Work on iterators and recycling of biosequences

pkg/obiseq/batchiterator.go

@@ -39,6 +39,7 @@ func (batch BioSequenceBatch) IsNil() bool {
 type __ibiosequencebatch__ struct {
 	channel     chan BioSequenceBatch
 	current     BioSequenceBatch
+	pushBack    bool
 	all_done    *sync.WaitGroup
 	buffer_size int
 	finished    bool
@@ -61,9 +62,11 @@ func MakeIBioSequenceBatch(sizes ...int) IBioSequenceBatch {
 	i := __ibiosequencebatch__{
 		channel:     make(chan BioSequenceBatch, buffsize),
 		current:     NilBioSequenceBatch,
+		pushBack:    false,
 		buffer_size: buffsize,
 		finished:    false,
-		p_finished:  nil}
+		p_finished:  nil,
+	}
 	i.p_finished = &i.finished
 	waiting := sync.WaitGroup{}
 	i.all_done = &waiting
@@ -99,6 +102,7 @@ func (iterator IBioSequenceBatch) Split() IBioSequenceBatch {
 	i := __ibiosequencebatch__{
 		channel:     iterator.pointer.channel,
 		current:     NilBioSequenceBatch,
+		pushBack:    false,
 		all_done:    iterator.pointer.all_done,
 		buffer_size: iterator.pointer.buffer_size,
 		finished:    false,
@@ -111,6 +115,12 @@ func (iterator IBioSequenceBatch) Next() bool {
 	if *(iterator.pointer.p_finished) {
 		return false
 	}
+
+	if iterator.pointer.pushBack {
+		iterator.pointer.pushBack = false
+		return true
+	}
+
 	next, ok := (<-iterator.pointer.channel)
 
 	if ok {
@@ -123,6 +133,12 @@ func (iterator IBioSequenceBatch) Next() bool {
 	return false
 }
 
+func (iterator IBioSequenceBatch) PushBack() {
+	if !iterator.pointer.current.IsNil() {
+		iterator.pointer.pushBack = true
+	}
+}
+
 // The 'Get' method returns the instance of BioSequenceBatch
 // currently pointed by the iterator. You have to use the
 // 'Next' method to move to the next entry before calling
@@ -303,14 +319,14 @@ func (iterator IBioSequenceBatch) Rebatch(size int, sizes ...int) IBioSequenceBa
 	return newIter
 }
 
-func (iterator IBioSequenceBatch) Destroy() {
+func (iterator IBioSequenceBatch) Recycle() {
 
 	log.Println("Start recycling of Bioseq objects")
 
 	for iterator.Next() {
 		batch := iterator.Get()
 		for _, seq := range batch.Slice() {
-			(&seq).Destroy()
+			(&seq).Recycle()
 		}
 	}
 	log.Println("End of the recycling of Bioseq objects")

pkg/obiseq/biosequence.go

@@ -44,7 +44,7 @@ func (s BioSequence) IsNil() bool {
 	return s.sequence == nil
 }
 
-func (s BioSequence) Reset() {
+func (s *BioSequence) Reset() {
 	s.sequence.id.Reset()
 	s.sequence.definition.Reset()
 	s.sequence.sequence.Reset()
@@ -168,6 +168,10 @@ func (s BioSequence) SetQualities(qualities Quality) {
 	s.sequence.qualities.Write(qualities)
 }
 
+func (s BioSequence) WriteQualities(data []byte) (int, error) {
+	return s.sequence.qualities.Write(data)
+}
+
 func (s BioSequence) Write(data []byte) (int, error) {
 	return s.sequence.sequence.Write(data)
 }

pkg/obiseq/iterator.go

@@ -10,10 +10,11 @@ import (
 type __ibiosequence__ struct {
 	channel     chan BioSequence
 	current     BioSequence
+	pushBack    bool
 	all_done    *sync.WaitGroup
 	buffer_size int
 	finished    bool
-	p_finished  *bool
+	pFinished   *bool
 }
 
 type IBioSequence struct {
@@ -55,10 +56,13 @@ func MakeIBioSequence(sizes ...int) IBioSequence {
 	i := __ibiosequence__{
 		channel:     make(chan BioSequence, buffsize),
 		current:     NilBioSequence,
+		pushBack:    false,
 		buffer_size: buffsize,
 		finished:    false,
-		p_finished:  nil}
-	i.p_finished = &i.finished
+		pFinished:   nil,
+	}
+
+	i.pFinished = &i.finished
 	waiting := sync.WaitGroup{}
 	i.all_done = &waiting
 	ii := IBioSequence{&i}
@@ -66,23 +70,32 @@ func MakeIBioSequence(sizes ...int) IBioSequence {
 }
 
 func (iterator IBioSequence) Split() IBioSequence {
+
 	i := __ibiosequence__{
 		channel:     iterator.pointer.channel,
 		current:     NilBioSequence,
+		pushBack:    false,
 		finished:    false,
 		all_done:    iterator.pointer.all_done,
 		buffer_size: iterator.pointer.buffer_size,
-		p_finished:  iterator.pointer.p_finished}
+		pFinished:   iterator.pointer.pFinished,
+	}
+
 	newIter := IBioSequence{&i}
 	return newIter
 }
 
 func (iterator IBioSequence) Next() bool {
-	if iterator.IsNil() || *(iterator.pointer.p_finished) {
+	if iterator.IsNil() || *(iterator.pointer.pFinished) {
 		iterator.pointer.current = NilBioSequence
 		return false
 	}
 
