2022-01-13 23:27:39 +01:00
|
|
|
package obiformats
|
|
|
|
|
|
|
|
import (
|
|
|
|
"bytes"
|
|
|
|
"fmt"
|
|
|
|
"io"
|
|
|
|
"os"
|
|
|
|
"strings"
|
2023-02-08 13:56:50 +01:00
|
|
|
"sync"
|
2023-02-08 13:14:26 +01:00
|
|
|
"time"
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-02-24 12:14:52 +01:00
|
|
|
log "github.com/sirupsen/logrus"
|
|
|
|
|
2022-02-24 07:08:40 +01:00
|
|
|
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
2022-01-13 23:43:01 +01:00
|
|
|
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
2023-03-24 10:25:12 +07:00
|
|
|
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiutils"
|
2022-01-13 23:27:39 +01:00
|
|
|
)
|
|
|
|
|
|
|
|
func min(x, y int) int {
|
|
|
|
if x < y {
|
|
|
|
return x
|
|
|
|
}
|
|
|
|
return y
|
|
|
|
}
|
|
|
|
|
2022-02-21 19:00:23 +01:00
|
|
|
func FormatFasta(seq *obiseq.BioSequence, formater FormatHeader) string {
|
2022-01-13 23:27:39 +01:00
|
|
|
var fragments strings.Builder
|
|
|
|
|
2022-02-24 07:08:40 +01:00
|
|
|
if seq == nil {
|
2022-02-21 19:00:23 +01:00
|
|
|
log.Panicln("try to format a nil BioSequence")
|
|
|
|
}
|
|
|
|
|
2022-01-13 23:27:39 +01:00
|
|
|
s := seq.Sequence()
|
|
|
|
l := len(s)
|
|
|
|
|
2022-11-16 09:22:55 +01:00
|
|
|
folded := ""
|
|
|
|
if l == 0 {
|
|
|
|
log.Println("Writing a BioSequence of length zero")
|
|
|
|
} else {
|
|
|
|
fragments.Grow(l + int(l/60)*2 + 100)
|
|
|
|
|
|
|
|
for i := 0; i < l; i += 60 {
|
|
|
|
to := min(i+60, l)
|
|
|
|
fmt.Fprintf(&fragments, "%s\n", string(s[i:to]))
|
|
|
|
}
|
|
|
|
|
|
|
|
folded = fragments.String()
|
|
|
|
folded = folded[:fragments.Len()-1]
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
info := formater(seq)
|
|
|
|
return fmt.Sprintf(">%s %s %s\n%s",
|
|
|
|
seq.Id(), info,
|
|
|
|
seq.Definition(),
|
|
|
|
folded)
|
|
|
|
}
|
|
|
|
|
2022-02-24 07:08:40 +01:00
|
|
|
func FormatFastaBatch(batch obiiter.BioSequenceBatch, formater FormatHeader) []byte {
|
2022-01-13 23:27:39 +01:00
|
|
|
var bs bytes.Buffer
|
|
|
|
for _, seq := range batch.Slice() {
|
|
|
|
bs.WriteString(FormatFasta(seq, formater))
|
|
|
|
bs.WriteString("\n")
|
|
|
|
}
|
|
|
|
return bs.Bytes()
|
|
|
|
}
|
|
|
|
|
2023-01-22 22:04:17 +01:00
|
|
|
func WriteFasta(iterator obiiter.IBioSequence,
|
2023-02-18 12:06:52 +01:00
|
|
|
file io.WriteCloser,
|
2023-01-22 22:04:17 +01:00
|
|
|
options ...WithOption) (obiiter.IBioSequence, error) {
|
2022-01-14 23:11:36 +01:00
|
|
|
opt := MakeOptions(options)
|
|
|
|
|
2023-03-28 19:37:05 +07:00
|
|
|
iterator = iterator.Rebatch(1000)
|
2023-03-24 10:25:12 +07:00
|
|
|
file, _ = obiutils.CompressStream(file, opt.CompressedFile(), opt.CloseFile())
|
2023-02-18 12:06:52 +01:00
|
|
|
|
2023-03-07 11:12:13 +07:00
|
|
|
newIter := obiiter.MakeIBioSequence()
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-01-14 23:11:36 +01:00
|
|
|
nwriters := opt.ParallelWorkers()
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2023-02-08 13:14:26 +01:00
|
|
|
obiiter.RegisterAPipe()
|
2022-01-13 23:27:39 +01:00
|
|
|
chunkchan := make(chan FileChunck)
|
|
|
|
|
|
|
|
header_format := opt.FormatFastSeqHeader()
|
|
|
|
|
2022-01-14 23:11:36 +01:00
|
|
|
newIter.Add(nwriters)
|
2023-02-08 13:56:50 +01:00
|
|
|
var waitWriter sync.WaitGroup
|
2022-01-13 23:27:39 +01:00
|
|
|
|
|
|
|
go func() {
|
2022-02-21 19:00:23 +01:00
|
|
|
newIter.WaitAndClose()
|
2023-02-08 13:14:26 +01:00
|
|
|
for len(chunkchan) > 0 {
|
|
|
|
time.Sleep(time.Millisecond)
|
|
|
|
}
|
2022-01-13 23:27:39 +01:00
|
|
|
close(chunkchan)
|
2023-02-08 13:56:50 +01:00
|
|
|
waitWriter.Wait()
|
2022-01-13 23:27:39 +01:00
|
|
|
}()
|
|
|
|
|
2023-01-22 22:04:17 +01:00
|
|
|
ff := func(iterator obiiter.IBioSequence) {
|
2022-01-13 23:27:39 +01:00
|
|
|
