Merge pull request #11 from h-2/wip

[feature] seq_io::reader and lots of required code

Author: h-2

include/bio/detail/in_file_iterator.hpp

@@ -0,0 +1,166 @@
+// -----------------------------------------------------------------------------------------------------
+// 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
+// -----------------------------------------------------------------------------------------------------
+
+/*!\file
+ * \brief Provides the seqan3::detail::in_file_iterator class template.
+ * \author Hannes Hauswedell <hannes.hauswedell AT fu-berlin.de>
+ */
+
+#pragma once
+
+#include <cassert>
+#include <ranges>
+
+#include <bio/platform.hpp>
+
+namespace bio::detail
+{
+
+/*!\brief Input iterator that provides a range-like interface for readers.
+ * \tparam file_type The data structure on which the iterator operates.
+ * \implements std::input_Iterator
+ * \ingroup bio
+ *
+ * This iterator is a single-pass input iterator for input files. All member types are resolved
+ * via `file_type`'s member types, dereference is implemented via file's `front()` member
+ * function, and increment calls the `buffer_next_record()` member of file.
+ *
+ * Note that since this is a single-pass input iterator, post-increment returns void because
+ * previous iterators are always invalid (all iterators point to the current position in single-pass
+ * ranges).
+ *
+ * This iterator may be compared against std::default_sentinel_t, this check delegates to
+ * calling the `eof()` member function on the file's stream.
+ */
+template <typename file_type>
+class in_file_iterator
+{
+    static_assert(!std::is_const_v<file_type>,
+                  "You cannot iterate over const files, because the iterator changes the file.");
+
+public:
+    /*!\name Member types
+     * \brief The associated types are derived from the `file_type`.
+     * \{
+     */
+
+    //!\brief The value type.
+    using value_type        = typename file_type::record_type;
+    //!\brief The reference type.
+    using reference         = typename file_type::record_type &;
+    //!\brief The const reference type.
+    using const_reference   = typename file_type::record_type &;
+    //!\brief The size type.
+    using size_type         = size_t;
+    //!\brief The difference type. A signed integer type, usually std::ptrdiff_t.
+    using difference_type   = ptrdiff_t;
+    //!\brief The pointer type.
+    using pointer           = typename file_type::record_type *;
+    //!\brief Tag this class as an input iterator.
+    using iterator_category = std::input_iterator_tag;
+    //!\}
+
+    /*!\name Constructors, destructor and assignment.
+     * \{
+     */
+    in_file_iterator()                         = default;             //!< Defaulted.
+    in_file_iterator(in_file_iterator const &) = default;             //!< Defaulted.
+    in_file_iterator(in_file_iterator &&)      = default;             //!< Defaulted.
+    ~in_file_iterator()                        = default;             //!< Defaulted.
+    in_file_iterator & operator=(in_file_iterator const &) = default; //!< Defaulted.
+    in_file_iterator & operator=(in_file_iterator &&) = default;      //!< Defaulted.
+
+    //!\brief Construct with reference to host.
+    in_file_iterator(file_type & _host) noexcept : host{&_host} {}
+    //!\}
+
+    /*!\name Iterator operations
+     * \{
+     */
+    //!\brief Move to the next record in the file and return a reference to it.
+    in_file_iterator & operator++()
+    {
+        assert(host != nullptr);
+        host->read_next_record();
+        return *this;
+    }
+
+    //!\brief Post-increment is the same as pre-increment, but returns void.
+    void operator++(int)
+    {
+        assert(host != nullptr);
+        ++(*this);
+    }
+
+    //!\brief Dereference returns the currently buffered record.
+    reference operator*() noexcept
+    {
+        assert(host != nullptr);
+        return host->record_buffer;
+    }
+
+    //!\brief Dereference returns the currently buffered record.
+    reference operator*() const noexcept
+    {
+        assert(host != nullptr);
+        return host->record_buffer;
+    }
+
+    //!\brief Dereference returns the currently buffered record.
+    value_type * operator->() noexcept
+    {
+        assert(host != nullptr);
+        return &host->record_buffer;
+    }
+
+    //!\brief Dereference returns the currently buffered record.
+    value_type const * operator->() const noexcept
+    {
+        assert(host != nullptr);
+        return &host->record_buffer;
+    }
+
+    //!\}
+
+    /*!\name Comparison operators
+     * \brief Only (in-)equality comparison of iterator with end() is supported.
+     * \{
+     */
+
+    //!\brief Checks whether `*this` is equal to the sentinel.
+    constexpr bool operator==(std::default_sentinel_t const &) const noexcept
+    {
+        assert(host != nullptr);
+        return host->at_end;
+    }
+
+    //!\brief Checks whether `*this` is not equal to the sentinel.
+    constexpr bool operator!=(std::default_sentinel_t const &) const noexcept
+    {
+        assert(host != nullptr);
+        return !host->at_end;
+    }
+
+    //!\brief Checks whether `it` is equal to the sentinel.
+    constexpr friend bool operator==(std::default_sentinel_t const &, in_file_iterator const & it) noexcept
+    {
+        return (it == std::default_sentinel);
+    }
+
+    //!\brief Checks whether `it` is not equal to the sentinel.
+    constexpr friend bool operator!=(std::default_sentinel_t const &, in_file_iterator const & it) noexcept
+    {
+        return (it != std::default_sentinel);
+    }
+    //!\}
+
+private:
+    //!\brief Pointer to file host.
+    file_type * host{};
+};
+
+} // namespace bio::detail

