c_api.cc

/**
 * Copyright 2014-2025, XGBoost Contributors
 */
#include "xgboost/c_api.h"

#include <algorithm>     // for copy, transform
#include <cinttypes>     // for strtoimax
#include <cmath>         // for nan
#include <cstring>       // for strcmp
#include <limits>        // for numeric_limits
#include <map>           // for operator!=, _Rb_tree_const_iterator, _Rb_tre...
#include <memory>        // for shared_ptr, allocator, __shared_ptr_access
#include <string>        // for char_traits, basic_string, operator==, string
#include <system_error>  // for errc
#include <utility>       // for pair
#include <vector>        // for vector

#include "../common/api_entry.h"         // for XGBAPIThreadLocalEntry
#include "../common/charconv.h"          // for from_chars, to_chars, NumericLimits, from_ch...
#include "../common/cuda_rt_utils.h"     // for MemoryPoolsSupported
#include "../common/error_msg.h"         // for NoFederated
#include "../common/hist_util.h"         // for HistogramCuts
#include "../common/io.h"                // for FileExtension, LoadSequentialFile, MemoryBuf...
#include "../common/threading_utils.h"   // for OmpGetNumThreads, ParallelFor
#include "../data/adapter.h"             // for ArrayAdapter, DenseAdapter
#include "../data/batch_utils.h"         // for MatchingPageBytes, CachePageRatio
#include "../data/cat_container.h"       // for CatContainer
#include "../data/ellpack_page.h"        // for EllpackPage
#include "../data/metainfo.h"            // for DispatchDType
#include "../data/proxy_dmatrix.h"       // for DMatrixProxy
#include "../data/simple_dmatrix.h"      // for SimpleDMatrix
#include "../encoder/types.h"            // for Overloaded
#include "c_api_error.h"                 // for xgboost_CHECK_C_ARG_PTR, API_END, API_BEGIN
#include "c_api_utils.h"                 // for RequiredArg, OptionalArg, GetMissing, CastDM...
#include "dmlc/base.h"                   // for BeginPtr
#include "dmlc/io.h"                     // for Stream
#include "dmlc/parameter.h"              // for FieldAccessEntry, FieldEntry, ParamManager
#include "dmlc/thread_local.h"           // for ThreadLocalStore
#include "xgboost/base.h"                // for bst_ulong, bst_float, GradientPair, bst_feat...
#include "xgboost/context.h"             // for Context
#include "xgboost/data.h"                // for DMatrix, MetaInfo, DataType, ExtSparsePage
#include "xgboost/feature_map.h"         // for FeatureMap
#include "xgboost/global_config.h"       // for GlobalConfiguration, GlobalConfigThreadLocal...
#include "xgboost/host_device_vector.h"  // for HostDeviceVector
#include "xgboost/json.h"                // for Json, get, Integer, IsA, Boolean, String
#include "xgboost/learner.h"             // for Learner, PredictionType
#include "xgboost/logging.h"             // for LOG_FATAL, LogMessageFatal, CHECK, LogCheck_EQ
#include "xgboost/predictor.h"           // for PredictionCacheEntry
#include "xgboost/span.h"                // for Span
#include "xgboost/string_view.h"         // for StringView, operator<<
#include "xgboost/version_config.h"      // for XGBOOST_VER_MAJOR, XGBOOST_VER_MINOR, XGBOOS...
#include "xgboost/windefs.h"             // for xgboost_IS_WIN

using namespace xgboost;  // NOLINT(*);

XGB_DLL void XGBoostVersion(int *major, int *minor, int *patch) {
  if (major) {
    *major = XGBOOST_VER_MAJOR;
  }
  if (minor) {
    *minor = XGBOOST_VER_MINOR;
  }
  if (patch) {
    *patch = XGBOOST_VER_PATCH;
  }
}

static_assert(DMLC_CXX11_THREAD_LOCAL, "XGBoost depends on thread-local storage.");
using GlobalConfigAPIThreadLocalStore = dmlc::ThreadLocalStore<XGBAPIThreadLocalEntry>;

#if !defined(XGBOOST_USE_CUDA)
namespace xgboost {
void XGBBuildInfoDevice(Json *p_info) {
  auto &info = *p_info;
  info["USE_CUDA"] = Boolean{false};
  info["USE_NCCL"] = Boolean{false};
  info["USE_RMM"] = Boolean{false};
  info["USE_DLOPEN_NCCL"] = Boolean{false};
}
}  // namespace xgboost
#endif

XGB_DLL int XGBuildInfo(char const **out) {
  API_BEGIN_UNGUARD()
  xgboost_CHECK_C_ARG_PTR(out);
  Json info{Object{}};

#if defined(XGBOOST_BUILTIN_PREFETCH_PRESENT)
  info["BUILTIN_PREFETCH_PRESENT"] = Boolean{true};
#else
  info["BUILTIN_PREFETCH_PRESENT"] = Boolean{false};
#endif

#if defined(XGBOOST_MM_PREFETCH_PRESENT)
  info["MM_PREFETCH_PRESENT"] = Boolean{true};
#else
  info["MM_PREFETCH_PRESENT"] = Boolean{false};
#endif

#if defined(_OPENMP)
  info["USE_OPENMP"] = Boolean{true};
#else
  info["USE_OPENMP"] = Boolean{false};
#endif

#if defined(__GNUC__) && !defined(__clang__)
  info["GCC_VERSION"] = std::vector<Json>{Json{Integer{__GNUC__}}, Json{Integer{__GNUC_MINOR__}},
                                          Json{Integer{__GNUC_PATCHLEVEL__}}};
#endif

#if defined(__GLIBC__)
  info["GLIBC_VERSION"] = std::vector<Json>{Json{__GLIBC__}, Json{__GLIBC_MINOR__}};
#endif  // defined(__GLIBC__)

#if defined(__clang__)
  info["CLANG_VERSION"] =
      std::vector<Json>{Json{Integer{__clang_major__}}, Json{Integer{__clang_minor__}},
                        Json{Integer{__clang_patchlevel__}}};
#endif

#if !defined(NDEBUG)
  info["DEBUG"] = Boolean{true};
#else
  info["DEBUG"] = Boolean{false};
#endif

#if defined(XGBOOST_USE_FEDERATED)
  info["USE_FEDERATED"] = Boolean{true};
#else
  info["USE_FEDERATED"] = Boolean{false};
#endif

#if defined(XGBOOST_GIT_HASH)
  char const *git_hash = XGBOOST_GIT_HASH;
  info["GIT_HASH"] = String{git_hash};
#endif

  XGBBuildInfoDevice(&info);

  auto &out_str = GlobalConfigAPIThreadLocalStore::Get()->ret_str;
  Json::Dump(info, &out_str);
  *out = out_str.c_str();

  API_END();
}

XGB_DLL int XGBRegisterLogCallback(void (*callback)(const char *)) {
  API_BEGIN_UNGUARD()
  LogCallbackRegistry *registry = LogCallbackRegistryStore::Get();
  registry->Register(callback);
  API_END();
}

XGB_DLL int XGBSetGlobalConfig(const char *json_str) {
  API_BEGIN_UNGUARD()

  xgboost_CHECK_C_ARG_PTR(json_str);
  Json config{Json::Load(StringView{json_str})};

  // handle nthread, it's not a dmlc parameter.
  auto &obj = get<Object>(config);
  auto it = obj.find("nthread");
  if (it != obj.cend()) {
    auto nthread = OptionalArg<Integer>(config, "nthread", Integer::Int{0});
    if (nthread > 0) {
      omp_set_num_threads(nthread);
      GlobalConfigThreadLocalStore::Get()->nthread = nthread;
    }
    get<Object>(config).erase("nthread");
  }

  for (auto &items : obj) {
    switch (items.second.GetValue().Type()) {
      case xgboost::Value::ValueKind::kInteger: {
        items.second = String{std::to_string(get<Integer const>(items.second))};
        break;
      }
      case xgboost::Value::ValueKind::kBoolean: {
        if (get<Boolean const>(items.second)) {
          items.second = String{"true"};
        } else {
          items.second = String{"false"};
        }
        break;
      }
      case xgboost::Value::ValueKind::kNumber: {
        auto n = get<Number const>(items.second);
        char chars[NumericLimits<float>::kToCharsSize];
        auto ec = to_chars(chars, chars + sizeof(chars), n).ec;
        CHECK(ec == std::errc());
        items.second = String{chars};
        break;
      }
      default:
        break;
    }
  }
  auto unknown = FromJson(config, GlobalConfigThreadLocalStore::Get());
  if (!unknown.empty()) {
    std::stringstream ss;
    ss << "Unknown global parameters: { ";
    size_t i = 0;
    for (auto const &item : unknown) {
      ss << item.first;
      i++;
      if (i != unknown.size()) {
        ss << ", ";
      }
    }
    LOG(FATAL) << ss.str() << " }";
  }

  // Check configuration is valid.
  bool use_async_pool = GlobalConfigThreadLocalStore::Get()->use_cuda_async_pool;
#if defined(XGBOOST_USE_RMM)
  CHECK(!use_async_pool) << "Cannot enable `use_cuda_async_pool` when compiled with RMM.";
#endif  // defined(XGBOOST_USE_RMM)
#if defined(xgboost_IS_WIN)
  CHECK(!use_async_pool) << "Cannot enable `use_cuda_async_pool` on Windows.";
#endif  // defined(XGBOOST_USE_RMM)
  if (use_async_pool && !curt::MemoryPoolsSupported(xgboost::curt::CurrentDevice())) {
    LOG(FATAL) << "CUDA async memory pool is not available for the current device.";
  }

