vortex_array/arrays/varbinview/
view.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4//! The 16-byte view struct stored in variable-length binary vectors.
5
6use std::fmt;
7use std::hash::Hash;
8use std::hash::Hasher;
9use std::ops::Range;
10
11use static_assertions::assert_eq_align;
12use static_assertions::assert_eq_size;
13use vortex_error::VortexExpect;
14
15/// A view over a variable-length binary value.
16///
17/// Either an inlined representation (for values <= 12 bytes) or a reference
18/// to an external buffer (for values > 12 bytes).
19#[derive(Clone, Copy)]
20#[repr(C, align(16))]
21pub union BinaryView {
22    /// Numeric representation. This is logically `u128`, but we split it into the high and low
23    /// bits to preserve the alignment.
24    pub(crate) le_bytes: [u8; 16],
25
26    /// Inlined representation: strings <= 12 bytes
27    pub(crate) inlined: Inlined,
28
29    /// Reference type: strings > 12 bytes.
30    pub(crate) _ref: Ref,
31}
32
33assert_eq_align!(BinaryView, u128);
34assert_eq_size!(BinaryView, [u8; 16]);
35assert_eq_size!(Inlined, [u8; 16]);
36assert_eq_size!(Ref, [u8; 16]);
37
38/// Variant of a [`BinaryView`] that holds an inlined value.
39#[derive(Clone, Copy, Debug, PartialEq, Eq)]
40#[repr(C, align(8))]
41pub struct Inlined {
42    /// The size of the full value.
43    pub size: u32,
44    /// The full inlined value.
45    pub data: [u8; BinaryView::MAX_INLINED_SIZE],
46}
47
48impl Inlined {
49    /// Creates a new inlined representation from the provided value of constant size.
50    fn new<const N: usize>(value: &[u8]) -> Self {
51        debug_assert_eq!(value.len(), N);
52        let mut inlined = Self {
53            size: N.try_into().vortex_expect("inlined size must fit in u32"),
54            data: [0u8; BinaryView::MAX_INLINED_SIZE],
55        };
56        inlined.data[..N].copy_from_slice(&value[..N]);
57        inlined
58    }
59
60    /// Returns the full inlined value.
61    #[inline]
62    pub fn value(&self) -> &[u8] {
63        &self.data[0..(self.size as usize)]
64    }
65}
66
67/// Variant of a [`BinaryView`] that holds a reference to an external buffer.
68#[derive(Clone, Copy, Debug)]
69#[repr(C, align(8))]
70pub struct Ref {
71    /// The size of the full value.
72    pub size: u32,
73    /// The prefix bytes of the value (first 4 bytes).
74    pub prefix: [u8; 4],
75    /// The index of the buffer where the full value is stored.
76    pub buffer_index: u32,
77    /// The offset within the buffer where the full value starts.
78    pub offset: u32,
79}
80
81impl Ref {
82    /// Returns the range within the buffer where the full value is stored.
83    #[inline]
84    pub fn as_range(&self) -> Range<usize> {
85        self.offset as usize..(self.offset + self.size) as usize
86    }
87
88    /// Replaces the buffer index and offset of the reference, returning a new `Ref`.
89    #[inline]
90    pub fn with_buffer_and_offset(&self, buffer_index: u32, offset: u32) -> Ref {
91        Self {
92            size: self.size,
93            prefix: self.prefix,
94            buffer_index,
95            offset,
96        }
97    }
98}
99
100impl BinaryView {
101    /// Maximum size of an inlined binary value.
102    pub const MAX_INLINED_SIZE: usize = 12;
103
104    /// Create a view from a value, block and offset
105    ///
106    /// Depending on the length of the provided value either a new inlined
107    /// or a reference view will be constructed.
