bitlist 0.0.5

Word-sized bit list implementation with bigint functionality
Documentation 
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

use crate::heap::HeapBitList;
use crate::inline::InlineBitList;
use std::hash::{Hash, Hasher};
use std::mem::{ManuallyDrop, align_of, size_of, transmute};
use std::num::NonZeroUsize;
use std::ptr::NonNull;

pub(crate) type NonZeroPtr = NonNull<usize>;
#[cfg_attr(feature = "align_16", repr(C, packed(2)))]
#[cfg_attr(all(feature = "align_32", not(feature = "align_16")), repr(C, packed(4)))]
#[cfg_attr(all(not(feature = "align_32"), not(feature = "align_16")), repr(transparent))]
pub struct BitList {
    //this is a pointer to keep memory provenance
    val: NonZeroPtr,
}
unsafe impl Send for BitList {}
unsafe impl Sync for BitList {}

//assert sizes of struct, and niche optimizations
const _: () = assert!(size_of::<BitList>() == size_of::<HeapBitList>());
const _: () = assert!(size_of::<BitList>() == size_of::<InlineBitList>());
const _: () = assert!(align_of::<BitList>() >= align_of::<HeapBitList>());
const _: () = assert!(align_of::<BitList>() >= align_of::<InlineBitList>());
const _: () = assert!(size_of::<BitList>() == size_of::<Option<BitList>>());
const _: () = assert!(size_of::<InlineBitList>() == size_of::<Option<InlineBitList>>());
const _: () = assert!(size_of::<HeapBitList>() == size_of::<Option<HeapBitList>>());
const _: () = assert!(size_of::<BitList>() >= 4);
const _: () = assert!(align_of::<usize>() >= 2);

impl BitList {
    #[inline]
    pub(crate) const fn from_inline(value: InlineBitList) -> Self {
        // SAFETY: we know that InlineBitList is transparent, and it must have INLINE_FLAG set so is never zero
        unsafe { transmute::<InlineBitList, BitList>(value) }
    }
    #[inline]
    pub(crate) fn from_heap(value: HeapBitList) -> Self {
        // SAFETY: we know that HeapBitList is transparent, and it must have INLINE_FLAG cleared, also pointer is never zero
        unsafe { transmute::<HeapBitList, BitList>(value) }
    }
    #[inline]
    pub(crate) const fn ref_inline(value: &InlineBitList) -> &Self {
        // SAFETY: we know that InlineBitList is transparent, and it must have INLINE_FLAG set so is never zero
        unsafe { transmute::<&InlineBitList, &BitList>(value) }
    }
    #[inline]
    pub(crate) fn ref_heap(value: &HeapBitList) -> &Self {
        // SAFETY: we know that HeapBitList is transparent, and it must have INLINE_FLAG cleared, also pointer is never zero
        unsafe { transmute::<&HeapBitList, &BitList>(value) }
    }
    #[inline]
    pub(crate) fn from_inner(inner: Repr) -> Self {
        match inner {
            Repr::Heap(v) => Self::from_heap(v),
            Repr::Inline(v) => Self::from_inline(v),
        }
    }
    #[inline]
    pub const fn is_inline(&self) -> bool {
        // SAFETY: NonNull<usize> has same alignment and size as usize
        // CONST SAFETY: Const values can only be created from inline repr, which is a integer
        // so since it's never and address, we can transmute it to type from which it came from
        let val = unsafe { transmute::<NonZeroPtr, usize>(self.val) };
        val & InlineBitList::INLINE_FLAG != 0
    }
    #[inline]
    pub const fn by_ref(&self) -> &Self {
        self
    }
    #[inline]
    pub(crate) const fn inner(&self) -> ReprRef<'_> {
        // SAFETY: transmuting references here is safe cause both types have #[repr(transparent)]
        // with same size and alignment as NonZeroUsize, or just usize
        // then we check upfront which representation is it
        // CONST SAFETY: Const values can only be created from inline repr, which is a integer
        // so since it's never and address, we can transmute it to type from which it came from
        if self.is_inline() {
            //this is transmuted as value, cause another pointer indirection would be wasteful
            ReprRef::Inline(unsafe { transmute::<NonZeroPtr, InlineBitList>(self.val) })
        } else {
            ReprRef::Heap(unsafe { transmute::<&BitList, &HeapBitList>(self) })
        }
    }
    #[inline]
    pub(crate) const fn inner_by_ref(&self) -> ReprByRef<'_> {
        // SAFETY: transmuting references here is safe cause both types have #[repr(transparent)]
        // with same size and alignment as NonZeroUsize, or just usize
        // then we check upfront which representation is it
        // CONST SAFETY: Const values can only be created from inline repr, which is a integer
        // so since it's never and address, we can transmute it to type from which it came from
        if self.is_inline() {
            //this is transmuted as value, cause another pointer indirection would be wasteful
            ReprByRef::Inline(unsafe { transmute::<&BitList, &InlineBitList>(self) })
        } else {
            ReprByRef::Heap(unsafe { transmute::<&BitList, &HeapBitList>(self) })
        }
    }
    #[inline]
    pub(crate) const fn inner_mut(&mut self) -> ReprMut<'_> {
        // SAFETY: transmuting references here is safe cause both types have #[repr(transparent)]
        // with same size and alignment as NonZeroUsize, or just usize
        // then we check upfront which representation is it
        if self.is_inline() {
            ReprMut::Inline(unsafe { transmute::<&mut BitList, &mut InlineBitList>(self) })
        } else {
            ReprMut::Heap(unsafe { transmute::<&mut BitList, &mut HeapBitList>(self) })
        }
    }
    ///transform inline to heap, without running any drop check
    #[inline]
    pub(crate) unsafe fn set_heap(&mut self, list: HeapBitList) {
        debug_assert!(self.is_inline());
        self.val = unsafe { transmute::<HeapBitList, NonZeroPtr>(list) };
    }
    #[inline]
    pub(crate) const fn into_inner(self) -> Repr {
        // SAFETY: transmuting values here is safe cause both types have #[repr(transparent)]
        // with same size and alignment as NonZeroUsize, or just usize
        // then we check upfront which representation is it
        // CONST SAFETY: Const values can only be created from inline repr, which is a integer
        // so since it's never and address, we can transmute it to type from which it came from
        if self.is_inline() {
            Repr::Inline(unsafe { transmute::<BitList, InlineBitList>(self) })
        } else {
            Repr::Heap(unsafe { transmute::<BitList, HeapBitList>(self) })
        }
    }

