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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703
//! Operator trait implementations. use super::api::BitSliceIndex; use crate::{ access::BitAccess, indices::Indexable, order::BitOrder, slice::BitSlice, store::BitStore, }; use either::Either; use core::{ ops::{ AddAssign, BitAndAssign, BitOrAssign, BitXorAssign, Index, IndexMut, Neg, Not, Range, RangeFrom, RangeFull, RangeInclusive, RangeTo, RangeToInclusive, ShlAssign, ShrAssign, }, ptr, }; /** Performs unsigned addition in place on a `BitSlice`. If the addend bitstream is shorter than `self`, the addend is zero-extended at the left (so that its final bit matches with `self`’s final bit). If the addend is longer, the excess front length is unused. Addition proceeds from the right ends of each slice towards the left. Because this trait is forbidden from returning anything, the final carry-out bit is discarded. Note that, unlike `BitVec`, there is no subtraction implementation until I find a subtraction algorithm that does not require modifying the subtrahend. Subtraction can be implemented by negating the intended subtrahend yourself and then using addition, or by using `BitVec`s instead of `BitSlice`s. # Type Parameters - `I: IntoIterator<Item=bool, IntoIter: DoubleEndedIterator>`: The bitstream to add into `self`. It must be finite and double-ended, since addition operates in reverse. **/ impl<O, T, I> AddAssign<I> for BitSlice<O, T> where O: BitOrder, T: BitStore, I: IntoIterator<Item=bool>, I::IntoIter: DoubleEndedIterator { /// Performs unsigned wrapping addition in place. /// /// # Examples /// /// This example shows addition of a slice wrapping from max to zero. /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = [0b1110_1111u8, 0b0000_0001]; /// let bits = src.bits_mut::<Msb0>(); /// let (nums, one) = bits.split_at_mut(12); /// let (accum, steps) = nums.split_at_mut(4); /// *accum += one.iter().copied(); /// assert_eq!(accum, &steps[.. 4]); /// *accum += one.iter().copied(); /// assert_eq!(accum, &steps[4 ..]); /// ``` // Clippy doesn’t like single-letter names (which is accurate) but this is // pretty standard mathematical notation in EE. #[allow(clippy::many_single_char_names)] fn add_assign(&mut self, addend: I) { use core::iter::repeat; // I don't, at this time, want to implement a carry-lookahead adder in // software, so this is going to be a plain ripple-carry adder with // O(n) runtime. Furthermore, until I think of an optimization // strategy, it is going to build up another bitvec to use as a stack. // // Computers are fast. Whatever. let mut c = false; // Reverse self, reverse addend and zero-extend, and zip both together. // This walks both slices from rightmost to leftmost, and considers an // early expiration of addend to continue with 0 bits. // // 100111 // + 0010 // ^^---- semantically zero let addend_iter = addend.into_iter().rev().chain(repeat(false)); for (i, b) in (0 .. self.len()).rev().zip(addend_iter) { // Bounds checks are performed in the loop header. let a = unsafe { *self.get_unchecked(i) }; let (y, z) = crate::rca1(a, b, c); unsafe { self.set_unchecked(i, y); } c = z; } } } /** Performs the Boolean `AND` operation against another bitstream and writes the result into `self`. If the other bitstream ends before `self,`, the remaining bits of `self` are cleared. # Type Parameters - `I: IntoIterator<Item=bool>`: A stream of bits, which may be a `BitSlice` or some other bit producer as desired. **/ impl<O, T, I> BitAndAssign<I> for BitSlice<O, T> where O: BitOrder, T: BitStore, I: IntoIterator<Item=bool> { /// `AND`s a bitstream into a slice. /// /// # Parameters /// /// - `&mut self` /// - `rhs`: The bitstream to `AND` into `self`. /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut store = [0b0101_0100u8]; /// let other = [0b0011_0000u8]; /// let lhs = store.bits_mut::<Msb0>(); /// let rhs = other.bits::<Msb0>(); /// lhs[.. 6] &= rhs[.. 4].iter().copied(); /// assert_eq!(store[0], 0b0001_0000); /// ``` fn bitand_assign(&mut self, rhs: I) { use core::iter; rhs.into_iter() .chain(iter::repeat(false)) .enumerate() .take(self.len()) .for_each(|(idx, bit)| unsafe { let val = *self.get_unchecked(idx); self.set_unchecked(idx, val & bit); }); } } /** Performs the Boolean `OR` operation against another bitstream and writes the result into `self`. If the other bitstream ends before `self`, the remaining bits of `self` are not affected. # Type Parameters - `I: IntoIterator<Item=bool>`: A stream of bits, which may be a `BitSlice` or some other bit producer as desired. **/ impl<O, T, I> BitOrAssign<I> for BitSlice<O, T> where O: BitOrder, T: BitStore, I: IntoIterator<Item=bool> { /// `OR`s a bitstream into a slice. /// /// # Parameters /// /// - `&mut self` /// - `rhs`: The bitstream to `OR` into `self`. /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut store = [0b0101_0100u8]; /// let other = [0b0011_0000u8]; /// let lhs = store.bits_mut::<Msb0>(); /// let rhs = other.bits::<Msb0>(); /// lhs[.. 