Initial implementation of the ordinal recoder. (dmlc#11098)

Author: trivialfis

src/common/device_helpers.cuh

@@ -453,10 +453,15 @@ xgboost::common::Span<T> ToSpan(VectorT &vec, IndexT offset = 0,
 }
 
 template <typename T>
-xgboost::common::Span<T> ToSpan(thrust::device_vector<T> &vec, size_t offset, size_t size) {
+xgboost::common::Span<T> ToSpan(device_vector<T> &vec, size_t offset, size_t size) {
   return ToSpan(vec, offset, size);
 }
 
+template <typename T>
+xgboost::common::Span<std::add_const_t<T>> ToSpan(device_vector<T> const &vec) {
+  return {thrust::raw_pointer_cast(vec.data()), vec.size()};
+}
+
 template <typename T>
 xgboost::common::Span<T> ToSpan(DeviceUVector<T> &vec) {
   return {vec.data(), vec.size()};

src/data/cat_container.cuh

@@ -0,0 +1,51 @@
+/**
+ * Copyright 2024, XGBoost Contributors
+ */
+#pragma once
+#include "../common/device_helpers.cuh"  // for ToSpan
+#include "../common/device_vector.cuh"   // for device_vector, XGBDeviceAllocator
+#include "../encoder/ordinal.h"          // for CatCharT
+
+namespace xgboost::cuda_impl {
+struct CatStrArray {
+  dh::device_vector<std::int32_t> offsets;
+  dh::device_vector<enc::CatCharT> values;
+
+  CatStrArray() = default;
+  CatStrArray(CatStrArray const& that) = delete;
+  CatStrArray& operator=(CatStrArray const& that) = delete;
+
+  CatStrArray(CatStrArray&& that) = default;
+  CatStrArray& operator=(CatStrArray&& that) = default;
+
+  [[nodiscard]] explicit operator enc::CatStrArrayView() const {
+    return {dh::ToSpan(offsets), dh::ToSpan(values)};
+  }
+  [[nodiscard]] std::size_t size() const {  // NOLINT
+    return enc::CatStrArrayView(*this).size();
+  }
+};
+
+template <typename T>
+struct ViewToStorageImpl;
+
+template <>
+struct ViewToStorageImpl<enc::CatStrArrayView> {
+  using Type = CatStrArray;
+};
+
+template <typename T>
+struct ViewToStorageImpl<common::Span<T const>> {
+  using Type = dh::device_vector<T>;
+};
+
+template <typename... Ts>
+struct ViewToStorage;
+
+template <typename... Ts>
+struct ViewToStorage<std::tuple<Ts...>> {
+  using Type = std::tuple<typename ViewToStorageImpl<Ts>::Type...>;
+};
+
+using CatIndexTypes = ViewToStorage<enc::CatIndexViewTypes>::Type;
+}  // namespace xgboost::cuda_impl

src/data/cat_container.h

@@ -0,0 +1,48 @@
+/**
+ * Copyright 2025, XGBoost Contributors
+ */
+#pragma once
+
+#include <cstdint>  // for int32_t, int8_t
+#include <tuple>    // for tuple
+#include <vector>   // for vector
+
+#include "../encoder/ordinal.h"  // for DictionaryView
+
+namespace xgboost {
+namespace cpu_impl {
+struct CatStrArray {
+  std::vector<std::int32_t> offsets;
+  std::vector<enc::CatCharT> values;
+
+  [[nodiscard]] explicit operator enc::CatStrArrayView() const { return {offsets, values}; }
+  [[nodiscard]] std::size_t size() const {  // NOLINT
+    return enc::CatStrArrayView(*this).size();
+  }
+};
+
+template <typename T>
+struct ViewToStorageImpl;
+
+template <>
+struct ViewToStorageImpl<enc::CatStrArrayView> {
+  using Type = CatStrArray;
+};
+
+template <typename T>
+struct ViewToStorageImpl<common::Span<T const>> {
+  using Type = std::vector<T>;
+};
+
+template <typename... Ts>
+struct ViewToStorage;
+
+template <typename... Ts>
+struct ViewToStorage<std::tuple<Ts...>> {
+  using Type = std::tuple<typename ViewToStorageImpl<Ts>::Type...>;
+};
+
+using CatIndexTypes = ViewToStorage<enc::CatIndexViewTypes>::Type;
+using ColumnType = enc::cpu_impl::TupToVarT<CatIndexTypes>;
+}  // namespace cpu_impl
+}  // namespace xgboost