  API_END();
}

XGB_DLL int XGBGetGlobalConfig(const char **json_str) {
  API_BEGIN_UNGUARD()
  auto const &global_config = *GlobalConfigThreadLocalStore::Get();
  Json config{ToJson(global_config)};
  auto const *mgr = global_config.__MANAGER__();

  for (auto &item : get<Object>(config)) {
    auto const &str = get<String const>(item.second);
    auto const &name = item.first;
    auto e = mgr->Find(name);
    CHECK(e);

    if (dynamic_cast<dmlc::parameter::FieldEntry<int32_t> const *>(e) ||
        dynamic_cast<dmlc::parameter::FieldEntry<int64_t> const *>(e) ||
        dynamic_cast<dmlc::parameter::FieldEntry<uint32_t> const *>(e) ||
        dynamic_cast<dmlc::parameter::FieldEntry<uint64_t> const *>(e)) {
      auto i = std::strtoimax(str.data(), nullptr, 10);
      CHECK_LE(i, static_cast<intmax_t>(std::numeric_limits<int64_t>::max()));
      item.second = Integer(static_cast<int64_t>(i));
    } else if (dynamic_cast<dmlc::parameter::FieldEntry<float> const *>(e) ||
               dynamic_cast<dmlc::parameter::FieldEntry<double> const *>(e)) {
      float f;
      auto ec = from_chars(str.data(), str.data() + str.size(), f).ec;
      CHECK(ec == std::errc());
      item.second = Number(f);
    } else if (dynamic_cast<dmlc::parameter::FieldEntry<bool> const *>(e)) {
      item.second = Boolean(str != "0");
    }
  }

  config["nthread"] = GlobalConfigThreadLocalStore::Get()->nthread;
  auto &local = *GlobalConfigAPIThreadLocalStore::Get();
  Json::Dump(config, &local.ret_str);

  xgboost_CHECK_C_ARG_PTR(json_str);
  *json_str = local.ret_str.c_str();
  API_END();
}

XGB_DLL int XGDMatrixCreateFromFile(const char *fname, int silent, DMatrixHandle *out) {
  xgboost_CHECK_C_ARG_PTR(fname);
  xgboost_CHECK_C_ARG_PTR(out);

  LOG(WARNING) << error::DeprecatedFunc(__func__, "2.0.0", "XGDMatrixCreateFromURI");

  Json config{Object()};
  config["uri"] = std::string{fname};
  config["silent"] = silent;
  std::string config_str;
  Json::Dump(config, &config_str);
  return XGDMatrixCreateFromURI(config_str.c_str(), out);
}

XGB_DLL int XGDMatrixCreateFromURI(const char *config, DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(config);
  xgboost_CHECK_C_ARG_PTR(out);

  auto jconfig = Json::Load(StringView{config});
  std::string uri = RequiredArg<String>(jconfig, "uri", __func__);
  auto silent = static_cast<bool>(OptionalArg<Integer, int64_t>(jconfig, "silent", 1));
  auto data_split_mode =
      static_cast<DataSplitMode>(OptionalArg<Integer, int64_t>(jconfig, "data_split_mode", 0));

  *out = new std::shared_ptr<DMatrix>(DMatrix::Load(uri, silent, data_split_mode));
  API_END();
}

XGB_DLL int XGDMatrixCreateFromDataIter(
    void *data_handle,                  // a Java iterator
    XGBCallbackDataIterNext *callback,  // C++ callback defined in xgboost4j.cpp
    const char *cache_info, float missing, DMatrixHandle *out) {
  API_BEGIN();

  std::string scache;
  if (cache_info != nullptr) {
    scache = cache_info;
  }
  xgboost::data::IteratorAdapter<DataIterHandle, XGBCallbackDataIterNext, XGBoostBatchCSR> adapter(
      data_handle, callback);
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>{DMatrix::Create(&adapter, missing, 1, scache)};
  API_END();
}

#ifndef XGBOOST_USE_CUDA
XGB_DLL int XGDMatrixCreateFromCudaColumnar(char const *, char const *, DMatrixHandle *) {
  API_BEGIN();
  common::AssertGPUSupport();
  API_END();
}

XGB_DLL int XGDMatrixCreateFromCudaArrayInterface(char const *, char const *, DMatrixHandle *) {
  API_BEGIN();
  common::AssertGPUSupport();
  API_END();
}

#endif

// Create from data iterator
XGB_DLL int XGDMatrixCreateFromCallback(DataIterHandle iter, DMatrixHandle proxy,
                                        DataIterResetCallback *reset, XGDMatrixCallbackNext *next,
                                        char const *config, DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(config);

  auto jconfig = Json::Load(StringView{config});
  auto missing = GetMissing(jconfig);
  std::string cache = RequiredArg<String>(jconfig, "cache_prefix", __func__);
  std::int32_t n_threads = OptionalArg<Integer, std::int64_t>(jconfig, "nthread", 0);
  auto on_host = OptionalArg<Boolean>(jconfig, "on_host", false);
  auto min_cache_page_bytes = OptionalArg<Integer, std::int64_t>(jconfig, "min_cache_page_bytes",
                                                                 cuda_impl::MatchingPageBytes());
  CHECK_EQ(min_cache_page_bytes, cuda_impl::MatchingPageBytes())
      << "Page concatenation is not supported by the DMatrix yet.";
  auto cache_host_ratio =
      OptionalArg<Number, float>(jconfig, "cache_host_ratio", cuda_impl::AutoHostRatio());

  xgboost_CHECK_C_ARG_PTR(next);
  xgboost_CHECK_C_ARG_PTR(reset);
  xgboost_CHECK_C_ARG_PTR(out);

  auto config =
      ExtMemConfig{cache, on_host, cache_host_ratio, min_cache_page_bytes, missing, n_threads};
  *out = new std::shared_ptr<xgboost::DMatrix>{
      xgboost::DMatrix::Create(iter, proxy, reset, next, config)};
  API_END();
}

namespace {
std::shared_ptr<DMatrix> GetRefDMatrix(DataIterHandle ref) {
  std::shared_ptr<DMatrix> _ref{nullptr};
  if (ref) {
    auto pp_ref = static_cast<std::shared_ptr<xgboost::DMatrix> *>(ref);
    StringView err{"Invalid handle to ref."};
    CHECK(pp_ref) << err;
    _ref = *pp_ref;
    CHECK(_ref) << err;
  }
  return _ref;
}

void WarnDeprecatedMaxQuantileBlocks(Json const &config) {
  auto const &obj = get<Object const>(config);
  auto it = obj.find("max_quantile_blocks");
  if (it != obj.cend() && !IsA<Null>(it->second)) {
    LOG(WARNING) << "`max_quantile_blocks` is deprecated and has no effect. "
                    "The parameter will be removed in a future release.";
  }
}
}  // namespace

XGB_DLL int XGQuantileDMatrixCreateFromCallback(DataIterHandle iter, DMatrixHandle proxy,
                                                DataIterHandle ref, DataIterResetCallback *reset,
                                                XGDMatrixCallbackNext *next, char const *config,
                                                DMatrixHandle *out) {
  API_BEGIN();
  std::shared_ptr<DMatrix> p_ref{GetRefDMatrix(ref)};

  xgboost_CHECK_C_ARG_PTR(config);
  auto jconfig = Json::Load(StringView{config});
  WarnDeprecatedMaxQuantileBlocks(jconfig);
  auto missing = GetMissing(jconfig);
  auto n_threads = OptionalArg<Integer, int64_t>(jconfig, "nthread", 0);
  auto max_bin = OptionalArg<Integer, int64_t>(jconfig, "max_bin", 256);

  xgboost_CHECK_C_ARG_PTR(next);
  xgboost_CHECK_C_ARG_PTR(reset);
  xgboost_CHECK_C_ARG_PTR(out);

  *out = new std::shared_ptr<xgboost::DMatrix>{
      xgboost::DMatrix::Create(iter, proxy, p_ref, reset, next, missing, n_threads, max_bin)};
  API_END();
}

XGB_DLL int XGExtMemQuantileDMatrixCreateFromCallback(DataIterHandle iter, DMatrixHandle proxy,
                                                      DataIterHandle ref,
                                                      DataIterResetCallback *reset,
                                                      XGDMatrixCallbackNext *next,
                                                      char const *config, DMatrixHandle *out) {
  API_BEGIN();
  std::shared_ptr<DMatrix> p_ref{GetRefDMatrix(ref)};

  xgboost_CHECK_C_ARG_PTR(config);
  auto jconfig = Json::Load(StringView{config});
  WarnDeprecatedMaxQuantileBlocks(jconfig);
  auto missing = GetMissing(jconfig);
  std::int32_t n_threads = OptionalArg<Integer, std::int64_t>(jconfig, "nthread", 0);
  auto max_bin = OptionalArg<Integer, std::int64_t>(jconfig, "max_bin", 256);
  auto on_host = OptionalArg<Boolean>(jconfig, "on_host", false);
  std::string cache = RequiredArg<String>(jconfig, "cache_prefix", __func__);
  auto min_cache_page_bytes = OptionalArg<Integer, std::int64_t>(jconfig, "min_cache_page_bytes",
                                                                 cuda_impl::AutoCachePageBytes());
  auto cache_host_ratio =
      OptionalArg<Number, float>(jconfig, "cache_host_ratio", cuda_impl::AutoHostRatio());

  xgboost_CHECK_C_ARG_PTR(next);
  xgboost_CHECK_C_ARG_PTR(reset);
  xgboost_CHECK_C_ARG_PTR(out);

  auto config =
      ExtMemConfig{cache, on_host, cache_host_ratio, min_cache_page_bytes, missing, n_threads};
  *out = new std::shared_ptr<xgboost::DMatrix>{
      xgboost::DMatrix::Create(iter, proxy, p_ref, reset, next, max_bin, config)};
  API_END();
}

XGB_DLL int XGProxyDMatrixCreate(DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<xgboost::DMatrix>(new xgboost::data::DMatrixProxy);
  API_END();
}

namespace {
[[nodiscard]] xgboost::data::DMatrixProxy *GetDMatrixProxy(DMatrixHandle handle) {
  auto p_m = static_cast<std::shared_ptr<xgboost::DMatrix> *>(handle);
  CHECK(p_m);
  auto m = static_cast<xgboost::data::DMatrixProxy *>(p_m->get());
  CHECK(m) << "Current DMatrix type does not support set data.";
  return m;
}
}  // namespace

XGB_DLL int XGProxyDMatrixSetDataCudaArrayInterface(DMatrixHandle handle, char const *data) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(data);
  GetDMatrixProxy(handle)->SetCudaArray(data);
  API_END();
}

XGB_DLL int XGProxyDMatrixSetDataCudaColumnar(DMatrixHandle handle, char const *data) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(data);
  GetDMatrixProxy(handle)->SetCudaColumnar(data);
  API_END();
}

XGB_DLL int XGProxyDMatrixSetDataColumnar(DMatrixHandle handle, char const *data) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(data);
  GetDMatrixProxy(handle)->SetColumnar(data);
  API_END();
}

XGB_DLL int XGProxyDMatrixSetDataDense(DMatrixHandle handle, char const *data) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(data);
  GetDMatrixProxy(handle)->SetArray(data);
  API_END();
}

