From f74c0bd517fadfc48ce4092dbb4e807d6268ba59 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Fri, 27 Jan 2023 11:35:39 +0100
Subject: [PATCH] First prelease of obiannotate

---
 cmd/obitools/obiannotate/main.go        | 42 ++++++++++++++
 pkg/obitools/obiannotate/obiannotate.go | 75 +++++++++++++++++++++++++
 pkg/obitools/obiannotate/options.go     | 36 +++++++++++-
 3 files changed, 150 insertions(+), 3 deletions(-)
 create mode 100644 cmd/obitools/obiannotate/main.go

diff --git a/cmd/obitools/obiannotate/main.go b/cmd/obitools/obiannotate/main.go
new file mode 100644
index 0000000..a7d72c3
--- /dev/null
+++ b/cmd/obitools/obiannotate/main.go
@@ -0,0 +1,42 @@
+package main
+
+import (
+	"os"
+	"runtime/pprof"
+
+	log "github.com/sirupsen/logrus"
+
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obioptions"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obitools/obiannotate"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obitools/obiconvert"
+)
+
+func main() {
+
+	defer obiseq.LogBioSeqStatus()
+
+	// go tool pprof -http=":8000" ./obipairing ./cpu.pprof
+	f, err := os.Create("cpu.pprof")
+	if err != nil {
+		log.Fatal(err)
+	}
+	pprof.StartCPUProfile(f)
+	defer pprof.StopCPUProfile()
+
+	// go tool trace cpu.trace
+	// ftrace, err := os.Create("cpu.trace")
+	// if err != nil {
+	// 	log.Fatal(err)
+	// }
+	// trace.Start(ftrace)
+	// defer trace.Stop()
+
+	optionParser := obioptions.GenerateOptionParser(obiannotate.OptionSet)
+
+	_, args, _ := optionParser(os.Args)
+
+	sequences, _ := obiconvert.ReadBioSequences(args...)
+	annotator := obiannotate.CLIAnnotationPipeline()
+	obiconvert.WriteBioSequences(sequences.Pipe(annotator), true)
+}
diff --git a/pkg/obitools/obiannotate/obiannotate.go b/pkg/obitools/obiannotate/obiannotate.go
index a760842..a1f5c37 100644
--- a/pkg/obitools/obiannotate/obiannotate.go
+++ b/pkg/obitools/obiannotate/obiannotate.go
@@ -1,2 +1,77 @@
 package obiannotate
 
+import (
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obitools/obigrep"
+)
+
+func DeleteAttributesWorker(toBeDeleted []string) obiseq.SeqWorker {
+	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+		for _, k := range toBeDeleted {
+			s.DeleteAttribute(k)
+		}
+		return s
+	}
+
+	return f
+}
+
+func ToBeKeptAttributesWorker(toBeKept map[string]bool) obiseq.SeqWorker {
+
+	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+		annot := s.Annotations()
+		for key := range annot {
+			if _, ok := toBeKept[key]; !ok {
+				s.DeleteAttribute(key)
+			}
+		}
+		return s
+	}
+
+	return f
+}
+
+func RenameAttributeWorker(toBeRenamed map[string]string) obiseq.SeqWorker {
+	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+		for newName, oldName := range toBeRenamed {
+			s.RenameAttribute(newName, oldName)
+		}
+		return s
+	}
+
+	return f
+}
+
+func CLIAnnotationWorker() obiseq.SeqWorker {
+	var annotator obiseq.SeqWorker
+	annotator = nil
+
+	if CLIHasAttributeToBeRenamed() {
+		w := RenameAttributeWorker(CLIAttributeToBeRenamed())
+		annotator = annotator.ChainWorkers(w)
+	}
+
+	if CLIHasAttibuteToDelete() {
+		w := DeleteAttributesWorker(CLIAttibuteToDelete())
+		annotator = annotator.ChainWorkers(w)
+	}
+
+	if CLIHasToBeKeptAttributes() {
+		w := ToBeKeptAttributesWorker(CLIToBeKeptAttributes())
+		annotator = annotator.ChainWorkers(w)
+	}
+
+	return annotator
+}
+
+func CLIAnnotationPipeline() obiiter.Pipeable {
+
+	predicate := obigrep.CLISequenceSelectionPredicate()
+	worker := CLIAnnotationWorker()
+
+	annotator := obiseq.SeqToSliceConditionalWorker(worker, predicate, true)
+	f := obiiter.SliceWorkerPipe(annotator)
+
+	return f
+}
diff --git a/pkg/obitools/obiannotate/options.go b/pkg/obitools/obiannotate/options.go
index 7c8fd20..fb0018f 100644
--- a/pkg/obitools/obiannotate/options.go
+++ b/pkg/obitools/obiannotate/options.go
@@ -16,13 +16,12 @@ var _clearAll = false
 var _setSeqLength = false
 var _uniqueID = false
 
-func SequenceSelectionOptionSet(options *getoptions.GetOpt) {
+func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	options.BoolVar(&_addRank, "seq-rank", _addRank,
 		options.Description("Adds a new attribute named seq_rank to the sequence record indicating its entry number in the sequence file."),
 	)
 
 	options.BoolVar(&_clearAll, "clear", _clearAll,
-		options.Alias("C"),
 		options.Description("Clears all attributes associated to the sequence records."),
 	)
 
@@ -65,7 +64,7 @@ func SequenceSelectionOptionSet(options *getoptions.GetOpt) {
 func OptionSet(options *getoptions.GetOpt) {
 	obiconvert.OptionSet(options)
 	obigrep.SequenceSelectionOptionSet(options)
-	SequenceSelectionOptionSet(options)
+	SequenceAnnotationOptionSet(options)
 }
 
 // -S <KEY>:<PYTHON_EXPRESSION>, --set-tag=<KEY>:<PYTHON_EXPRESSION>
@@ -91,3 +90,34 @@ func OptionSet(options *getoptions.GetOpt) {
 
 // --uniq-id
 // Forces sequence record ids to be unique.
+
+func CLIHasAttributeToBeRenamed() bool {
+	return len(_toBeRenamed) > 0
+}
+
+func CLIAttributeToBeRenamed() map[string]string {
+	return _toBeRenamed
+}
+
+func CLIHasAttibuteToDelete() bool {
+	return len(_toBeDeleted) > 0
+}
+
+func CLIAttibuteToDelete() []string {
+	return _toBeDeleted
+}
+
+func CLIHasToBeKeptAttributes() bool {
+	return len(_keepOnly) > 0
+}
+
+func CLIToBeKeptAttributes() map[string]bool {
+	d := make(map[string]bool,len(_keepOnly))
+
+	for _,v := range _keepOnly {
+		d[v]=true
+	}
+
+	return d
+}
+