Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
pola-rs
GitHub Repository: pola-rs/polars
Path: blob/main/crates/polars-arrow/src/io/ipc/read/array/binary.rs
6940 views
1
use std::collections::VecDeque;
2
use std::io::{Read, Seek};
3
4
use polars_error::{PolarsResult, polars_err};
5
6
use super::super::read_basic::*;
7
use super::super::{Compression, IpcBuffer, Node};
8
use crate::array::BinaryArray;
9
use crate::buffer::Buffer;
10
use crate::datatypes::ArrowDataType;
11
use crate::io::ipc::read::array::{try_get_array_length, try_get_field_node};
12
use crate::offset::Offset;
13
14
#[allow(clippy::too_many_arguments)]
15
pub fn read_binary<O: Offset, R: Read + Seek>(
16
field_nodes: &mut VecDeque<Node>,
17
dtype: ArrowDataType,
18
buffers: &mut VecDeque<IpcBuffer>,
19
reader: &mut R,
20
block_offset: u64,
21
is_little_endian: bool,
22
compression: Option<Compression>,
23
limit: Option<usize>,
24
scratch: &mut Vec<u8>,
25
) -> PolarsResult<BinaryArray<O>> {
26
let field_node = try_get_field_node(field_nodes, &dtype)?;
27
28
let validity = read_validity(
29
buffers,
30
field_node,
31
reader,
32
block_offset,
33
is_little_endian,
34
compression,
35
limit,
36
scratch,
37
)?;
38
39
let length = try_get_array_length(field_node, limit)?;
40
41
let offsets: Buffer<O> = read_buffer(
42
buffers,
43
1 + length,
44
reader,
45
block_offset,
46
is_little_endian,
47
compression,
48
scratch,
49
)
50
// Older versions of the IPC format sometimes do not report an offset
51
.or_else(|_| PolarsResult::Ok(Buffer::<O>::from(vec![O::default()])))?;
52
53
let last_offset = offsets.last().unwrap().to_usize();
54
let values = read_buffer(
55
buffers,
56
last_offset,
57
reader,
58
block_offset,
59
is_little_endian,
60
compression,
61
scratch,
62
)?;
63
64
BinaryArray::<O>::try_new(dtype, offsets.try_into()?, values, validity)
65
}
66
67
pub fn skip_binary(
68
field_nodes: &mut VecDeque<Node>,
69
buffers: &mut VecDeque<IpcBuffer>,
70
) -> PolarsResult<()> {
71
let _ = field_nodes.pop_front().ok_or_else(|| {
72
polars_err!(
73
oos = "IPC: unable to fetch the field for binary. The file or stream is corrupted."
74
)
75
})?;
76
77
let _ = buffers
78
.pop_front()
79
.ok_or_else(|| polars_err!(oos = "IPC: missing validity buffer."))?;
80
let _ = buffers
81
.pop_front()
82
.ok_or_else(|| polars_err!(oos = "IPC: missing offsets buffer."))?;
83
let _ = buffers
84
.pop_front()
85
.ok_or_else(|| polars_err!(oos = "IPC: missing values buffer."))?;
86
Ok(())
87
}
88
89