fqgrep_lib/
mod.rs

1#![deny(unsafe_code)]
2#![allow(
3    clippy::must_use_candidate,
4    clippy::missing_panics_doc,
5    clippy::missing_errors_doc,
6    clippy::module_name_repetitions
7)]
8pub mod color;
9pub mod matcher;
10use lazy_static::lazy_static;
11use std::{borrow::Borrow, path::Path};
12
13pub const DNA_BASES: [u8; 5] = *b"ACGTN";
14pub const IUPAC_BASES: [u8; 15] = *b"AGCTYRWSKMDVHBN";
15pub const IUPAC_BASES_COMPLEMENT: [u8; 15] = *b"TCGARYWSMKHBDVN";
16
17lazy_static! {
18    pub static ref COMPLEMENT: [u8; 256] = {
19        let mut comp = [0; 256];
20        for (v, a) in comp.iter_mut().enumerate() {
21            *a = v as u8;
22        }
23        for (&a, &b) in IUPAC_BASES.iter().zip(IUPAC_BASES_COMPLEMENT.iter()) {
24            comp[a as usize] = b;
25            comp[a as usize + 32] = b + 32;  // lowercase variants
26        }
27        comp
28    };
29}
30
31fn complement(a: u8) -> u8 {
32    COMPLEMENT[a as usize]
33}
34
35fn reverse_complement<C, T>(text: T) -> Vec<u8>
36where
37    C: Borrow<u8>,
38    T: IntoIterator<Item = C>,
39    T::IntoIter: DoubleEndedIterator,
40{
41    text.into_iter()
42        .rev()
43        .map(|a| complement(*a.borrow()))
44        .collect()
45}
46
47/// Returns true if the path ends with a recognized GZIP file extension
48fn is_path_with_extension<P: AsRef<Path>>(p: &P, extensions: [&str; 2]) -> bool {
49    if let Some(ext) = p.as_ref().extension() {
50        match ext.to_str() {
51            Some(x) => extensions.contains(&x),
52            None => false,
53        }
54    } else {
55        false
56    }
57}
58
59/// The set of file extensions to treat as GZIPPED
60const GZIP_EXTENSIONS: [&str; 2] = ["gz", "bgz"];
61
62/// Returns true if the path ends with a recognized GZIP file extension
63pub fn is_gzip_path<P: AsRef<Path>>(p: &P) -> bool {
64    is_path_with_extension(p, GZIP_EXTENSIONS)
65}
66
67/// The set of file extensions to treat as FASTQ
68const FASTQ_EXTENSIONS: [&str; 2] = ["fastq", "fq"];
69
70/// Returns true if the path ends with a recognized FASTQ file extension
71pub fn is_fastq_path<P: AsRef<Path>>(p: &P) -> bool {
72    is_path_with_extension(p, FASTQ_EXTENSIONS)
73}
74
75// Tests
76#[cfg(test)]
77pub mod tests {
78    use crate::*;
79    use rstest::rstest;
80    use std::str;
81    use tempfile::TempDir;
82
83    // ############################################################################################
84    // Tests reverse_complement()
85    // ############################################################################################
86
87    #[rstest]
88    #[case("ACGT", "ACGT")] // Reverse complement with even length string
89    #[case("ACG", "CGT")] // Reverse complement with odd length string (tests for off by one error)
90    fn test_reverse_complement(#[case] seq: &str, #[case] expected: &str) {
91        let result = reverse_complement(seq.as_bytes());
92        let string_result = str::from_utf8(&result).unwrap();
93        assert_eq!(&string_result, &expected);
94    }
95
96    // ############################################################################################
97    // Tests is_gzip_path()
98    // ############################################################################################
99
100    #[rstest]
101    #[case("test_fastq.fq.gz", true)] // .fq.gz is valid gzip
102    #[case("test_fastq.fq.bgz", true)] // .fq.bgz is valid gzip
103    #[case("test_fastq.fq.tar", false)] // .fq.tar is invalid gzip
104    fn test_is_gzip_path(#[case] file_name: &str, #[case] expected: bool) {
105        let dir = TempDir::new().unwrap();
106        let file_path = dir.path().join(file_name);
107        let result = is_gzip_path(&file_path);
108        assert_eq!(result, expected);
109    }
110    // ############################################################################################
111    // Tests is_fastq_path()
112    // ############################################################################################
113
114    #[rstest]
115    #[case("test_fastq.fq", true)] // .fq is valid fastq
116    #[case("test_fastq.fastq", true)] // .fastq is valid fastq
117    #[case("test_fastq.sam", false)] // .sam is invalid fastq
118    fn test_is_fastq_path(#[case] file_name: &str, #[case] expected: bool) {
119        let dir = TempDir::new().unwrap();
120        let file_path = dir.path().join(file_name);
121        let result = is_fastq_path(&file_path);
122        assert_eq!(result, expected);
123    }
124}
fqgrep_lib/mod.rs

fqgrep_lib/
mod.rs
