debug_matrix.hpp

Go to the documentation of this file.

// SPDX-FileCopyrightText: 2006-2024 Knut Reinert & Freie Universität Berlin
// SPDX-FileCopyrightText: 2016-2024 Knut Reinert & MPI für molekulare Genetik
// SPDX-License-Identifier: BSD-3-Clause
 
#pragma once
 
#include <iomanip>
 
#include <seqan3/alignment/matrix/detail/matrix_concept.hpp>
#include <seqan3/alignment/matrix/detail/row_wise_matrix.hpp>
#include <seqan3/alignment/matrix/detail/trace_directions.hpp>
#include <seqan3/alphabet/detail/debug_stream_alphabet.hpp>
#include <seqan3/core/debug_stream/debug_stream_type.hpp>
#include <seqan3/core/debug_stream/optional.hpp>
#include <seqan3/core/debug_stream/range.hpp>
#include <seqan3/core/detail/template_inspection.hpp>
 
namespace seqan3::detail
{
 
template <matrix matrix_t, typename first_sequence_t = std::nullopt_t, typename second_sequence_t = std::nullopt_t>
class debug_matrix
{
protected:
    static constexpr bool has_first_sequence = !std::is_same_v<std::decay_t<first_sequence_t>, std::nullopt_t>;
    static constexpr bool has_second_sequence = !std::is_same_v<std::decay_t<second_sequence_t>, std::nullopt_t>;
    using entry_t = typename std::remove_reference_t<matrix_t>::value_type;
    static constexpr bool is_traceback_matrix = std::is_same_v<std::decay_t<entry_t>, trace_directions>;
    static constexpr bool is_optional_score = is_type_specialisation_of_v<entry_t, std::optional>;
 
public:
    using value_type = std::conditional_t<is_traceback_matrix || is_optional_score, entry_t, std::optional<entry_t>>;
    using reference = value_type;
    using const_reference = reference;
    using size_type = typename std::remove_reference_t<matrix_t>::size_type;
 
    debug_matrix() = default;                                 
    debug_matrix(debug_matrix const &) = default;             
    debug_matrix(debug_matrix &&) = default;                  
    debug_matrix & operator=(debug_matrix const &) = default; 
    debug_matrix & operator=(debug_matrix &&) = default;      
    ~debug_matrix() = default;                                
 
    debug_matrix(matrix_t matrix) : debug_matrix(std::forward<matrix_t>(matrix), std::nullopt, std::nullopt)
    {}
 
    debug_matrix(matrix_t matrix, first_sequence_t first_sequence, second_sequence_t second_sequence) :
        _matrix{std::forward<matrix_t>(matrix)},
        _first_sequence{std::forward<first_sequence_t>(first_sequence)},
        _second_sequence{std::forward<second_sequence_t>(second_sequence)}
    {
        if constexpr (has_first_sequence)
        {
            assert(_matrix.cols() <= _first_sequence.size() + 1u);
        }
 
        if constexpr (has_second_sequence)
        {
            assert(_matrix.rows() <= _second_sequence.size() + 1u);
        }
    }
 
    size_t rows() const noexcept
    {
        if (!_transpose)
            return _rows.value_or(_matrix.rows());
        else
            return _cols.value_or(_matrix.cols());
    }
 
    size_t cols() const noexcept
    {
        if (!_transpose)
            return _cols.value_or(_matrix.cols());
        else
            return _rows.value_or(_matrix.rows());
    }
 
    first_sequence_t const & first_sequence() const noexcept
    {
        if (!_transpose)
            return _first_sequence;
        else
            return _second_sequence;
    }
 
    second_sequence_t const & second_sequence() const noexcept
    {
        if (!_transpose)
            return _second_sequence;
        else
            return _first_sequence;
    }
 
    const_reference at(matrix_coordinate const & coordinate) const noexcept
    {
        size_t row = coordinate.row;
        size_t col = coordinate.col;
 
        assert(row < rows() && col < cols());
 