XGB_DLL int XGProxyDMatrixSetDataCSR(DMatrixHandle handle, char const *indptr, char const *indices,
                                     char const *data, xgboost::bst_ulong ncol) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(indptr);
  xgboost_CHECK_C_ARG_PTR(indices);
  xgboost_CHECK_C_ARG_PTR(data);
  GetDMatrixProxy(handle)->SetCsr(indptr, indices, data, ncol, true);
  API_END();
}

// End Create from data iterator

XGB_DLL int XGDMatrixCreateFromColumnar(char const *data, char const *c_json_config,
                                        DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  xgboost_CHECK_C_ARG_PTR(data);

  auto config = Json::Load(c_json_config);
  float missing = GetMissing(config);
  auto n_threads = OptionalArg<Integer, std::int64_t>(config, "nthread", 0);
  auto data_split_mode =
      static_cast<DataSplitMode>(OptionalArg<Integer, int64_t>(config, "data_split_mode", 0));

  data::ColumnarAdapter adapter{data};
  *out = new std::shared_ptr<DMatrix>(
      DMatrix::Create(&adapter, missing, n_threads, "", data_split_mode));

  API_END();
}

XGB_DLL int XGDMatrixCreateFromCSR(char const *indptr, char const *indices, char const *data,
                                   xgboost::bst_ulong ncol, char const *c_json_config,
                                   DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(indptr);
  xgboost_CHECK_C_ARG_PTR(indices);
  xgboost_CHECK_C_ARG_PTR(data);
  data::CSRArrayAdapter adapter(StringView{indptr}, StringView{indices}, StringView{data}, ncol);
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  auto config = Json::Load(StringView{c_json_config});
  float missing = GetMissing(config);
  auto n_threads = OptionalArg<Integer, int64_t>(config, "nthread", 0);
  auto data_split_mode =
      static_cast<DataSplitMode>(OptionalArg<Integer, int64_t>(config, "data_split_mode", 0));
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>(
      DMatrix::Create(&adapter, missing, n_threads, "", data_split_mode));
  API_END();
}

XGB_DLL int XGDMatrixCreateFromDense(char const *data, char const *c_json_config,
                                     DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(data);
  xgboost::data::ArrayAdapter adapter{xgboost::data::ArrayAdapter(StringView{data})};
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  auto config = Json::Load(StringView{c_json_config});
  float missing = GetMissing(config);
  auto n_threads = OptionalArg<Integer, int64_t>(config, "nthread", 0);
  auto data_split_mode =
      static_cast<DataSplitMode>(OptionalArg<Integer, int64_t>(config, "data_split_mode", 0));
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>(
      DMatrix::Create(&adapter, missing, n_threads, "", data_split_mode));
  API_END();
}

XGB_DLL int XGDMatrixCreateFromCSC(char const *indptr, char const *indices, char const *data,
                                   xgboost::bst_ulong nrow, char const *c_json_config,
                                   DMatrixHandle *out) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(indptr);
  xgboost_CHECK_C_ARG_PTR(indices);
  xgboost_CHECK_C_ARG_PTR(data);
  data::CSCArrayAdapter adapter{StringView{indptr}, StringView{indices}, StringView{data},
                                static_cast<std::size_t>(nrow)};
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  auto config = Json::Load(StringView{c_json_config});
  float missing = GetMissing(config);
  auto n_threads = OptionalArg<Integer, int64_t>(config, "nthread", common::OmpGetNumThreads(0));
  auto data_split_mode =
      static_cast<DataSplitMode>(OptionalArg<Integer, int64_t>(config, "data_split_mode", 0));
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>(
      DMatrix::Create(&adapter, missing, n_threads, "", data_split_mode));

  API_END();
}

XGB_DLL int XGDMatrixCreateFromMat(const bst_float *data, xgboost::bst_ulong nrow,
                                   xgboost::bst_ulong ncol, bst_float missing, DMatrixHandle *out) {
  API_BEGIN();
  data::DenseAdapter adapter(data, nrow, ncol);
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>(DMatrix::Create(&adapter, missing, 1));
  API_END();
}

XGB_DLL int XGDMatrixCreateFromMat_omp(const bst_float *data,  // NOLINT
                                       xgboost::bst_ulong nrow, xgboost::bst_ulong ncol,
                                       bst_float missing, DMatrixHandle *out, int nthread) {
  API_BEGIN();
  data::DenseAdapter adapter(data, nrow, ncol);
  xgboost_CHECK_C_ARG_PTR(out);
  *out = new std::shared_ptr<DMatrix>(DMatrix::Create(&adapter, missing, nthread));
  API_END();
}

XGB_DLL int XGDMatrixSliceDMatrix(DMatrixHandle handle, const int *idxset, xgboost::bst_ulong len,
                                  DMatrixHandle *out) {
  xgboost_CHECK_C_ARG_PTR(out);
  return XGDMatrixSliceDMatrixEx(handle, idxset, len, out, 0);
}

XGB_DLL int XGDMatrixSliceDMatrixEx(DMatrixHandle handle, const int *idxset, xgboost::bst_ulong len,
                                    DMatrixHandle *out, int allow_groups) {
  API_BEGIN();
  CHECK_HANDLE();
  if (!allow_groups) {
    CHECK_EQ(static_cast<std::shared_ptr<DMatrix> *>(handle)->get()->Info().group_ptr_.size(), 0U)
        << "slice does not support group structure";
  }
  DMatrix *dmat = static_cast<std::shared_ptr<DMatrix> *>(handle)->get();
  *out = new std::shared_ptr<DMatrix>(dmat->Slice({idxset, static_cast<std::size_t>(len)}));
  API_END();
}

XGB_DLL int XGDMatrixFree(DMatrixHandle handle) {
  API_BEGIN();
  CHECK_HANDLE();
  delete static_cast<std::shared_ptr<DMatrix> *>(handle);
  API_END();
}

XGB_DLL int XGDMatrixSaveBinary(DMatrixHandle handle, const char *fname, int) {
  API_BEGIN();
  CHECK_HANDLE();
  auto dmat = static_cast<std::shared_ptr<DMatrix> *>(handle)->get();
  xgboost_CHECK_C_ARG_PTR(fname);
  if (data::SimpleDMatrix *derived = dynamic_cast<data::SimpleDMatrix *>(dmat)) {
    derived->SaveToLocalFile(fname);
  } else {
    LOG(FATAL) << "binary saving only supported by SimpleDMatrix";
  }
  API_END();
}

XGB_DLL int XGDMatrixSetFloatInfo(DMatrixHandle handle, const char *field, const bst_float *info,
                                  xgboost::bst_ulong len) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(field);
  auto const &p_fmat = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  p_fmat->SetInfo(field, linalg::Make1dInterface(info, len));
  API_END();
}

XGB_DLL int XGDMatrixSetInfoFromInterface(DMatrixHandle handle, char const *field,
                                          char const *interface_c_str) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(field);
  auto const &p_fmat = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  p_fmat->SetInfo(field, interface_c_str);
  API_END();
}

XGB_DLL int XGDMatrixSetUIntInfo(DMatrixHandle handle, const char *field, const unsigned *info,
                                 xgboost::bst_ulong len) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(field);
  LOG(WARNING) << error::DeprecatedFunc(__func__, "2.1.0", "XGDMatrixSetInfoFromInterface");
  auto const &p_fmat = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  p_fmat->SetInfo(field, linalg::Make1dInterface(info, len));
  API_END();
}

XGB_DLL int XGDMatrixSetStrFeatureInfo(DMatrixHandle handle, const char *field, const char **c_info,
                                       const xgboost::bst_ulong size) {
  API_BEGIN();
  CHECK_HANDLE();
  auto &info = static_cast<std::shared_ptr<DMatrix> *>(handle)->get()->Info();
  xgboost_CHECK_C_ARG_PTR(field);
  info.SetFeatureInfo(field, c_info, size);
  API_END();
}

XGB_DLL int XGDMatrixGetStrFeatureInfo(DMatrixHandle handle, const char *field,
                                       xgboost::bst_ulong *len, const char ***out_features) {
  API_BEGIN();
  CHECK_HANDLE();
  auto m = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  auto &info = static_cast<std::shared_ptr<DMatrix> *>(handle)->get()->Info();

  std::vector<const char *> &charp_vecs = m->GetThreadLocal().ret_vec_charp;
  std::vector<std::string> &str_vecs = m->GetThreadLocal().ret_vec_str;

  xgboost_CHECK_C_ARG_PTR(field);
  info.GetFeatureInfo(field, &str_vecs);

  charp_vecs.resize(str_vecs.size());
  for (size_t i = 0; i < str_vecs.size(); ++i) {
    charp_vecs[i] = str_vecs[i].c_str();
  }
  xgboost_CHECK_C_ARG_PTR(out_features);
  xgboost_CHECK_C_ARG_PTR(len);
  *out_features = dmlc::BeginPtr(charp_vecs);
  *len = static_cast<xgboost::bst_ulong>(charp_vecs.size());
  API_END();
}

namespace {
template <typename FidxT>
void GetCategoriesImpl(enc::HostColumnsView const &cats, FidxT n_features,
                       std::string *p_out_storage, char const **out) {
  auto &ret_str = *p_out_storage;
  ret_str.clear();

  // We can directly use the storage in the cat container instead of allocating temporary storage.
  Json jout{Array{}};
  for (decltype(n_features) f_idx = 0; f_idx < n_features; ++f_idx) {
    if (cats.Empty()) {
      get<Array>(jout).emplace_back();
      continue;
    }
    auto const &col = cats[f_idx];
    if (std::visit([](auto &&arg) { return arg.empty(); }, col)) {
      get<Array>(jout).emplace_back();
      continue;
    }
    std::visit(enc::Overloaded{[&](enc::CatStrArrayView const &str) {
                                 auto const &offsets = str.offsets;
                                 auto ovec = linalg::MakeVec(offsets.data(), offsets.size());
                                 auto jovec = linalg::ArrayInterface(ovec);

                                 auto const &values = str.values;
                                 auto dvec = linalg::MakeVec(values.data(), values.size());
                                 auto jdvec = linalg::ArrayInterface(dvec);

                                 get<Array>(jout).emplace_back(Object{});
                                 get<Array>(jout).back()["offsets"] = std::move(jovec);
                                 get<Array>(jout).back()["values"] = std::move(jdvec);
                               },
                               [&](auto &&values) {
                                 auto vec = linalg::MakeVec(values.data(), values.size());
                                 auto jvec = linalg::ArrayInterface(vec);
                                 get<Array>(jout).emplace_back(std::move(jvec));
                               }},
               col);
  }
  auto str = Json::Dump(jout);
  ret_str = std::move(str);

