VCF2duck.rs

use vcf::{VCFReader,VCFRecord};
use std::fs::File;
use std::fs::remove_file;
use std::io::BufReader;
use std::str::from_utf8;


use duckdb::{params, Connection};

#[derive(Debug)]
struct Variant {

    chrom: String,
    pos: u64,
    reference: String,
    alternative: String
}

fn main() {

    let file = File::open("test.snpeff.vcf").expect("cannot read file");

    let mut reader = VCFReader::new(BufReader::new(file)).expect("cannot read vcf");

    remove_file("/tmp/test.db").expect("File don't exists");

    let conn = Connection::open("/tmp/test.db").expect("cannot open database");

    conn.execute_batch("
      CREATE TABLE variant (
      chrom     TEXT,
      pos       INTEGER,
      reference TEXT,
      alternative TEXT
      );
      ").expect("cannot create table");


    let mut record: VCFRecord =  reader.empty_record(); 

    loop {
        let success = reader.next_record(&mut record).unwrap_or(false);

        let v = Variant {
            chrom :from_utf8(&record.chromosome).unwrap_or("N").to_string(),
            pos :record.position,
            reference :from_utf8(&record.reference).unwrap_or("N").to_string(),
            alternative :from_utf8(&record.alternative[0]).unwrap_or("N").to_string(), 
        };

        conn.execute(
            "INSERT INTO variant (chrom, pos, reference, alternative) VALUES (?, ?, ?, ?)",
            params![v.chrom, v.pos, v.reference, v.alternative],
            ).expect("cannot insert");

        if !success {break};
    }
}