mirror of
https://github.com/metabarcoding/obitools4.git
synced 2025-12-08 16:50:27 +00:00
Some code refactoring, a new version of obiuniq more efficient in memory and a first make file allowing to build obitools
This commit is contained in:
@@ -5,15 +5,15 @@ import (
|
||||
"log"
|
||||
"sync"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
)
|
||||
|
||||
type SequenceBatchWriterToFile func(iterator obiseq.IBioSequenceBatch,
|
||||
type SequenceBatchWriterToFile func(iterator obiiter.IBioSequenceBatch,
|
||||
filename string,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error)
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error)
|
||||
|
||||
func WriterDispatcher(prototypename string,
|
||||
dispatcher obiseq.IDistribute,
|
||||
dispatcher obiiter.IDistribute,
|
||||
formater SequenceBatchWriterToFile,
|
||||
options ...WithOption) {
|
||||
|
||||
|
||||
@@ -10,6 +10,7 @@ import (
|
||||
"strconv"
|
||||
"strings"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@@ -117,7 +118,7 @@ func __read_ecopcr_bioseq__(file *__ecopcr_file__) (*obiseq.BioSequence, error)
|
||||
return bseq, nil
|
||||
}
|
||||
|
||||
func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenceBatch {
|
||||
func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiiter.IBioSequenceBatch {
|
||||
tag := make([]byte, 11)
|
||||
n, _ := reader.Read(tag)
|
||||
|
||||
@@ -163,7 +164,7 @@ func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenc
|
||||
|
||||
opt := MakeOptions(options)
|
||||
|
||||
newIter := obiseq.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter := obiiter.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
@@ -181,7 +182,7 @@ func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenc
|
||||
slice = append(slice, seq)
|
||||
ii++
|
||||
if ii >= opt.BatchSize() {
|
||||
newIter.Push(obiseq.MakeBioSequenceBatch(i, slice))
|
||||
newIter.Push(obiiter.MakeBioSequenceBatch(i, slice))
|
||||
slice = obiseq.MakeBioSequenceSlice()
|
||||
i++
|
||||
ii = 0
|
||||
@@ -191,7 +192,7 @@ func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenc
|
||||
}
|
||||
|
||||
if len(slice) > 0 {
|
||||
newIter.Push(obiseq.MakeBioSequenceBatch(i, slice))
|
||||
newIter.Push(obiiter.MakeBioSequenceBatch(i, slice))
|
||||
}
|
||||
|
||||
newIter.Done()
|
||||
@@ -205,12 +206,12 @@ func ReadEcoPCRBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenc
|
||||
return newIter
|
||||
}
|
||||
|
||||
func ReadEcoPCR(reader io.Reader, options ...WithOption) obiseq.IBioSequence {
|
||||
func ReadEcoPCR(reader io.Reader, options ...WithOption) obiiter.IBioSequence {
|
||||
ib := ReadEcoPCRBatch(reader, options...)
|
||||
return ib.SortBatches().IBioSequence()
|
||||
}
|
||||
|
||||
func ReadEcoPCRBatchFromFile(filename string, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func ReadEcoPCRBatchFromFile(filename string, options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
var reader io.Reader
|
||||
var greader io.Reader
|
||||
var err error
|
||||
@@ -218,7 +219,7 @@ func ReadEcoPCRBatchFromFile(filename string, options ...WithOption) (obiseq.IBi
|
||||
reader, err = os.Open(filename)
|
||||
if err != nil {
|
||||
log.Printf("open file error: %+v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
// Test if the flux is compressed by gzip
|
||||
@@ -230,7 +231,7 @@ func ReadEcoPCRBatchFromFile(filename string, options ...WithOption) (obiseq.IBi
|
||||
return ReadEcoPCRBatch(reader, options...), nil
|
||||
}
|
||||
|
||||
func ReadEcoPCRFromFile(filename string, options ...WithOption) (obiseq.IBioSequence, error) {
|
||||
func ReadEcoPCRFromFile(filename string, options ...WithOption) (obiiter.IBioSequence, error) {
|
||||
ib, err := ReadEcoPCRBatchFromFile(filename, options...)