  *out = ret_str.c_str();
}

CatContainer *CopyCatContainer(Context const *ctx, CatContainer const *cats,
                               bst_feature_t n_features) {
  CatContainer *new_cats = new CatContainer{};
  new_cats->Copy(ctx, *cats);
  CHECK_EQ(new_cats->Empty(), cats->Empty());
  if (!new_cats->Empty()) {
    CHECK_EQ(new_cats->NumFeatures(), n_features);
    CHECK_EQ(new_cats->NumFeatures(), cats->NumFeatures());
  }
  return new_cats;
}
}  // anonymous namespace

/**
 * No actual container method is exposed through the C API. It's just an opaque handle at
 * the moment. This way we get to reuse the methods and the context from the DMatrix and
 * Booster.
 */
XGB_DLL int XGDMatrixGetCategories(DMatrixHandle handle, char const * /*config*/,
                                   CategoriesHandle *out) {
  API_BEGIN()
  CHECK_HANDLE()

  auto const p_fmat = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  auto const cats = p_fmat->Cats();
  xgboost_CHECK_C_ARG_PTR(out);
  if (cats->Empty()) {
    out = nullptr;
  } else {
    auto new_cats = CopyCatContainer(p_fmat->Ctx(), cats, p_fmat->Info().num_col_);
    *out = new_cats;
  }

  API_END()
}

XGB_DLL int XGDMatrixGetCategoriesExportToArrow(DMatrixHandle handle, char const * /*config*/,
                                                CategoriesHandle *out, char const **export_out) {
  API_BEGIN();
  CHECK_HANDLE()

  auto const p_fmat = *static_cast<std::shared_ptr<DMatrix> *>(handle);
  auto const cats = p_fmat->Cats();
  auto n_features = p_fmat->Info().num_col_;

  xgboost_CHECK_C_ARG_PTR(out);
  xgboost_CHECK_C_ARG_PTR(export_out);

  if (cats->Empty()) {
    *out = nullptr;
    *export_out = nullptr;
  } else {
    // Create a new container
    auto new_cats = CopyCatContainer(p_fmat->Ctx(), cats, n_features);
    *out = new_cats;
    // Export to arrow
    auto &ret_str = p_fmat->GetThreadLocal().ret_str;
    GetCategoriesImpl(new_cats->HostView(), n_features, &ret_str, export_out);
  }

  API_END();
}

XGB_DLL int XGBCategoriesFree(CategoriesHandle handle) {
  API_BEGIN();
  xgboost_CHECK_C_ARG_PTR(handle);
  auto p_cats = static_cast<CatContainer *>(handle);
  CHECK(p_cats);
  delete p_cats;
  API_END();
}

XGB_DLL int XGDMatrixSetDenseInfo(DMatrixHandle handle, const char *field, void const *data,
                                  xgboost::bst_ulong size, int type) {
  API_BEGIN();
  CHECK_HANDLE();
  LOG(WARNING) << error::DeprecatedFunc(__func__, "2.1.0", "XGDMatrixSetInfoFromInterface");
  CHECK(type >= 1 && type <= 4);
  xgboost_CHECK_C_ARG_PTR(field);

  Context ctx;
  auto dtype = static_cast<DataType>(type);
  auto p_fmat = CastDMatrixHandle(handle);

  // Legacy code using XGBoost dtype, which is a small subset of array interface types.
  data::DispatchDType(dtype, [&](auto dtype) {
    using DType = decltype(dtype);
    auto cast_d_ptr = reinterpret_cast<const DType *>(data);
    auto t = linalg::TensorView<DType const, 1>(
        common::Span<DType const>{cast_d_ptr,
                                  static_cast<typename common::Span<DType>::index_type>(size)},
        {size}, DeviceOrd::CPU());
    CHECK(t.CContiguous());
    Json iface{linalg::ArrayInterface(t)};
    CHECK(ArrayInterface<1>{iface}.is_contiguous);
    std::string str = Json::Dump(iface);
    p_fmat->Info().SetInfo(ctx, field, StringView{str});
  });
  API_END();
}

XGB_DLL int XGDMatrixGetInfoRef(DMatrixHandle handle, char const *field, char const **out_array) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(field);
  xgboost_CHECK_C_ARG_PTR(out_array);

  auto p_fmat = CastDMatrixHandle(handle);
  MetaInfo const &info = p_fmat->Info();
  auto aif = info.GetInfo(p_fmat->Ctx(), StringView{field});

  auto &res = p_fmat->GetThreadLocal().ret_str;
  res = aif.ArrayInterfaceStr();
  *out_array = res.c_str();

  API_END();
}

namespace {
template <typename T>
int OldGetInfoImpl(const DMatrixHandle handle, const char *field, xgboost::bst_ulong *out_len,
                   const T **out_dptr, DataType dtype) {
  API_BEGIN();
  CHECK_HANDLE();
  auto p_fmat = CastDMatrixHandle(handle);
  const MetaInfo &info = p_fmat->Info();

  xgboost_CHECK_C_ARG_PTR(field);
  xgboost_CHECK_C_ARG_PTR(out_len);
  xgboost_CHECK_C_ARG_PTR(out_dptr);

  auto aif = info.GetInfo(p_fmat->Ctx(), StringView{field});
  CHECK(aif.dtype == dtype) << "Invalid dtype for the requested field: `" << field << "`";
  *out_len = aif.Size();
  *out_dptr = static_cast<T const *>(aif.data);
  API_END();
}
}  // namespace

XGB_DLL int XGDMatrixGetFloatInfo(const DMatrixHandle handle, const char *field,
                                  xgboost::bst_ulong *out_len, const float **out_dptr) {
  return OldGetInfoImpl(handle, field, out_len, out_dptr, DataType::kFloat32);
}

XGB_DLL int XGDMatrixGetUIntInfo(const DMatrixHandle handle, const char *field,
                                 xgboost::bst_ulong *out_len, const unsigned **out_dptr) {
  return OldGetInfoImpl(handle, field, out_len, out_dptr, DataType::kUInt32);
}

namespace {
// out is using xgboost::bst_ulong to make sure the defs of bst_ulong match.
template <typename Fn>
std::enable_if_t<std::is_integral_v<std::invoke_result_t<Fn, DMatrix const *>>, int>
GetDMatrixIntegralInfo(DMatrixHandle handle, xgboost::bst_ulong *out, Fn &&fn) {
  API_BEGIN();
  CHECK_HANDLE();
  auto p_m = CastDMatrixHandle(handle);
  xgboost_CHECK_C_ARG_PTR(out);
  *out = fn(p_m.get());
  API_END();
}
}  // namespace

XGB_DLL int XGDMatrixNumRow(DMatrixHandle handle, bst_ulong *out) {
  return GetDMatrixIntegralInfo(handle, out, [](DMatrix const *p_fmat) {
    return static_cast<bst_ulong>(p_fmat->Info().num_row_);
  });
}

XGB_DLL int XGDMatrixNumCol(DMatrixHandle handle, bst_ulong *out) {
  return GetDMatrixIntegralInfo(handle, out, [](DMatrix const *p_fmat) {
    return static_cast<bst_ulong>(p_fmat->Info().num_col_);
  });
}

// We name the function non-missing instead of non-zero since zero is perfectly valid for XGBoost.
XGB_DLL int XGDMatrixNumNonMissing(DMatrixHandle handle, bst_ulong *out) {
  return GetDMatrixIntegralInfo(handle, out, [](DMatrix const *p_fmat) {
    return static_cast<bst_ulong>(p_fmat->Info().num_nonzero_);
  });
}

XGB_DLL int XGDMatrixDataSplitMode(DMatrixHandle handle, bst_ulong *out) {
  return GetDMatrixIntegralInfo(handle, out, [](DMatrix const *p_fmat) {
    return static_cast<bst_ulong>(p_fmat->Info().data_split_mode);
  });
}

XGB_DLL int XGDMatrixGetDataAsCSR(DMatrixHandle const handle, char const *config,
                                  xgboost::bst_ulong *out_indptr, unsigned *out_indices,
                                  float *out_data) {
  API_BEGIN();
  CHECK_HANDLE();

  xgboost_CHECK_C_ARG_PTR(config);
  auto jconfig = Json::Load(StringView{config});

  auto p_m = CastDMatrixHandle(handle);

  xgboost_CHECK_C_ARG_PTR(out_indptr);
  xgboost_CHECK_C_ARG_PTR(out_indices);
  xgboost_CHECK_C_ARG_PTR(out_data);

  CHECK_LE(p_m->Info().num_col_, std::numeric_limits<unsigned>::max());

  for (auto const &page : p_m->GetBatches<ExtSparsePage>(p_m->Ctx(), BatchParam{})) {
    CHECK(page.page);
    auto const &h_offset = page.page->offset.ConstHostVector();
    std::copy(h_offset.cbegin(), h_offset.cend(), out_indptr);
    auto pv = page.page->GetView();
    common::ParallelFor(page.page->data.Size(), p_m->Ctx()->Threads(), [&](std::size_t i) {
      auto fvalue = pv.data[i].fvalue;
      auto findex = pv.data[i].index;
      out_data[i] = fvalue;
      out_indices[i] = findex;
    });
  }

  API_END();
}

namespace {
template <typename Page>
void GetCutImpl(Context const *ctx, std::shared_ptr<DMatrix> p_m,
                std::vector<std::uint64_t> *p_indptr, std::vector<float> *p_data) {
  auto &indptr = *p_indptr;
  auto &data = *p_data;
  for (auto const &page : p_m->GetBatches<Page>(ctx, {})) {
    auto const &cut = page.Cuts();
    auto const &ptrs = cut.Ptrs();
    auto const &vals = cut.Values();
    auto ft = p_m->Info().feature_types.ConstHostSpan();

    indptr.resize(ptrs.size());
    data.clear();

    for (bst_feature_t fidx = 0; fidx < p_m->Info().num_col_; ++fidx) {
      indptr[fidx] = data.size();

      if (!common::IsCat(ft, fidx)) {
        data.push_back(common::HistogramCuts::NumericBinLowerBound(ptrs, vals, fidx, ptrs[fidx]));
      }

      auto beg = ptrs[fidx];
      auto end = ptrs[fidx + 1];
      data.insert(data.end(), vals.cbegin() + beg, vals.cbegin() + end);
    }
    indptr.back() = data.size();
    break;
  }
}
}  // namespace

