GH-48588 [C++] Migrate to stdlib span (#49492)

### Rationale for this change
C++ 20 already includes a span implementation so there is no reason to maintain a custom one in utils as described in #48588

### What changes are included in this PR?
arrow::utils::span definition and tests are removed and usages are replaced with std::span. Some span comparisons in tests are replaced from ASSERT_EQ to ASSERT_TRUE(std::ranges::equal(span1, span2)) because ASSERT_EQ inernally relies on == operator which is not defined for some types that we used to compare.

### Are these changes tested?
Yes

### Are there any user-facing changes?
No
* GitHub Issue: #48588

Lead-authored-by: Paweł Biegun <biegunpawel900@gmail.com>
Co-authored-by: Will Ayd <william.ayd@icloud.com>
Signed-off-by: Antoine Pitrou <antoine@python.org>

Author: Anakin100100

cpp/src/arrow/array/concatenate.cc

@@ -22,6 +22,7 @@
 #include <cstdint>
 #include <limits>
 #include <memory>
+#include <span>
 #include <utility>
 #include <vector>
 
@@ -424,7 +425,7 @@ class ConcatenateImpl {
     out_->buffers.resize(2);
 
     for (const auto& in_data : in_) {
-      for (const auto& buf : util::span(in_data->buffers).subspan(2)) {
+      for (const auto& buf : std::span(in_data->buffers).subspan(2)) {
         out_->buffers.push_back(buf);
       }
     }

cpp/src/arrow/array/data.cc

@@ -21,6 +21,7 @@
 #include <cstddef>
 #include <cstdint>
 #include <memory>
+#include <span>
 #include <string>
 #include <utility>
 #include <vector>
@@ -102,7 +103,7 @@ bool DictionaryMayHaveLogicalNulls(const ArrayData& data) {
 
 namespace {
 
-BufferSpan PackVariadicBuffers(util::span<const std::shared_ptr<Buffer>> buffers) {
+BufferSpan PackVariadicBuffers(std::span<const std::shared_ptr<Buffer>> buffers) {
   return {const_cast<uint8_t*>(reinterpret_cast<const uint8_t*>(buffers.data())),
           static_cast<int64_t>(buffers.size() * sizeof(std::shared_ptr<Buffer>))};
 }
@@ -322,7 +323,7 @@ void ArraySpan::SetMembers(const ArrayData& data) {
 
   if (type_id == Type::STRING_VIEW || type_id == Type::BINARY_VIEW) {
     // store the span of data buffers in the third buffer
-    this->buffers[2] = internal::PackVariadicBuffers(util::span(data.buffers).subspan(2));
+    this->buffers[2] = internal::PackVariadicBuffers(std::span(data.buffers).subspan(2));
   }
 
   if (type_id == Type::DICTIONARY) {
@@ -679,7 +680,7 @@ std::shared_ptr<ArrayData> ArraySpan::ToArrayData() const {
   return result;
 }
 
-util::span<const std::shared_ptr<Buffer>> ArraySpan::GetVariadicBuffers() const {
+std::span<const std::shared_ptr<Buffer>> ArraySpan::GetVariadicBuffers() const {
   DCHECK(HasVariadicBuffers());
   return {buffers[2].data_as<std::shared_ptr<Buffer>>(),
           static_cast<size_t>(buffers[2].size) / sizeof(std::shared_ptr<Buffer>)};

cpp/src/arrow/array/data.h

@@ -21,6 +21,7 @@
 #include <cassert>
 #include <cstdint>
 #include <memory>
+#include <span>
 #include <utility>
 #include <vector>
 
@@ -31,7 +32,6 @@
 #include "arrow/type_fwd.h"
 #include "arrow/util/bit_util.h"
 #include "arrow/util/macros.h"
-#include "arrow/util/span.h"
 #include "arrow/util/visibility.h"
 
 namespace arrow {
@@ -577,11 +577,11 @@ struct ARROW_EXPORT ArraySpan {
   /// this array type
   /// \return A span<const T> of the requested length
   template <typename T>
-  util::span<const T> GetSpan(int i, int64_t length) const {
+  std::span<const T> GetSpan(int i, int64_t length) const {
     const int64_t buffer_length = buffers[i].size / static_cast<int64_t>(sizeof(T));
     assert(i > 0 && length + offset <= buffer_length);
     ARROW_UNUSED(buffer_length);
-    return util::span<const T>(buffers[i].data_as<T>() + this->offset, length);
+    return std::span<const T>(buffers[i].data_as<T>() + this->offset, length);
   }
 