6] |= rhs[.. 4].iter().copied(); /// assert_eq!(store[0], 0b0111_0100); /// ``` fn bitor_assign(&mut self, rhs: I) { rhs.into_iter() .enumerate() .take(self.len()) .for_each(|(idx, bit)| unsafe { let val = *self.get_unchecked(idx); self.set_unchecked(idx, val | bit); }); } } /** Performs the Boolean `XOR` operation against another bitstream and writes the result into `self`. If the other bitstream ends before `self`, the remaining bits of `self` are not affected. # Type Parameters - `I: IntoIterator<Item=bool>`: A stream of bits, which may be a `BitSlice` or some other bit producer as desired. **/ impl<O, T, I> BitXorAssign<I> for BitSlice<O, T> where O: BitOrder, T: BitStore, I: IntoIterator<Item=bool> { /// `XOR`s a bitstream into a slice. /// /// # Parameters /// /// - `&mut self` /// - `rhs`: The bitstream to `XOR` into `self`. /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut store = [0b0101_0100u8]; /// let other = [0b0011_0000u8]; /// let lhs = store.bits_mut::<Msb0>(); /// let rhs = other.bits::<Msb0>(); /// lhs[.. 6] ^= rhs[.. 4].iter().copied(); /// assert_eq!(store[0], 0b0110_0100); /// ``` fn bitxor_assign(&mut self, rhs: I) { rhs.into_iter() .enumerate() .take(self.len()) .for_each(|(idx, bit)| unsafe { let val = *self.get_unchecked(idx); self.set_unchecked(idx, val ^ bit); }) } } impl<O, T> Index<usize> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = bool; fn index(&self, place: usize) -> &Self::Output { place.index(self) } } impl<O, T> Index<Range<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, range: Range<usize>) -> &Self { range.index(self) } } impl<O, T> IndexMut<Range<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, range: Range<usize>) -> &mut Self { range.index_mut(self) } } impl<O, T> Index<RangeInclusive<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, range: RangeInclusive<usize>) -> &Self { range.index(self) } } impl<O, T> IndexMut<RangeInclusive<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, range: RangeInclusive<usize>) -> &mut Self { range.index_mut(self) } } impl<O, T> Index<RangeFrom<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, range: RangeFrom<usize>) -> &Self { range.index(self) } } impl<O, T> IndexMut<RangeFrom<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, range: RangeFrom<usize>) -> &mut Self { range.index_mut(self) } } impl<O, T> Index<RangeFull> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, _: RangeFull) -> &Self { self } } impl<O, T> IndexMut<RangeFull> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, _: RangeFull) -> &mut Self { self } } impl<O, T> Index<RangeTo<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, range: RangeTo<usize>) -> &Self { range.index(self) } } impl<O, T> IndexMut<RangeTo<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, range: RangeTo<usize>) -> &mut Self { range.index_mut(self) } } impl<O, T> Index<RangeToInclusive<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { type Output = Self; fn index(&self, range: RangeToInclusive<usize>) -> &Self { range.index(self) } } impl<O, T> IndexMut<RangeToInclusive<usize>> for BitSlice<O, T> where O: BitOrder, T: BitStore { fn index_mut(&mut self, range: RangeToInclusive<usize>) -> &mut Self { range.index_mut(self) } } /** Performs fixed-width 2’s-complement negation of a `BitSlice`. Unlike the `!` operator (`Not` trait), the unary `-` operator treats the `BitSlice` as if it represents a signed 2’s-complement integer of fixed width. The negation of a number in 2’s complement is defined as its inversion (using `!`) plus one, and on fixed-width numbers has the following discontinuities: - A slice whose bits are all zero is considered to represent the number zero which negates as itself. - A slice whose bits are all one is considered to represent the most negative number, which has no correpsonding positive number, and thus negates as zero. This behavior was chosen so that all possible values would have *some* output, and so that repeated application converges at idempotence. The most negative input can never be reached by negation, but `--MOST_NEG` converges at the least unreasonable fallback value, 0. Because `BitSlice` cannot move, the negation is performed in place. **/ impl<'a, O, T> Neg for &'a mut BitSlice<O, T> where O: BitOrder, T: 'a + BitStore { type Output = Self; /// Perform 2’s-complement fixed-width negation. /// /// Negation is