XGB_DLL int XGDMatrixGetQuantileCut(DMatrixHandle const handle, char const *config,
                                    char const **out_indptr, char const **out_data) {
  API_BEGIN();
  CHECK_HANDLE();

  auto p_m = CastDMatrixHandle(handle);

  xgboost_CHECK_C_ARG_PTR(config);
  xgboost_CHECK_C_ARG_PTR(out_indptr);
  xgboost_CHECK_C_ARG_PTR(out_data);

  auto jconfig = Json::Load(StringView{config});

  if (!p_m->PageExists<GHistIndexMatrix>() && !p_m->PageExists<EllpackPage>()) {
    LOG(FATAL) << "The quantile cut hasn't been generated yet. Unless this is a `QuantileDMatrix`, "
                  "quantile cut is generated during training.";
  }
  // Get return buffer
  auto &data = p_m->GetThreadLocal().ret_vec_float;
  auto &indptr = p_m->GetThreadLocal().ret_vec_u64;

  if (p_m->PageExists<GHistIndexMatrix>()) {
    auto ctx = p_m->Ctx()->IsCPU() ? *p_m->Ctx() : p_m->Ctx()->MakeCPU();
    GetCutImpl<GHistIndexMatrix>(&ctx, p_m, &indptr, &data);
  } else {
    auto ctx = p_m->Ctx()->IsCUDA() ? *p_m->Ctx() : p_m->Ctx()->MakeCUDA(0);
    GetCutImpl<EllpackPage>(&ctx, p_m, &indptr, &data);
  }

  // Create a CPU context
  Context ctx;
  // Get return buffer
  auto &ret_vec_str = p_m->GetThreadLocal().ret_vec_str;
  ret_vec_str.clear();

  ret_vec_str.emplace_back(linalg::ArrayInterfaceStr(
      linalg::MakeTensorView(&ctx, common::Span{indptr.data(), indptr.size()}, indptr.size())));
  ret_vec_str.emplace_back(linalg::ArrayInterfaceStr(
      linalg::MakeTensorView(&ctx, common::Span{data.data(), data.size()}, data.size())));

  auto &charp_vecs = p_m->GetThreadLocal().ret_vec_charp;
  charp_vecs.resize(ret_vec_str.size());
  std::transform(ret_vec_str.cbegin(), ret_vec_str.cend(), charp_vecs.begin(),
                 [](auto const &str) { return str.c_str(); });

  *out_indptr = charp_vecs[0];
  *out_data = charp_vecs[1];
  API_END();
}

// xgboost implementation
XGB_DLL int XGBoosterCreate(const DMatrixHandle dmats[], xgboost::bst_ulong len,
                            BoosterHandle *out) {
  API_BEGIN();
  std::vector<std::shared_ptr<DMatrix>> mats;
  for (xgboost::bst_ulong i = 0; i < len; ++i) {
    xgboost_CHECK_C_ARG_PTR(dmats);
    mats.push_back(*static_cast<std::shared_ptr<DMatrix> *>(dmats[i]));
  }
  xgboost_CHECK_C_ARG_PTR(out);
  *out = Learner::Create(mats);
  API_END();
}

XGB_DLL int XGBoosterFree(BoosterHandle handle) {
  API_BEGIN();
  CHECK_HANDLE();
  delete static_cast<Learner *>(handle);
  API_END();
}

XGB_DLL int XGBoosterReset(BoosterHandle handle) {
  API_BEGIN();
  CHECK_HANDLE();
  static_cast<Learner *>(handle)->Reset();
  API_END();
}

XGB_DLL int XGBoosterSetParam(BoosterHandle handle, const char *name, const char *value) {
  API_BEGIN();
  CHECK_HANDLE();
  static_cast<Learner *>(handle)->SetParam(name, value);
  API_END();
}

XGB_DLL int XGBoosterGetNumFeature(BoosterHandle handle, xgboost::bst_ulong *out) {
  API_BEGIN();
  CHECK_HANDLE();
  static_cast<Learner *>(handle)->Configure();
  xgboost_CHECK_C_ARG_PTR(out);
  *out = static_cast<Learner *>(handle)->GetNumFeature();
  API_END();
}

XGB_DLL int XGBoosterBoostedRounds(BoosterHandle handle, int *out) {
  API_BEGIN();
  CHECK_HANDLE();
  static_cast<Learner *>(handle)->Configure();
  xgboost_CHECK_C_ARG_PTR(out);
  *out = static_cast<Learner *>(handle)->BoostedRounds();
  API_END();
}

XGB_DLL int XGBoosterLoadJsonConfig(BoosterHandle handle, char const *json_parameters) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(json_parameters);
  Json config{Json::Load(StringView{json_parameters})};
  static_cast<Learner *>(handle)->LoadConfig(config);
  API_END();
}

XGB_DLL int XGBoosterSaveJsonConfig(BoosterHandle handle, xgboost::bst_ulong *out_len,
                                    char const **out_str) {
  API_BEGIN();
  CHECK_HANDLE();
  Json config{Object()};
  auto *learner = static_cast<Learner *>(handle);
  learner->Configure();
  learner->SaveConfig(&config);
  std::string &raw_str = learner->GetThreadLocal().ret_str;
  Json::Dump(config, &raw_str);

  xgboost_CHECK_C_ARG_PTR(out_str);
  xgboost_CHECK_C_ARG_PTR(out_len);

  *out_str = raw_str.c_str();
  *out_len = static_cast<xgboost::bst_ulong>(raw_str.length());
  API_END();
}

XGB_DLL int XGBoosterUpdateOneIter(BoosterHandle handle, int iter, DMatrixHandle dtrain) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *bst = static_cast<Learner *>(handle);
  xgboost_CHECK_C_ARG_PTR(dtrain);
  auto *dtr = static_cast<std::shared_ptr<DMatrix> *>(dtrain);
  CHECK(dtr);
  bst->UpdateOneIter(iter, *dtr);
  API_END();
}

XGB_DLL int XGBoosterBoostOneIter(BoosterHandle handle, DMatrixHandle dtrain, bst_float *grad,
                                  bst_float *hess, xgboost::bst_ulong len) {
  API_BEGIN();
  CHECK_HANDLE();
  LOG(WARNING) << error::DeprecatedFunc(__func__, "2.1.0", "XGBoosterTrainOneIter");
  auto *learner = static_cast<Learner *>(handle);
  auto ctx = learner->Ctx()->MakeCPU();

  auto t_grad = linalg::MakeTensorView(&ctx, common::Span{grad, static_cast<size_t>(len)}, len);
  auto t_hess = linalg::MakeTensorView(&ctx, common::Span{hess, static_cast<size_t>(len)}, len);

  auto s_grad = linalg::ArrayInterfaceStr(t_grad);
  auto s_hess = linalg::ArrayInterfaceStr(t_hess);

  return XGBoosterTrainOneIter(handle, dtrain, 0, s_grad.c_str(), s_hess.c_str());
  API_END();
}

namespace xgboost {
// copy user-supplied CUDA gradient arrays
void CopyGradientFromCudaArrays(Context const *, ArrayInterface<2, false> const &,
                                ArrayInterface<2, false> const &, linalg::Matrix<GradientPair> *)
#if !defined(XGBOOST_USE_CUDA)
{
  common::AssertGPUSupport();
}
#else
    ;  // NOLINT
#endif

// Helper function to copy gradient from array interface to linalg::Matrix
void CopyGradientFromArrays(Context const *ctx, ArrayInterface<2, false> const &i_grad,
                            ArrayInterface<2, false> const &i_hess,
                            linalg::Matrix<GradientPair> *out_gpair) {
  auto grad_is_cuda = ArrayInterfaceHandler::IsCudaPtr(i_grad.data);
  auto hess_is_cuda = ArrayInterfaceHandler::IsCudaPtr(i_hess.data);
  CHECK_EQ(grad_is_cuda, hess_is_cuda) << "gradient and hessian should be on the same device.";

  if (!grad_is_cuda) {
    out_gpair->Reshape(i_grad.Shape<0>(), i_grad.Shape<1>());
    auto h_gpair = out_gpair->HostView();
    DispatchDType(i_grad, DeviceOrd::CPU(), [&](auto &&t_grad) {
      DispatchDType(i_hess, DeviceOrd::CPU(), [&](auto &&t_hess) {
        common::ParallelFor(h_gpair.Size(), ctx->Threads(),
                            detail::CustomGradHessOp{t_grad, t_hess, h_gpair});
      });
    });
  } else {
    CopyGradientFromCudaArrays(ctx, i_grad, i_hess, out_gpair);
  }
}
}  // namespace xgboost

XGB_DLL int XGBoosterTrainOneIter(BoosterHandle handle, DMatrixHandle dtrain, int iter,
                                  char const *grad, char const *hess) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(grad);
  xgboost_CHECK_C_ARG_PTR(hess);
  auto p_fmat = CastDMatrixHandle(dtrain);
  ArrayInterface<2, false> i_grad{StringView{grad}};
  ArrayInterface<2, false> i_hess{StringView{hess}};
  StringView msg{"Mismatched shape between the gradient and hessian."};
  CHECK_EQ(i_grad.Shape<0>(), i_hess.Shape<0>()) << msg;
  CHECK_EQ(i_grad.Shape<1>(), i_hess.Shape<1>()) << msg;
  CHECK_EQ(i_grad.Shape<0>(), p_fmat->Info().num_row_)
      << "Mismatched size between the gradient and training data.";
  auto *learner = static_cast<Learner *>(handle);
  auto ctx = learner->Ctx();
  GradientContainer gpair;
  CopyGradientFromArrays(ctx, i_grad, i_hess, &gpair.gpair);
  learner->BoostOneIter(iter, p_fmat, &gpair);
  API_END();
}

typedef char const *JArrayStr;  // NOLINT

// Hidden, working-in-progress support for reduced gradient.
/**
 * @brief Use a different type of gradient for tree split.
 *
 * @param split_grad Gradient for finding tree splits.
 * @param split_hess Hessian for finding tree splits.
 * @param value_grad Gradient for calculating tree leaf weights.
 * @param value_hess Hessian for calculating tree leaf weights.
 */
XGB_DLL int XGBoosterTrainOneIterWithSplitGrad(BoosterHandle handle, DMatrixHandle dtrain, int iter,
                                               JArrayStr split_grad, JArrayStr split_hess,
                                               JArrayStr value_grad, JArrayStr value_hess) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *learner = static_cast<Learner *>(handle);
  GradientContainer gpair;
  auto ctx = learner->Ctx();
  {
    ArrayInterface<2, false> i_grad{StringView{split_grad}};
    ArrayInterface<2, false> i_hess{StringView{split_hess}};
    StringView msg{"Mismatched shape between the gradient and hessian."};
    CHECK_EQ(i_grad.Shape<0>(), i_hess.Shape<0>()) << msg;
    CHECK_EQ(i_grad.Shape<1>(), i_hess.Shape<1>()) << msg;
    CopyGradientFromArrays(ctx, i_grad, i_hess, &gpair.gpair);
  }
  {
    ArrayInterface<2, false> i_grad{StringView{value_grad}};
    ArrayInterface<2, false> i_hess{StringView{value_hess}};
    StringView msg{"Mismatched shape between the gradient and hessian."};
    CHECK_EQ(i_grad.Shape<0>(), i_hess.Shape<0>()) << msg;
    CHECK_EQ(i_grad.Shape<1>(), i_hess.Shape<1>()) << msg;
    CopyGradientFromArrays(ctx, i_grad, i_hess, &gpair.value_gpair);
  }

  auto p_fmat = CastDMatrixHandle(dtrain);
  learner->BoostOneIter(iter, p_fmat, &gpair);

