From d46f6b06c51512c139ea4fef3355f37941b0b247 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Wed, 7 Jun 2023 17:50:10 +0200
Subject: [PATCH] several small changes

Former-commit-id: c1cdb95885e44fd6ee7d1c963860d7ab41230c96
---
 pkg/obioptions/options.go               |  5 ++++-
 pkg/obiseq/attributes.go                |  4 ++++
 pkg/obiseq/language.go                  |  4 ++++
 pkg/obitools/obiannotate/obiannotate.go |  6 ++++++
 pkg/obitools/obiannotate/options.go     | 14 ++++++++++++++
 pkg/obitools/obifind/options.go         |  7 -------
 6 files changed, 32 insertions(+), 8 deletions(-)

diff --git a/pkg/obioptions/options.go b/pkg/obioptions/options.go
index 6e810a7..f368bb2 100644
--- a/pkg/obioptions/options.go
+++ b/pkg/obioptions/options.go
@@ -39,6 +39,10 @@ func GenerateOptionParser(optionset ...func(*getoptions.GetOpt)) ArgumentParser
 		options.GetEnv("OBIMAXCPU"),
 		options.Description("Number of parallele threads computing the result"))
 
+	options.IntVar(&_BatchSize, "batch-size", _BatchSize,
+		options.GetEnv("OBIBATCHSIZE"),
+		options.Description("Number of sequence per batch for paralelle processing"))
+
 	for _, o := range optionset {
 		o(options)
 	}
@@ -100,7 +104,6 @@ func CLIReadParallelWorkers() int {
 	return int(float64(_MaxAllowedCPU) * float64(_ReadWorkerPerCore))
 }
 
-
 // CLIParallelWorkers returns the number of parallel workers requested by
 // the command line option --workers|-w.
 func CLIMaxCPU() int {
diff --git a/pkg/obiseq/attributes.go b/pkg/obiseq/attributes.go
index f34287f..c9be3d6 100644
--- a/pkg/obiseq/attributes.go
+++ b/pkg/obiseq/attributes.go
@@ -198,3 +198,7 @@ func (s *BioSequence) OBITagRefIndex() map[int]string {
 
 	return val
 }
+
+func (s *BioSequence) SetOBITagRefIndex(idx map[int]string) {
+	s.SetAttribute("obitag_ref_index", idx)
+}
\ No newline at end of file
diff --git a/pkg/obiseq/language.go b/pkg/obiseq/language.go
index b0abaf6..80f47eb 100644
--- a/pkg/obiseq/language.go
+++ b/pkg/obiseq/language.go
@@ -155,6 +155,10 @@ var OBILang = gval.NewLanguage(
 		text := fmt.Sprintf(args[0].(string), args[1:]...)
 		return text, nil
 	}),
+	gval.Function("gsub", func(args ...interface{}) (interface{}, error) {
+		text := strings.ReplaceAll(args[0].(string), args[1].(string), args[2].(string))
+		return text, nil
+	}),
 	gval.Function("subspc", func(args ...interface{}) (interface{}, error) {
 		text := strings.ReplaceAll(args[0].(string), " ", "_")
 		return text, nil
diff --git a/pkg/obitools/obiannotate/obiannotate.go b/pkg/obitools/obiannotate/obiannotate.go
index caff95e..7f19984 100644
--- a/pkg/obitools/obiannotate/obiannotate.go
+++ b/pkg/obitools/obiannotate/obiannotate.go
@@ -81,6 +81,7 @@ func EvalAttributeWorker(expression map[string]string) obiseq.SeqWorker {
 	return w
 }
 
+
 func AddTaxonAtRankWorker(taxonomy *obitax.Taxonomy, ranks ...string) obiseq.SeqWorker {
 	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
 		for _, r := range ranks {
@@ -111,6 +112,11 @@ func CLIAnnotationWorker() obiseq.SeqWorker {
 		annotator = annotator.ChainWorkers(w)
 	}
 
+	if CLIHasSetId() {
+		w := obiseq.EditIdWorker(CLSetIdExpression())
+		annotator = annotator.ChainWorkers(w)
+	}
+
 	if CLIHasAttibuteToDelete() {
 		w := DeleteAttributesWorker(CLIAttibuteToDelete())
 		annotator = annotator.ChainWorkers(w)
diff --git a/pkg/obitools/obiannotate/options.go b/pkg/obitools/obiannotate/options.go
index 3ba5fcf..8132be0 100644
--- a/pkg/obitools/obiannotate/options.go
+++ b/pkg/obitools/obiannotate/options.go
@@ -24,6 +24,7 @@ var _uniqueID = false
 var _ahoCorazick = ""
 var _lcaSlot = ""
 var _lcaError = 0.0
+var _setId = ""
 
 func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	// options.BoolVar(&_addRank, "seq-rank", _addRank,
@@ -47,6 +48,10 @@ func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 			"a new slot named <SLOT_NAME> is added with the taxid of the lowest common ancester corresponding "+
 			"to the current annotation."))
 
+	options.StringVar(&_setId, "set-identifier", _setId,
+		options.ArgName("EXPRESSION"),
+		options.Description("An expression used to assigned the new id of the sequence"))
+
 	options.Float64Var(&_lcaError, "lca-error", _lcaError,
 		options.ArgName("#.###"),
 		options.Description("Error rate tolerated on the taxonomical discription during the lowest common "+
@@ -123,6 +128,15 @@ func OptionSet(options *getoptions.GetOpt) {
 // --uniq-id
 // Forces sequence record ids to be unique.
 
+func CLIHasSetId() bool {
+	return _setId != ""
+}
+
+func CLSetIdExpression() string {
+	return _setId 
+}
+
+
 func CLIHasAttributeToBeRenamed() bool {
 	return len(_toBeRenamed) > 0
 }
diff --git a/pkg/obitools/obifind/options.go b/pkg/obitools/obifind/options.go
index d250ac2..285745c 100644
--- a/pkg/obitools/obifind/options.go
+++ b/pkg/obitools/obifind/options.go
@@ -36,13 +36,6 @@ func LoadTaxonomyOptionSet(options *getoptions.GetOpt, required, alternatiive bo
 			options.Alias("a"),
 			options.Description("Enable the search on all alternative names and not only scientific names."))
 	}
-	options.BoolVar(&__rank_list__, "rank-list", false,
-		options.Alias("l"),
-		options.Description("List every taxonomic rank available in the taxonomy."))
-
-	options.IntSliceVar(&__taxonomical_restriction__, "restrict-to-taxon", 1, 1,
-		options.Alias("r"),
-		options.Description("Restrict output to some subclades."))
 }
 
 func FilterTaxonomyOptionSet(options *getoptions.GetOpt) {