|
||||
return ib.SortBatches().IBioSequence(), err
|
||||
|
||||
|
||||
@@ -10,6 +10,7 @@ import (
|
||||
"strconv"
|
||||
"strings"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@@ -80,7 +81,7 @@ func _EndOfLastEntry(buff []byte) int {
|
||||
return -1
|
||||
}
|
||||
|
||||
func _ParseEmblFile(input <-chan _FileChunk, out obiseq.IBioSequenceBatch) {
|
||||
func _ParseEmblFile(input <-chan _FileChunk, out obiiter.IBioSequenceBatch) {
|
||||
|
||||
for chunks := range input {
|
||||
scanner := bufio.NewScanner(chunks.raw)
|
||||
@@ -139,7 +140,7 @@ func _ParseEmblFile(input <-chan _FileChunk, out obiseq.IBioSequenceBatch) {
|
||||
seqBytes = new(bytes.Buffer)
|
||||
}
|
||||
}
|
||||
out.Push(obiseq.MakeBioSequenceBatch(order, sequences))
|
||||
out.Push(obiiter.MakeBioSequenceBatch(order, sequences))
|
||||
}
|
||||
|
||||
out.Done()
|
||||
@@ -176,11 +177,11 @@ func _ReadFlatFileChunk(reader io.Reader, readers chan _FileChunk) {
|
||||
|
||||
// 6 5 43 2 1
|
||||
// <CR>?<LF>//<CR>?<LF>
|
||||
func ReadEMBLBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenceBatch {
|
||||
func ReadEMBLBatch(reader io.Reader, options ...WithOption) obiiter.IBioSequenceBatch {
|
||||
opt := MakeOptions(options)
|
||||
entry_channel := make(chan _FileChunk, opt.BufferSize())
|
||||
|
||||
newIter := obiseq.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter := obiiter.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
|
||||
nworkers := opt.ParallelWorkers()
|
||||
newIter.Add(nworkers)
|
||||
@@ -199,12 +200,12 @@ func ReadEMBLBatch(reader io.Reader, options ...WithOption) obiseq.IBioSequenceB
|
||||
return newIter
|
||||
}
|
||||
|
||||
func ReadEMBL(reader io.Reader, options ...WithOption) obiseq.IBioSequence {
|
||||
func ReadEMBL(reader io.Reader, options ...WithOption) obiiter.IBioSequence {
|
||||
ib := ReadEMBLBatch(reader, options...)
|
||||
return ib.SortBatches().IBioSequence()
|
||||
}
|
||||
|
||||
func ReadEMBLBatchFromFile(filename string, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func ReadEMBLBatchFromFile(filename string, options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
var reader io.Reader
|
||||
var greader io.Reader
|
||||
var err error
|
||||
@@ -212,7 +213,7 @@ func ReadEMBLBatchFromFile(filename string, options ...WithOption) (obiseq.IBioS
|
||||
reader, err = os.Open(filename)
|
||||
if err != nil {
|
||||
log.Printf("open file error: %+v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
// Test if the flux is compressed by gzip
|
||||
@@ -224,7 +225,7 @@ func ReadEMBLBatchFromFile(filename string, options ...WithOption) (obiseq.IBioS
|
||||
return ReadEMBLBatch(reader, options...), nil
|
||||
}
|
||||
|
||||
func ReadEMBLFromFile(filename string, options ...WithOption) (obiseq.IBioSequence, error) {
|
||||
func ReadEMBLFromFile(filename string, options ...WithOption) (obiiter.IBioSequence, error) {
|
||||
ib, err := ReadEMBLBatchFromFile(filename, options...)