  API_END();
}

XGB_DLL int XGBoosterEvalOneIter(BoosterHandle handle, int iter, DMatrixHandle dmats[],
                                 const char *evnames[], xgboost::bst_ulong len,
                                 const char **out_str) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *bst = static_cast<Learner *>(handle);
  std::string &eval_str = bst->GetThreadLocal().ret_str;

  std::vector<std::shared_ptr<DMatrix>> data_sets;
  std::vector<std::string> data_names;

  for (xgboost::bst_ulong i = 0; i < len; ++i) {
    xgboost_CHECK_C_ARG_PTR(dmats);
    data_sets.push_back(*static_cast<std::shared_ptr<DMatrix> *>(dmats[i]));
    xgboost_CHECK_C_ARG_PTR(evnames);
    data_names.emplace_back(evnames[i]);
  }

  eval_str = bst->EvalOneIter(iter, data_sets, data_names);
  xgboost_CHECK_C_ARG_PTR(out_str);
  *out_str = eval_str.c_str();
  API_END();
}

XGB_DLL int XGBoosterPredict(BoosterHandle handle, DMatrixHandle dmat, int option_mask,
                             unsigned ntree_limit, int training, xgboost::bst_ulong *len,
                             const bst_float **out_result) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *learner = static_cast<Learner *>(handle);
  auto &entry = learner->GetThreadLocal().prediction_entry;
  auto iteration_end = GetIterationFromTreeLimit(ntree_limit, learner);
  learner->Predict(*static_cast<std::shared_ptr<DMatrix> *>(dmat), (option_mask & 1) != 0,
                   &entry.predictions, 0, iteration_end, static_cast<bool>(training),
                   (option_mask & 2) != 0, (option_mask & 4) != 0, (option_mask & 8) != 0,
                   (option_mask & 16) != 0);

  xgboost_CHECK_C_ARG_PTR(len);
  xgboost_CHECK_C_ARG_PTR(out_result);

  *out_result = dmlc::BeginPtr(entry.predictions.ConstHostVector());
  *len = static_cast<xgboost::bst_ulong>(entry.predictions.Size());
  API_END();
}

XGB_DLL int XGBoosterPredictFromDMatrix(BoosterHandle handle, DMatrixHandle dmat,
                                        char const *c_json_config,
                                        xgboost::bst_ulong const **out_shape,
                                        xgboost::bst_ulong *out_dim, bst_float const **out_result) {
  API_BEGIN();
  if (handle == nullptr) {
    LOG(FATAL) << "Booster has not been initialized or has already been disposed.";
  }
  if (dmat == nullptr) {
    LOG(FATAL) << "DMatrix has not been initialized or has already been disposed.";
  }
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  auto config = Json::Load(StringView{c_json_config});

  auto *learner = static_cast<Learner *>(handle);
  auto &entry = learner->GetThreadLocal().prediction_entry;
  auto p_m = *static_cast<std::shared_ptr<DMatrix> *>(dmat);

  auto type = PredictionType(RequiredArg<Integer>(config, "type", __func__));
  auto iteration_begin = RequiredArg<Integer>(config, "iteration_begin", __func__);
  auto iteration_end = RequiredArg<Integer>(config, "iteration_end", __func__);

  auto const &j_config = get<Object const>(config);
  auto ntree_limit_it = j_config.find("ntree_limit");
  if (ntree_limit_it != j_config.cend() && !IsA<Null>(ntree_limit_it->second) &&
      get<Integer const>(ntree_limit_it->second) != 0) {
    CHECK(iteration_end == 0)
        << "Only one of the `ntree_limit` or `iteration_range` can be specified.";
    LOG(WARNING) << "`ntree_limit` is deprecated, use `iteration_range` instead.";
    iteration_end = GetIterationFromTreeLimit(get<Integer const>(ntree_limit_it->second), learner);
  }

  bool approximate =
      type == PredictionType::kApproxContribution || type == PredictionType::kApproxInteraction;
  bool contribs =
      type == PredictionType::kContribution || type == PredictionType::kApproxContribution;
  bool interactions =
      type == PredictionType::kInteraction || type == PredictionType::kApproxInteraction;
  bool training = RequiredArg<Boolean>(config, "training", __func__);
  learner->Predict(p_m, type == PredictionType::kMargin, &entry.predictions, iteration_begin,
                   iteration_end, training, type == PredictionType::kLeaf, contribs, approximate,
                   interactions);

  xgboost_CHECK_C_ARG_PTR(out_result);
  *out_result = dmlc::BeginPtr(entry.predictions.ConstHostVector());

  auto &shape = learner->GetThreadLocal().prediction_shape;
  auto chunksize = p_m->Info().num_row_ == 0 ? 0 : entry.predictions.Size() / p_m->Info().num_row_;
  auto rounds = iteration_end - iteration_begin;
  rounds = rounds == 0 ? learner->BoostedRounds() : rounds;
  // Determine shape
  bool strict_shape = RequiredArg<Boolean>(config, "strict_shape", __func__);

  xgboost_CHECK_C_ARG_PTR(out_dim);
  xgboost_CHECK_C_ARG_PTR(out_shape);

  CalcPredictShape(strict_shape, type, p_m->Info().num_row_, p_m->Info().num_col_, chunksize,
                   learner->Groups(), rounds, &shape, out_dim);
  *out_shape = dmlc::BeginPtr(shape);
  API_END();
}

void InplacePredictImpl(std::shared_ptr<DMatrix> p_m, char const *c_json_config, Learner *learner,
                        xgboost::bst_ulong const **out_shape, xgboost::bst_ulong *out_dim,
                        const float **out_result) {
  xgboost_CHECK_C_ARG_PTR(c_json_config);
  auto config = Json::Load(StringView{c_json_config});

  HostDeviceVector<float> *p_predt{nullptr};
  auto type = PredictionType(RequiredArg<Integer>(config, "type", __func__));
  float missing = GetMissing(config);
  learner->InplacePredict(p_m, type, missing, &p_predt,
                          RequiredArg<Integer>(config, "iteration_begin", __func__),
                          RequiredArg<Integer>(config, "iteration_end", __func__));
  CHECK(p_predt);
  auto &shape = learner->GetThreadLocal().prediction_shape;
  auto const &info = p_m->Info();
  auto n_samples = info.num_row_;
  auto n_features = info.num_col_;
  auto chunksize = n_samples == 0 ? 0 : p_predt->Size() / n_samples;
  bool strict_shape = RequiredArg<Boolean>(config, "strict_shape", __func__);

  xgboost_CHECK_C_ARG_PTR(out_dim);
  CalcPredictShape(strict_shape, type, n_samples, n_features, chunksize, learner->Groups(),
                   learner->BoostedRounds(), &shape, out_dim);
  CHECK_GE(p_predt->Size(), n_samples);

  xgboost_CHECK_C_ARG_PTR(out_result);
  xgboost_CHECK_C_ARG_PTR(out_shape);

  *out_result = dmlc::BeginPtr(p_predt->HostVector());
  *out_shape = dmlc::BeginPtr(shape);
}

XGB_DLL int XGBoosterPredictFromDense(BoosterHandle handle, char const *data,
                                      char const *c_json_config, DMatrixHandle m,
                                      xgboost::bst_ulong const **out_shape,
                                      xgboost::bst_ulong *out_dim, const float **out_result) {
  API_BEGIN();
  CHECK_HANDLE();
  std::shared_ptr<DMatrix> p_m{nullptr};
  if (!m) {
    p_m.reset(new data::DMatrixProxy);
  } else {
    p_m = *static_cast<std::shared_ptr<DMatrix> *>(m);
  }
  auto proxy = dynamic_cast<data::DMatrixProxy *>(p_m.get());
  CHECK(proxy) << "Invalid input type for inplace predict.";
  xgboost_CHECK_C_ARG_PTR(data);
  proxy->SetArray(data);
  auto *learner = static_cast<xgboost::Learner *>(handle);
  InplacePredictImpl(p_m, c_json_config, learner, out_shape, out_dim, out_result);
  API_END();
}

XGB_DLL int XGBoosterPredictFromColumnar(BoosterHandle handle, char const *array_interface,
                                         char const *c_json_config, DMatrixHandle m,
                                         xgboost::bst_ulong const **out_shape,
                                         xgboost::bst_ulong *out_dim, const float **out_result) {
  API_BEGIN();
  CHECK_HANDLE();
  std::shared_ptr<DMatrix> p_m{nullptr};
  if (!m) {
    p_m.reset(new data::DMatrixProxy);
  } else {
    p_m = *static_cast<std::shared_ptr<DMatrix> *>(m);
  }
  auto proxy = dynamic_cast<data::DMatrixProxy *>(p_m.get());
  CHECK(proxy) << "Invalid input type for inplace predict.";
  xgboost_CHECK_C_ARG_PTR(array_interface);
  proxy->SetColumnar(array_interface);
  auto *learner = static_cast<xgboost::Learner *>(handle);
  InplacePredictImpl(p_m, c_json_config, learner, out_shape, out_dim, out_result);
  API_END();
}