        row_index_type const _row{!_transpose ? row : col};
        column_index_type const _col{!_transpose ? col : row};
        row_index_type const _mask_row{_transpose == _transpose_mask ? row : col};
        column_index_type const _mask_col{_transpose == _transpose_mask ? col : row};
 
        if (!_masking_matrix.has_value() || _masking_matrix.value().at({_mask_row, _mask_col}))
        {
            entry_t const & entry = _matrix.at({_row, _col});
 
            if (!is_traceback_matrix || !_transpose)
                return entry;
 
            if constexpr (is_traceback_matrix)
            {
                trace_directions reverse{};
                if ((entry & trace_directions::left) == trace_directions::left)
                    reverse |= trace_directions::up;
                if ((entry & trace_directions::up) == trace_directions::up)
                    reverse |= trace_directions::left;
                if ((entry & trace_directions::diagonal) == trace_directions::diagonal)
                    reverse |= trace_directions::diagonal;
                return reverse;
            }
        }
 
        if constexpr (is_traceback_matrix)
            return trace_directions::none;
        else
            return std::nullopt;
    }
 
    debug_matrix & mask_matrix(row_wise_matrix<bool> masking_matrix) noexcept
    {
        assert(masking_matrix.rows() == rows());
        assert(masking_matrix.cols() == cols());
        _transpose_mask = _transpose;
        _masking_matrix = std::move(masking_matrix);
        return *this;
    }
 
    debug_matrix & mask_matrix(std::vector<bool> masking_vector) noexcept
    {
        return mask_matrix(row_wise_matrix<bool>{number_rows{rows()}, number_cols{cols()}, std::move(masking_vector)});
    }
 
    debug_matrix & sub_matrix(size_t const new_rows, size_t const new_cols) noexcept
    {
        assert(new_rows <= rows());
        assert(new_cols <= cols());
        if (!_transpose)
        {
            _rows = new_rows;
            _cols = new_cols;
        }
        else
        {
            _rows = new_cols;
            _cols = new_rows;
        }
        return *this;
    }
 
    debug_matrix & transpose_matrix() noexcept
    {
        _transpose = !_transpose;
        return *this;
    }
 
protected:
    struct format_type; // forward declaration
 
public:
    template <typename ostream_t>
    void stream_matrix(ostream_t & cout, fmtflags2 const flags) const noexcept
    {
        format_type const & symbols = (flags & fmtflags2::utf8) == fmtflags2::utf8 ? unicode : csv;
        size_t const column_width =
            this->column_width.has_value() ? this->column_width.value() : auto_column_width(flags);
 
        auto char_first_sequence = [&]([[maybe_unused]] size_t const i) -> std::string
        {
            if constexpr (!has_first_sequence)
                return " ";
            else
                return as_string(first_sequence()[i], flags);
        };
 
        auto char_second_sequence = [&]([[maybe_unused]] size_t const i) -> std::string
        {
            if constexpr (!has_second_sequence)
                return " ";
            else
                return as_string(second_sequence()[i], flags);
        };
 
        auto print_cell = [&](std::string const & symbol)
        {
            // deal with unicode chars that mess up std::setw
            size_t const length_bytes = symbol.size();
            size_t const length = unicode_str_length(symbol);
            size_t const offset = length_bytes - length;
 
            cout << std::left << std::setw(column_width + offset) << symbol << symbols.col_sep;
        };
 
        auto print_first_cell = [&](std::string const & symbol)
        {
            cout << symbol << symbols.col_sep;
        };
 
        // |_|d|a|t|a|b|a|s|e|
        auto print_first_row = [&]
        {
            print_first_cell(" ");
            print_cell(symbols.epsilon);
 
            for (size_t col = 0; col < cols() - 1; ++col)
                print_cell(char_first_sequence(col));
 
            cout << "\n";
        };
 
        // |-|-|-|-|-|-|-|-|-|
        auto print_divider = [&]
        {
            cout << " " << symbols.row_col_sep;
            for (size_t col = 0; col < cols(); ++col)
            {
                for (size_t i = 0; i < column_width; ++i)
                    cout << symbols.row_sep;
 
                cout << symbols.row_col_sep;
            }
            cout << "\n";
        };
 
        print_first_row();
        for (size_t row = 0; row < rows(); ++row)
        {
            if (symbols.row_sep[0] != '\0')
                print_divider();
 
