|
| 1 | +// SPDX-License-Identifier: Apache-2.0 |
| 2 | +// SPDX-FileCopyrightText: Copyright the Vortex contributors |
| 3 | + |
| 4 | +use std::fmt::Display; |
| 5 | +use std::fmt::Formatter; |
| 6 | +use std::hash::Hasher; |
| 7 | +use std::sync::Arc; |
| 8 | + |
| 9 | +use smallvec::smallvec; |
| 10 | +use vortex_error::VortexExpect; |
| 11 | +use vortex_error::VortexResult; |
| 12 | +use vortex_error::vortex_ensure; |
| 13 | + |
| 14 | +use crate::ArrayEq; |
| 15 | +use crate::ArrayHash; |
| 16 | +use crate::ArrayRef; |
| 17 | +use crate::EqMode; |
| 18 | +use crate::IntoArray; |
| 19 | +use crate::array::Array; |
| 20 | +use crate::array::ArrayParts; |
| 21 | +use crate::array::ArrayView; |
| 22 | +use crate::array::TypedArrayRef; |
| 23 | +use crate::arrays::ListView; |
| 24 | +use crate::arrays::ListViewArray; |
| 25 | +use crate::arrays::listview::ListViewArrayExt; |
| 26 | +use crate::arrays::map::Map; |
| 27 | +use crate::dtype::DType; |
| 28 | +use crate::dtype::MapDType; |
| 29 | +use crate::validity::Validity; |
| 30 | + |
| 31 | +/// The one child slot holding a [`ListViewArray`] of map entries. |
| 32 | +pub(super) const ENTRIES_SLOT: usize = 0; |
| 33 | +pub(super) const NUM_SLOTS: usize = 1; |
| 34 | +pub(super) const SLOT_NAMES: [&str; NUM_SLOTS] = ["entries"]; |
| 35 | + |
| 36 | +/// Encoding-specific metadata for [`crate::arrays::MapArray`]. |
| 37 | +/// |
| 38 | +/// All map metadata is represented by the outer [`DType::Map`] and the entries child, so this |
| 39 | +/// value is intentionally empty. |
| 40 | +#[derive(Clone, Debug, Default)] |
| 41 | +pub struct MapData; |
| 42 | + |
| 43 | +impl Display for MapData { |
| 44 | + fn fmt(&self, _f: &mut Formatter<'_>) -> std::fmt::Result { |
| 45 | + Ok(()) |
| 46 | + } |
| 47 | +} |
| 48 | + |
| 49 | +impl ArrayEq for MapData { |
| 50 | + fn array_eq(&self, _other: &Self, _accuracy: EqMode) -> bool { |
| 51 | + true |
| 52 | + } |
| 53 | +} |
| 54 | + |
| 55 | +impl ArrayHash for MapData { |
| 56 | + fn array_hash<H: Hasher>(&self, _state: &mut H, _accuracy: EqMode) {} |
| 57 | +} |
| 58 | + |
| 59 | +/// The logical and physical inputs used to construct a [`crate::arrays::MapArray`]. |
| 60 | +pub struct MapDataParts { |
| 61 | + /// The key/value type and sortedness assertion for the map. |
| 62 | + pub map_dtype: MapDType, |
| 63 | + /// The physical list-view storage of `{key, value}` entry structs. |
| 64 | + pub entries: ListViewArray, |
| 65 | +} |
| 66 | + |
| 67 | +/// Accessors for the canonical map representation. |
| 68 | +pub trait MapArrayExt: TypedArrayRef<Map> { |
| 69 | + /// Returns the list-view storage of map entry structs. |
| 70 | + fn entries(&self) -> ArrayView<'_, ListView> { |
| 71 | + self.as_ref().slots()[ENTRIES_SLOT] |
| 72 | + .as_ref() |
| 73 | + .vortex_expect("MapArray entries slot") |
| 74 | + .as_::<ListView>() |
| 75 | + } |
| 76 | + |
| 77 | + /// Returns the entry structs for one map row. |
| 78 | + fn entries_at(&self, index: usize) -> VortexResult<ArrayRef> { |
| 79 | + self.entries().list_elements_at(index) |
| 80 | + } |
| 81 | + |
| 82 | + /// Returns the number of entries in one map row. |
| 83 | + fn entry_count_at(&self, index: usize) -> usize { |
| 84 | + self.entries().size_at(index) |
| 85 | + } |
| 86 | + |
| 87 | + /// Returns the outer map validity delegated from the entries list-view. |
| 88 | + fn map_validity(&self) -> Validity { |
| 89 | + self.entries().listview_validity() |
| 90 | + } |
| 91 | + |
| 92 | + /// Returns this map's key/value type information. |
| 93 | + fn map_dtype(&self) -> &MapDType { |
| 94 | + self.as_ref() |
| 95 | + .dtype() |
| 96 | + .as_map_opt() |
| 97 | + .vortex_expect("MapArray requires a map dtype") |
| 98 | + } |
| 99 | + |
| 100 | + /// Returns whether producers assert sorted keys within each map value. |
| 101 | + fn keys_sorted(&self) -> bool { |