XGB_DLL int XGBoosterPredictFromCSR(BoosterHandle handle, char const *indptr, char const *indices,
                                    char const *data, xgboost::bst_ulong cols,
                                    char const *c_json_config, DMatrixHandle m,
                                    xgboost::bst_ulong const **out_shape,
                                    xgboost::bst_ulong *out_dim, const float **out_result) {
  API_BEGIN();
  CHECK_HANDLE();
  std::shared_ptr<DMatrix> p_m{nullptr};
  if (!m) {
    p_m.reset(new data::DMatrixProxy);
  } else {
    p_m = *static_cast<std::shared_ptr<DMatrix> *>(m);
  }
  auto proxy = dynamic_cast<data::DMatrixProxy *>(p_m.get());
  CHECK(proxy) << "Invalid input type for inplace predict.";
  xgboost_CHECK_C_ARG_PTR(indptr);
  proxy->SetCsr(indptr, indices, data, cols, true);
  auto *learner = static_cast<xgboost::Learner *>(handle);
  InplacePredictImpl(p_m, c_json_config, learner, out_shape, out_dim, out_result);
  API_END();
}

#if !defined(XGBOOST_USE_CUDA)
XGB_DLL int XGBoosterPredictFromCudaArray(BoosterHandle handle, char const *, char const *,
                                          DMatrixHandle, xgboost::bst_ulong const **,
                                          xgboost::bst_ulong *, const float **) {
  API_BEGIN();
  CHECK_HANDLE();
  common::AssertGPUSupport();
  API_END();
}

XGB_DLL int XGBoosterPredictFromCudaColumnar(BoosterHandle handle, char const *, char const *,
                                             DMatrixHandle, xgboost::bst_ulong const **,
                                             xgboost::bst_ulong *, const float **) {
  API_BEGIN();
  CHECK_HANDLE();
  common::AssertGPUSupport();
  API_END();
}
#endif  // !defined(XGBOOST_USE_CUDA)

namespace {
template <typename Buffer, typename Iter = typename Buffer::const_iterator>
Json DispatchModelType(Buffer const &buffer, StringView ext, bool warn) {
  auto first_non_space = [&](Iter beg, Iter end) {
    for (auto i = beg; i != end; ++i) {
      if (!std::isspace(*i)) {
        return i;
      }
    }
    return end;
  };

  Json model;
  auto it = first_non_space(buffer.cbegin() + 1, buffer.cend());
  if (it != buffer.cend() && *it == '"') {
    if (warn) {
      LOG(WARNING) << "Unknown file format: `" << ext << "`. Using JSON (`json`) as a guess.";
    }
    model = Json::Load(StringView{buffer.data(), buffer.size()});
  } else if (it != buffer.cend() && std::isalpha(*it)) {
    if (warn) {
      LOG(WARNING) << "Unknown file format: `" << ext << "`. Using UBJSON (`ubj`) as a guess.";
    }
    model = Json::Load(StringView{buffer.data(), buffer.size()}, std::ios::binary);
  } else {
    LOG(FATAL) << "Invalid model format. Expecting UBJSON (`ubj`) or JSON (`json`), got `" << ext
               << "`";
  }
  return model;
}
}  // namespace

XGB_DLL int XGBoosterLoadModel(BoosterHandle handle, const char *fname) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(fname);
  auto read_file = [&]() {
    auto str = common::LoadSequentialFile(fname);
    // "{}"
    CHECK_GE(str.size(), 2) << error::InvalidModel(fname);
    // The old binary format has the starting bytes "binf".
    if (str.size() >= 4 && StringView{str.data(), 4} == "binf") {  // NOLINT
      LOG(FATAL) << error::OldBinaryModel(fname);
    }
    CHECK_EQ(str[0], '{') << error::InvalidModel(fname);
    return str;
  };
  auto ext = common::FileExtension(fname);
  if (ext == "json") {
    auto buffer = read_file();
    Json in{Json::Load(StringView{buffer.data(), buffer.size()})};
    static_cast<Learner *>(handle)->LoadModel(in);
  } else if (ext == "ubj") {
    auto buffer = read_file();
    Json in = Json::Load(StringView{buffer.data(), buffer.size()}, std::ios::binary);
    static_cast<Learner *>(handle)->LoadModel(in);
  } else {
    auto buffer = read_file();
    auto in = DispatchModelType(buffer, ext, true);
    static_cast<Learner *>(handle)->LoadModel(in);
  }
  API_END();
}

XGB_DLL int XGBoosterSaveModel(BoosterHandle handle, const char *fname) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(fname);

  std::unique_ptr<dmlc::Stream> fo(dmlc::Stream::Create(fname, "w"));
  auto *learner = static_cast<Learner *>(handle);
  learner->Configure();
  auto save_json = [&](std::ios::openmode mode) {
    Json out{Object()};
    learner->SaveModel(&out);
    std::vector<char> str;
    Json::Dump(out, &str, mode);
    fo->Write(str.data(), str.size());
  };
  auto ext = common::FileExtension(fname);
  if (ext == "json") {
    save_json(std::ios::out);
  } else if (ext == "ubj") {
    save_json(std::ios::binary);
  } else {
    LOG(WARNING) << "Saving model in the UBJSON format as default.  You can use a file extension:"
                    " `json` or `ubj` to choose between formats.";
    save_json(std::ios::binary);
  }
  API_END();
}

XGB_DLL int XGBoosterLoadModelFromBuffer(BoosterHandle handle, const void *buf,
                                         xgboost::bst_ulong len) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(buf);
  using CharT = std::add_const_t<char>;
  using IdxType = common::Span<CharT>::index_type;
  auto buffer = common::Span{static_cast<CharT *>(buf), static_cast<IdxType>(len)};
  // Don't warn, we have to guess the format with buffer input.
  auto in = DispatchModelType(buffer, "", false);
  common::MemoryFixSizeBuffer fs((void *)buf, len);  // NOLINT(*)
  static_cast<Learner *>(handle)->LoadModel(in);
  API_END();
}

XGB_DLL int XGBoosterSaveModelToBuffer(BoosterHandle handle, char const *json_config,
                                       xgboost::bst_ulong *out_len, char const **out_dptr) {
  API_BEGIN();
  CHECK_HANDLE();

  xgboost_CHECK_C_ARG_PTR(json_config);
  xgboost_CHECK_C_ARG_PTR(out_dptr);
  xgboost_CHECK_C_ARG_PTR(out_len);

  auto config = Json::Load(StringView{json_config});
  auto format = RequiredArg<String>(config, "format", __func__);

  auto *learner = static_cast<Learner *>(handle);
  learner->Configure();

  auto save_json = [&](std::ios::openmode mode) {
    std::vector<char> &raw_char_vec = learner->GetThreadLocal().ret_char_vec;
    Json out{Object{}};
    learner->SaveModel(&out);
    Json::Dump(out, &raw_char_vec, mode);
    *out_dptr = dmlc::BeginPtr(raw_char_vec);
    *out_len = static_cast<xgboost::bst_ulong>(raw_char_vec.size());
  };

  Json out{Object{}};
  if (format == "json") {
    save_json(std::ios::out);
  } else if (format == "ubj") {
    save_json(std::ios::binary);
  } else {
    LOG(FATAL) << "Unknown model format: `" << format
               << "`. Expecting UBJSON (`ubj`) or JSON (`json`).";
  }

  API_END();
}

// The following two functions are `Load` and `Save` for memory based
// serialization methods. E.g. Python pickle.
XGB_DLL int XGBoosterSerializeToBuffer(BoosterHandle handle, xgboost::bst_ulong *out_len,
                                       const char **out_dptr) {
  API_BEGIN();
  CHECK_HANDLE();

  auto *learner = static_cast<Learner *>(handle);
  std::string &raw_str = learner->GetThreadLocal().ret_str;
  raw_str.resize(0);
  common::MemoryBufferStream fo(&raw_str);
  learner->Configure();
  learner->Save(&fo);

  xgboost_CHECK_C_ARG_PTR(out_dptr);
  xgboost_CHECK_C_ARG_PTR(out_len);

  *out_dptr = dmlc::BeginPtr(raw_str);
  *out_len = static_cast<xgboost::bst_ulong>(raw_str.length());
  API_END();
}

XGB_DLL int XGBoosterUnserializeFromBuffer(BoosterHandle handle, const void *buf,
                                           xgboost::bst_ulong len) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(buf);

  common::MemoryFixSizeBuffer fs((void *)buf, len);  // NOLINT(*)
  static_cast<Learner *>(handle)->Load(&fs);
  API_END();
}

XGB_DLL int XGBoosterSlice(BoosterHandle handle, int begin_layer, int end_layer, int step,
                           BoosterHandle *out) {
  API_BEGIN();
  CHECK_HANDLE();
  xgboost_CHECK_C_ARG_PTR(out);

  auto *learner = static_cast<Learner *>(handle);
  bool out_of_bound = false;
  auto p_out = learner->Slice(begin_layer, end_layer, step, &out_of_bound);
  if (out_of_bound) {
    return -2;
  }
  CHECK(p_out);
  *out = p_out;
  API_END();
}

inline void XGBoostDumpModelImpl(BoosterHandle handle, FeatureMap *fmap, int with_stats,
                                 const char *format, xgboost::bst_ulong *len,
                                 const char ***out_models) {
  auto *bst = static_cast<Learner *>(handle);
  bst->Configure();
  GenerateFeatureMap(bst, {}, bst->GetNumFeature(), fmap);

  std::vector<std::string> &str_vecs = bst->GetThreadLocal().ret_vec_str;
  std::vector<const char *> &charp_vecs = bst->GetThreadLocal().ret_vec_charp;
  str_vecs = bst->DumpModel(*fmap, with_stats != 0, format);
  charp_vecs.resize(str_vecs.size());
  for (size_t i = 0; i < str_vecs.size(); ++i) {
    charp_vecs[i] = str_vecs[i].c_str();
  }

  xgboost_CHECK_C_ARG_PTR(out_models);
  xgboost_CHECK_C_ARG_PTR(len);

  *out_models = dmlc::BeginPtr(charp_vecs);
  *len = static_cast<xgboost::bst_ulong>(charp_vecs.size());
}

XGB_DLL int XGBoosterDumpModel(BoosterHandle handle, const char *fmap, int with_stats,
                               xgboost::bst_ulong *len, const char ***out_models) {
  API_BEGIN();
  CHECK_HANDLE();
  return XGBoosterDumpModelEx(handle, fmap, with_stats, "text", len, out_models);
  API_END();
}

