format_vienna.hpp

Go to the documentation of this file.

// -----------------------------------------------------------------------------------------------------
// Copyright (c) 2006-2020, Knut Reinert & Freie Universität Berlin
// Copyright (c) 2016-2020, Knut Reinert & MPI für molekulare Genetik
// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
// shipped with this file and also available at: https://github.com/seqan/seqan3/blob/master/LICENSE.md
// -----------------------------------------------------------------------------------------------------
 
#pragma once
 
#include <cstdio>
#include <iterator>
#include <stack>
#include <string>
#include <string_view>
#include <type_traits>
#include <vector>
 
#include <seqan3/alphabet/adaptation/char.hpp>
#include <seqan3/alphabet/structure/wuss.hpp>
#include <seqan3/core/char_operations/predicate.hpp>
#include <seqan3/core/type_traits/range.hpp>
#include <seqan3/io/detail/misc.hpp>
#include <seqan3/io/stream/iterator.hpp>
#include <seqan3/io/structure_file/detail.hpp>
#include <seqan3/io/structure_file/input_format_concept.hpp>
#include <seqan3/io/structure_file/input_options.hpp>
#include <seqan3/io/structure_file/output_format_concept.hpp>
#include <seqan3/io/structure_file/output_options.hpp>
#include <seqan3/range/detail/misc.hpp>
#include <seqan3/range/views/char_to.hpp>
#include <seqan3/range/views/istreambuf.hpp>
#include <seqan3/range/views/take.hpp>
#include <seqan3/range/views/take_line.hpp>
#include <seqan3/range/views/take_until.hpp>
#include <seqan3/range/views/to_char.hpp>
#include <seqan3/range/views/to.hpp>
#include <seqan3/std/algorithm>
#include <seqan3/std/ranges>
 
namespace seqan3
{
class format_vienna
{
public:
    format_vienna() noexcept = default; 
    format_vienna(format_vienna const &) noexcept = default; 
    format_vienna & operator=(format_vienna const &) noexcept = default; 
    format_vienna(format_vienna &&) noexcept = default; 
    format_vienna & operator=(format_vienna &&) noexcept = default; 
    ~format_vienna() noexcept = default; 
 
    static inline std::vector<std::string> file_extensions
    {
        { "dbn" },
        { "fasta" },
        { "fa" }
    };
 
protected:
    template <typename stream_type,     // constraints checked by file
              typename seq_legal_alph_type,
              bool     structured_seq_combined,
              typename seq_type,        // other constraints checked inside function
              typename id_type,
              typename bpp_type,
              typename structure_type,
              typename energy_type,
              typename react_type,
              typename comment_type,
              typename offset_type>
    void read_structure_record(stream_type & stream,
                               structure_file_input_options<seq_legal_alph_type, structured_seq_combined> const & options,
                               seq_type & seq,
                               id_type & id,
                               bpp_type & bpp,
                               structure_type & structure,
                               energy_type & energy,
                               react_type & SEQAN3_DOXYGEN_ONLY(react),
                               react_type & SEQAN3_DOXYGEN_ONLY(react_err),
                               comment_type & SEQAN3_DOXYGEN_ONLY(comment),
                               offset_type & SEQAN3_DOXYGEN_ONLY(offset))
    {
        auto stream_view = views::istreambuf(stream);
 
        // READ ID (if present)
        auto constexpr is_id = is_char<'>'>;
        if (is_id(*begin(stream_view)))
        {
            if constexpr (!detail::decays_to_ignore_v<id_type>)
            {
                if (options.truncate_ids)
                {
                    std::ranges::copy(stream_view | std::views::drop_while(is_id || is_blank) // skip leading >
                                                  | views::take_until_or_throw(is_cntrl || is_blank)
                                                  | views::char_to<std::ranges::range_value_t<id_type>>,
                                      std::cpp20::back_inserter(id));
                    detail::consume(stream_view | views::take_line_or_throw);
                }
                else
                {
                    std::ranges::copy(stream_view | std::views::drop_while(is_id || is_blank) // skip leading >
                                                  | views::take_line_or_throw
                                                  | views::char_to<std::ranges::range_value_t<id_type>>,
                                      std::cpp20::back_inserter(id));
                }
            }
            else
            {
                detail::consume(stream_view | views::take_line_or_throw);
            }
        }
        else if constexpr (!detail::decays_to_ignore_v<id_type>)
        {
            auto constexpr is_legal_seq = is_in_alphabet<seq_legal_alph_type>;
            if (!is_legal_seq(*begin(stream_view))) // if neither id nor seq found: throw
            {
                throw parse_error{std::string{"Expected to be on beginning of ID or sequence, but "} +
                                  is_id.msg + " and " + is_legal_seq.msg +
                                  " evaluated to false on " + detail::make_printable(*begin(stream_view))};
            }
        }
 