| 102 | + self.map_dtype().keys_sorted() |
| 103 | + } |
| 104 | +} |
| 105 | +impl<T: TypedArrayRef<Map>> MapArrayExt for T {} |
| 106 | + |
| 107 | +impl Array<Map> { |
| 108 | + /// Creates a canonical map array from its map dtype and list-view entry storage. |
| 109 | + /// |
| 110 | + /// # Panics |
| 111 | + /// |
| 112 | + /// Panics if `entries` is not a list of the map dtype's non-nullable `{key, value}` entry |
| 113 | + /// struct with matching outer nullability. |
| 114 | + pub fn new(map_dtype: MapDType, entries: ListViewArray) -> Self { |
| 115 | + Self::try_new(map_dtype, entries).vortex_expect("MapArray construction failed") |
| 116 | + } |
| 117 | + |
| 118 | + /// Constructs a canonical map array from its map dtype and list-view entry storage. |
| 119 | + /// |
| 120 | + /// # Errors |
| 121 | + /// |
| 122 | + /// Returns an error when the entry child is not `ListView<Struct<key, value>>`, has a |
| 123 | + /// different outer nullability, or has a different length than the outer map array. |
| 124 | + pub fn try_new(map_dtype: MapDType, entries: ListViewArray) -> VortexResult<Self> { |
| 125 | + let nullability = entries.nullability(); |
| 126 | + let dtype = DType::Map(map_dtype, nullability); |
| 127 | + let len = entries.len(); |
| 128 | + let parts = ArrayParts::new(Map, dtype, len, MapData) |
| 129 | + .with_slots(smallvec![Some(entries.into_array())]); |
| 130 | + Self::try_from_parts(parts) |
| 131 | + } |
| 132 | + |
| 133 | + /// Creates a canonical map array without validating its entry storage. |
| 134 | + /// |
| 135 | + /// # Safety |
| 136 | + /// |
| 137 | + /// The caller must ensure that `entries` has dtype |
| 138 | + /// `List(Struct { key, value }, entries.nullability())`, where the struct exactly matches |
| 139 | + /// `map_dtype.entries_dtype()`. |
| 140 | + pub unsafe fn new_unchecked(map_dtype: MapDType, entries: ListViewArray) -> Self { |
| 141 | + let nullability = entries.nullability(); |
| 142 | + let dtype = DType::Map(map_dtype, nullability); |
| 143 | + let len = entries.len(); |
| 144 | + let parts = ArrayParts::new(Map, dtype, len, MapData) |
| 145 | + .with_slots(smallvec![Some(entries.into_array())]); |
| 146 | + unsafe { Self::from_parts_unchecked(parts) } |
| 147 | + } |
| 148 | + |
| 149 | + /// Decomposes this map array into its logical dtype and physical entries child. |
| 150 | + pub fn into_data_parts(self) -> MapDataParts { |
| 151 | + let map_dtype = self |
| 152 | + .dtype() |
| 153 | + .as_map_opt() |
| 154 | + .vortex_expect("MapArray requires a map dtype") |
| 155 | + .clone(); |
| 156 | + let entries = self.entries().into_owned(); |
| 157 | + MapDataParts { map_dtype, entries } |
| 158 | + } |
| 159 | +} |
| 160 | + |
| 161 | +fn expected_entries_dtype(map_dtype: &MapDType, nullability: crate::dtype::Nullability) -> DType { |
| 162 | + DType::List(Arc::new(map_dtype.entries_dtype()), nullability) |
| 163 | +} |
| 164 | + |
| 165 | +pub(super) fn validate_entries( |
| 166 | + map_dtype: &MapDType, |
| 167 | + nullability: crate::dtype::Nullability, |
| 168 | + len: usize, |
| 169 | + entries: &ArrayRef, |
| 170 | +) -> VortexResult<()> { |
| 171 | + vortex_ensure!( |
| 172 | + entries.is::<ListView>(), |
| 173 | + "MapArray entries must use vortex.listview encoding, got {}", |
| 174 | + entries.encoding_id() |
| 175 | + ); |
| 176 | + vortex_ensure!( |
| 177 | + entries.len() == len, |
| 178 | + "MapArray entries length {} does not match outer length {len}", |
| 179 | + entries.len() |
| 180 | + ); |
| 181 | + |
| 182 | + let expected_dtype = expected_entries_dtype(map_dtype, nullability); |
| 183 | + vortex_ensure!( |
| 184 | + entries.dtype() == &expected_dtype, |
| 185 | + "MapArray entries dtype {} does not match expected {expected_dtype}", |
| 186 | + entries.dtype() |
| 187 | + ); |
| 188 | + |
| 189 | + Ok(()) |
| 190 | +} |
0 commit comments