From c0ecaf90ab2f45f0d7cf78e23e4f3bf03ffea95a Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric@coissac.eu>
Date: Tue, 22 Apr 2025 18:35:51 +0200
Subject: [PATCH] Add the --number option to obiannotate

---
 cmd/obitools/obiannotate/main.go    |  5 +++
 pkg/obiiter/numbering.go            | 62 +++++++++++++++++++++++++++++
 pkg/obioptions/version.go           |  2 +-
 pkg/obiseq/language.go              |  2 +-
 pkg/obitools/obiannotate/options.go |  9 +++++
 5 files changed, 78 insertions(+), 2 deletions(-)
 create mode 100644 pkg/obiiter/numbering.go

diff --git a/cmd/obitools/obiannotate/main.go b/cmd/obitools/obiannotate/main.go
index b83dfc2..6006073 100644
--- a/cmd/obitools/obiannotate/main.go
+++ b/cmd/obitools/obiannotate/main.go
@@ -42,6 +42,11 @@ func main() {
 	obiconvert.OpenSequenceDataErrorMessage(args, err)
 
 	annotator := obiannotate.CLIAnnotationPipeline()
+
+	if obiannotate.CLIHasSetNumberFlag() {
+		sequences = sequences.NumberSequences(1, !obiconvert.CLINoInputOrder())
+	}
+
 	obiconvert.CLIWriteBioSequences(sequences.Pipe(annotator), true)
 
 	obiutils.WaitForLastPipe()
diff --git a/pkg/obiiter/numbering.go b/pkg/obiiter/numbering.go
new file mode 100644
index 0000000..4908b0d
--- /dev/null
+++ b/pkg/obiiter/numbering.go
@@ -0,0 +1,62 @@
+package obiiter
+
+import (
+	"sync"
+	"sync/atomic"
+
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
+)
+
+func (iter IBioSequence) NumberSequences(start int, forceReordering bool) IBioSequence {
+
+	next_first := &atomic.Int64{}
+	next_first.Store(int64(start))
+	lock := &sync.Mutex{}
+
+	w := obidefault.ParallelWorkers()
+	if forceReordering {
+		iter = iter.SortBatches()
+		w = 1
+	}
+
+	newIter := MakeIBioSequence()
+	newIter.Add(w)
+
+	is_paired := false
+
+	if iter.IsPaired() {
+		is_paired = true
+		newIter.MarkAsPaired()
+	}
+
+	number := func(iter IBioSequence) {
+		for iter.Next() {
+			batch := iter.Get()
+			seqs := batch.Slice()
+			lock.Lock()
+			start := int(next_first.Load())
+			next_first.Store(int64(start + len(seqs)))
+			lock.Unlock()
+			for i, seq := range seqs {
+				num := start + i
+				seq.SetAttribute("seq_number", num)
+				if is_paired {
+					seq.PairedWith().SetAttribute("seq_number", num)
+				}
+			}
+			newIter.Push(batch)
+		}
+		newIter.Done()
+	}
+
+	go func() {
+		newIter.WaitAndClose()
+	}()
+
+	for i := 1; i < w; i++ {
+		go number(iter.Split())
+	}
+	go number(iter)
+
+	return newIter
+}
diff --git a/pkg/obioptions/version.go b/pkg/obioptions/version.go
index 4415d4f..f18b5eb 100644
--- a/pkg/obioptions/version.go
+++ b/pkg/obioptions/version.go
@@ -8,7 +8,7 @@ import (
 // corresponds to the last commit, and not the one when the file will be
 // commited
 
-var _Commit = "0aec5ba"
+var _Commit = "a57cfda"
 var _Version = "Release 4.4.0"
 
 // Version returns the version of the obitools package.
diff --git a/pkg/obiseq/language.go b/pkg/obiseq/language.go
index 9292092..40e1a75 100644
--- a/pkg/obiseq/language.go
+++ b/pkg/obiseq/language.go
@@ -145,7 +145,7 @@ var OBILang = gval.NewLanguage(
 		ismap := obiutils.IsAMap(args[0])
 		return ismap, nil
 	}),
-	gval.Function("printf", func(args ...interface{}) (interface{}, error) {
+	gval.Function("sprintf", func(args ...interface{}) (interface{}, error) {
 		text := fmt.Sprintf(args[0].(string), args[1:]...)
 		return text, nil
 	}),
diff --git a/pkg/obitools/obiannotate/options.go b/pkg/obitools/obiannotate/options.go
index b3c02c6..437e124 100644
--- a/pkg/obitools/obiannotate/options.go
+++ b/pkg/obitools/obiannotate/options.go
@@ -32,6 +32,7 @@ var _cut = ""
 var _taxonomicPath = false
 var _withRank = false
 var _withScientificName = false
+var _withNumbering = false
 
 func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	// options.BoolVar(&_addRank, "seq-rank", _addRank,
@@ -46,6 +47,10 @@ func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 		options.Description("Adds attribute with seq_length as a key and sequence length as a value."),
 	)
 
+	options.BoolVar(&_withNumbering, "number", _withNumbering,
+		options.Description("Adds an attribute with seq_number as a key and an ordinal number starting from 1 as a value."),
+	)
+
 	options.StringVar(&_ahoCorazick, "aho-corasick", _ahoCorazick,
 		options.Description("Adds an aho-corasick attribut with the count of matches of the provided patterns."))
 
@@ -203,6 +208,10 @@ func CLIHasSetLengthFlag() bool {
 	return _setSeqLength
 }
 
+func CLIHasSetNumberFlag() bool {
+	return _withNumbering
+}
+
 func CLIHasClearAllFlag() bool {
 	return _clearAll
 }