XGB_DLL int XGBoosterDumpModelEx(BoosterHandle handle, const char *fmap, int with_stats,
                                 const char *format, xgboost::bst_ulong *len,
                                 const char ***out_models) {
  API_BEGIN();
  CHECK_HANDLE();

  xgboost_CHECK_C_ARG_PTR(fmap);
  std::string uri{fmap};
  FeatureMap featmap = LoadFeatureMap(uri);
  XGBoostDumpModelImpl(handle, &featmap, with_stats, format, len, out_models);
  API_END();
}

XGB_DLL int XGBoosterDumpModelWithFeatures(BoosterHandle handle, int fnum, const char **fname,
                                           const char **ftype, int with_stats,
                                           xgboost::bst_ulong *len, const char ***out_models) {
  return XGBoosterDumpModelExWithFeatures(handle, fnum, fname, ftype, with_stats, "text", len,
                                          out_models);
}

XGB_DLL int XGBoosterDumpModelExWithFeatures(BoosterHandle handle, int fnum, const char **fname,
                                             const char **ftype, int with_stats, const char *format,
                                             xgboost::bst_ulong *len, const char ***out_models) {
  API_BEGIN();
  CHECK_HANDLE();
  FeatureMap featmap;
  for (int i = 0; i < fnum; ++i) {
    xgboost_CHECK_C_ARG_PTR(fname);
    xgboost_CHECK_C_ARG_PTR(ftype);
    featmap.PushBack(i, fname[i], ftype[i]);
  }
  XGBoostDumpModelImpl(handle, &featmap, with_stats, format, len, out_models);
  API_END();
}

XGB_DLL int XGBoosterGetCategories(BoosterHandle handle, char const * /*config*/,
                                   CategoriesHandle *out) {
  API_BEGIN()
  CHECK_HANDLE()

  auto *bst = static_cast<Learner *>(handle);
  auto const cats = bst->Cats();
  xgboost_CHECK_C_ARG_PTR(out);
  if (cats->Empty()) {
    out = nullptr;
  } else {
    auto new_cats = CopyCatContainer(bst->Ctx(), cats, bst->GetNumFeature());
    *out = new_cats;
  }

  API_END()
}

XGB_DLL int XGBoosterGetCategoriesExportToArrow(BoosterHandle handle, char const * /*config*/,
                                                CategoriesHandle *out, char const **export_out) {
  API_BEGIN()
  CHECK_HANDLE()

  auto *bst = static_cast<Learner *>(handle);
  auto const cats = bst->Cats();
  auto n_features = bst->GetNumFeature();

  xgboost_CHECK_C_ARG_PTR(out);
  xgboost_CHECK_C_ARG_PTR(export_out);

  if (cats->Empty()) {
    *out = nullptr;
    *export_out = nullptr;
  } else {
    // Create a new container
    auto new_cats = CopyCatContainer(bst->Ctx(), cats, n_features);
    *out = new_cats;
    // Export to arrow
    auto &ret_str = bst->GetThreadLocal().ret_str;
    GetCategoriesImpl(new_cats->HostView(), n_features, &ret_str, export_out);
  }

  API_END()
}

XGB_DLL int XGBoosterGetAttr(BoosterHandle handle, const char *key, const char **out,
                             int *success) {
  auto *bst = static_cast<Learner *>(handle);
  std::string &ret_str = bst->GetThreadLocal().ret_str;
  API_BEGIN();
  CHECK_HANDLE();

  xgboost_CHECK_C_ARG_PTR(out);
  xgboost_CHECK_C_ARG_PTR(success);

  if (bst->GetAttr(key, &ret_str)) {
    *out = ret_str.c_str();
    *success = 1;
  } else {
    *out = nullptr;
    *success = 0;
  }
  API_END();
}

XGB_DLL int XGBoosterSetAttr(BoosterHandle handle, const char *key, const char *value) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *bst = static_cast<Learner *>(handle);
  xgboost_CHECK_C_ARG_PTR(key);
  if (value == nullptr) {
    bst->DelAttr(key);
  } else {
    xgboost_CHECK_C_ARG_PTR(value);
    bst->SetAttr(key, value);
  }
  API_END();
}

XGB_DLL int XGBoosterGetAttrNames(BoosterHandle handle, xgboost::bst_ulong *out_len,
                                  const char ***out) {
  API_BEGIN();
  CHECK_HANDLE();

  auto *learner = static_cast<Learner *>(handle);
  std::vector<std::string> &str_vecs = learner->GetThreadLocal().ret_vec_str;
  std::vector<const char *> &charp_vecs = learner->GetThreadLocal().ret_vec_charp;
  str_vecs = learner->GetAttrNames();
  charp_vecs.resize(str_vecs.size());
  for (size_t i = 0; i < str_vecs.size(); ++i) {
    charp_vecs[i] = str_vecs[i].c_str();
  }

  xgboost_CHECK_C_ARG_PTR(out);
  xgboost_CHECK_C_ARG_PTR(out_len);

  *out = dmlc::BeginPtr(charp_vecs);
  *out_len = static_cast<xgboost::bst_ulong>(charp_vecs.size());
  API_END();
}

XGB_DLL int XGBoosterSetStrFeatureInfo(BoosterHandle handle, const char *field,
                                       const char **features, const xgboost::bst_ulong size) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *learner = static_cast<Learner *>(handle);
  std::vector<std::string> feature_info;
  if (size > 0) {
    xgboost_CHECK_C_ARG_PTR(features);
  }
  for (size_t i = 0; i < size; ++i) {
    feature_info.emplace_back(features[i]);
  }

  xgboost_CHECK_C_ARG_PTR(field);
  if (!std::strcmp(field, "feature_name")) {
    learner->SetFeatureNames(feature_info);
  } else if (!std::strcmp(field, "feature_type")) {
    learner->SetFeatureTypes(feature_info);
  } else {
    LOG(FATAL) << "Unknown field for Booster feature info:" << field;
  }
  API_END();
}

XGB_DLL int XGBoosterGetStrFeatureInfo(BoosterHandle handle, const char *field,
                                       xgboost::bst_ulong *len, const char ***out_features) {
  API_BEGIN();
  CHECK_HANDLE();
  auto const *learner = static_cast<Learner const *>(handle);
  std::vector<const char *> &charp_vecs = learner->GetThreadLocal().ret_vec_charp;
  std::vector<std::string> &str_vecs = learner->GetThreadLocal().ret_vec_str;
  if (!std::strcmp(field, "feature_name")) {
    learner->GetFeatureNames(&str_vecs);
  } else if (!std::strcmp(field, "feature_type")) {
    learner->GetFeatureTypes(&str_vecs);
  } else {
    LOG(FATAL) << "Unknown field for Booster feature info:" << field;
  }
  charp_vecs.resize(str_vecs.size());
  for (size_t i = 0; i < str_vecs.size(); ++i) {
    charp_vecs[i] = str_vecs[i].c_str();
  }

  xgboost_CHECK_C_ARG_PTR(out_features);
  xgboost_CHECK_C_ARG_PTR(len);

  *out_features = dmlc::BeginPtr(charp_vecs);
  *len = static_cast<xgboost::bst_ulong>(charp_vecs.size());
  API_END();
}

XGB_DLL int XGBoosterFeatureScore(BoosterHandle handle, char const *config,
                                  xgboost::bst_ulong *out_n_features, char const ***out_features,
                                  bst_ulong *out_dim, bst_ulong const **out_shape,
                                  float const **out_scores) {
  API_BEGIN();
  CHECK_HANDLE();
  auto *learner = static_cast<Learner *>(handle);
  xgboost_CHECK_C_ARG_PTR(config);
  auto jconfig = Json::Load(StringView{config});

  auto importance = RequiredArg<String>(jconfig, "importance_type", __func__);
  std::string feature_map_uri;
  if (!IsA<Null>(jconfig["feature_map"])) {
    feature_map_uri = get<String const>(jconfig["feature_map"]);
  }
  FeatureMap feature_map = LoadFeatureMap(feature_map_uri);
  std::vector<Json> custom_feature_names;
  if (!IsA<Null>(jconfig["feature_names"])) {
    custom_feature_names = get<Array const>(jconfig["feature_names"]);
  }

  std::vector<int32_t> tree_idx;
  if (!IsA<Null>(jconfig["tree_idx"])) {
    auto j_tree_idx = get<Array const>(jconfig["tree_idx"]);
    for (auto const &idx : j_tree_idx) {
      tree_idx.push_back(get<Integer const>(idx));
    }
  }

  auto &scores = learner->GetThreadLocal().ret_vec_float;
  std::vector<bst_feature_t> features;
  learner->CalcFeatureScore(importance, common::Span<int32_t const>(tree_idx), &features, &scores);

  auto n_features = learner->GetNumFeature();
  GenerateFeatureMap(learner, custom_feature_names, n_features, &feature_map);

  auto &feature_names = learner->GetThreadLocal().ret_vec_str;
  feature_names.resize(features.size());
  auto &feature_names_c = learner->GetThreadLocal().ret_vec_charp;
  feature_names_c.resize(features.size());

  for (bst_feature_t i = 0; i < features.size(); ++i) {
    feature_names[i] = feature_map.Name(features[i]);
    feature_names_c[i] = feature_names[i].data();
  }
  xgboost_CHECK_C_ARG_PTR(out_n_features);
  *out_n_features = feature_names.size();

  CHECK_LE(features.size(), scores.size());
  auto &shape = learner->GetThreadLocal().prediction_shape;

  xgboost_CHECK_C_ARG_PTR(out_dim);
  if (scores.size() > features.size()) {
    // Linear model multi-class model
    CHECK_EQ(scores.size() % features.size(), 0ul);
    auto n_classes = scores.size() / features.size();
    *out_dim = 2;
    shape = {n_features, n_classes};
  } else {
    CHECK_EQ(features.size(), scores.size());
    *out_dim = 1;
    shape.resize(1);
    shape.front() = scores.size();
  }

  xgboost_CHECK_C_ARG_PTR(out_shape);
  xgboost_CHECK_C_ARG_PTR(out_scores);
  xgboost_CHECK_C_ARG_PTR(out_features);

  *out_shape = dmlc::BeginPtr(shape);
  *out_scores = scores.data();
  *out_features = dmlc::BeginPtr(feature_names_c);
  API_END();
}