   /// \brief Access a buffer's data as a span
@@ -593,11 +593,11 @@ struct ARROW_EXPORT ArraySpan {
   /// this array type
   /// \return A span<T> of the requested length
   template <typename T>
-  util::span<T> GetSpan(int i, int64_t length) {
+  std::span<T> GetSpan(int i, int64_t length) {
     const int64_t buffer_length = buffers[i].size / static_cast<int64_t>(sizeof(T));
     assert(i > 0 && length + offset <= buffer_length);
     ARROW_UNUSED(buffer_length);
-    return util::span<T>(buffers[i].mutable_data_as<T>() + this->offset, length);
+    return std::span<T>(buffers[i].mutable_data_as<T>() + this->offset, length);
   }
 
   inline bool IsNull(int64_t i) const { return !IsValid(i); }
@@ -709,7 +709,7 @@ struct ARROW_EXPORT ArraySpan {
   /// sizeof(shared_ptr<Buffer>).
   ///
   /// \see HasVariadicBuffers
-  util::span<const std::shared_ptr<Buffer>> GetVariadicBuffers() const;
+  std::span<const std::shared_ptr<Buffer>> GetVariadicBuffers() const;
   bool HasVariadicBuffers() const;
 
  private:

cpp/src/arrow/array/util.cc

@@ -23,6 +23,7 @@
 #include <cstring>
 #include <limits>
 #include <memory>
+#include <span>
 #include <type_traits>
 #include <utility>
 #include <vector>
@@ -44,7 +45,6 @@
 #include "arrow/util/endian.h"
 #include "arrow/util/logging_internal.h"
 #include "arrow/util/sort_internal.h"
-#include "arrow/util/span.h"
 #include "arrow/visit_data_inline.h"
 #include "arrow/visit_type_inline.h"
 

cpp/src/arrow/array/validate.cc

@@ -17,6 +17,7 @@
 
 #include "arrow/array/validate.h"
 
+#include <span>
 #include <vector>
 
 #include "arrow/array.h"  // IWYU pragma: keep
@@ -650,9 +651,9 @@ struct ValidateArrayImpl {
                              HexEncode(data, BinaryViewType::kPrefixSize));
     };
 
-    util::span views(data.GetValues<BinaryViewType::c_type>(1),
-                     static_cast<size_t>(data.length));
-    util::span data_buffers(data.buffers.data() + 2, data.buffers.size() - 2);
+    std::span views(data.GetValues<BinaryViewType::c_type>(1),
+                    static_cast<size_t>(data.length));
+    std::span data_buffers(data.buffers.data() + 2, data.buffers.size() - 2);
 
     for (size_t i = 0; i < static_cast<size_t>(data.length); ++i) {
       if (data.IsNull(i)) continue;
@@ -663,7 +664,7 @@ struct ValidateArrayImpl {
       }
 
       if (views[i].is_inline()) {
-        auto padding_bytes = util::span(views[i].inlined.data).subspan(views[i].size());
+        auto padding_bytes = std::span(views[i].inlined.data).subspan(views[i].size());
         for (auto padding_byte : padding_bytes) {
           if (padding_byte != 0) {
             return Status::Invalid("View at slot ", i, " was inline with size ",

cpp/src/arrow/buffer.h

@@ -21,6 +21,7 @@
 #include <cstring>
 #include <memory>
 #include <optional>
+#include <span>
 #include <string>
 #include <string_view>
 #include <utility>
@@ -30,7 +31,6 @@
 #include "arrow/status.h"
 #include "arrow/type_fwd.h"
 #include "arrow/util/macros.h"
-#include "arrow/util/span.h"
 #include "arrow/util/visibility.h"
 
 namespace arrow {
@@ -236,8 +236,8 @@ class ARROW_EXPORT Buffer {
 
   /// \brief Return the buffer's data as a span
   template <typename T>
-  util::span<const T> span_as() const {
-    return util::span(data_as<T>(), static_cast<size_t>(size() / sizeof(T)));
+  std::span<const T> span_as() const {
+    return std::span(data_as<T>(), static_cast<size_t>(size() / sizeof(T)));
   }
 
   /// \brief Return a writable pointer to the buffer's data
@@ -269,8 +269,8 @@ class ARROW_EXPORT Buffer {
 
   /// \brief Return the buffer's mutable data as a span
   template <typename T>
-  util::span<T> mutable_span_as() {
-    return util::span(mutable_data_as<T>(), static_cast<size_t>(size() / sizeof(T)));
+  std::span<T> mutable_span_as() {
+    return std::span(mutable_data_as<T>(), static_cast<size_t>(size() / sizeof(T)));
   }
 