|
||||
return ib.SortBatches().IBioSequence(), err
|
||||
|
||||
|
||||
@@ -3,6 +3,7 @@ package obiformats
|
||||
import (
|
||||
"strings"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@@ -14,14 +15,16 @@ func ParseGuessedFastSeqHeader(sequence *obiseq.BioSequence) {
|
||||
}
|
||||
}
|
||||
|
||||
func IParseFastSeqHeaderBatch(iterator obiseq.IBioSequenceBatch, options ...WithOption) obiseq.IBioSequenceBatch {
|
||||
func IParseFastSeqHeaderBatch(iterator obiiter.IBioSequenceBatch,
|
||||
options ...WithOption) obiiter.IBioSequenceBatch {
|
||||
opt := MakeOptions(options)
|
||||
return iterator.MakeIWorker(obiseq.AnnotatorToSeqWorker(opt.ParseFastSeqHeader()),
|
||||
return iterator.MakeIWorker(obiiter.AnnotatorToSeqWorker(opt.ParseFastSeqHeader()),
|
||||
opt.ParallelWorkers(),
|
||||
opt.BufferSize())
|
||||
}
|
||||
|
||||
func IParseFastSeqHeader(iterator obiseq.IBioSequence, options ...WithOption) obiseq.IBioSequence {
|
||||
func IParseFastSeqHeader(iterator obiiter.IBioSequence,
|
||||
options ...WithOption) obiiter.IBioSequence {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
return IParseFastSeqHeaderBatch(iterator.IBioSequenceBatch(opt.BatchSize(),
|
||||
|
||||
@@ -13,11 +13,12 @@ import (
|
||||
"unsafe"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/cutils"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
func _FastseqReader(seqfile C.fast_kseq_p,
|
||||
iterator obiseq.IBioSequenceBatch,
|
||||
iterator obiiter.IBioSequenceBatch,
|
||||
batch_size int) {
|
||||
var comment string
|
||||
i := 0
|
||||
@@ -63,7 +64,7 @@ func _FastseqReader(seqfile C.fast_kseq_p,
|
||||
// log.Printf("\n==> Pushing sequence batch\n")
|
||||
// start := time.Now()
|
||||
|
||||
iterator.Push(obiseq.MakeBioSequenceBatch(i, slice))
|
||||
iterator.Push(obiiter.MakeBioSequenceBatch(i, slice))
|
||||
// elapsed := time.Since(start)
|
||||
// log.Printf("\n==>sequences pushed after %s\n", elapsed)
|
||||
|
||||
@@ -73,13 +74,13 @@ func _FastseqReader(seqfile C.fast_kseq_p,
|
||||
}
|
||||
}
|
||||
if len(slice) > 0 {
|
||||
iterator.Push(obiseq.MakeBioSequenceBatch(i, slice))
|
||||
iterator.Push(obiiter.MakeBioSequenceBatch(i, slice))
|
||||
}
|
||||
iterator.Done()
|
||||
|
||||
}
|
||||
|
||||
func ReadFastSeqBatchFromFile(filename string, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func ReadFastSeqBatchFromFile(filename string, options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
name := C.CString(filename)
|
||||
@@ -92,7 +93,7 @@ func ReadFastSeqBatchFromFile(filename string, options ...WithOption) (obiseq.IB
|
||||
|
||||
if pointer == nil {
|
||||
err = fmt.Errorf("cannot open file %s", filename)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
size := int64(-1)
|
||||
@@ -104,7 +105,7 @@ func ReadFastSeqBatchFromFile(filename string, options ...WithOption) (obiseq.IB
|
||||
size = -1
|
||||
}
|
||||
|
||||
newIter := obiseq.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter := obiiter.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
@@ -124,14 +125,14 @@ func ReadFastSeqBatchFromFile(filename string, options ...WithOption) (obiseq.IB
|
||||
return newIter, err
|
||||
}
|
||||
|
||||
func ReadFastSeqFromFile(filename string, options ...WithOption) (obiseq.IBioSequence, error) {
|
||||
func ReadFastSeqFromFile(filename string, options ...WithOption) (obiiter.IBioSequence, error) {
|
||||
ib, err := ReadFastSeqBatchFromFile(filename, options...)