        // READ SEQUENCE
        if constexpr (!detail::decays_to_ignore_v<seq_type>)
        {
            auto constexpr is_legal_seq = is_in_alphabet<seq_legal_alph_type>;
            std::ranges::copy(stream_view | views::take_line_or_throw                      // until end of line
                                          | std::views::filter(!(is_space || is_digit)) // ignore whitespace and numbers
                                          | std::views::transform([is_legal_seq](char const c)
                                            {
                                                if (!is_legal_seq(c))                    // enforce legal alphabet
                                                {
                                                    throw parse_error{std::string{"Encountered an unexpected letter: "} +
                                                                      is_legal_seq.msg +
                                                                      " evaluated to false on " +
                                                                      detail::make_printable(c)};
                                                }
                                              return c;
                                            })
                                          | views::char_to<std::ranges::range_value_t<seq_type>>, // convert to actual target alphabet
                              std::cpp20::back_inserter(seq));
        }
        else
        {
            detail::consume(stream_view | views::take_line_or_throw);
        }
 
        // READ STRUCTURE (if present)
        [[maybe_unused]] int64_t structure_length{};
        if constexpr (!detail::decays_to_ignore_v<structure_type>)
        {
            if constexpr (structured_seq_combined)
            {
                assert(std::addressof(seq) == std::addressof(structure));
                using alph_type = typename std::ranges::range_value_t<structure_type>::structure_alphabet_type;
                // We need the structure_length parameter to count the length of the structure while reading
                // because we cannot infer it from the (already resized) structure_seq object.
                auto res = std::ranges::copy(read_structure<alph_type>(stream_view), std::ranges::begin(structure));
                structure_length = std::ranges::distance(std::ranges::begin(structure), res.out);
 
                if constexpr (!detail::decays_to_ignore_v<bpp_type>)
                    detail::bpp_from_rna_structure<alph_type>(bpp, structure);
            }
            else
            {
                using alph_type = std::ranges::range_value_t<structure_type>;
                std::ranges::copy(read_structure<alph_type>(stream_view), std::cpp20::back_inserter(structure));
                structure_length = std::ranges::distance(structure);
 
                if constexpr (!detail::decays_to_ignore_v<bpp_type>)
                    detail::bpp_from_rna_structure<alph_type>(bpp, structure);
            }
        }
        else if constexpr (!detail::decays_to_ignore_v<bpp_type>)
        {
            detail::bpp_from_rna_structure<wuss51>(bpp, read_structure<wuss51>(stream_view));
            structure_length = std::ranges::distance(bpp);
        }
        else
        {
            detail::consume(stream_view | views::take_until(is_space)); // until whitespace
        }
 
        if constexpr (!detail::decays_to_ignore_v<seq_type> &&
                      !(detail::decays_to_ignore_v<structure_type> && detail::decays_to_ignore_v<bpp_type>))
        {
            if (std::ranges::distance(seq) != structure_length)
                throw parse_error{"Found sequence and associated structure of different length."};
        }
 
        // READ ENERGY (if present)
        if constexpr (!detail::decays_to_ignore_v<energy_type>)
        {
            std::string e_str = stream_view | views::take_line
                                            | std::views::filter(!(is_space || is_char<'('> || is_char<')'>))
                                            | views::to<std::string>;
            if (!e_str.empty())
            {
                size_t num_processed;
                energy = std::stod(e_str, &num_processed);
                if (num_processed != e_str.size()) // [[unlikely]]
                {
                    throw parse_error{std::string{"Failed to parse energy value '"} + e_str + "'."};
                }
            }
        }
        else
        {
            detail::consume(stream_view | views::take_line);
        }
        detail::consume(stream_view | views::take_until(!is_space));
    }
 
