)->second.c_str(), buffer_size);

        std::ofstream bwt_out_buf(bwt_tmp_file_name.c_str(), std::ios::binary | std::ios::trunc | std::ios::out); // open out file stream

        int_vector<8> bwt_buf(buffer_size);
        size_type bit_size = n*8;
        bwt_out_buf.write((char*) &(bit_size), sizeof(int_vector_size_type));   // write size
        size_type wb = 0; // written bytes
        size_type to_add[2] = {-1,n-1};
        for (size_type i=0, r_sum=0, r = 0; r_sum < n;) {
            for (; i<r_sum+r; ++i) {
                bwt_buf[i-r_sum] = text[ sa_buf[i-r_sum]+to_add[sa_buf[i-r_sum]==0] ]; 
            }
            if (r > 0) {
                bwt_out_buf.write((const char*)bwt_buf.data(), r);
                wb += r;
            }
            r_sum += r; r = sa_buf.load_next_block();
        }
        if (wb%8) {
            bwt_out_buf.write("\0\0\0\0\0\0\0\0", 8-wb%8);
        }
        bwt_out_buf.close();
		util::clear(text);
		util::load_from_file(bwt, bwt_tmp_file_name.c_str()); // load bwt in memory
		std::remove(bwt_tmp_file_name.c_str()); // remove temporary file
        write_R_output("csa", "construct BWT", "end", 1, 0);
    }else{
		util::load_from_file(bwt, key->second.c_str()); // load bwt in memory
	}
}