polars_arrow/array/growable/
list.rs

1use std::sync::Arc;
2
3use super::{make_growable, Growable};
4use crate::array::growable::utils::{extend_validity, prepare_validity};
5use crate::array::{Array, ListArray};
6use crate::bitmap::BitmapBuilder;
7use crate::offset::{Offset, Offsets};
8
9unsafe fn extend_offset_values<O: Offset>(
10    growable: &mut GrowableList<'_, O>,
11    index: usize,
12    start: usize,
13    len: usize,
14) {
15    let array = growable.arrays.get_unchecked(index);
16    let offsets = array.offsets();
17
18    growable
19        .offsets
20        .try_extend_from_slice(offsets, start, len)
21        .unwrap();
22
23    let end = offsets.buffer().get_unchecked(start + len).to_usize();
24    let start = offsets.buffer().get_unchecked(start).to_usize();
25    let len = end - start;
26    growable.values.extend(index, start, len);
27}
28
29/// Concrete [`Growable`] for the [`ListArray`].
30pub struct GrowableList<'a, O: Offset> {
31    arrays: Vec<&'a ListArray<O>>,
32    validity: Option<BitmapBuilder>,
33    values: Box<dyn Growable<'a> + 'a>,
34    offsets: Offsets<O>,
35}
36
37impl<'a, O: Offset> GrowableList<'a, O> {
38    /// Creates a new [`GrowableList`] bound to `arrays` with a pre-allocated `capacity`.
39    /// # Panics
40    /// If `arrays` is empty.
41    pub fn new(arrays: Vec<&'a ListArray<O>>, mut use_validity: bool, capacity: usize) -> Self {
42        // if any of the arrays has nulls, insertions from any array requires setting bits
43        // as there is at least one array with nulls.
44        if !use_validity & arrays.iter().any(|array| array.null_count() > 0) {
45            use_validity = true;
46        };
47
48        let inner = arrays
49            .iter()
50            .map(|array| array.values().as_ref())
51            .collect::<Vec<_>>();
52        let values = make_growable(&inner, use_validity, 0);
53
54        Self {
55            arrays,
56            offsets: Offsets::with_capacity(capacity),
57            values,
58            validity: prepare_validity(use_validity, capacity),
59        }
60    }
61
62    pub fn to(&mut self) -> ListArray<O> {
63        let validity = std::mem::take(&mut self.validity);
64        let offsets = std::mem::take(&mut self.offsets);
65        let values = self.values.as_box();
66
67        ListArray::<O>::new(
68            self.arrays[0].dtype().clone(),
69            offsets.into(),
70            values,
71            validity.map(|v| v.freeze()),
72        )
73    }
74}
75
76impl<'a, O: Offset> Growable<'a> for GrowableList<'a, O> {
77    unsafe fn extend(&mut self, index: usize, start: usize, len: usize) {
78        let array = *self.arrays.get_unchecked(index);
79        extend_validity(&mut self.validity, array, start, len);
80        extend_offset_values::<O>(self, index, start, len);
81    }
82
83    fn extend_validity(&mut self, additional: usize) {
84        self.offsets.extend_constant(additional);
85        if let Some(validity) = &mut self.validity {
86            validity.extend_constant(additional, false);
87        }
88    }
89
90    #[inline]
91    fn len(&self) -> usize {
92        self.offsets.len() - 1
93    }
94
95    fn as_arc(&mut self) -> Arc<dyn Array> {
96        Arc::new(self.to())
97    }
98
99    fn as_box(&mut self) -> Box<dyn Array> {
100        Box::new(self.to())
101    }
102}
103
104impl<'a, O: Offset> From<GrowableList<'a, O>> for ListArray<O> {
105    fn from(mut val: GrowableList<'a, O>) -> Self {
106        val.to()
107    }
108}