arrow_buffer/util/bit_iterator.rs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
//! Types for iterating over packed bitmasks
use crate::bit_chunk_iterator::{UnalignedBitChunk, UnalignedBitChunkIterator};
use crate::bit_util::{ceil, get_bit_raw};
/// Iterator over the bits within a packed bitmask
///
/// To efficiently iterate over just the set bits see [`BitIndexIterator`] and [`BitSliceIterator`]
pub struct BitIterator<'a> {
buffer: &'a [u8],
current_offset: usize,
end_offset: usize,
}
impl<'a> BitIterator<'a> {
/// Create a new [`BitIterator`] from the provided `buffer`,
/// and `offset` and `len` in bits
///
/// # Panic
///
/// Panics if `buffer` is too short for the provided offset and length
pub fn new(buffer: &'a [u8], offset: usize, len: usize) -> Self {
let end_offset = offset.checked_add(len).unwrap();
let required_len = ceil(end_offset, 8);
assert!(
buffer.len() >= required_len,
"BitIterator buffer too small, expected {required_len} got {}",
buffer.len()
);
Self {
buffer,
current_offset: offset,
end_offset,
}
}
}
impl Iterator for BitIterator<'_> {
type Item = bool;
fn next(&mut self) -> Option<Self::Item> {
if self.current_offset == self.end_offset {
return None;
}
// Safety:
// offsets in bounds
let v = unsafe { get_bit_raw(self.buffer.as_ptr(), self.current_offset) };
self.current_offset += 1;
Some(v)
}
fn size_hint(&self) -> (usize, Option<usize>) {
let remaining_bits = self.end_offset - self.current_offset;
(remaining_bits, Some(remaining_bits))
}
}
impl ExactSizeIterator for BitIterator<'_> {}
impl DoubleEndedIterator for BitIterator<'_> {
fn next_back(&mut self) -> Option<Self::Item> {
if self.current_offset == self.end_offset {
return None;
}
self.end_offset -= 1;
// Safety:
// offsets in bounds
let v = unsafe { get_bit_raw(self.buffer.as_ptr(), self.end_offset) };
Some(v)
}
}
/// Iterator of contiguous ranges of set bits within a provided packed bitmask
///
/// Returns `(usize, usize)` each representing an interval where the corresponding
/// bits in the provides mask are set
///
#[derive(Debug)]
pub struct BitSliceIterator<'a> {
iter: UnalignedBitChunkIterator<'a>,
len: usize,
current_offset: i64,
current_chunk: u64,
}
impl<'a> BitSliceIterator<'a> {
/// Create a new [`BitSliceIterator`] from the provided `buffer`,
/// and `offset` and `len` in bits
pub fn new(buffer: &'a [u8], offset: usize, len: usize) -> Self {
let chunk = UnalignedBitChunk::new(buffer, offset, len);
let mut iter = chunk.iter();
let current_offset = -(chunk.lead_padding() as i64);
let current_chunk = iter.next().unwrap_or(0);
Self {
iter,
len,
current_offset,
current_chunk,
}
}
/// Returns `Some((chunk_offset, bit_offset))` for the next chunk that has at
/// least one bit set, or None if there is no such chunk.
