Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 1 | // The Computer Language Benchmarks Game |
Haibo Huang | 47619dd | 2021-01-08 17:05:43 -0800 | [diff] [blame^] | 2 | // https://benchmarksgame-team.pages.debian.net/benchmarksgame/ |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 3 | // |
| 4 | // contributed by the Rust Project Developers |
| 5 | // contributed by TeXitoi |
| 6 | // contributed by BurntSushi |
| 7 | |
| 8 | extern crate regex; |
| 9 | |
| 10 | use std::io::{self, Read}; |
| 11 | use std::sync::Arc; |
| 12 | use std::thread; |
| 13 | |
| 14 | macro_rules! regex { |
| 15 | ($re:expr) => { |
| 16 | ::regex::Regex::new($re).unwrap() |
| 17 | }; |
| 18 | } |
| 19 | |
| 20 | fn main() { |
| 21 | let mut seq = String::with_capacity(51 * (1 << 20)); |
| 22 | io::stdin().read_to_string(&mut seq).unwrap(); |
| 23 | let ilen = seq.len(); |
| 24 | |
| 25 | seq = regex!(">[^\n]*\n|\n").replace_all(&seq, "").into_owned(); |
| 26 | let clen = seq.len(); |
| 27 | let seq_arc = Arc::new(seq.clone()); |
| 28 | |
| 29 | let variants = vec![ |
| 30 | regex!("agggtaaa|tttaccct"), |
| 31 | regex!("[cgt]gggtaaa|tttaccc[acg]"), |
| 32 | regex!("a[act]ggtaaa|tttacc[agt]t"), |
| 33 | regex!("ag[act]gtaaa|tttac[agt]ct"), |
| 34 | regex!("agg[act]taaa|ttta[agt]cct"), |
| 35 | regex!("aggg[acg]aaa|ttt[cgt]ccct"), |
| 36 | regex!("agggt[cgt]aa|tt[acg]accct"), |
| 37 | regex!("agggta[cgt]a|t[acg]taccct"), |
| 38 | regex!("agggtaa[cgt]|[acg]ttaccct"), |
| 39 | ]; |
| 40 | let mut counts = vec![]; |
| 41 | for variant in variants { |
| 42 | let seq = seq_arc.clone(); |
| 43 | let restr = variant.to_string(); |
| 44 | let future = thread::spawn(move || variant.find_iter(&seq).count()); |
| 45 | counts.push((restr, future)); |
| 46 | } |
| 47 | |
| 48 | let substs = vec![ |
| 49 | (regex!("B"), "(c|g|t)"), |
| 50 | (regex!("D"), "(a|g|t)"), |
| 51 | (regex!("H"), "(a|c|t)"), |
| 52 | (regex!("K"), "(g|t)"), |
| 53 | (regex!("M"), "(a|c)"), |
| 54 | (regex!("N"), "(a|c|g|t)"), |
| 55 | (regex!("R"), "(a|g)"), |
| 56 | (regex!("S"), "(c|g)"), |
| 57 | (regex!("V"), "(a|c|g)"), |
| 58 | (regex!("W"), "(a|t)"), |
| 59 | (regex!("Y"), "(c|t)"), |
| 60 | ]; |
| 61 | let mut seq = seq; |
| 62 | for (re, replacement) in substs { |
| 63 | seq = re.replace_all(&seq, replacement).into_owned(); |
| 64 | } |
| 65 | |
| 66 | for (variant, count) in counts { |
| 67 | println!("{} {}", variant, count.join().unwrap()); |
| 68 | } |
| 69 | println!("\n{}\n{}\n{}", ilen, clen, seq.len()); |
| 70 | } |