    // pub const fn len(&self) -> usize { (actually latest tests on rust 1.93 shows that hand optimized len is same as this)
    //     match self.inner() {
    //         ReprRef::Heap(v) => v.len(),
    //         ReprRef::Inline(v) => v.len(),
    //     }
    // }
    pub const fn len(&self) -> usize {
        //this has better assembly than normal len (2 instructions less = in total 8 ops)
        let ptr = self.val;
        let val: usize = unsafe { transmute::<NonZeroPtr, usize>(ptr) };
        if (val) & InlineBitList::INLINE_FLAG != 0 {
            (val & InlineBitList::MASK_COUNT_BITS).wrapping_shr(InlineBitList::COUNT_SHIFT)
        } else {
            unsafe { ptr.read() }
        }
    }

    // pub const fn is_empty(&self) -> bool {
    //     match self.inner() {
    //         ReprRef::Heap(v) => v.is_empty(),
    //         ReprRef::Inline(v) => v.is_empty(),
    //     }
    // }
    pub const fn is_empty(&self) -> bool {
        //this has better assembly than normal is_empty (2 instructions less = in total 11 ops)
        let ptr = self.val;
        let val: usize = unsafe { transmute::<NonZeroPtr, usize>(ptr) };
        if (val) & InlineBitList::INLINE_FLAG != 0 {
            (val & InlineBitList::MASK_COUNT_BITS) == 0
        } else {
            unsafe { ptr.read() == 0 }
        }
    }
}
#[derive(Clone, Eq, PartialEq)]
pub enum Repr {
    Heap(HeapBitList),
    Inline(InlineBitList),
}
#[derive(Copy, Clone, Eq, PartialEq)]
pub enum ReprRef<'a> {
    Heap(&'a HeapBitList),
    Inline(InlineBitList),
}
#[derive(Copy, Clone, Eq, PartialEq)]
pub enum ReprByRef<'a> {
    Heap(&'a HeapBitList),
    Inline(&'a InlineBitList),
}
#[derive(Eq, PartialEq)]
pub enum ReprMut<'a> {
    Heap(&'a mut HeapBitList),
    Inline(&'a mut InlineBitList),
}

impl Clone for BitList {
    fn clone(&self) -> Self {
        match self.inner() {
            ReprRef::Inline(v) => Self::from_inline(v),
            ReprRef::Heap(v) => {
                // clone is an opportunity to trim capacity, and also
                // reduce to no allocation at all
                let len = v.len();
                if len <= Self::MAX_INLINE_BITS {
                    //downgrade to inline, in case of small amount of bits
                    Self::from_inline(InlineBitList::new(v.first_word_init().unwrap_or(0), len as _))
                } else {
                    Self::from_heap(v.clone())
                }
            }
        }
    }
}

impl PartialEq for BitList {
    fn eq(&self, other: &Self) -> bool {
        match (self.inner(), other.inner()) {
            (ReprRef::Inline(a), ReprRef::Inline(b)) => a == b,
            (ReprRef::Heap(a), ReprRef::Heap(b)) => a == b,
            (ReprRef::Inline(a), ReprRef::Heap(b)) | (ReprRef::Heap(b), ReprRef::Inline(a)) => {
                let len = a.len();
                if len != b.len() {
                    return false;
                }
                if len != 0 {
                    //unwrap should never be reached
                    return a.data() == b.first_word_init().unwrap_or(0);
                }
                true
            }
        }
    }
}
impl Eq for BitList {}
impl Hash for BitList {
    fn hash<H: Hasher>(&self, state: &mut H) {
        match self.inner() {
            ReprRef::Heap(v) => {
                state.write_usize(v.len());
                for v in v.init_data() {
                    state.write_usize(*v);
                }
            }
            ReprRef::Inline(v) => {
                state.write_usize(v.len());
                state.write_usize(v.data());
            }
        }
    }
}

impl Drop for BitList {
    fn drop(&mut self) {
        if !self.is_inline() {
            unsafe {
                drop(transmute::<NonZeroPtr, HeapBitList>(self.val));
            }
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use rand::prelude::*;

    #[test]
    fn test_list_eq() {
        let rng = &mut StdRng::seed_from_u64(1234566543);
        for _ in 0..1000 {
            let to_push = rng.random_range(0..100);
            let to_pop = rng.random_range(0..100);
            let mut list = BitList::NO_BITS;
            for _ in 0..to_push {
                list.push_bit(rng.random_bool(0.5));
            }
            for _ in 0..to_pop {
                list.pop_bit();
            }
            let bits = list.iter().collect::<Vec<_>>();
            let cmp_list = BitList::from_bits(bits.iter().copied());
            assert_eq!(list, cmp_list);
        }
    }
}