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.len() <= usize::BITS as usize {
36            Ok(slice.bs.load_le::<usize>())
37        } else {
38            let len: usize = slice.bs.len() / A::BITS as usize;
39            let maximum: usize = usize::BITS as usize / A::BITS as usize;
40            Err(ParseBioError::SequenceTooLong(maximum, len))
41        }
42    }
43}
44
45impl<A: Codec> From<&SeqSlice<A>> for u8 {
46    fn from(slice: &SeqSlice<A>) -> u8 {
47        debug_assert!(slice.bs.len() <= u8::BITS as usize);
48        slice.bs.load_le::<u8>()
49    }
50}
51
52impl<A: Codec> SeqSlice<A> {
53    /// unsafely index into the `i`th position of a sequence
54    pub fn nth(&self, i: usize) -> A {
55        A::unsafe_from_bits(self[i].into())
56    }
57
58    pub fn len(&self) -> usize {
59        self.bs.len() / A::BITS as usize
60    }
61
62    /// Get the `i`th element of a `Seq`. Returns `None` if index out of range.
63    pub fn get(&self, i: usize) -> Option<A> {
64        if i >= self.bs.len() / A::BITS as usize {
65            None
66        } else {
67            Some(A::unsafe_from_bits(self[i].into()))
68        }
69    }
70
71    pub fn is_empty(&self) -> bool {
72        self.len() == 0
73    }
74}
75
76impl<A: Codec> From<&SeqSlice<A>> for String {
77    fn from(seq: &SeqSlice<A>) -> Self {
78        seq.into_iter().map(Codec::to_char).collect()
79    }
80}
81
82impl<A: Codec> PartialEq<SeqSlice<A>> for SeqSlice<A> {
83    fn eq(&self, other: &SeqSlice<A>) -> bool {
84        self.bs == other.bs
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<Seq<A>> for SeqSlice<A> {
95    fn eq(&self, other: &Seq<A>) -> bool {
96        self == other.as_ref()
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<&str> for SeqSlice<A> {
107    fn eq(&self, other: &&str) -> bool {
108        let bs = other.as_bytes();
109        if bs.len() != self.len() {
110            return false;
111        }
112        for (a, c) in self.iter().zip(bs) {
113            match A::try_from_ascii(*c) {
114                Some(b) => {
115                    if a != b {
116                        return false;
117                    }
118                }
119                None => return false,
120            }
121        }
122        true
123    }
124}
125
126impl<A: Codec> Hash for SeqSlice<A> {
127    fn hash<H: Hasher>(&self, state: &mut H) {
128        // Prefix with length as 8 little-endian bytes so the byte stream is
129        // identical on 32-/64-bit targets and across endianness. Going through
130        // `u64::hash` would route via `Hasher::write_u64`, whose default impl
131        // emits native-endian bytes.
132        state.write(&(self.len() as u64).to_le_bytes());
133        crate::hash::hash_bits(&self.bs, state);
134    }
135}
136
137/// Clone a borrowed slice of a sequence into an owned version.
138///
139/// ```
140/// use bio_seq::prelude::*;
141///
142/// let seq = dna!("CATCGATCGATCG");
143/// let slice = &seq[2..7]; // TCGAT
144/// let owned = slice.to_owned();
145///
146/// assert_eq!(&owned, &seq[2..7]);
147/// ```
148///
149impl<A: Codec> ToOwned for SeqSlice<A> {
150    type Owned = Seq<A>;
151
152    fn to_owned(&self) -> Self::Owned {
153        Seq {
154            _p: PhantomData,
155            bv: self.bs.into(),
156        }
157    }
158}
159
160impl<A: Codec> AsRef<SeqSlice<A>> for SeqSlice<A> {
161    fn as_ref(&self) -> &SeqSlice<A> {
162        self
163    }
164}
165
166impl<A: Codec> fmt::Display for SeqSlice<A> {
167    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
168        write!(f, "{}", String::from(self))
169    }
170}
171
172impl<A: Codec> BitAnd for &SeqSlice<A> {
173    type Output = Seq<A>;
174
175    fn bitand(self, rhs: Self) -> Self::Output {
176        let mut bv = self.bs.to_bitvec();
177        bv &= &rhs.bs;
178        Seq::<A> {
179            bv,
180            _p: PhantomData,
181        }
182    }
183}
184
185impl<A: Codec> BitOr for &SeqSlice<A> {
186    type Output = Seq<A>;
187
188    fn bitor(self, rhs: Self) -> Self::Output {
189        let mut bv = self.bs.to_bitvec();
190        bv |= &rhs.bs;
191
192        Seq::<A> {
193            bv,
194            _p: PhantomData,
195        }
196    }
197}
198
199impl<A: Codec> ReverseMut for SeqSlice<A> {
200    fn rev(&mut self) {
201        self.bs.reverse();
202        for chunk in self.bs.rchunks_exact_mut(A::BITS as usize) {
203            chunk.reverse();
204        }
205    }
206}
207
208impl<A: Codec + ComplementMut> ComplementMut for SeqSlice<A> {
209    fn comp(&mut self) {
210        unsafe {
211            for base in self.bs.chunks_exact_mut(A::BITS as usize).remove_alias() {
212                let mut bc = A::unsafe_from_bits(base.load_le::<u8>());
213                bc.comp();
214                base.store(bc.to_bits() as usize);
215            }
216        }
217    }
218}
219
220impl<A: Codec + ComplementMut> ReverseComplementMut for SeqSlice<A> where
221    SeqSlice<A>: ComplementMut + ReverseMut
222{
223}
224
225impl<A: Codec> Reverse for SeqSlice<A> {}
226
227impl<A: Codec + ComplementMut> Complement for SeqSlice<A> {}
228
229impl<A: Codec + ComplementMut> ReverseComplement for SeqSlice<A> where
230    SeqSlice<A>: ComplementMut + ReverseMut
231{
232}