Implement N-gram indexer

CMakeLists.txt

@@ -185,6 +185,12 @@ if(MSVC)
         _iris_cxx_best_practices
         INTERFACE
             /W4 /analyze /analyze:external-
+            $<$<CONFIG:Release>:/Zi /Zo>
+    )
+    target_link_options(
+        _iris_cxx_best_practices
+        INTERFACE
+            $<$<CONFIG:Release>:/DEBUG:FULL>
     )
 
 else()

include/iris/format.hpp

@@ -0,0 +1,27 @@
+#ifndef IRIS_FORMAT_HPP
+#define IRIS_FORMAT_HPP
+
+// SPDX-License-Identifier: MIT
+
+#include <iris/format_traits.hpp>
+
+#include <format>
+
+namespace iris {
+
+template<class CharT>
+struct no_spec_formatter
+{
+    static constexpr std::basic_format_parse_context<CharT>::const_iterator
+    parse(std::basic_format_parse_context<CharT>& ctx)
+    {
+        auto it = ctx.begin();
+        if (it == ctx.end()) return it;
+        if (*it == format_traits<CharT>::brace_close) return it;
+        throw std::format_error("unknown format specifier");
+    }
+};
+
+} // iris
+
+#endif

include/iris/format_traits.hpp

@@ -1,4 +1,4 @@
-#ifndef IRIS_ZZ_FORMAT_TRAITS_HPP
+#ifndef IRIS_ZZ_FORMAT_TRAITS_HPP
 #define IRIS_ZZ_FORMAT_TRAITS_HPP
 
 // SPDX-License-Identifier: MIT
@@ -16,8 +16,11 @@ template<>
 struct format_traits<char>
 {
     using char_type = char;
+    static constexpr char_type square_brace_open = '[';
+    static constexpr char_type paren_close = ')';
     static constexpr char_type brace_open = '{';
     static constexpr char_type brace_close = '}';
+    static constexpr char_type comma = ',';
 
     template<class T>
     static constexpr std::basic_format_string<char_type, std::type_identity_t<T>>
@@ -28,8 +31,11 @@ template<>
 struct format_traits<wchar_t>
 {
     using char_type = wchar_t;
+    static constexpr char_type square_brace_open = L'[';
+    static constexpr char_type paren_close = L')';
     static constexpr char_type brace_open = L'{';
     static constexpr char_type brace_close = L'}';
+    static constexpr char_type comma = L',';
 
     template<class T>
     static constexpr std::basic_format_string<char_type, std::type_identity_t<T>>

include/iris/interval.hpp

@@ -0,0 +1,129 @@
+#ifndef IRIS_INTERVAL_HPP
+#define IRIS_INTERVAL_HPP
+
+// SPDX-License-Identifier: MIT
+
+#include <iris/config.hpp>
+#include <iris/format_traits.hpp>
+
+#include <iterator>
+#include <format>
+#include <concepts>
+#include <compare>
+#include <type_traits>
+#include <utility>
+
+namespace iris {
+
+template<std::signed_integral T>
+struct interval
+{
+    using value_type = T;
+    T left, right;
+
+    [[nodiscard]]
+    constexpr bool operator==(interval const&) const noexcept = default;
+
+    [[nodiscard]]
+    constexpr std::strong_ordering operator<=>(interval const&) const noexcept = default;
+};
+
+template<std::size_t I, class T>
+[[nodiscard]] constexpr T& get(interval<T>& iv) noexcept
+{
+    static_assert(I == 0 || I == 1);
+    if constexpr (I == 0) { return iv.left; } else { return iv.right; }
+}
+template<std::size_t I, class T>
+[[nodiscard]] constexpr T const& get(interval<T> const& iv) noexcept
+{
+    static_assert(I == 0 || I == 1);
+    if constexpr (I == 0) { return iv.left; } else { return iv.right; }
+}
+template<std::size_t I, class T>
+[[nodiscard]] constexpr T&& get(interval<T>&& iv) noexcept
+{
+    static_assert(I == 0 || I == 1);
+    if constexpr (I == 0) { return std::move(iv).left; } else { return std::move(iv).right; }
+}
+template<std::size_t I, class T>
+[[nodiscard]] constexpr T const&& get(interval<T> const&& iv) noexcept
+{
+    static_assert(I == 0 || I == 1);
+    if constexpr (I == 0) { return std::move(iv).left; } else { return std::move(iv).right; }
+}
+
+} // iris
+
+template<class T>
+struct std::tuple_size<iris::interval<T>>
+    : std::integral_constant<std::size_t, 2>
+{};
+
+template<std::size_t I, class T>
+struct std::tuple_element<I, iris::interval<T>>
+{
+    using type = T;
+};
+
+template<class T, class CharT>
+struct std::formatter<iris::interval<T>, CharT>
+{
+    [[nodiscard]] constexpr std::basic_format_parse_context<CharT>::const_iterator
+    parse(std::basic_format_parse_context<CharT>& ctx)
+    {
+        auto const first = ctx.begin();
+        if (first == ctx.end()) return first;
+        if (*first == iris::format_traits<CharT>::brace_close) return first;
+
+        // Bound the search to this replacement field
+        auto const close_it = std::find(
+            first, ctx.end(),
+            iris::format_traits<CharT>::brace_close
+        );
+        if (close_it == ctx.end()) {
+            throw std::format_error("unterminated format specifier");
+        }
+
+        auto const comma_it = std::find(
+            first, close_it,
+            iris::format_traits<CharT>::comma
+        );
+        if (comma_it == close_it) {
+            throw std::format_error("expected ',' in format specifier");
+        }
+
+        {
+            std::basic_format_parse_context<CharT> left_ctx{
+                std::basic_string_view<CharT>{first, comma_it}
+            };
+            if (left_fmt_.parse(left_ctx) != left_ctx.end()) {
+                throw std::format_error("trailing characters in left format specifier");
+            }
+        }
+        {
+            std::basic_format_parse_context<CharT> right_ctx{
+                std::basic_string_view<CharT>{std::next(comma_it), close_it}
+            };
+            if (right_fmt_.parse(right_ctx) != right_ctx.end()) {
+                throw std::format_error("trailing characters in right format specifier");
+            }
+        }
+        return close_it;
+    }
+
+    template<class Ctx>
+    Ctx::iterator format(iris::interval<T> const& iv, Ctx& ctx) const
+    {
+        ctx.advance_to(std::format_to(ctx.out(), "{}", iris::format_traits<CharT>::square_brace_open));
+        left_fmt_.format(iv.left, ctx);
+        ctx.advance_to(std::format_to(ctx.out(), "{}", iris::format_traits<CharT>::comma));
+        right_fmt_.format(iv.right, ctx);
+        return std::format_to(ctx.out(), "{}", iris::format_traits<CharT>::paren_close);
+    }
+
+private:
+    std::formatter<T, CharT> left_fmt_, right_fmt_;
+};
+
+#endif