|
||||
return ib.SortBatches().IBioSequence(), err
|
||||
}
|
||||
|
||||
func ReadFastSeqBatchFromStdin(options ...WithOption) obiseq.IBioSequenceBatch {
|
||||
func ReadFastSeqBatchFromStdin(options ...WithOption) obiiter.IBioSequenceBatch {
|
||||
opt := MakeOptions(options)
|
||||
newIter := obiseq.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
newIter := obiiter.MakeIBioSequenceBatch(opt.BufferSize())
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
@@ -139,12 +140,13 @@ func ReadFastSeqBatchFromStdin(options ...WithOption) obiseq.IBioSequenceBatch {
|
||||
newIter.WaitAndClose()
|
||||
}()
|
||||
|
||||
go _FastseqReader(C.open_fast_sek_stdin(C.int32_t(opt.QualityShift())), newIter, opt.BatchSize())
|
||||
go _FastseqReader(C.open_fast_sek_stdin(C.int32_t(opt.QualityShift())),
|
||||
newIter, opt.BatchSize())
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func ReadFastSeqFromStdin(options ...WithOption) obiseq.IBioSequence {
|
||||
func ReadFastSeqFromStdin(options ...WithOption) obiiter.IBioSequence {
|
||||
ib := ReadFastSeqBatchFromStdin(options...)
|
||||
return ib.SortBatches().IBioSequence()
|
||||
}
|
||||
|
||||
@@ -8,6 +8,7 @@ import (
|
||||
"os"
|
||||
"strings"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@@ -21,7 +22,7 @@ func min(x, y int) int {
|
||||
func FormatFasta(seq *obiseq.BioSequence, formater FormatHeader) string {
|
||||
var fragments strings.Builder
|
||||
|
||||
if seq==nil {
|
||||
if seq == nil {
|
||||
log.Panicln("try to format a nil BioSequence")
|
||||
}
|
||||
|
||||
@@ -44,7 +45,7 @@ func FormatFasta(seq *obiseq.BioSequence, formater FormatHeader) string {
|
||||
folded)
|
||||
}
|
||||
|
||||
func FormatFastaBatch(batch obiseq.BioSequenceBatch, formater FormatHeader) []byte {
|
||||
func FormatFastaBatch(batch obiiter.BioSequenceBatch, formater FormatHeader) []byte {
|
||||
var bs bytes.Buffer
|
||||
for _, seq := range batch.Slice() {
|
||||
bs.WriteString(FormatFasta(seq, formater))
|
||||
@@ -53,7 +54,7 @@ func FormatFastaBatch(batch obiseq.BioSequenceBatch, formater FormatHeader) []by
|
||||
return bs.Bytes()
|
||||
}
|
||||
|
||||
func WriteFasta(iterator obiseq.IBioSequence, file io.Writer, options ...WithOption) error {
|
||||
func WriteFasta(iterator obiiter.IBioSequence, file io.Writer, options ...WithOption) error {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
header_format := opt.FormatFastSeqHeader()
|
||||
@@ -73,7 +74,7 @@ func WriteFasta(iterator obiseq.IBioSequence, file io.Writer, options ...WithOpt
|
||||
return nil
|
||||
}
|
||||
|
||||
func WriteFastaToFile(iterator obiseq.IBioSequence,
|
||||
func WriteFastaToFile(iterator obiiter.IBioSequence,
|
||||
filename string,
|
||||
options ...WithOption) error {
|
||||
|
||||
@@ -89,16 +90,18 @@ func WriteFastaToFile(iterator obiseq.IBioSequence,
|
||||
return WriteFasta(iterator, file, options...)
|
||||
}
|
||||
|
||||
func WriteFastaToStdout(iterator obiseq.IBioSequence, options ...WithOption) error {
|
||||
func WriteFastaToStdout(iterator obiiter.IBioSequence, options ...WithOption) error {
|
||||
options = append(options, OptionDontCloseFile())
|
||||
return WriteFasta(iterator, os.Stdout, options...)
|
||||
}
|
||||
|
||||
func WriteFastaBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func WriteFastaBatch(iterator obiiter.IBioSequenceBatch,
|
||||
file io.Writer,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
buffsize := iterator.BufferSize()
|
||||
newIter := obiseq.MakeIBioSequenceBatch(buffsize)
|
||||
newIter := obiiter.MakeIBioSequenceBatch(buffsize)
|
||||
|
||||
nwriters := opt.ParallelWorkers()
|
||||
|
||||
@@ -113,7 +116,7 @@ func WriteFastaBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options
|
||||
close(chunkchan)
|
||||
}()
|
||||
|
||||
ff := func(iterator obiseq.IBioSequenceBatch) {
|
||||
ff := func(iterator obiiter.IBioSequenceBatch) {
|
||||
for iterator.Next() {
|
||||
batch := iterator.Get()
|
||||
chunkchan <- FileChunck{
|
||||
@@ -164,20 +167,21 @@ func WriteFastaBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options
|
||||
return newIter, nil
|
||||
}
|
||||
|
||||
func WriteFastaBatchToStdout(iterator obiseq.IBioSequenceBatch, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func WriteFastaBatchToStdout(iterator obiiter.IBioSequenceBatch,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
options = append(options, OptionDontCloseFile())
|
||||
return WriteFastaBatch(iterator, os.Stdout, options...)