108    ///
109    /// Adapted from arrow-rs <https://github.com/apache/arrow-rs/blob/f4fde769ab6e1a9b75f890b7f8b47bc22800830b/arrow-array/src/builder/generic_bytes_view_builder.rs#L524>
110    /// Explicitly enumerating inlined view produces code that avoids calling generic `ptr::copy_non_interleave` that's slower than explicit stores
111    #[inline(never)]
112    pub fn make_view(value: &[u8], block: u32, offset: u32) -> Self {
113        match value.len() {
114            0 => Self {
115                inlined: Inlined::new::<0>(value),
116            },
117            1 => Self {
118                inlined: Inlined::new::<1>(value),
119            },
120            2 => Self {
121                inlined: Inlined::new::<2>(value),
122            },
123            3 => Self {
124                inlined: Inlined::new::<3>(value),
125            },
126            4 => Self {
127                inlined: Inlined::new::<4>(value),
128            },
129            5 => Self {
130                inlined: Inlined::new::<5>(value),
131            },
132            6 => Self {
133                inlined: Inlined::new::<6>(value),
134            },
135            7 => Self {
136                inlined: Inlined::new::<7>(value),
137            },
138            8 => Self {
139                inlined: Inlined::new::<8>(value),
140            },
141            9 => Self {
142                inlined: Inlined::new::<9>(value),
143            },
144            10 => Self {
145                inlined: Inlined::new::<10>(value),
146            },
147            11 => Self {
148                inlined: Inlined::new::<11>(value),
149            },
150            12 => Self {
151                inlined: Inlined::new::<12>(value),
152            },
153            _ => Self {
154                _ref: Ref {
155                    size: u32::try_from(value.len()).vortex_expect("value length must fit in u32"),
156                    prefix: value[0..4]
157                        .try_into()
158                        .ok()
159                        .vortex_expect("prefix must be exactly 4 bytes"),
160                    buffer_index: block,
161                    offset,
162                },
163            },
164        }
165    }
166
167    /// Create a new empty view
168    #[inline]
169    pub fn empty_view() -> Self {
170        Self { le_bytes: [0; 16] }
171    }
172
173    /// Create a new inlined binary view
174    ///
175    /// # Panics
176    ///
177    /// Panics if the provided string is too long to inline.
178    #[inline]
179    pub fn new_inlined(value: &[u8]) -> Self {
180        assert!(
181            value.len() <= Self::MAX_INLINED_SIZE,
182            "expected inlined value to be <= 12 bytes, was {}",
183            value.len()
184        );
185
186        Self::make_view(value, 0, 0)
187    }
188
189    /// Returns the length of the binary value.
190    #[inline]
191    pub fn len(&self) -> u32 {
192        unsafe { self.inlined.size }
193    }
194
195    /// Returns true if the binary value is empty.
196    #[inline]
197    pub fn is_empty(&self) -> bool {
198        self.len() == 0
199    }
200
201    /// Returns true if the binary value is inlined.
202    #[inline]
203    #[expect(
204        clippy::cast_possible_truncation,
205        reason = "MAX_INLINED_SIZE is a small constant"
206    )]
207    pub fn is_inlined(&self) -> bool {
208        self.len() <= (Self::MAX_INLINED_SIZE as u32)
209    }
210
211    /// Returns the inlined representation of the binary value.
212    pub fn as_inlined(&self) -> &Inlined {
213        debug_assert!(self.is_inlined());
214        unsafe { &self.inlined }
215    }
216
217    /// Returns the reference representation of the binary value.
218    pub fn as_view(&self) -> &Ref {
219        debug_assert!(!self.is_inlined());
220        unsafe { &self._ref }
221    }
222
223    /// Returns a mutable reference to the reference representation of the binary value.
224    pub fn as_view_mut(&mut self) -> &mut Ref {
225        unsafe { &mut self._ref }
226    }
227
228    /// Returns the binary view as u128 representation.
229    pub fn as_u128(&self) -> u128 {
230        // SAFETY: binary view always safe to read as u128 LE bytes
231        unsafe { u128::from_le_bytes(self.le_bytes) }
232    }
233}
234
235impl From<u128> for BinaryView {
236    fn from(value: u128) -> Self {
237        BinaryView {
238            le_bytes: value.to_le_bytes(),
239        }
240    }
241}
242
243impl From<Ref> for BinaryView {
244    fn from(value: Ref) -> Self {
245        BinaryView { _ref: value }
246    }
247}
248
249impl PartialEq for BinaryView {
250    fn eq(&self, other: &Self) -> bool {
251        let a = unsafe { std::mem::transmute::<&BinaryView, &u128>(self) };
252        let b = unsafe { std::mem::transmute::<&BinaryView, &u128>(other) };
253        a == b
254    }
255}
256impl Eq for BinaryView {}
257
258impl Hash for BinaryView {
259    fn hash<H: Hasher>(&self, state: &mut H) {
260        unsafe { std::mem::transmute::<&BinaryView, &u128>(self) }.hash(state);
261    }
262}
263
264impl Default for BinaryView {
265    fn default() -> Self {
266        Self::make_view(&[], 0, 0)
267    }
268}
269
270impl fmt::Debug for BinaryView {
271    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
272        let mut s = f.debug_struct("BinaryView");
273        if self.is_inlined() {
274            s.field("inline", &self.as_inlined());
275        } else {
276            s.field("ref", &self.as_view());
277        }
278        s.finish()
279    }
280}
vortex_array/arrays/varbinview/view.rs

vortex_array/arrays/varbinview/
view.rs
