Make the replace function of the eval language accepting regex

pkg/obicorazick/worker.go

@@ -2,21 +2,75 @@ package obicorazick
 
 import (
 	log "github.com/sirupsen/logrus"
-
+	"sync"
+	"os"
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiseq"
+	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obidefault"
 	"github.com/rrethy/ahocorasick"
+	"github.com/schollz/progressbar/v3"
 )
 
 func AhoCorazickWorker(slot string, patterns []string) obiseq.SeqWorker {
 
-	matcher := ahocorasick.CompileStrings(patterns)
+	sizebatch:=10000000
+	nmatcher := len(patterns) / sizebatch + 1
+	log.Infof("Building AhoCorasick %d matcher for %d patterns in slot %s", 
+			 nmatcher, len(patterns), slot)
+
+	if nmatcher == 0 {
+		log.Errorln("No patterns provided")
+	}
+	
+	matchers := make([]*ahocorasick.Matcher, nmatcher)
+	ieme := make(chan int) 
+	mutex := &sync.WaitGroup{}
+	npar := min(obidefault.ParallelWorkers(), nmatcher)
+	mutex.Add(npar)
+
+	pbopt := make([]progressbar.Option, 0, 5)
+	pbopt = append(pbopt,
+		progressbar.OptionSetWriter(os.Stderr),
+		progressbar.OptionSetWidth(15),
+		progressbar.OptionShowCount(),
+		progressbar.OptionShowIts(),
+		progressbar.OptionSetDescription("Building AhoCorasick matcher..."),
+	)
+
+	bar := progressbar.NewOptions(nmatcher, pbopt...)
+	bar.Add(0)
+
+	builder := func() {
+		for i := range ieme  {		
+		matchers[i] = ahocorasick.CompileStrings(patterns[i*sizebatch:min((i+1)*sizebatch,len(patterns))])
+		bar.Add(1)
+		}
+		mutex.Done()
+	}
+
+	for i := 0; i < npar; i++ {
+		go builder()
+	}
+
+	for i := 0; i < nmatcher; i++ {
+		ieme <- i
+	}
+
+	close(ieme)
+	mutex.Wait()
 
 	fslot := slot + "_Fwd"
 	rslot := slot + "_Rev"
 
 	f := func(s *obiseq.BioSequence) (obiseq.BioSequenceSlice, error) {
-		matchesF := len(matcher.FindAllByteSlice(s.Sequence()))
+		matchesF := 0
-		matchesR := len(matcher.FindAllByteSlice(s.ReverseComplement(false).Sequence()))
+		matchesR := 0
+		b := s.Sequence()
+		bc := s.ReverseComplement(false).Sequence()
+
+		for _, matcher := range matchers {
+			matchesF += len(matcher.FindAllByteSlice(b))
+			matchesR += len(matcher.FindAllByteSlice(bc))
+		}
 
 		log.Debugln("Macthes = ", matchesF, matchesR)
 		matches := matchesF + matchesR

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 = "67e5b6e"
+var _Commit = "0aec5ba"
 var _Version = "Release 4.4.0"
 
 // Version returns the version of the obitools package.

pkg/obiseq/language.go

@@ -3,6 +3,7 @@ package obiseq
 import (
 	"fmt"
 	"reflect"
+	"regexp"
 	"strings"
 
 	log "github.com/sirupsen/logrus"
@@ -204,6 +205,8 @@ var OBILang = gval.NewLanguage(
 		return scomp, nil
 	}),
 	gval.Function("replace", func(args ...interface{}) (interface{}, error) {
-		return strings.ReplaceAll(args[0].(string), args[1].(string), args[2].(string)), nil
+		pattern := regexp.MustCompile(args[1].(string))
+		results := pattern.ReplaceAllString(args[0].(string), args[2].(string))
+		return results, nil
 	}),
 )

pkg/obiutils/cast_interface.go

@@ -3,6 +3,7 @@ package obiutils
 import (
 	"fmt"
 	"reflect"
+	"strconv"
 
 	log "github.com/sirupsen/logrus"
 )
@@ -125,6 +126,12 @@ func InterfaceToInt(i interface{}) (val int, err error) {
 		val = int(t) // standardizes across systems
 	case uint64:
 		val = int(t) // standardizes across systems
+	case string:
+		rep, err := strconv.ParseInt(t, 10, 64)
+		if err != nil {
+			err = &NotAnFloat64{"value attribute cannot be casted to an int value"}
+		}
+		val = int(rep)
 	default:
 		err = &NotAnInteger{"value attribute cannot be casted to an integer"}
 	}
@@ -162,6 +169,11 @@ func InterfaceToFloat64(i interface{}) (val float64, err error) {
 		val = float64(t) // standardizes across systems
 	case uint64:
 		val = float64(t) // standardizes across systems
+	case string:
+		val, err = strconv.ParseFloat(t, 10)
+		if err != nil {
+			err = &NotAnFloat64{"value attribute cannot be casted to a float value"}
+		}
 	default:
 		err = &NotAnFloat64{"value attribute cannot be casted to a float value"}
 	}