///
/// Where `chunk_offset` is the bit offset to the current `u64` chunk
/// and `bit_offset` is the offset of the first `1` bit in that chunk
fn advance_to_set_bit(&mut self) -> Option<(i64, u32)> {
loop {
if self.current_chunk != 0 {
// Find the index of the first 1
let bit_pos = self.current_chunk.trailing_zeros();
return Some((self.current_offset, bit_pos));
}
self.current_chunk = self.iter.next()?;
self.current_offset += 64;
}
}
}
impl Iterator for BitSliceIterator<'_> {
type Item = (usize, usize);
fn next(&mut self) -> Option<Self::Item> {
// Used as termination condition
if self.len == 0 {
return None;
}
let (start_chunk, start_bit) = self.advance_to_set_bit()?;
// Set bits up to start
self.current_chunk |= (1 << start_bit) - 1;
loop {
if self.current_chunk != u64::MAX {
// Find the index of the first 0
let end_bit = self.current_chunk.trailing_ones();
// Zero out up to end_bit
self.current_chunk &= !((1 << end_bit) - 1);
return Some((
(start_chunk + start_bit as i64) as usize,
(self.current_offset + end_bit as i64) as usize,
));
}
match self.iter.next() {
Some(next) => {
self.current_chunk = next;
self.current_offset += 64;
}
None => {
return Some((
(start_chunk + start_bit as i64) as usize,
std::mem::replace(&mut self.len, 0),
));
}
}
}
}
}
/// An iterator of `usize` whose index in a provided bitmask is true
///
/// This provides the best performance on most masks, apart from those which contain
/// large runs and therefore favour [`BitSliceIterator`]
#[derive(Debug)]
pub struct BitIndexIterator<'a> {
current_chunk: u64,
chunk_offset: i64,
iter: UnalignedBitChunkIterator<'a>,
}
impl<'a> BitIndexIterator<'a> {
/// Create a new [`BitIndexIterator`] from the provide `buffer`,
/// and `offset` and `len` in bits
pub fn new(buffer: &'a [u8], offset: usize, len: usize) -> Self {
let chunks = UnalignedBitChunk::new(buffer, offset, len);
let mut iter = chunks.iter();
let current_chunk = iter.next().unwrap_or(0);
let chunk_offset = -(chunks.lead_padding() as i64);
Self {
current_chunk,
chunk_offset,
iter,
}
}
}
impl Iterator for BitIndexIterator<'_> {
type Item = usize;
fn next(&mut self) -> Option<Self::Item> {
loop {
if self.current_chunk != 0 {
let bit_pos = self.current_chunk.trailing_zeros();
self.current_chunk ^= 1 << bit_pos;
return Some((self.chunk_offset + bit_pos as i64) as usize);
}
self.current_chunk = self.iter.next()?;
self.chunk_offset += 64;
}
}
}
/// Calls the provided closure for each index in the provided null mask that is set,
/// using an adaptive strategy based on the null count
///
/// Ideally this would be encapsulated in an [`Iterator`] that would determine the optimal
/// strategy up front, and then yield indexes based on this.
///
/// Unfortunately, external iteration based on the resulting [`Iterator`] would match the strategy
/// variant on each call to [`Iterator::next`], and LLVM generally cannot eliminate this.
///
/// One solution to this might be internal iteration, e.g. [`Iterator::try_fold`], however,
/// it is currently [not possible] to override this for custom iterators in stable Rust.
///
/// As such this is the next best option
///
/// [not possible]: https://github.com/rust-lang/rust/issues/69595
#[inline]
pub fn try_for_each_valid_idx<E, F: FnMut(usize) -> Result<(), E>>(
len: usize,
offset: usize,
null_count: usize,
nulls: Option<&[u8]>,
f: F,
) -> Result<(), E> {
let valid_count = len - null_count;
if valid_count == len {
(0..len).try_for_each(f)
} else if null_count != len {
BitIndexIterator::new(nulls.unwrap(), offset, len).try_for_each(f)
} else {
Ok(())
}
}
// Note: further tests located in arrow_select::filter module
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_bit_iterator_size_hint() {
let mut b = BitIterator::new(&[0b00000011], 0, 2);
assert_eq!(
b.size_hint(),
(2, Some(2)),
"Expected size_hint to be (2, Some(2))"
);
b.next();
assert_eq!(
b.size_hint(),
(1, Some(1)),
"Expected size_hint to be (1, Some(1)) after one bit consumed"
);
b.next();
assert_eq!(
b.size_hint(),
(0, Some(0)),
"Expected size_hint to be (0, Some(0)) after all bits consumed"
);
}
#[test]
fn test_bit_iterator() {
let mask = &[0b00010010, 0b00100011, 0b00000101, 0b00010001, 0b10010011];
let actual: Vec<_> = BitIterator::new(mask, 0, 5).collect();
assert_eq!(actual, &[false, true, false, false, true]);
let actual: Vec<_> = BitIterator::new(mask, 4, 5).collect();
assert_eq!(actual, &[true, false, false, false, true]);
let actual: Vec<_> = BitIterator::new(mask, 12, 14).collect();
assert_eq!(
actual,
&[
false, true, false, false, true, false, true, false, false, false, false, false,
true, false
]
);
assert_eq!(BitIterator::new(mask, 0, 0).count(), 0);
assert_eq!(BitIterator::new(mask, 40, 0).count(), 0);
}
#[test]
#[should_panic(expected = "BitIterator buffer too small, expected 3 got 2")]
fn test_bit_iterator_bounds() {
let mask = &[223, 23];
BitIterator::new(mask, 17, 0);
}
}