Preserve whitespace

entropy.go

@@ -20,8 +20,8 @@ const (
 
 // EntropyScrubber splits the input by Whitespace and matches each part's entropy with the configured threshold.
 type EntropyScrubber struct {
-	// Whitespace characters.
-	Whitespace string
+	// Whitespace runes.
+	Whitespace []rune
 
 	// Threshold for scrubbing. If not set the ideal entropy is calculated based on the length of the input string.
 	Threshold float64
@@ -38,8 +38,8 @@ func (es EntropyScrubber) Scrub(s string) string {
 		threshold  = es.Threshold
 		correction = es.Correction
 	)
-	if whitespace == "" {
-		whitespace = DefaultWhitespace
+	if len(whitespace) == 0 {
+		whitespace = []rune(DefaultWhitespace)
 	}
 	if threshold == 0 {
 		if correction == 0 {
@@ -51,7 +51,7 @@ func (es EntropyScrubber) Scrub(s string) string {
 	f := splitAfter(s, []rune(whitespace))
 	for i, p := range f {
 		if e := entropy(p); e >= threshold {
-			f[i] = Replacement
+			f[i] = replaceNoWhitespace(f[i], Replacement, whitespace)
 		}
 	}
 	return strings.Join(f, "")
@@ -76,7 +76,7 @@ func EntropyWithThreshold(s string, threshold float64) string {
 	f := strings.Fields(s)
 	for i, p := range f {
 		if entropy(strings.TrimSpace(p)) >= threshold {
-			f[i] = Replacement
+			f[i] = replaceNoWhitespace(f[i], Replacement, []rune(DefaultWhitespace))
 		}
 	}
 	return strings.Join(f, " ")
@@ -131,3 +131,21 @@ func splitAfter(s string, separators []rune) (out []string) {
 	}
 	return
 }
+
+func replaceNoWhitespace(s, r string, whitespace []rune) string {
+	var (
+		out = make([]rune, 0, len(s))
+		j   int
+		t   = []rune(r)
+		l   = len(t)
+	)
+	for _, c := range s {
+		if strings.ContainsRune(string(whitespace), c) {
+			out = append(out, c)
+		} else if j < l {
+			out = append(out, t[j])
+			j++
+		}
+	}
+	return string(out)
+}

entropy_test.go

@@ -2,20 +2,21 @@ package scrub
 
 import "testing"
 
+var testEntropy = []struct {
+	Test, Want string
+}{
+	{"", ""},
+	{"1223334444", "1223334444"},
+	{"password123", Replacement},
+	{"Tr0ub4dour&3", Replacement},
+	{"test correcthorsebatterystaple", "test correcthorsebatterystaple"},
+	{"test coRrecth0rseba++ery9.23.2007staple$", "test " + Replacement},
+	{"rWibMFACxAUGZmxhVncy test", Replacement + " test"},
+	{"test Ba9ZyWABu99[BK#6MBgbH88Tofv)vs$ test", "test " + Replacement + " test"},
+}
+
 func TestEntropy(t *testing.T) {
-	tests := []struct {
-		Test, Want string
-	}{
-		{"", ""},
-		{"1223334444", "1223334444"},
-		{"password123", Replacement},
-		{"Tr0ub4dour&3", Replacement},
-		{"test correcthorsebatterystaple", "test correcthorsebatterystaple"},
-		{"test coRrecth0rseba++ery9.23.2007staple$", "test " + Replacement},
-		{"rWibMFACxAUGZmxhVncy test", Replacement + " test"},
-		{"test Ba9ZyWABu99[BK#6MBgbH88Tofv)vs$ test", "test " + Replacement + " test"},
-	}
-	for _, test := range tests {
+	for _, test := range testEntropy {
 		t.Run(test.Test, func(it *testing.T) {
 			if v := Entropy(test.Test); v != test.Want {
 				it.Errorf("expected %q to return %q, got %q", test.Test, test.Want, v)
@@ -24,9 +25,20 @@ func TestEntropy(t *testing.T) {
 	}
 }
 
+func TestEntropyScrubber_Scrub(t *testing.T) {
+	var scrubber EntropyScrubber // using defaults
+	for _, test := range testEntropy {
+		t.Run(test.Test, func(it *testing.T) {
+			if v := scrubber.Scrub(test.Test); v != test.Want {
+				it.Errorf("expected %q to return %q, got %q", test.Test, test.Want, v)
+			}
+		})
+	}
+}
+
 func TestEntropyScrubber(t *testing.T) {
 	scrubber := EntropyScrubber{
-		Whitespace: DefaultWhitespace,
+		Whitespace: []rune(DefaultWhitespace),
 		Threshold:  DefaultEntropyThreshold,
 	}
 	tests := []struct {

password.go

@@ -8,6 +8,8 @@ const (
 	reCryptHash = `(?s)\$[-0-9a-z]+\$((?:rounds=\d+\$)?(?:` + reBase64 + `+)+\$?)`
 )
 
+// Good source of examples can be found here: https://hashcat.net/wiki/doku.php?id=example_hashes
+
 var (
 	// CryptHash can scrub hashes in common crypt formats, such as Apache and IANA crypt hashes.
 	CryptHash Scrubber = regexpScrubber{

scrub.go

@@ -45,7 +45,7 @@ var All = Scrubbers{
 	PEMDHParameters,
 	PEMPrivateKey,
 	EntropyScrubber{
-		Whitespace: DefaultWhitespace,
+		Whitespace: []rune(DefaultWhitespace),
 		Threshold:  DefaultEntropyThreshold,
 	},
 }