heuristics regexp engine configurable #2, skip rules at runtime

regex/oniguruma.go

@@ -7,6 +7,8 @@ import (
 	rubex "github.com/go-enry/go-oniguruma"
 )
 
+const Name = Oniguruma
+
 type EnryRegexp = *rubex.Regexp
 
 func MustCompile(s string) EnryRegexp {

regex/regex.go

@@ -0,0 +1,9 @@
+package regex
+
+// Package regex abstracts regular expression engine
+// that can be chosen at compile-time by a build tag.
+
+const (
+	RE2       = "RE2"
+	Oniguruma = "Oniguruma"
+)

regex/standard.go

@@ -7,6 +7,8 @@ import (
 	"regexp"
 )
 
+const Name = RE2
+
 type EnryRegexp = *regexp.Regexp
 
 func MustCompile(str string) EnryRegexp {
@@ -21,9 +23,13 @@ func MustCompileMultiline(s string) EnryRegexp {
 }
 
 // MustCompileRuby used for expressions with syntax not supported by RE2.
+// Now it's confusing as we use the result as [data/rule.Matcher] and
+//
+//	(*Matcher)(nil) != nil
+//
+// What is a better way for an expression to indicate unsupported syntax?
+// e.g. add .IsValidSyntax() to both, Matcher interface and EnryRegexp implementations?
 func MustCompileRuby(s string) EnryRegexp {
-	// TODO(bzz): find a bettee way?
-	// This will only trigger a panic on .Match() for the clients
 	return nil
 }
 

regex/standard_test.go

@@ -0,0 +1,27 @@
+//go:build !oniguruma
+// +build !oniguruma
+
+package regex
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestMustCompileMultiline(t *testing.T) {
+	const re = `^\.(.*)!$`
+	want := MustCompileMultiline(re)
+	assert.Equal(t, "(?m)"+re, want.String())
+
+	const s = `.one
+.two!
+thre!`
+	if !want.MatchString(s) {
+		t.Fatalf("MustCompileMultiline(`%s`) must match multiline %q\n", re, s)
+	}
+}
+
+func TestMustCompileRuby(t *testing.T) {
+	assert.Nil(t, MustCompileRuby(``))
+}