mirror of
https://github.com/metabarcoding/obitools4.git
synced 2025-06-29 16:20:46 +00:00
Remove single sequence ierators. Only batch iterators persist
This commit is contained in:
@ -244,37 +244,6 @@ func (iterator IBioSequenceBatch) Finished() bool {
|
||||
return iterator.pointer.finished.IsSet()
|
||||
}
|
||||
|
||||
func (iterator IBioSequenceBatch) IBioSequence(sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.Channel())
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for iterator.Next() {
|
||||
batch := iterator.Get()
|
||||
|
||||
for batch.NotEmpty() {
|
||||
newIter.Channel() <- batch.Pop0()
|
||||
}
|
||||
batch.Recycle()
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequenceBatch) SortBatches(sizes ...int) IBioSequenceBatch {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
@ -472,7 +441,6 @@ func (iterator IBioSequenceBatch) Consume() {
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
func (iterator IBioSequenceBatch) Count(recycle bool) (int, int, int) {
|
||||
variants := 0
|
||||
reads := 0
|
||||
|
@ -1,342 +0,0 @@
|
||||
package obiiter
|
||||
|
||||
import (
|
||||
"sync"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
// Private structure implementing an iterator over
|
||||
// bioseq.BioSequence based on a channel.
|
||||
type __ibiosequence__ struct {
|
||||
channel chan *obiseq.BioSequence
|
||||
current *obiseq.BioSequence
|
||||
pushBack bool
|
||||
all_done *sync.WaitGroup
|
||||
buffer_size int
|
||||
finished bool
|
||||
pFinished *bool
|
||||
}
|
||||
|
||||
type IBioSequence struct {
|
||||
pointer *__ibiosequence__
|
||||
}
|
||||
|
||||
var NilIBioSequence = IBioSequence{pointer: nil}
|
||||
|
||||
func (iterator IBioSequence) IsNil() bool {
|
||||
return iterator.pointer == nil
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Add(n int) {
|
||||
iterator.pointer.all_done.Add(n)
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Done() {
|
||||
iterator.pointer.all_done.Done()
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Wait() {
|
||||
iterator.pointer.all_done.Wait()
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Channel() chan *obiseq.BioSequence {
|
||||
return iterator.pointer.channel
|
||||
}
|
||||
func (iterator IBioSequence) PChannel() *chan *obiseq.BioSequence {
|
||||
return &(iterator.pointer.channel)
|
||||
}
|
||||
|
||||
func MakeIBioSequence(sizes ...int) IBioSequence {
|
||||
buffsize := 1
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
i := __ibiosequence__{
|
||||
channel: make(chan *obiseq.BioSequence, buffsize),
|
||||
current: nil,
|
||||
pushBack: false,
|
||||
buffer_size: buffsize,
|
||||
finished: false,
|
||||
pFinished: nil,
|
||||
}
|
||||
|
||||
i.pFinished = &i.finished
|
||||
waiting := sync.WaitGroup{}
|
||||
i.all_done = &waiting
|
||||
ii := IBioSequence{&i}
|
||||
return ii
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Split() IBioSequence {
|
||||
|
||||
i := __ibiosequence__{
|
||||
channel: iterator.pointer.channel,
|
||||
current: nil,
|
||||
pushBack: false,
|
||||
finished: false,
|
||||
all_done: iterator.pointer.all_done,
|
||||
buffer_size: iterator.pointer.buffer_size,
|
||||
pFinished: iterator.pointer.pFinished,
|
||||
}
|
||||
|
||||
newIter := IBioSequence{&i}
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Next() bool {
|
||||
if iterator.IsNil() || *(iterator.pointer.pFinished) {
|
||||
iterator.pointer.current = nil
|
||||
return false
|
||||
}
|
||||
|
||||
if iterator.pointer.pushBack {
|
||||
iterator.pointer.pushBack = false
|
||||
return true
|
||||
}
|
||||
|
||||
next, ok := (<-iterator.pointer.channel)
|
||||
|
||||
if ok {
|
||||
iterator.pointer.current = next
|
||||
return true
|
||||
}
|
||||
|
||||
iterator.pointer.current = nil
|
||||
*iterator.pointer.pFinished = true
|
||||
return false
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) PushBack() {
|
||||
if !(iterator.pointer.current == nil) {
|
||||
iterator.pointer.pushBack = true
|
||||
}
|
||||
}
|
||||
|
||||
// The 'Get' method returns the instance of BioSequence
|
||||
// currently pointed by the iterator. You have to use the
|
||||
// 'Next' method to move to the next entry before calling
|
||||
// 'Get' to retreive the following instance.
|
||||
func (iterator IBioSequence) Get() *obiseq.BioSequence {
|
||||
return iterator.pointer.current
|
||||
}
|
||||
|
||||
// Finished returns 'true' value if no more data is available
|
||||
// from the iterator.
