blob: 70a979c6d445c9207d963a2c4db13ba3d69dfd58 [file] [log] [blame]
Chih-Hung Hsiehe42c5052020-04-16 10:44:21 -07001// The Computer Language Benchmarks Game
Haibo Huang47619dd2021-01-08 17:05:43 -08002// https://benchmarksgame-team.pages.debian.net/benchmarksgame/
Chih-Hung Hsiehe42c5052020-04-16 10:44:21 -07003//
4// contributed by the Rust Project Developers
5// contributed by TeXitoi
6// contributed by BurntSushi
7
Chih-Hung Hsiehe42c5052020-04-16 10:44:21 -07008use std::io::{self, Read};
9
10macro_rules! regex {
11 ($re:expr) => {
12 ::regex::Regex::new($re).unwrap()
13 };
14}
15
16fn main() {
17 let mut seq = String::with_capacity(50 * (1 << 20));
18 io::stdin().read_to_string(&mut seq).unwrap();
19 let ilen = seq.len();
20
21 seq = regex!(">[^\n]*\n|\n").replace_all(&seq, "").into_owned();
22 let clen = seq.len();
23
24 let variants = vec![
25 regex!("agggtaaa|tttaccct"),
26 regex!("[cgt]gggtaaa|tttaccc[acg]"),
27 regex!("a[act]ggtaaa|tttacc[agt]t"),
28 regex!("ag[act]gtaaa|tttac[agt]ct"),
29 regex!("agg[act]taaa|ttta[agt]cct"),
30 regex!("aggg[acg]aaa|ttt[cgt]ccct"),
31 regex!("agggt[cgt]aa|tt[acg]accct"),
32 regex!("agggta[cgt]a|t[acg]taccct"),
33 regex!("agggtaa[cgt]|[acg]ttaccct"),
34 ];
35 for re in variants {
36 println!("{} {}", re.to_string(), re.find_iter(&seq).count());
37 }
38
39 let substs = vec![
40 (b'B', "(c|g|t)"),
41 (b'D', "(a|g|t)"),
42 (b'H', "(a|c|t)"),
43 (b'K', "(g|t)"),
44 (b'M', "(a|c)"),
45 (b'N', "(a|c|g|t)"),
46 (b'R', "(a|g)"),
47 (b'S', "(c|g)"),
48 (b'V', "(a|c|g)"),
49 (b'W', "(a|t)"),
50 (b'Y', "(c|t)"),
51 ]; // combined into one regex in `replace_all`
52 let seq = replace_all(&seq, substs);
53
54 println!("\n{}\n{}\n{}", ilen, clen, seq.len());
55}
56
57fn replace_all(text: &str, substs: Vec<(u8, &str)>) -> String {
58 let mut replacements = vec![""; 256];
59 let mut alternates = vec![];
60 for (re, replacement) in substs {
61 replacements[re as usize] = replacement;
62 alternates.push((re as char).to_string());
63 }
64
65 let re = regex!(&alternates.join("|"));
66 let mut new = String::with_capacity(text.len());
67 let mut last_match = 0;
68 for m in re.find_iter(text) {
69 new.push_str(&text[last_match..m.start()]);
70 new.push_str(replacements[text.as_bytes()[m.start()] as usize]);
71 last_match = m.end();
72 }
73 new.push_str(&text[last_match..]);
74 new
75}