include/bio/detail/misc.hpp

@@ -0,0 +1,76 @@
+// -----------------------------------------------------------------------------------------------------
+// Copyright (c) 2006-2021, Knut Reinert & Freie Universität Berlin
+// Copyright (c) 2016-2021, Knut Reinert & MPI für molekulare Genetik
+// Copyright (c) 2020-2021, deCODE Genetics
+// 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/bio/blob/master/LICENSE.md
+// -----------------------------------------------------------------------------------------------------
+
+/*!\file
+ * \brief Provides miscellaneous utilities.
+ * \author Hannes Hauswedell <hannes.hauswedell AT decode.is>
+ */
+
+#pragma once
+
+#include <algorithm>
+#include <concepts>
+#include <filesystem>
+#include <ranges>
+#include <string>
+
+#include <seqan3/core/detail/template_inspection.hpp>
+#include <seqan3/utility/type_list/detail/type_list_algorithm.hpp>
+#include <seqan3/utility/type_list/type_list.hpp>
+
+#include <bio/exception.hpp>
+
+namespace bio::detail
+{
+
+/*!\addtogroup bio
+ * \{
+ */
+
+/*!\brief Sets the file format according to the file name extension.
+ * \param[out] format    The format to set.
+ * \param[in]  file_name The file name to extract the extension from.
+ *
+ * \throws seqan3::unhandled_extension_error If the extension in file_name does
+ *         not occur in any valid extensions of the formats specified in the
+ *         \p format_variant_type template argument list.
+ */
+void set_format(auto & format, std::filesystem::path const & file_name)
+{
+    using format_variant_type = std::remove_cvref_t<decltype(format)>;
+    using valid_formats       = seqan3::detail::transfer_template_args_onto_t<format_variant_type, seqan3::type_list>;
+
+    bool        format_found = false;
+    std::string extension    = file_name.extension().string();
+    if (extension.size() > 1)
+    {
+        extension = extension.substr(1); // drop leading "."
+        seqan3::detail::for_each<valid_formats>(
+          [&](auto fmt)
+          {
+              using fm_type = typename decltype(fmt)::type; // remove type_identity wrapper
+
+              for (auto const & ext : fm_type::file_extensions)
+              {
+                  if (std::ranges::equal(ext, extension))
+                  {
+                      format.template emplace<fm_type>();
+                      format_found = true;
+                      return;
+                  }
+              }
+          });
+    }
+
+    if (!format_found)
+        throw unhandled_extension_error("No valid format found for this extension.");
+}
+
+//!\}
+
+} // namespace bio::detail

include/bio/detail/range.hpp

@@ -13,8 +13,10 @@
 
 #pragma once
 
+#include <ranges>
+#include <span>
+
 #include <seqan3/alphabet/concept.hpp>
-#include <seqan3/utility/concept/container.hpp>
 
