Skip to main content

bio_seq/seq/
slice.rs

1// Copyright 2021-2024 Jeff Knaggs
2// Licensed under the MIT license (http://opensource.org/licenses/MIT)
3// This file may not be copied, modified, or distributed
4// except according to those terms.
5
6use crate::codec::Codec;
7use crate::error::ParseBioError;
8use crate::seq::Seq;
9use crate::{
10    Complement, ComplementMut, Reverse, ReverseComplement, ReverseComplementMut, ReverseMut,
11};
12
13use crate::Bs;
14use bitvec::field::BitField;
15
16use core::fmt;
17use core::hash::{Hash, Hasher};
18use core::marker::PhantomData;
19use core::str;
20
21use core::ops::{BitAnd, BitOr};
22
23/// An unsized, read-only window into part of a sequence
24#[derive(Debug, Eq)]
25#[repr(transparent)]
26pub struct SeqSlice<A: Codec> {
27    pub(crate) _p: PhantomData<A>,
28    pub(crate) bs: Bs,
29}
30
31impl<A: Codec> TryFrom<&SeqSlice<A>> for usize {
32    type Error = ParseBioError;
33
34    fn try_from(slice: &SeqSlice<A>) -> Result<usize, Self::Error> {
35        if slice.bs.is_empty() {
36            Ok(0)
37        } else if slice.bs.len() <= usize::BITS as usize {
38            Ok(slice.bs.load_le::<usize>())
39        } else {
40            let len: usize = slice.bs.len() / A::BITS as usize;
41            let maximum: usize = usize::BITS as usize / A::BITS as usize;
42            Err(ParseBioError::SequenceTooLong(maximum, len))
43        }
44    }
45}
46
47impl<A: Codec> From<&SeqSlice<A>> for u8 {
48    fn from(slice: &SeqSlice<A>) -> u8 {
49        debug_assert!(slice.bs.len() <= u8::BITS as usize);
50        if slice.bs.is_empty() {
51            0
52        } else {
53            slice.bs.load_le::<u8>()
54        }
55    }
56}
57
58impl<A: Codec> SeqSlice<A> {
59    /// unsafely index into the `i`th position of a sequence
60    pub fn nth(&self, i: usize) -> A {
61        A::unsafe_from_bits(self[i].into())
62    }
63
64    pub fn len(&self) -> usize {
65        self.bs.len() / A::BITS as usize
66    }
67
68    /// Get the `i`th element of a `Seq`. Returns `None` if index out of range.
69    pub fn get(&self, i: usize) -> Option<A> {
70        if i >= self.bs.len() / A::BITS as usize {
71            None
72        } else {
73            Some(A::unsafe_from_bits(self[i].into()))
74        }
75    }
76
77    pub fn is_empty(&self) -> bool {
78        self.len() == 0
79    }
80}
81
82impl<A: Codec> From<&SeqSlice<A>> for String {
83    fn from(seq: &SeqSlice<A>) -> Self {
84        seq.into_iter().map(Codec::to_char).collect()
85    }
86}
87
88impl<A: Codec> PartialEq<SeqSlice<A>> for SeqSlice<A> {
89    fn eq(&self, other: &SeqSlice<A>) -> bool {
90        self.bs == other.bs
91    }
92}
93
94impl<A: Codec> PartialEq<SeqSlice<A>> for &SeqSlice<A> {
95    fn eq(&self, other: &SeqSlice<A>) -> bool {
96        self.bs == other.bs
97    }
98}
99
100impl<A: Codec> PartialEq<Seq<A>> for SeqSlice<A> {
101    fn eq(&self, other: &Seq<A>) -> bool {
102        self == other.as_ref()
103    }
104}
105
106impl<A: Codec> PartialEq<Seq<A>> for &SeqSlice<A> {
107    fn eq(&self, other: &Seq<A>) -> bool {
108        *self == other.as_ref()
109    }
110}
111
112impl<A: Codec> PartialEq<&str> for SeqSlice<A> {
113    fn eq(&self, other: &&str) -> bool {
114        let bs = other.as_bytes();
115        if bs.len() != self.len() {
116            return false;
117        }
118        for (a, c) in self.iter().zip(bs) {
119            match A::try_from_ascii(*c) {
120                Some(b) => {
121                    if a != b {
122                        return false;
123                    }
124                }
125                None => return false,
126            }
127        }
128        true
129    }
130}
131
132impl<A: Codec> Hash for SeqSlice<A> {
133    fn hash<H: Hasher>(&self, state: &mut H) {
134        // Prefix with length as 8 little-endian bytes so the byte stream is
135        // identical on 32-/64-bit targets and across endianness. Going through
136        // `u64::hash` would route via `Hasher::write_u64`, whose default impl
137        // emits native-endian bytes.