accomplished by inverting the bits and adding one. This has /// one edge case: `1000…`, the most negative number for its width, will /// negate to zero instead of itself. It thas no corresponding positive /// number to which it can negate. /// /// # Parameters /// /// - `self` /// /// # Examples /// /// The contortions shown here are a result of this operator applying to a /// mutable reference, and this example balancing access to the original /// `BitVec` for comparison with aquiring a mutable borrow *as a slice* to /// ensure that the `BitSlice` implementation is used, not the `BitVec`. /// /// Negate an arbitrary positive number (first bit unset). /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = 0b0110_1010u8; /// let bits = src.bits_mut::<Msb0>(); /// eprintln!("{:?}", bits.split_at(4)); /// let num = &mut bits[.. 4]; /// -num; /// eprintln!("{:?}", bits.split_at(4)); /// assert_eq!(&bits[.. 4], &bits[4 ..]); /// ``` /// /// Negate an arbitrary negative number. This example will use the above /// result to demonstrate round-trip correctness. /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = 0b1010_0110u8; /// let bits = src.bits_mut::<Msb0>(); /// let num = &mut bits[.. 4]; /// -num; /// assert_eq!(&bits[.. 4], &bits[4 ..]); /// ``` /// /// Negate the most negative number, which will become zero, and show /// convergence at zero. /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = 128u8; /// let bits = src.bits_mut::<Msb0>(); /// let num = &mut bits[..]; /// -num; /// assert!(bits.not_any()); /// let num = &mut bits[..]; /// -num; /// assert!(bits.not_any()); /// ``` fn neg(self) -> Self::Output { // negative zero is zero. The invert-and-add will result in zero, but // this case can be detected quickly. if self.is_empty() || self.not_any() { return self; } // The most negative number (leading one, all zeroes else) negates to // zero. if unsafe { *self.get_unchecked(0) } { // Testing the whole range, rather than [1 ..], is more likely to // hit the fast path for `not_any`. unsafe { self.set_unchecked(0, false); } if self.not_any() { return self; } unsafe { self.set_unchecked(0, true); } } let this = !self; *this += core::iter::once(true); this } } /// Flips all bits in the slice, in place. impl<'a, O, T> Not for &'a mut BitSlice<O, T> where O: BitOrder, T: 'a + BitStore { type Output = Self; /// Inverts all bits in the slice. /// /// This will not affect bits outside the slice in slice storage elements. /// /// # Parameters /// /// - `self` /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = [0u8; 2]; /// let bits = &mut src.bits_mut::<Msb0>()[2 .. 14]; /// let _ = !bits; /// // The `bits` binding is consumed by the `!` operator, and a new /// // reference is returned. /// // assert_eq!(bits.as_ref(), &[!0, !0]); /// assert_eq!(src, [0x3F, 0xFC]); /// ``` fn not(self) -> Self::Output { match self.bitptr().domain().splat() { Either::Right((h, e, t)) => for n in *h .. *t { e.invert_bit::<O>(n.idx()); }, Either::Left((h, b, t)) => { if let Some((h, head)) = h { for n in *h .. T::BITS { head.invert_bit::<O>(n.idx()) } } if let Some(body) = b { for elt in body { elt.store(!elt.load()); } } if let Some((tail, t)) = t { for n in 0 .. *t { tail.invert_bit::<O>(n.idx()) } } }, } self } } __bitslice_shift!(u8, u16, u32, u64, i8, i16, i32, i64); /** Shifts all bits in the array to the left — **DOWN AND TOWARDS THE FRONT**. On fundamentals, the left-shift operator `<<` moves bits away from the origin and towards the ceiling. This is because we label the bits in a primitive with the minimum on the right and the maximum on the left, which is big-endian bit order. This increases the value of the primitive being shifted. **THAT IS NOT HOW `BitSlice` WORKS!** `BitSlice` defines its layout with the minimum on the left and the maximum on the right! Thus, left-shifting moves bits towards the **minimum**. In `Msb0` order, the effect in memory will be what you expect the `<<` operator to do. **In `Lsb0` order, the effect will be equivalent to using `>>` on the** **fundamentals in memory!** # Notes In order to preserve the effecs in memory that this operator traditionally expects, the bits that are emptied by this operation are zeroed rather than left to their old value. The shift amount is modulated against the array length, so it is not an error to pass a shift amount greater than the array length. A shift amount of zero is a no-op, and returns immediately. **/ impl<O, T> ShlAssign<usize> for BitSlice<O, T> where O: BitOrder, T: BitStore { /// Shifts a slice left, in place. /// /// # Parameters /// /// - `&mut self` /// - `shamt`: The shift amount. If this is greater than the length, then /// the slice is zeroed immediately. /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = [0x4Bu8, 0xA5]; /// let bits = &mut src.bits_mut::<Msb0>()[2 .. 