|
||||
func (iterator IBioSequence) Finished() bool {
|
||||
return *iterator.pointer.pFinished
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) BufferSize() int {
|
||||
return iterator.pointer.buffer_size
|
||||
}
|
||||
|
||||
// The IBioSequenceBatch converts a IBioSequence iterator
|
||||
// into an iterator oveer batches oof sequences. By default
|
||||
// the size of a batch is of 100 sequences and the iterator
|
||||
// implements a buffer equal to that of the source iterator.
|
||||
// These defaults can be overriden by specifying one or two
|
||||
// optional parametters at the method call. The first one
|
||||
// indicates the batch size. The second optional parametter
|
||||
// indicates the size of the buffer.
|
||||
func (iterator IBioSequence) IBioSequenceBatch(sizes ...int) IBioSequenceBatch {
|
||||
batchsize := 100
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
batchsize = sizes[0]
|
||||
}
|
||||
if len(sizes) > 1 {
|
||||
buffsize = sizes[1]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequenceBatch(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.WaitAndClose()
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for j := 0; !iterator.Finished(); j++ {
|
||||
batch := BioSequenceBatch{
|
||||
slice: obiseq.MakeBioSequenceSlice(),
|
||||
order: j}
|
||||
for i := 0; i < batchsize && iterator.Next(); i++ {
|
||||
seq := iterator.Get()
|
||||
batch.slice = append(batch.slice, seq)
|
||||
}
|
||||
newIter.pointer.channel <- batch
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) IBioSequence(sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for iterator.Next() {
|
||||
s := iterator.Get()
|
||||
newIter.pointer.channel <- s
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Skip(n int, sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for i := 0; iterator.Next(); i++ {
|
||||
if i >= n {
|
||||
s := iterator.Get()
|
||||
newIter.pointer.channel <- s
|
||||
}
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Head(n int, sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
not_done := true
|
||||
for i := 0; iterator.Next(); i++ {
|
||||
if i < n {
|
||||
s := iterator.Get()
|
||||
newIter.pointer.channel <- s
|
||||
} else {
|
||||
if not_done {
|
||||
newIter.Done()
|
||||
not_done = false
|
||||
}
|
||||
}
|
||||
}
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
// The 'Tail' method discard every data from the source iterator
|
||||
// except the 'n' last ones.
|
||||
func (iterator IBioSequence) Tail(n int, sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
buffseq := obiseq.MakeBioSequenceSlice()
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
var i int
|
||||
for i = 0; iterator.Next(); i++ {
|
||||
buffseq[i%n] = iterator.Get()
|
||||
}
|
||||
if i > n {
|
||||
for j := 0; j < n; j++ {
|
||||
newIter.Channel() <- buffseq[(i+j)%n]
|
||||
}
|
||||
|
||||
} else {
|
||||
for j := 0; j < i; j++ {
|
||||
newIter.Channel() <- buffseq[j]
|
||||
}
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) Concat(iterators ...IBioSequence) IBioSequence {
|
||||
|
||||
if len(iterators) == 0 {
|
||||
return iterator
|
||||
}
|
||||
|
||||
buffsize := iterator.BufferSize()
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for iterator.Next() {
|
||||
s := iterator.Get()
|
||||
newIter.pointer.channel <- s
|
||||
}
|
||||
|
||||
for _, iter := range iterators {
|
||||
for iter.Next() {
|
||||
s := iter.Get()
|
||||
newIter.pointer.channel <- s
|
||||
}
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
@ -153,34 +153,6 @@ func (iterator IBioSequenceBatch) MakeISliceWorker(worker SeqSliceWorker, sizes
|
||||
return newIter
|
||||
}
|
||||
|
||||
func (iterator IBioSequence) MakeIWorker(worker SeqWorker, sizes ...int) IBioSequence {
|
||||
buffsize := iterator.BufferSize()
|
||||
|
||||
if len(sizes) > 0 {
|
||||
buffsize = sizes[0]
|
||||
}
|
||||
|
||||
newIter := MakeIBioSequence(buffsize)
|
||||
|
||||
newIter.Add(1)
|
||||
|
||||
go func() {
|
||||
newIter.Wait()
|
||||
close(newIter.pointer.channel)
|
||||
}()
|
||||
|
||||
go func() {
|
||||
for iterator.Next() {
|
||||
seq := iterator.Get()
|
||||
seq = worker(seq)
|
||||
newIter.pointer.channel <- seq
|
||||
}
|
||||
newIter.Done()
|
||||
}()
|
||||
|
||||
return newIter
|
||||
}
|
||||
|
||||
func WorkerPipe(worker SeqWorker, sizes ...int) Pipeable {
|
||||
f := func(iterator IBioSequenceBatch) IBioSequenceBatch {
|
||||
return iterator.MakeIWorker(worker, sizes...)
|
||||
|
Reference in New Issue
Block a user