sparrow-ipc 0.3.0
Loading...
Searching...
No Matches
deserialize_variable_size_binary_array.hpp
Go to the documentation of this file.
1#pragma once
2
3#include <span>
4#include <unordered_set>
5
6#include <sparrow/arrow_interface/arrow_array_schema_proxy.hpp>
7#include <sparrow/variable_size_binary_array.hpp>
8
13
14namespace sparrow_ipc
15{
16 template <typename T>
19 const field_descriptor& field_desc
20 )
21 {
22 const std::string_view format = sparrow::data_type_to_format(sparrow::detail::get_data_type_from_array<T>::get());
23
24 ArrowSchema schema = make_non_owning_arrow_schema(
25 format,
26 field_desc.name,
27 field_desc.metadata,
28 field_desc.flags,
29 0,
30 nullptr,
31 nullptr
32 );
33
34 const auto compression = context.record_batch.compression();
35 std::vector<arrow_array_private_data::optionally_owned_buffer> buffers;
36 constexpr auto nb_buffers = 3;
37 buffers.reserve(nb_buffers);
38
39 {
40 auto validity_buffer_span = utils::get_buffer(context.record_batch, context.body, context.buffer_index);
41 auto offset_buffer_span = utils::get_buffer(context.record_batch, context.body, context.buffer_index);
42 auto data_buffer_span = utils::get_buffer(context.record_batch, context.body, context.buffer_index);
43
44 if (compression)
45 {
46 buffers.push_back(utils::get_decompressed_buffer(validity_buffer_span, compression));
47 buffers.push_back(utils::get_decompressed_buffer(offset_buffer_span, compression));
48 buffers.push_back(utils::get_decompressed_buffer(data_buffer_span, compression));
49 }
50 else
51 {
52 buffers.push_back(std::move(validity_buffer_span));
53 buffers.push_back(std::move(offset_buffer_span));
54 buffers.push_back(std::move(data_buffer_span));
55 }
56 }
57
58 const auto null_count = std::visit(
59 [length = field_desc.length](const auto& arg) {
60 std::span<const uint8_t> span(arg.data(), arg.size());
61 return utils::get_bitmap_pointer_and_null_count(span, length).second;
62 },
63 buffers[0]
64 );
65
67 field_desc.length,
68 null_count,
69 0,
70 0,
71 nullptr,
72 nullptr,
73 std::move(buffers)
74 );
75
76 sparrow::arrow_proxy ap{std::move(array), std::move(schema)};
77 return T{std::move(ap)};
78 }
79}
std::span< const uint8_t > get_buffer(const org::apache::arrow::flatbuf::RecordBatch &record_batch, std::span< const uint8_t > body, size_t &buffer_index)
Extracts a buffer from a RecordBatch's body.
std::variant< sparrow::buffer< std::uint8_t >, std::span< const std::uint8_t > > get_decompressed_buffer(std::span< const uint8_t > buffer_span, const org::apache::arrow::flatbuf::BodyCompression *compression)
Retrieves a decompressed buffer or a view of the original buffer.
T deserialize_variable_size_binary_array(deserialization_context &context, const field_descriptor &field_desc)
ArrowSchema make_non_owning_arrow_schema(std::string_view format, std::string_view name, std::optional< M > metadata, std::optional< std::unordered_set< sparrow::ArrowFlag > > flags, size_t children_count, ArrowSchema **children, ArrowSchema *dictionary)
ArrowArray make_arrow_array(int64_t length, int64_t null_count, int64_t offset, size_t children_count, ArrowArray **children, ArrowArray *dictionary, Arg &&private_data_arg)
Encapsulates the context required for deserialization.
const org::apache::arrow::flatbuf::RecordBatch & record_batch
Encapsulates the description of a field to be deserialized.
std::optional< std::vector< sparrow::metadata_pair > > metadata
std::optional< std::unordered_set< sparrow::ArrowFlag > > flags