|
||||
}
|
||||
|
||||
func WriteFastaBatchToFile(iterator obiseq.IBioSequenceBatch,
|
||||
func WriteFastaBatchToFile(iterator obiiter.IBioSequenceBatch,
|
||||
filename string,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
|
||||
file, err := os.Create(filename)
|
||||
|
||||
if err != nil {
|
||||
log.Fatalf("open file error: %v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
options = append(options, OptionCloseFile())
|
||||
|
||||
@@ -8,6 +8,7 @@ import (
|
||||
"os"
|
||||
"time"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@@ -34,7 +35,7 @@ func FormatFastq(seq *obiseq.BioSequence, quality_shift int, formater FormatHead
|
||||
)
|
||||
}
|
||||
|
||||
func FormatFastqBatch(batch obiseq.BioSequenceBatch, quality_shift int,
|
||||
func FormatFastqBatch(batch obiiter.BioSequenceBatch, quality_shift int,
|
||||
formater FormatHeader) []byte {
|
||||
var bs bytes.Buffer
|
||||
for _, seq := range batch.Slice() {
|
||||
@@ -44,7 +45,7 @@ func FormatFastqBatch(batch obiseq.BioSequenceBatch, quality_shift int,
|
||||
return bs.Bytes()
|
||||
}
|
||||
|
||||
func WriteFastq(iterator obiseq.IBioSequence, file io.Writer, options ...WithOption) error {
|
||||
func WriteFastq(iterator obiiter.IBioSequence, file io.Writer, options ...WithOption) error {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
header_format := opt.FormatFastSeqHeader()
|
||||
@@ -65,7 +66,7 @@ func WriteFastq(iterator obiseq.IBioSequence, file io.Writer, options ...WithOpt
|
||||
return nil
|
||||
}
|
||||
|
||||
func WriteFastqToFile(iterator obiseq.IBioSequence,
|
||||
func WriteFastqToFile(iterator obiiter.IBioSequence,
|
||||
filename string,
|
||||
options ...WithOption) error {
|
||||
|
||||
@@ -80,7 +81,7 @@ func WriteFastqToFile(iterator obiseq.IBioSequence,
|
||||
return WriteFastq(iterator, file, options...)
|
||||
}
|
||||
|
||||
func WriteFastqToStdout(iterator obiseq.IBioSequence, options ...WithOption) error {
|
||||
func WriteFastqToStdout(iterator obiiter.IBioSequence, options ...WithOption) error {
|
||||
options = append(options, OptionDontCloseFile())
|
||||
return WriteFastq(iterator, os.Stdout, options...)
|
||||
}
|
||||
@@ -90,11 +91,13 @@ type FileChunck struct {
|
||||
order int
|
||||
}
|
||||
|
||||
func WriteFastqBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func WriteFastqBatch(iterator obiiter.IBioSequenceBatch,
|
||||
file io.Writer,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
opt := MakeOptions(options)
|
||||
|
||||
buffsize := iterator.BufferSize()
|
||||
newIter := obiseq.MakeIBioSequenceBatch(buffsize)
|
||||
newIter := obiiter.MakeIBioSequenceBatch(buffsize)
|
||||
|
||||
nwriters := opt.ParallelWorkers()
|
||||
|
||||
@@ -113,7 +116,7 @@ func WriteFastqBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options
|
||||
close(chunkchan)
|
||||
}()
|
||||
|
||||
ff := func(iterator obiseq.IBioSequenceBatch) {
|
||||
ff := func(iterator obiiter.IBioSequenceBatch) {
|
||||
for iterator.Next() {
|
||||
batch := iterator.Get()
|
||||
chunk := FileChunck{
|
||||
@@ -165,20 +168,21 @@ func WriteFastqBatch(iterator obiseq.IBioSequenceBatch, file io.Writer, options
|
||||
return newIter, nil
|
||||
}
|
||||
|
||||
func WriteFastqBatchToStdout(iterator obiseq.IBioSequenceBatch, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func WriteFastqBatchToStdout(iterator obiiter.IBioSequenceBatch,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
options = append(options, OptionDontCloseFile())
|
||||
return WriteFastqBatch(iterator, os.Stdout, options...)