    template <typename stream_type,     // constraints checked by file
              typename seq_type,        // other constraints checked inside function
              typename id_type,
              typename bpp_type,
              typename structure_type,
              typename energy_type,
              typename react_type,
              typename comment_type,
              typename offset_type>
    void write_structure_record(stream_type & stream,
                                structure_file_output_options const & options,
                                seq_type && seq,
                                id_type && id,
                                bpp_type && SEQAN3_DOXYGEN_ONLY(bpp),
                                structure_type && structure,
                                energy_type && energy,
                                react_type && SEQAN3_DOXYGEN_ONLY(react),
                                react_type && SEQAN3_DOXYGEN_ONLY(react_err),
                                comment_type && SEQAN3_DOXYGEN_ONLY(comment),
                                offset_type && SEQAN3_DOXYGEN_ONLY(offset))
    {
        std::cpp20::ostreambuf_iterator stream_it{stream};
 
        // WRITE ID (optional)
        if constexpr (!detail::decays_to_ignore_v<id_type>)
        {
            if (!std::ranges::empty(id))
            {
                stream_it = '>';
                stream_it = ' ';
                std::ranges::copy(id, stream_it);
                detail::write_eol(stream_it, options.add_carriage_return);
            }
        }
 
        // WRITE SEQUENCE
        if constexpr (!detail::decays_to_ignore_v<seq_type>)
        {
            if (std::ranges::empty(seq)) //[[unlikely]]
                throw std::runtime_error{"The SEQ field may not be empty when writing Vienna files."};
 
            std::ranges::copy(seq | views::to_char, stream_it);
            detail::write_eol(stream_it, options.add_carriage_return);
        }
        else
        {
            throw std::logic_error{"The SEQ and STRUCTURED_SEQ fields may not both be set to ignore "
                                   "when writing Vienna files."};
        }
 
        // WRITE STRUCTURE (optional)
        if constexpr (!detail::decays_to_ignore_v<structure_type>)
        {
            if (!std::ranges::empty(structure))
                std::ranges::copy(structure | views::to_char, stream_it);
 
            // WRITE ENERGY (optional)
            if constexpr (!detail::decays_to_ignore_v<energy_type>)
            {
                if (energy)
                {
// TODO(joergi-w) enable the following when std::to_chars is implemented for float types
//                    auto [endptr, ec] = std::to_chars(str.data(),
//                                                      str.data() + str.size(),
//                                                      energy,
//                                                      std::chars_format::fixed,
//                                                      options.precision);
//                    if (ec == std::errc())
//                        std::ranges::copy(str.data(), endptr, stream_it);
//                    else
//                        throw std::runtime_error{"The energy could not be transformed into a string."};
 
                    stream_it = ' ';
                    stream_it = '(';
 
                    std::array<char, 100> str;
                    int len = std::snprintf(str.data(), 100, "%.*f", options.precision, energy);
                    if (len < 0 || len >= 100)
                        throw std::runtime_error{"The energy could not be transformed into a string."};
                    std::ranges::copy(str.data(), str.data() + len, stream_it);
 
                    stream_it = ')';
                }
            }
            detail::write_eol(stream_it, options.add_carriage_return);
        }
        else if constexpr (!detail::decays_to_ignore_v<energy_type>)
        {
            throw std::logic_error{"The ENERGY field cannot be written to a Vienna file without providing STRUCTURE."};
        }
    }
 
private:
    template <typename alph_type, typename stream_view_type>
    auto read_structure(stream_view_type & stream_view)
    {
        auto constexpr is_legal_structure = is_in_alphabet<alph_type>;
        return stream_view | views::take_until(is_space) // until whitespace
                           | std::views::transform([is_legal_structure](char const c)
                             {
                                 if (!is_legal_structure(c))
                                 {
                                     throw parse_error{
                                         std::string{"Encountered an unexpected letter: "} +
                                         is_legal_structure.msg +
                                         " evaluated to false on " + detail::make_printable(c)};
                                 }
                                 return c;
                             })                                  // enforce legal alphabet
                           | views::char_to<alph_type>;           // convert to actual target alphabet
    }
};
 
} // namespace seqan3::detail