You could ditch the control channel and use a sync.WaitGroup
:
package main
import (
"encoding/csv"
"fmt"
"io"
"log"
"os"
"regexp"
"sync"
)
var (
cleanRe *regexp.Regexp = regexp.MustCompile("[^0-9]+")
comma rune = '\t'
fieldsPerRecord = -1
)
func clean(s string) string {
clean := cleanRe.ReplaceAllLiteralString(s, "")
if len(clean) < 6 {
return ""
}
return clean
}
func uniqueChannel(inputChan chan []string) {
uniq := make(map[string]map[string]bool)
i := 0
for record := range inputChan {
i++
id, v := record[0], record[1]
if uniq[id] == nil {
uniq[id] = make(map[string]bool)
} else if !uniq[id][v] {
uniq[id][v] = true
fmt.Println(id, string(comma), v)
}
}
log.Println("digest ", i)
}
func processFile(fileName string, outputChan chan []string) {
f, err := os.Open(fileName)
if err != nil {
log.Fatal(err)
}
r := csv.NewReader(f)
r.FieldsPerRecord = fieldsPerRecord
r.Comma = comma
// Process the records
for record, err := r.Read(); err != io.EOF; record, err = r.Read() {
if err != nil {
continue
}
id := record[0]
for _, v := range record[1:] {
if cleanV := clean(v); cleanV != "" {
outputChan <- []string{id, cleanV}
}
}
}
}
func main() {
inputs := []string{"ex.tsv"}
recordChan := make(chan []string)
var wg sync.WaitGroup
// Ingest the inputs
for _, fName := range inputs {
wg.Add(1)
go func() {
processFile(fName, recordChan)
wg.Done()
}()
}
go func() {
wg.Wait()
close(recordChan)
}()
// This is the loop to ensure it's all unique
uniqueChannel(recordChan)
}