func queryGi2Taxid(dbFilePath string, queryType string, gis []string) {
db, err := bolt.Open(dbFilePath, 0600, nil)
defer db.Close()
checkError(err)
taxids, err := taxon.QueryGi2Taxid(db, queryType, gis)
checkError(err)
for i, gi := range gis {
fmt.Printf("%s\t%s\n", gi, taxids[i])
}
}
func queryGi2TaxidByFile(dbFilePath string, queryType string, dataFile string, chunkSize int, threads int) {
if chunkSize <= 0 {
chunkSize = 10000
}
fn := func(line string) (interface{}, bool, error) {
line = strings.TrimSpace(strings.TrimRight(line, "\n"))
if line == "" {
return "", false, nil
}
return line, true, nil
}
reader, err := breader.NewBufferedReader(dataFile, runtime.NumCPU(), chunkSize, fn)
checkError(err)
pool := taxon.NewDBPool(dbFilePath, threads)
chResults := make(chan [][]string, threads)
// receive result and print
chDone := make(chan int)
go func() {
for s := range chResults {
gis, taxids := s[0], s[1]
for i, gi := range gis {
fmt.Printf("%s\t%s\n", gi, taxids[i])
}
}
chDone <- 1
}()
// querying
var wg sync.WaitGroup
tokens := make(chan int, threads)
for chunk := range reader.Ch {
if chunk.Err != nil {
checkError(chunk.Err)
break
}
tokens <- 1
wg.Add(1)
gis := make([]string, len(chunk.Data))
for i, data := range chunk.Data {
gis[i] = data.(string)
}
go func(gis []string) {
db := pool.GetDB()
defer func() {
pool.ReleaseDB(db)
wg.Done()
<-tokens
}()
taxids, err := taxon.QueryGi2Taxid(db, queryType, gis)
checkError(err)
chResults <- [][]string{gis, taxids}
}(gis)
}
wg.Wait()
close(chResults)
<-chDone
}