138        state.write(&(self.len() as u64).to_le_bytes());
139        crate::hash::hash_bits(&self.bs, state);
140    }
141}
142
143/// Clone a borrowed slice of a sequence into an owned version.
144///
145/// ```
146/// use bio_seq::prelude::*;
147///
148/// let seq = dna!("CATCGATCGATCG");
149/// let slice = &seq[2..7]; // TCGAT
150/// let owned = slice.to_owned();
151///
152/// assert_eq!(&owned, &seq[2..7]);
153/// ```
154///
155impl<A: Codec> ToOwned for SeqSlice<A> {
156    type Owned = Seq<A>;
157
158    fn to_owned(&self) -> Self::Owned {
159        Seq {
160            _p: PhantomData,
161            bv: self.bs.into(),
162        }
163    }
164}
165
166impl<A: Codec> AsRef<SeqSlice<A>> for SeqSlice<A> {
167    fn as_ref(&self) -> &SeqSlice<A> {
168        self
169    }
170}
171
172impl<A: Codec> fmt::Display for SeqSlice<A> {
173    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
174        for symbol in self {
175            write!(f, "{}", symbol.to_char())?;
176        }
177        Ok(())
178    }
179}
180
181impl<A: Codec> BitAnd for &SeqSlice<A> {
182    type Output = Seq<A>;
183
184    fn bitand(self, rhs: Self) -> Self::Output {
185        let mut bv = self.bs.to_bitvec();
186        bv &= &rhs.bs;
187        Seq::<A> {
188            bv,
189            _p: PhantomData,
190        }
191    }
192}
193
194impl<A: Codec> BitOr for &SeqSlice<A> {
195    type Output = Seq<A>;
196
197    fn bitor(self, rhs: Self) -> Self::Output {
198        let mut bv = self.bs.to_bitvec();
199        bv |= &rhs.bs;
200
201        Seq::<A> {
202            bv,
203            _p: PhantomData,
204        }
205    }
206}
207
208impl<A: Codec> ReverseMut for SeqSlice<A> {
209    fn rev(&mut self) {
210        self.bs.reverse();
211        for chunk in self.bs.rchunks_exact_mut(A::BITS as usize) {
212            chunk.reverse();
213        }
214    }
215}
216
217impl<A: Codec + ComplementMut> ComplementMut for SeqSlice<A> {
218    fn comp(&mut self) {
219        unsafe {
220            for base in self.bs.chunks_exact_mut(A::BITS as usize).remove_alias() {
221                let mut bc = A::unsafe_from_bits(base.load_le::<u8>());
222                bc.comp();
223                base.store(bc.to_bits() as usize);
224            }
225        }
226    }
227}
228
229impl<A: Codec + ComplementMut> ReverseComplementMut for SeqSlice<A> where
230    SeqSlice<A>: ComplementMut + ReverseMut
231{
232}
233
234impl<A: Codec> Reverse for SeqSlice<A> {}
235
236impl<A: Codec + ComplementMut> Complement for SeqSlice<A> {}
237
238impl<A: Codec + ComplementMut> ReverseComplement for SeqSlice<A> where
239    SeqSlice<A>: ComplementMut + ReverseMut
240{
241}