|
||||
}
|
||||
|
||||
func WriteFastqBatchToFile(iterator obiseq.IBioSequenceBatch,
|
||||
func WriteFastqBatchToFile(iterator obiiter.IBioSequenceBatch,
|
||||
filename string,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
|
||||
file, err := os.Create(filename)
|
||||
|
||||
if err != nil {
|
||||
log.Fatalf("open file error: %v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
options = append(options, OptionCloseFile())
|
||||
|
||||
@@ -1,11 +1,12 @@
|
||||
package obiformats
|
||||
|
||||
import (
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
type __options__ struct {
|
||||
fastseq_header_parser obiseq.SeqAnnotator
|
||||
fastseq_header_parser obiiter.SeqAnnotator
|
||||
fastseq_header_writer func(*obiseq.BioSequence) string
|
||||
with_progress_bar bool
|
||||
buffer_size int
|
||||
@@ -58,7 +59,7 @@ func (opt Options) ParallelWorkers() int {
|
||||
return opt.pointer.parallel_workers
|
||||
}
|
||||
|
||||
func (opt Options) ParseFastSeqHeader() obiseq.SeqAnnotator {
|
||||
func (opt Options) ParseFastSeqHeader() obiiter.SeqAnnotator {
|
||||
return opt.pointer.fastseq_header_parser
|
||||
}
|
||||
|
||||
@@ -123,7 +124,7 @@ func OptionsQualitySolexa() WithOption {
|
||||
return OptionsQualityShift(64)
|
||||
}
|
||||
|
||||
func OptionsFastSeqHeaderParser(parser obiseq.SeqAnnotator) WithOption {
|
||||
func OptionsFastSeqHeaderParser(parser obiiter.SeqAnnotator) WithOption {
|
||||
f := WithOption(func(opt Options) {
|
||||
opt.pointer.fastseq_header_parser = parser
|
||||
})
|
||||
|
||||
@@ -8,7 +8,7 @@ import (
|
||||
"os"
|
||||
"strings"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
)
|
||||
|
||||
func GuessSeqFileType(firstline string) string {
|
||||
@@ -36,7 +36,8 @@ func GuessSeqFileType(firstline string) string {
|
||||
}
|
||||
}
|
||||
|
||||
func ReadSequencesBatchFromFile(filename string, options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func ReadSequencesBatchFromFile(filename string,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
var file *os.File
|
||||
var reader io.Reader
|
||||
var greader io.Reader
|
||||
@@ -46,7 +47,7 @@ func ReadSequencesBatchFromFile(filename string, options ...WithOption) (obiseq.
|
||||
|
||||
if err != nil {
|
||||
log.Fatalf("open file error: %v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
reader = file
|
||||
@@ -65,7 +66,7 @@ func ReadSequencesBatchFromFile(filename string, options ...WithOption) (obiseq.
|
||||
tag, _ := breader.Peek(30)
|
||||
|
||||
if len(tag) < 30 {
|
||||
newIter := obiseq.MakeIBioSequenceBatch()
|
||||
newIter := obiiter.MakeIBioSequenceBatch()
|
||||
newIter.Close()
|
||||
return newIter, nil
|
||||
}
|
||||
@@ -89,10 +90,11 @@ func ReadSequencesBatchFromFile(filename string, options ...WithOption) (obiseq.