 #include <bio/platform.hpp>
 
@@ -25,6 +27,57 @@ namespace bio::detail
  * \{
  */
 
+// ----------------------------------------------------------------------------
+// concepts
+// ----------------------------------------------------------------------------
+
+/*!\interface bio::detail::back_insertable_with <>
+ * \extends std::ranges::output_range
+ * \tparam rng_t The container type.
+ * \tparam val_t The type to append to the container.
+ * \brief Describes range types that can grow in amortised constant time by appending an element of type val_t.
+ */
+//!\cond
+template <typename rng_t, typename val_t>
+concept back_insertable_with = std::ranges::output_range<rng_t, val_t> && requires(rng_t & v)
+{
+    v.push_back(std::declval<val_t>());
+};
+//!\endcond
+
+/*!\interface bio::detail::back_insertable <>
+ * \extends std::ranges::output_range
+ * \extends std::ranges::input_range
+ * \tparam rng_t The container type.
+ * \brief Describes range types that can grow in amortised constant time by appending an element.
+ */
+//!\cond
+template <typename rng_t>
+concept back_insertable =
+  std::ranges::input_range<rng_t> && back_insertable_with<rng_t, std::ranges::range_reference_t<rng_t>>;
+//!\endcond
+
+//!\brief A seqan3::alphabet that is **not** a character or number (any std::integral).
+template <typename t>
+concept deliberate_alphabet = seqan3::alphabet<t> && !std::integral<std::remove_cvref_t<t>>;
+
+//!\brief A range whose value type is `char`.
+template <typename t>
+concept char_range = std::ranges::range<t> && std::same_as<char, std::remove_cvref_t<std::ranges::range_value_t<t>>>;
+
+//!\brief A range whose value type is an integral type other than `char`.
+template <typename t>
+concept int_range = std::ranges::range<t> && std::integral<std::remove_cvref_t<std::ranges::range_value_t<t>>> &&
+  !std::same_as<char, std::remove_cvref_t<std::ranges::range_value_t<t>>>;
+
+//!\brief A type that is not std::span<std::byte const>.
+template <typename t>
+concept not_a_byte_span = !std::same_as<t, std::span<std::byte const>>;
+
+// ----------------------------------------------------------------------------
+// copy functions
+// ----------------------------------------------------------------------------
+
 /*!\brief Copy elements from the first range into the second range.
  * \param[in] in The range to copy from.
  * \param[out] out The range to copy to.
@@ -36,8 +89,8 @@ namespace bio::detail
  * If the input range is sized and the target range offers a `.resize()` member, this function uses
  * resize and assignment instead of back-insertion.
  */
-void sized_range_copy(std::ranges::input_range auto &&                                                   in,
-                      seqan3::back_insertable_with<std::ranges::range_reference_t<decltype(in)>> auto && out)
+void sized_range_copy(std::ranges::input_range auto &&                                           in,
+                      back_insertable_with<std::ranges::range_reference_t<decltype(in)>> auto && out)
 {
     using in_t  = decltype(in);
     using out_t = decltype(out);
@@ -64,23 +117,6 @@ void string_copy(std::string_view const in, auto & out)
         sized_range_copy(in, out);
 }
 
-//!\brief A seqan3::alphabet that is **not** a character or number (any std::integral).
-template <typename t>
-concept deliberate_alphabet = seqan3::alphabet<t> && !std::integral<std::remove_cvref_t<t>>;
-
-//!\brief A range whose value type is `char`.
-template <typename t>
-concept char_range = std::ranges::range<t> && std::same_as<char, std::remove_cvref_t<std::ranges::range_value_t<t>>>;
-
-//!\brief A range whose value type is an integral type other than `char`.
-template <typename t>
-concept int_range = std::ranges::range<t> && std::integral<std::remove_cvref_t<std::ranges::range_value_t<t>>> &&
-  !std::same_as<char, std::remove_cvref_t<std::ranges::range_value_t<t>>>;
-
-//!\brief A type that is not std::span<std::byte const>.
-template <typename t>
-concept not_a_byte_span = !std::same_as<t, std::span<std::byte const>>;
-
 //!\}
 
 } // namespace bio::detail