            // one query letter + one row of scores / traces
            if (row == 0)
                print_first_cell(symbols.epsilon);
            else
                print_first_cell(char_second_sequence(row - 1));
 
            for (size_t col = 0; col < cols(); ++col)
                print_cell(entry_at({row_index_type{row}, column_index_type{col}}, flags));
 
            cout << "\n";
        }
    }
 
    size_t auto_column_width(fmtflags2 const flags) const noexcept
    {
        size_t col_width = 1;
        for (size_t row = 0; row < rows(); ++row)
            for (size_t col = 0; col < cols(); ++col)
                col_width =
                    std::max(col_width,
                             unicode_str_length(entry_at({row_index_type{row}, column_index_type{col}}, flags)));
 
        return col_width;
    }
 
protected:
    std::string entry_at(matrix_coordinate const coordinate, fmtflags2 flags) const noexcept
    {
        format_type const & symbols = (flags & fmtflags2::utf8) == fmtflags2::utf8 ? unicode : csv;
 
        value_type const & entry = at(coordinate);
        if (!is_traceback_matrix && entry == matrix_inf<value_type>)
            return symbols.inf;
 
        return as_string(entry, flags);
    }
 
    template <typename value_type>
    static std::string as_string(value_type && entry, fmtflags2 const flags) noexcept
    {
        std::stringstream strstream;
        debug_stream_type stream{strstream};
        stream << flags << entry;
        return strstream.str();
    }
 
    static size_t unicode_str_length(std::string const & str) noexcept
    {
        size_t length = 0u;
        for (auto it = str.cbegin(); it < str.cend(); ++it, ++length)
        {
            uint8_t v = *it;
            if ((v & 0b11100000) == 0b11000000)
                ++it;
            else if ((v & 0b11110000) == 0b11100000)
                it += 2;
            else if ((v & 0b11111000) == 0b11110000)
                it += 3;
        }
        return length;
    }
 
    struct format_type
    {
        char const * epsilon{};
        char const * col_sep{};
        char const * row_sep{};
        char const * row_col_sep{};
        char const * inf{};
    };
 
    static constexpr format_type csv{" ", ";", "", "", ""};
    static constexpr format_type unicode{"ε", "║", "═", "╬", "∞"};
 
public:
    std::optional<size_t> column_width{std::nullopt};
 
protected:
    matrix_t _matrix;
    first_sequence_t _first_sequence;
    second_sequence_t _second_sequence;
    std::optional<size_t> _rows{};
    std::optional<size_t> _cols{};
    std::optional<row_wise_matrix<bool>> _masking_matrix{};
    bool _transpose{};
    bool _transpose_mask{};
};
 
template <matrix matrix_t>
debug_matrix(matrix_t &&) -> debug_matrix<matrix_t>;
 
template <matrix matrix_t, typename first_sequence_t, typename second_sequence_t>
debug_matrix(matrix_t &&, first_sequence_t &&, second_sequence_t &&)
    -> debug_matrix<matrix_t, first_sequence_t, second_sequence_t>;
 
} // namespace seqan3::detail
 
namespace seqan3
{
template <typename char_t, detail::matrix alignment_matrix_t>
inline debug_stream_type<char_t> & operator<<(debug_stream_type<char_t> & s, alignment_matrix_t && matrix)
{
    detail::debug_matrix debug{std::forward<alignment_matrix_t>(matrix)};
 
    std::stringstream sstream{};
    debug.stream_matrix(sstream, s.flags2());
    s << sstream.str();
    return s;
}
 
template <typename char_t, std::ranges::input_range alignment_matrix_t>
    requires detail::debug_stream_range_guard<alignment_matrix_t> && detail::matrix<alignment_matrix_t>
inline debug_stream_type<char_t> & operator<<(debug_stream_type<char_t> & s, alignment_matrix_t && matrix)
{
    return s << detail::debug_matrix{std::forward<alignment_matrix_t>(matrix)};
}
 
} // namespace seqan3