|
||||
filename, filetype)
|
||||
}
|
||||
|
||||
return obiseq.NilIBioSequenceBatch, nil
|
||||
return obiiter.NilIBioSequenceBatch, nil
|
||||
}
|
||||
|
||||
func ReadSequencesFromFile(filename string, options ...WithOption) (obiseq.IBioSequence, error) {
|
||||
func ReadSequencesFromFile(filename string,
|
||||
options ...WithOption) (obiiter.IBioSequence, error) {
|
||||
ib, err := ReadSequencesBatchFromFile(filename, options...)
|
||||
return ib.SortBatches().IBioSequence(), err
|
||||
|
||||
|
||||
@@ -6,10 +6,10 @@ import (
|
||||
"log"
|
||||
"os"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
|
||||
)
|
||||
|
||||
func WriteSequences(iterator obiseq.IBioSequence,
|
||||
func WriteSequences(iterator obiiter.IBioSequence,
|
||||
file io.Writer,
|
||||
options ...WithOption) error {
|
||||
|
||||
@@ -34,7 +34,7 @@ func WriteSequences(iterator obiseq.IBioSequence,
|
||||
return nil
|
||||
}
|
||||
|
||||
func WriteSequencesToFile(iterator obiseq.IBioSequence,
|
||||
func WriteSequencesToFile(iterator obiiter.IBioSequence,
|
||||
filename string,
|
||||
options ...WithOption) error {
|
||||
|
||||
@@ -48,13 +48,13 @@ func WriteSequencesToFile(iterator obiseq.IBioSequence,
|
||||
return WriteSequences(iterator, file, options...)
|
||||
}
|
||||
|
||||
func WriteSequencesToStdout(iterator obiseq.IBioSequence, options ...WithOption) error {
|
||||
func WriteSequencesToStdout(iterator obiiter.IBioSequence, options ...WithOption) error {
|
||||
return WriteSequences(iterator, os.Stdout, options...)
|
||||
}
|
||||
|
||||
func WriteSequenceBatch(iterator obiseq.IBioSequenceBatch,
|
||||
func WriteSequenceBatch(iterator obiiter.IBioSequenceBatch,
|
||||
file io.Writer,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
|
||||
iterator = iterator.Rebatch(1000)
|
||||
|
||||
@@ -64,7 +64,7 @@ func WriteSequenceBatch(iterator obiseq.IBioSequenceBatch,
|
||||
batch := iterator.Get()
|
||||
iterator.PushBack()
|
||||
|
||||
var newIter obiseq.IBioSequenceBatch
|
||||
var newIter obiiter.IBioSequenceBatch
|
||||
var err error
|
||||
|
||||
if len(batch.Slice()) > 0 {
|
||||
@@ -84,24 +84,24 @@ func WriteSequenceBatch(iterator obiseq.IBioSequenceBatch,
|
||||
return iterator, nil
|
||||
}
|
||||
|
||||
return obiseq.NilIBioSequenceBatch, fmt.Errorf("input iterator not ready")
|
||||
return obiiter.NilIBioSequenceBatch, fmt.Errorf("input iterator not ready")
|
||||
}
|
||||
|
||||
func WriteSequencesBatchToStdout(iterator obiseq.IBioSequenceBatch,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
func WriteSequencesBatchToStdout(iterator obiiter.IBioSequenceBatch,
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
options = append(options, OptionDontCloseFile())
|
||||
return WriteSequenceBatch(iterator, os.Stdout, options...)
|
||||
}
|
||||
|
||||
func WriteSequencesBatchToFile(iterator obiseq.IBioSequenceBatch,
|
||||
func WriteSequencesBatchToFile(iterator obiiter.IBioSequenceBatch,
|
||||
filename string,
|
||||
options ...WithOption) (obiseq.IBioSequenceBatch, error) {
|
||||
options ...WithOption) (obiiter.IBioSequenceBatch, error) {
|
||||
|
||||
file, err := os.Create(filename)
|
||||
|
||||
if err != nil {
|
||||
log.Fatalf("open file error: %v", err)
|
||||
return obiseq.NilIBioSequenceBatch, err
|
||||
return obiiter.NilIBioSequenceBatch, err
|
||||
}
|
||||
|
||||
options = append(options, OptionCloseFile())
|
||||
|
||||
Reference in New Issue
Block a user