14]; /// *bits <<= 3; /// assert_eq!(src, [0b01_011_101, 0b001_000_01]); /// ``` #[allow(clippy::suspicious_op_assign_impl)] // These functions require math fn shl_assign(&mut self, shamt: usize) { if shamt == 0 { return; } let len = self.len(); if shamt >= len { self.set_all(false); return; } // If the slice fully owns its memory, then a fast path is available // with element-wise `memmove`. if self.bitptr().domain().is_spanning() { // Compute the shift distance measured in elements. let offset = shamt >> T::INDX; // Compute the number of elements that will remain. let rem = self.bitptr().elements() - offset; /* Memory model: suppose we have this slice of sixteen elements, that is shifted five elements to the left. We have three pointers and two lengths to manage. - rem is 11 (len - offset) - offset is 5 - to is &[0 .. 11] - from is &[5 .. 16] - tail is &[11] _ _ _ _ _ v-------before------v [ 0 1 2 3 4 5 6 7 8 9 a b c d e f ] ^-------after-------^ 0 0 0 0 0 */ // Pointer to the front of the slice. let to: *mut T = self.as_mut_ptr(); // Pointer to the front of the section that will move and be // retained. let from: *const T = &self.as_slice()[offset]; // Pointer to the back of the slice that will be zero-filled. let tail: *mut T = &mut self.as_mut_slice()[rem]; unsafe { ptr::copy(from, to, rem); ptr::write_bytes(tail, 0, offset); } // Any remaining shift amount only needs to shift the `after` block // above. self[.. rem << T::INDX] <<= shamt & T::INDX as usize; return; } // Otherwise, crawl. for (to, from) in (shamt .. len).enumerate() { unsafe { self.copy_unchecked(from, to); } } self[len - shamt ..].set_all(false); } } /** Shifts all bits in the array to the right — **UP AND TOWARDS THE BACK**. On fundamentals, the right-shift operator `>>` moves bits towards the origin and away from the ceiling. This is because we label the bits in a primitive with the minimum on the right and the maximum on the left, which is big-endian bit order. This decreases the value of the primitive being shifted. **THAT IS NOT HOW `BitSlice` WORKS!** `BitSlice` defines its layout with the minimum on the left and the maximum on the right! Thus, right-shifting moves bits towards the **maximum**. In `Msb0` order, the effect in memory will be what you expect the `>>` operator to do. **In `Lsb0` order, the effect will be equivalent to using `<<` on the** **fundamentals in memory!** # Notes In order to preserve the effects in memory that this operator traditionally expects, the bits that are emptied by this operation are zeroed rather than left to their old value. The shift amount is modulated against the array length, so it is not an error to pass a shift amount greater than the array length. A shift amount of zero is a no-op, and returns immediately. **/ impl<O, T> ShrAssign<usize> for BitSlice<O, T> where O: BitOrder, T: BitStore { /// Shifts a slice right, in place. /// /// # Parameters /// /// - `&mut self` /// - `shamt`: The shift amount. If this is greater than the length, then /// the slice is zeroed immediately. /// /// # Examples /// /// ```rust /// use bitvec::prelude::*; /// /// let mut src = [0x4Bu8, 0xA5]; /// let bits = &mut src.bits_mut::<Msb0>()[2 .. 14]; /// *bits >>= 3; /// assert_eq!(src, [0b01_000_00_1, 0b011_101_01]) /// ``` #[allow(clippy::suspicious_op_assign_impl)] // These functions require math fn shr_assign(&mut self, shamt: usize) { if shamt == 0 { return; } let len = self.len(); if shamt >= len { self.set_all(false); return; } // If the slice fully owns its memory, then a fast path is available // with element-wise `memmove`. if self.bitptr().domain().is_spanning() { // Compute the shift amount measured in elements. let offset = shamt >> T::INDX; // Compute the number of elements that will remain. let rem = self.bitptr().elements() - offset; /* Memory model: suppose we have this slice of sixteen elements, that is shifted five elements to the right. We have two pointers and two lengths to manage. - rem is 11 (len - offset) - offset is 5 - from is &[0 .. 11] - to is &[5 .. 16] v-------before------v [ 0 1 2 3 4 5 6 7 8 9 a b c d e f ] 0 0 0 0 0 ^-------after-------^ */ let from: *mut T = self.as_mut_ptr(); let to: *mut T = &mut self.as_mut_slice()[offset]; unsafe { ptr::copy(from, to, rem); ptr::write_bytes(from, 0, offset); } // Any remaining shift amount only needs to shift the `after` block // above. self[offset << T::INDX ..] >>= shamt & T::INDX as usize; return; } // Otherwise, crawl. for (from, to) in (shamt .. len).enumerate().rev() { unsafe { self.copy_unchecked(from, to); } } self[.. shamt].set_all(false); } }