for iterator.Next() {
|
2022-08-23 15:07:06 +02:00
|
|
|
|
2022-01-13 23:27:39 +01:00
|
|
|
batch := iterator.Get()
|
2022-08-23 15:07:06 +02:00
|
|
|
|
2022-01-13 23:27:39 +01:00
|
|
|
chunkchan <- FileChunck{
|
|
|
|
FormatFastaBatch(batch, header_format),
|
|
|
|
batch.Order(),
|
|
|
|
}
|
2022-02-21 19:00:23 +01:00
|
|
|
newIter.Push(batch)
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
2022-01-14 17:32:12 +01:00
|
|
|
newIter.Done()
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
|
2022-02-24 12:14:52 +01:00
|
|
|
log.Debugln("Start of the fasta file writing")
|
2022-01-18 13:09:32 +01:00
|
|
|
go ff(iterator)
|
2022-01-14 23:11:36 +01:00
|
|
|
for i := 0; i < nwriters-1; i++ {
|
2022-01-13 23:27:39 +01:00
|
|
|
go ff(iterator.Split())
|
|
|
|
}
|
|
|
|
|
|
|
|
next_to_send := 0
|
|
|
|
received := make(map[int]FileChunck, 100)
|
|
|
|
|
2023-02-08 13:56:50 +01:00
|
|
|
waitWriter.Add(1)
|
2022-01-13 23:27:39 +01:00
|
|
|
go func() {
|
|
|
|
for chunk := range chunkchan {
|
|
|
|
if chunk.order == next_to_send {
|
|
|
|
file.Write(chunk.text)
|
|
|
|
next_to_send++
|
|
|
|
chunk, ok := received[next_to_send]
|
|
|
|
for ok {
|
|
|
|
file.Write(chunk.text)
|
|
|
|
delete(received, next_to_send)
|
|
|
|
next_to_send++
|
|
|
|
chunk, ok = received[next_to_send]
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
received[chunk.order] = chunk
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
2022-02-21 19:00:23 +01:00
|
|
|
|
2023-02-18 12:06:52 +01:00
|
|
|
file.Close()
|
2023-02-23 23:35:58 +01:00
|
|
|
|
2023-02-17 22:52:53 +01:00
|
|
|
log.Debugln("End of the fasta file writing")
|
|
|
|
obiiter.UnregisterPipe()
|
2023-02-08 13:56:50 +01:00
|
|
|
waitWriter.Done()
|
2022-02-15 00:47:02 +01:00
|
|
|
|
2022-01-13 23:27:39 +01:00
|
|
|
}()
|
|
|
|
|
2022-01-14 23:11:36 +01:00
|
|
|
return newIter, nil
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
|
2023-01-22 22:04:17 +01:00
|
|
|
func WriteFastaToStdout(iterator obiiter.IBioSequence,
|
|
|
|
options ...WithOption) (obiiter.IBioSequence, error) {
|
2022-02-15 00:47:02 +01:00
|
|
|
options = append(options, OptionDontCloseFile())
|
2022-11-16 17:13:03 +01:00
|
|
|
return WriteFasta(iterator, os.Stdout, options...)
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
|
2023-01-22 22:04:17 +01:00
|
|
|
func WriteFastaToFile(iterator obiiter.IBioSequence,
|
2022-01-13 23:27:39 +01:00
|
|
|
filename string,
|
2023-01-22 22:04:17 +01:00
|
|
|
options ...WithOption) (obiiter.IBioSequence, error) {
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2023-02-16 16:13:13 +01:00
|
|
|
opt := MakeOptions(options)
|
2023-02-18 12:06:52 +01:00
|
|
|
flags := os.O_WRONLY | os.O_CREATE
|
2023-02-16 16:13:13 +01:00
|
|
|
|
2023-02-18 12:06:52 +01:00
|
|
|
if opt.AppendFile() {
|
|
|
|
flags |= os.O_APPEND
|
|
|
|
}
|
|
|
|
file, err := os.OpenFile(filename, flags, 0660)
|
2022-01-13 23:27:39 +01:00
|
|
|
|
|
|
|
if err != nil {
|
|
|
|
log.Fatalf("open file error: %v", err)
|
2023-01-22 22:04:17 +01:00
|
|
|
return obiiter.NilIBioSequence, err
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
|
2022-02-15 00:47:02 +01:00
|
|
|
options = append(options, OptionCloseFile())
|
|
|
|
|
2023-02-23 23:35:58 +01:00
|
|
|
iterator, err = WriteFasta(iterator, file, options...)
|
|
|
|
|
|
|
|
if opt.HaveToSavePaired() {
|
|
|
|
var revfile *os.File
|
|
|
|
|
|
|
|
revfile, err = os.OpenFile(opt.PairedFileName(), flags, 0660)
|
|
|
|
if err != nil {
|
|
|
|
log.Fatalf("open file error: %v", err)
|
|
|
|
return obiiter.NilIBioSequence, err
|
|
|
|
}
|
|
|
|
iterator, err = WriteFasta(iterator.PairedWith(), revfile, options...)
|
|
|
|
}
|
|
|
|
|
|
|
|
return iterator, err
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|