Path: blob/main/crates/polars-arrow/src/io/ipc/read/array/utf8.rs
8420 views
use std::io::{Read, Seek};12use polars_buffer::Buffer;3use polars_error::polars_err;4use polars_utils::bool::UnsafeBool;56use super::super::read_basic::*;7use super::*;8use crate::array::Utf8Array;9use crate::offset::Offset;1011#[allow(clippy::too_many_arguments)]12pub fn read_utf8<O: Offset, R: Read + Seek>(13field_nodes: &mut VecDeque<Node>,14dtype: ArrowDataType,15buffers: &mut VecDeque<IpcBuffer>,16reader: &mut R,17block_offset: u64,18is_little_endian: bool,19compression: Option<Compression>,20limit: Option<usize>,21scratch: &mut Vec<u8>,22checked: UnsafeBool,23) -> PolarsResult<Utf8Array<O>> {24let field_node = try_get_field_node(field_nodes, &dtype)?;2526let validity = read_validity(27buffers,28field_node,29reader,30block_offset,31is_little_endian,32compression,33limit,34scratch,35)?;3637let length = try_get_array_length(field_node, limit)?;3839let offsets: Buffer<O> = read_buffer(40buffers,411 + length,42reader,43block_offset,44is_little_endian,45compression,46scratch,47)48// Older versions of the IPC format sometimes do not report an offset49.or_else(|_| PolarsResult::Ok(Buffer::<O>::from(vec![O::default()])))?;5051let last_offset = offsets.last().unwrap().to_usize();52let values = read_buffer(53buffers,54last_offset,55reader,56block_offset,57is_little_endian,58compression,59scratch,60)?;6162if *checked {63Utf8Array::<O>::try_new(dtype, offsets.try_into()?, values, validity)64} else {65// SAFETY:66// Invariant of the `checked` state that this is valid.67unsafe {68Ok(Utf8Array::<O>::new_unchecked(69dtype,70offsets.try_into()?,71values,72validity,73))74}75}76}7778pub fn skip_utf8(79field_nodes: &mut VecDeque<Node>,80buffers: &mut VecDeque<IpcBuffer>,81) -> PolarsResult<()> {82let _ = field_nodes.pop_front().ok_or_else(|| {83polars_err!(84oos = "IPC: unable to fetch the field for utf8. The file or stream is corrupted."85)86})?;8788let _ = buffers89.pop_front()90.ok_or_else(|| polars_err!(oos = "IPC: missing validity buffer."))?;91let _ = buffers92.pop_front()93.ok_or_else(|| polars_err!(oos = "IPC: missing offsets buffer."))?;94let _ = buffers95.pop_front()96.ok_or_else(|| polars_err!(oos = "IPC: missing values buffer."))?;97Ok(())98}99100101