+	if iterator.pointer.pushBack {
+		iterator.pointer.pushBack = false
+		return true
+	}
+
 	next, ok := (<-iterator.pointer.channel)
 
 	if ok {
@@ -91,10 +104,16 @@ func (iterator IBioSequence) Next() bool {
 	}
 
 	iterator.pointer.current = NilBioSequence
-	*iterator.pointer.p_finished = true
+	*iterator.pointer.pFinished = true
 	return false
 }
 
+func (iterator IBioSequence) PushBack() {
+	if !iterator.pointer.current.IsNil() {
+		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
@@ -106,7 +125,7 @@ func (iterator IBioSequence) Get() BioSequence {
 // Finished returns 'true' value if no more data is available
 // from the iterator.
 func (iterator IBioSequence) Finished() bool {
-	return *iterator.pointer.p_finished
+	return *iterator.pointer.pFinished
 }
 
 func (iterator IBioSequence) BufferSize() int {

pkg/obiseq/pairedbatchiterator.go

@@ -55,6 +55,7 @@ func (batch PairedBioSequenceBatch) IsNil() bool {
 type __ipairedbiosequencebatch__ struct {
 	channel     chan PairedBioSequenceBatch
 	current     PairedBioSequenceBatch
+	pushBack    bool
 	all_done    *sync.WaitGroup
 	buffer_size int
 	finished    bool
@@ -77,9 +78,12 @@ func MakeIPairedBioSequenceBatch(sizes ...int) IPairedBioSequenceBatch {
 	i := __ipairedbiosequencebatch__{
 		channel:     make(chan PairedBioSequenceBatch, buffsize),
 		current:     NilPairedBioSequenceBatch,
+		pushBack:    false,
 		buffer_size: buffsize,
 		finished:    false,
-		p_finished:  nil}
+		p_finished:  nil,
+	}
+
 	i.p_finished = &i.finished
 	waiting := sync.WaitGroup{}
 	i.all_done = &waiting
@@ -115,6 +119,7 @@ func (iterator IPairedBioSequenceBatch) Split() IPairedBioSequenceBatch {
 	i := __ipairedbiosequencebatch__{
 		channel:     iterator.pointer.channel,
 		current:     NilPairedBioSequenceBatch,
+		pushBack:    false,
 		all_done:    iterator.pointer.all_done,
 		buffer_size: iterator.pointer.buffer_size,
 		finished:    false,
@@ -127,6 +132,12 @@ func (iterator IPairedBioSequenceBatch) Next() bool {
 	if *(iterator.pointer.p_finished) {
 		return false
 	}
+
+	if iterator.pointer.pushBack {
+		iterator.pointer.pushBack = false
+		return true
+	}
+
 	next, ok := (<-iterator.pointer.channel)
 
 	if ok {
@@ -139,6 +150,12 @@ func (iterator IPairedBioSequenceBatch) Next() bool {
 	return false
 }
 
+func (iterator IPairedBioSequenceBatch) PushBack() {
+	if !iterator.pointer.current.IsNil() {
+		iterator.pointer.pushBack = true
+	}
+}
+
 // The 'Get' method returns the instance of BioSequenceBatch
 // currently pointed by the iterator. You have to use the
 // 'Next' method to move to the next entry before calling

pkg/obiseq/pool.go

@@ -14,7 +14,6 @@ var __bioseq__pool__ = sync.Pool{
 
 func MakeEmptyBioSequence() BioSequence {
 	bs := BioSequence{__bioseq__pool__.Get().(*__sequence__)}
-	bs.Reset()
 	return bs
 }
 
@@ -23,12 +22,13 @@ func MakeBioSequence(id string,
 	definition string) BioSequence {
 	bs := MakeEmptyBioSequence()
 	bs.SetId(id)
-	bs.SetSequence(sequence)
+	bs.Write(sequence)
 	bs.SetDefinition(definition)
 	return bs
 }
 
-func (sequence *BioSequence) Destroy() {
+func (sequence *BioSequence) Recycle() {
+	sequence.Reset()
 	__bioseq__pool__.Put(sequence.sequence)
 	sequence.sequence = nil
 }

pkg/obiseq/workers.go

@@ -84,9 +84,10 @@ func (iterator IBioSequenceBatch) MakeIWorker(worker SeqWorker, sizes ...int) IB
 	}
 
 	log.Println("Start of the batch workers")
-	for i := 0; i < nworkers; i++ {
+	for i := 0; i < nworkers-1; i++ {
 		go f(iterator.Split())
 	}
+	go f(iterator)
 
 	return newIter
 }
@@ -126,9 +127,10 @@ func (iterator IBioSequenceBatch) MakeISliceWorker(worker SeqSliceWorker, sizes
 	}
 
 	log.Println("Start of the batch slice workers")
-	for i := 0; i < nworkers; i++ {
+	for i := 0; i < nworkers - 1; i++ {
 		go f(iterator.Split())
 	}
+	go f(iterator)
 
 	return newIter
 }