The Computer Language
Benchmarks Game

k-nucleotide Rust #5 program

source code

// The Computer Language Benchmarks Game
// http://benchmarksgame.alioth.debian.org/
//
// contributed by the Rust Project Developers
// contributed by TeXitoi
// Rust #3 modified to use same get_seq as Rust #4

extern crate futures;
extern crate futures_cpupool;

use std::sync::Arc;
use std::hash::{Hasher, BuildHasherDefault};
use std::collections::HashMap;
use futures::Future;
use futures_cpupool::CpuPool;
use Item::*;

struct NaiveHasher(u64);
impl Default for NaiveHasher {
    fn default() -> Self {
        NaiveHasher(0)
    }
}
impl Hasher for NaiveHasher {
    fn finish(&self) -> u64 {
        self.0
    }
    fn write(&mut self, _: &[u8]) {
        unimplemented!()
    }
    fn write_u64(&mut self, i: u64) {
        self.0 = i ^ i >> 7;
    }
}
type NaiveBuildHasher = BuildHasherDefault<NaiveHasher>;
type NaiveHashMap<K, V> = HashMap<K, V, NaiveBuildHasher>;
type Map = NaiveHashMap<Code, u32>;

#[derive(Hash, PartialEq, PartialOrd, Ord, Eq, Clone, Copy)]
struct Code(u64);
impl Code {
    fn push(&mut self, c: u8, mask: u64) {
        self.0 <<= 2;
        self.0 |= c as u64;
        self.0 &= mask;
    }
    fn from_str(s: &str) -> Code {
        let mask = Code::make_mask(s.len());
        let mut res = Code(0);
        for c in s.as_bytes() {
            res.push(Code::encode(*c), mask);
        }
        res
    }
    fn to_string(&self, frame: usize) -> String {
        let mut res = vec![];
        let mut code = self.0;
        for _ in 0..frame {
            let c = match code as u8 & 0b11 {
                c if c == Code::encode(b'A') => b'A',
                c if c == Code::encode(b'T') => b'T',
                c if c == Code::encode(b'G') => b'G',
                c if c == Code::encode(b'C') => b'C',
                _ => unreachable!(),
            };
            res.push(c);
            code >>= 2;
        }
        res.reverse();
        String::from_utf8(res).unwrap()
    }
    fn make_mask(frame: usize) -> u64 {
        (1u64 << (2 * frame)) - 1
    }
    fn encode(c: u8) -> u8 {
        (c & 0b110) >> 1
    }
}

struct Iter<'a> {
    iter: std::slice::Iter<'a, u8>,
    code: Code,
    mask: u64,
}
impl<'a> Iter<'a> {
    fn new(input: &[u8], frame: usize) -> Iter {
        let mut iter = input.iter();
        let mut code = Code(0);
        let mask = Code::make_mask(frame);
        for c in iter.by_ref().take(frame - 1) {
            code.push(*c, mask);
        }
        Iter {
            iter: iter,
            code: code,
            mask: mask,
        }
    }
}
impl<'a> Iterator for Iter<'a> {
    type Item = Code;
    fn next(&mut self) -> Option<Self::Item> {
        self.iter.next().map(|&c| {
            self.code.push(c, self.mask);
            self.code
        })
    }
}

fn gen_freq(input: &[u8], frame: usize) -> Map {
    let mut freq = Map::default();
    for code in Iter::new(input, frame) {
        *freq.entry(code).or_insert(0) += 1;
    }
    freq
}

#[derive(Clone, Copy)]
enum Item {
    Freq(usize),
    Occ(&'static str),
}
impl Item {
    fn print(&self, freq: &Map) {
        match *self {
            Freq(frame) => {
                let mut v: Vec<_> = freq.iter().map(|(&code, &count)| (count, code)).collect();
                v.sort();
                let total = v.iter().map(|&(count, _)| count).sum::<u32>() as f32;
                for &(count, key) in v.iter().rev() {
                    println!("{} {:.3}", key.to_string(frame), (count as f32 * 100.) / total);
                }
                println!("");
            }
            Occ(occ) => println!("{}\t{}", freq[&Code::from_str(occ)], occ),
        }
    }
    fn gen_freq(&self, input: &[u8]) -> Map {
        match *self {
            Freq(frame) => gen_freq(input, frame),
            Occ(occ) => gen_freq(input, occ.len()),
        }
    }
}
static ITEMS: [Item; 7] = [
    Freq(1),
    Freq(2),
    Occ("GGT"),
    Occ("GGTA"),
    Occ("GGTATT"),
    Occ("GGTATTTTAATT"),
    Occ("GGTATTTTAATTTATAGT"),
];


fn get_seq<R: std::io::BufRead>(mut r: R, key: &[u8]) -> Vec<u8> {
    let mut res = Vec::with_capacity(65536);
    let mut line = Vec::with_capacity(64);

    loop {
        match r.read_until(b'\n', &mut line) {
            Ok(b) if b > 0 => if line.starts_with(key) { break },
            _ => break,
        }
        line.clear();
    }

    loop {
        line.clear();
        match r.read_until(b'\n', &mut line) {
            Ok(b) if b > 0 => res.extend(line[..line.len()-1].iter().cloned().map(Code::encode)),
            _ => break,
        }
    }

    res
}

fn main() {
    let stdin = std::io::stdin();
    let input = get_seq(stdin.lock(), b">THREE");
    let input = Arc::new(input);
    let pool = CpuPool::new_num_cpus();

    // In reverse to spawn big tasks first
    let items: Vec<_> = ITEMS.iter().rev().map(|&item| {
        let input = input.clone();
        let future_freq = pool.spawn_fn(move || Ok::<_, ()>(item.gen_freq(&input)));
        (item, future_freq)
    }).collect();

    for (item, future_freq) in items.into_iter().rev() {
        item.print(&future_freq.wait().unwrap());
    }
}
    

notes, command-line, and program output

NOTES:
64-bit Ubuntu quad core
rustc 1.16.0 (30cf806ef 2017-03-10)


Thu, 16 Mar 2017 17:58:29 GMT

MAKE:
/usr/local/src/rust-1.16.0/bin/rustc -C opt-level=3 -C target-cpu=core2 -C lto -L /usr/local/src/rust-libs knucleotide.rs -o knucleotide.rust-5.rust_run
7.90s to complete and log all make actions

COMMAND LINE:
./knucleotide.rust-5.rust_run 0 < knucleotide-input25000000.txt

PROGRAM OUTPUT:
A 30.295
T 30.151
C 19.800
G 19.754

AA 9.177
TA 9.132
AT 9.131
TT 9.091
CA 6.002
AC 6.001
AG 5.987
GA 5.984
CT 5.971
TC 5.971
GT 5.957
TG 5.956
CC 3.917
GC 3.911
CG 3.909
GG 3.902

1471758	GGT
446535	GGTA
47336	GGTATT
893	GGTATTTTAATT
893	GGTATTTTAATTTATAGT