   /// \brief Return the device address of the buffer's data

cpp/src/arrow/c/bridge.cc

@@ -24,6 +24,7 @@
 #include <memory>
 #include <mutex>
 #include <queue>
+#include <span>
 #include <string>
 #include <string_view>
 #include <utility>
@@ -603,7 +604,7 @@ struct ArrayExporter {
                    });
 
     if (need_variadic_buffer_sizes) {
-      auto variadic_buffers = util::span(data->buffers).subspan(2);
+      auto variadic_buffers = std::span(data->buffers).subspan(2);
       export_.variadic_buffer_sizes_.resize(variadic_buffers.size());
       size_t i = 0;
       for (const auto& buf : variadic_buffers) {

cpp/src/arrow/c/bridge_test.cc

@@ -18,6 +18,7 @@
 #include <cerrno>
 #include <deque>
 #include <functional>
+#include <span>
 #include <string>
 #include <string_view>
 #include <type_traits>
@@ -592,8 +593,8 @@ struct ArrayExportChecker {
       ASSERT_EQ(c_export->buffers[i], expected_ptr);
     }
     if (has_variadic_buffer_sizes) {
-      auto variadic_buffers = util::span(expected_data.buffers).subspan(2);
-      auto variadic_buffer_sizes = util::span(
+      auto variadic_buffers = std::span(expected_data.buffers).subspan(2);
+      auto variadic_buffer_sizes = std::span(
           static_cast<const int64_t*>(c_export->buffers[c_export->n_buffers - 1]),
           variadic_buffers.size());
       for (auto [buf, size] : Zip(variadic_buffers, variadic_buffer_sizes)) {

cpp/src/arrow/chunk_resolver.cc

@@ -21,15 +21,14 @@
 #include <cstdint>
 #include <limits>
 #include <memory>
+#include <span>
 #include <vector>
 
 #include "arrow/array.h"
 #include "arrow/record_batch.h"
 
 namespace arrow {
 
-using util::span;
-
 namespace {
 template <typename T>
 int64_t GetLength(const T& array) {
@@ -44,7 +43,7 @@ int64_t GetLength<std::shared_ptr<RecordBatch>>(
 }
 
 template <typename T>
-inline std::vector<int64_t> MakeChunksOffsets(span<T> chunks) {
+inline std::vector<int64_t> MakeChunksOffsets(std::span<T> chunks) {
   std::vector<int64_t> offsets(chunks.size() + 1);
   int64_t offset = 0;
   std::transform(chunks.begin(), chunks.end(), offsets.begin(),
@@ -114,13 +113,13 @@ void ResolveManyInline(uint32_t num_offsets, const int64_t* signed_offsets,
 }  // namespace
 
 ChunkResolver::ChunkResolver(const ArrayVector& chunks) noexcept
-    : offsets_(MakeChunksOffsets(span(chunks))), cached_chunk_(0) {}
+    : offsets_(MakeChunksOffsets(std::span(chunks))), cached_chunk_(0) {}
 
-ChunkResolver::ChunkResolver(span<const Array* const> chunks) noexcept
+ChunkResolver::ChunkResolver(std::span<const Array* const> chunks) noexcept
     : offsets_(MakeChunksOffsets(chunks)), cached_chunk_(0) {}
 
 ChunkResolver::ChunkResolver(const RecordBatchVector& batches) noexcept
-    : offsets_(MakeChunksOffsets(span(batches))), cached_chunk_(0) {}
+    : offsets_(MakeChunksOffsets(std::span(batches))), cached_chunk_(0) {}
 
 ChunkResolver::ChunkResolver(ChunkResolver&& other) noexcept
     : offsets_(std::move(other.offsets_)),

cpp/src/arrow/chunk_resolver.h

@@ -21,12 +21,12 @@
 #include <cassert>
 #include <cstdint>
 #include <limits>
+#include <span>
 #include <type_traits>
 #include <vector>
 
 #include "arrow/type_fwd.h"
 #include "arrow/util/macros.h"
-#include "arrow/util/span.h"
 
 namespace arrow {
 
@@ -77,7 +77,7 @@ class ARROW_EXPORT ChunkResolver {
 
  public:
   explicit ChunkResolver(const ArrayVector& chunks) noexcept;
-  explicit ChunkResolver(util::span<const Array* const> chunks) noexcept;
+  explicit ChunkResolver(std::span<const Array* const> chunks) noexcept;
   explicit ChunkResolver(const RecordBatchVector& batches) noexcept;
 
   /// \brief Construct a ChunkResolver from a vector of chunks.size() + 1 offsets.