Correct the number of workers

Former-commit-id: febbccfb853263e0761ecfccb0f09c8c1bf88475

pkg/obitools/obiannotate/obiannotate.go

@@ -1,7 +1,7 @@
 package obiannotate
 
 import (
-	"log"
+	log "github.com/sirupsen/logrus"
 
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obicorazick"
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
@@ -22,6 +22,15 @@ func DeleteAttributesWorker(toBeDeleted []string) obiseq.SeqWorker {
 	return f
 }
 
+// func MatchPatternWorker(pattern string, errormax int, allowsIndel bool) obiseq.SeqWorker {
+// 	pat, err := obiapat.MakeApatPattern(pattern, errormax, allowsIndel)
+// 	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+// 		apats := obiapat.MakeApatSequence(s, false)
+// 		pat.BestMatch(apats, 0)
+// 		return s
+// 	}
+// }
+
 func ToBeKeptAttributesWorker(toBeKept []string) obiseq.SeqWorker {
 
 	d := make(map[string]bool, len(_keepOnly))
@@ -43,6 +52,58 @@ func ToBeKeptAttributesWorker(toBeKept []string) obiseq.SeqWorker {
 	return f
 }
 
+func CutSequenceWorker(from, to int, breakOnError bool) obiseq.SeqWorker {
+
+	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+		var f, t int
+
+		switch {
+		case from < 0:
+			f = s.Len() + from + 1
+		case from > 0:
+			f = from
+		}
+
+		switch {
+		case to < 0:
+			t = s.Len() + to + 1
+		case to > 0:
+			t = to
+		}
+
+		if from < 0 {
+			from = 0
+		}
+
+		if to >= s.Len() {
+			to = s.Len()
+		}
+
+		rep, err := s.Subsequence(f, t, false)
+		if err != nil {
+			if breakOnError {
+				log.Fatalf("Cannot cut sequence %s (%v)", s.Id(), err)
+			} else {
+				log.Warnf("Cannot cut sequence %s (%v), sequence discarded", s.Id(), err)
+				return nil
+			}
+		}
+		return rep
+	}
+
+	if from == 0 && to == 0 {
+		f = func(s *obiseq.BioSequence) *obiseq.BioSequence {
+			return s
+		}
+	}
+
+	if from > 0 {
+		from--
+	}
+
+	return f
+}
+
 func ClearAllAttributesWorker() obiseq.SeqWorker {
 	f := func(s *obiseq.BioSequence) *obiseq.BioSequence {
 		annot := s.Annotations()
@@ -81,7 +142,6 @@ 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 {
@@ -162,6 +222,13 @@ func CLIAnnotationWorker() obiseq.SeqWorker {
 		annotator = annotator.ChainWorkers(w)
 	}
 
+	if CLIHasCut() {
+		from, to := CLICut()
+		w := CutSequenceWorker(from, to, false)
+
+		annotator = annotator.ChainWorkers(w)
+	}
+
 	return annotator
 }
 
@@ -170,7 +237,7 @@ func CLIAnnotationPipeline() obiiter.Pipeable {
 	predicate := obigrep.CLISequenceSelectionPredicate()
 	worker := CLIAnnotationWorker()
 
-	annotator := obiseq.SeqToSliceConditionalWorker(worker, predicate, true)
+	annotator := obiseq.SeqToSliceConditionalWorker(worker, predicate, true, false)
 	f := obiiter.SliceWorkerPipe(annotator, obioptions.CLIParallelWorkers())
 
 	return f

pkg/obitools/obiannotate/options.go

@@ -1,11 +1,12 @@
 package obiannotate
 
 import (
-	"io/ioutil"
-	"log"
 	"os"
+	"strconv"
 	"strings"
 
+	log "github.com/sirupsen/logrus"
+
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obitools/obiconvert"
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obitools/obigrep"
 	"github.com/DavidGamba/go-getoptions"
@@ -22,9 +23,11 @@ var _clearAll = false
 var _setSeqLength = false
 var _uniqueID = false
 var _ahoCorazick = ""
+var _pattern = ""
 var _lcaSlot = ""
 var _lcaError = 0.0
 var _setId = ""
+var _cut = ""
 
 func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	// options.BoolVar(&_addRank, "seq-rank", _addRank,
@@ -42,6 +45,13 @@ func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	options.StringVar(&_ahoCorazick, "aho-corasick", _ahoCorazick,
 		options.Description("Adds an aho-corasick attribut with the count of matches of the provided patterns."))
 
+	options.StringVar(&_pattern, "pattern", _pattern,
+		options.Description("Adds a pattern attribut containing the pattern, a pattern_match slot "+
+			"indicating the matched sequence, "+
+			"and a pattern_error slot indicating the number difference between the pattern and the match "+
+			"to the sequence.",
+		))
+
 	options.StringVar(&_lcaSlot, "add-lca-in", _lcaSlot,
 		options.ArgName("SLOT_NAME"),
 		options.Description("From the taxonomic annotation of the sequence (taxid slot or merged_taxid slot), "+
@@ -59,6 +69,10 @@ func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 			"estimated LCA."),
 	)
 
+	options.StringVar(&_cut, "cut", _cut,
+		options.ArgName("###:###"),
+		options.Description("A pattern decribing how to cut the sequence"))
+
 	// options.BoolVar(&_uniqueID, "uniq-id", _uniqueID,
 	// 	options.Description("Forces sequence record ids to be unique."),
 	// )
@@ -133,10 +147,9 @@ func CLIHasSetId() bool {
 }
 
 func CLSetIdExpression() string {
-	return _setId 
+	return _setId
 }
 
-
 func CLIHasAttributeToBeRenamed() bool {
 	return len(_toBeRenamed) > 0
 }
@@ -191,7 +204,7 @@ func CLIHasAhoCorasick() bool {
 }
 
 func CLIAhoCorazick() []string {
-	content, err := ioutil.ReadFile(_ahoCorazick)
+	content, err := os.ReadFile(_ahoCorazick)
 	if err != nil {
 		log.Fatalln("Cannot open file ", _ahoCorazick)
 	}
@@ -221,3 +234,33 @@ func CLIHasAddLCA() bool {
 func CLILCAThreshold() float64 {
 	return 1 - _lcaError
 }
+
+func CLICut() (int, int) {
+	if _cut == "" {
+		return 0, 0
+	}
+	values := strings.Split(_cut, ":")
+
+	if len(values) != 2 {
+		log.Fatalf("Invalid cut value %s. value should be of the form start:end", _cut)
+	}
+
+	start, err := strconv.Atoi(values[0])
+
+	if err != nil {
+		log.Fatalf("Invalid cut value %s. value %s should be an integer", _cut, values[0])
+	}
+	end, err := strconv.Atoi(values[1])
+
+	if err != nil {
+		log.Fatalf("Invalid cut value %s. value %s should be an integer", _cut, values[1])
+	}
+
+	return start, end
+}
+
+func CLIHasCut() bool {
+	f, t := CLICut()
+
+	return f != 0 && t != 0
+}