main.go

package main

import (
	"crypto/sha1"
	"encoding/base64"
	"fmt"
	"log"
	"runtime"
	"sync"

	arg "github.com/alexflint/go-arg"
	"github.com/brentp/vcfgo"
	"github.com/brentp/xopen"
)

func digest(bv []byte) string {
	hasher := sha1.New()
	hasher.Write(bv)

	bs := hasher.Sum(nil)
	sha := base64.StdEncoding.EncodeToString(bs)
	return sha
}

func builder(file string, wg *sync.WaitGroup, dchan chan string) {

	// open VCF file.
	f, err := xopen.Ropen(file)
	echeck("Can't open VCF file.", err)
	defer f.Close()

	r, err := vcfgo.NewReader(f, false)
	echeck("Can't access vcf file.", err)
	defer r.Close()

	for {
		read := r.Read()
		if read == nil {
			break
		}
		wg.Add(1)

		go func(record *vcfgo.Variant) {
			site := fmt.Sprintf("%s:%d:%d:%s:%s", record.Chromosome, record.Start(), record.End(), record.Reference, record.Alt())
			siteDigest := digest([]byte(site))
			dchan <- siteDigest
			wg.Done()
		}(read)
	}
}

func echeck(m string, e error) {
	if e != nil {
		log.Println(m)
		panic(e)
	}
}

func main() {
	var args struct {
		VCF  []string `help:"VCF file to collect concordance from. Space separated."`
		CPUS int      `help:"Number of CPUS workers to allow."`
	}
	args.CPUS = runtime.NumCPU()
	arg.MustParse(&args)

	var wg sync.WaitGroup
	dchan := make(chan string)

	for _, file := range args.VCF {
		wg.Add(1)
		go builder(file, &wg, dchan)

		for income := range dchan {
			fmt.Println("income: ", income)
		}
		wg.Done()
	}
	wg.Wait()
	close(dchan)

	fmt.Println("out!!")
}