Merge pull request #632 from MichaelMure/data-input-cleanup

make sure every text input is safe and validated

2 files changed, 39 insertions, 3 deletions

diff --git a/util/text/transform.go b/util/text/transform.go
index 59dc4e03..395a57be 100644
--- a/util/text/transform.go
+++ b/util/text/transform.go
@@ -8,7 +8,7 @@ import (
 	"golang.org/x/text/transform"
 )
 
-func Cleanup(text string) (string, error) {
+func Cleanup(text string) string {
 	// windows new line, Github, really ?
 	text = strings.Replace(text, "\r\n", "\n", -1)
 
@@ -23,9 +23,33 @@ func Cleanup(text string) (string, error) {
 	}))
 	sanitized, _, err := transform.String(t, text)
 	if err != nil {
-		return "", err
+		// transform.String should never return an error as our transformer doesn't returns one.
+		// Confirmed with fuzzing.
+		panic(err)
 	}
 
 	// trim extra new line not displayed in the github UI but still present in the data
-	return strings.TrimSpace(sanitized), nil
+	return strings.TrimSpace(sanitized)
+}
+
+func CleanupOneLine(text string) string {
+	// remove all unicode control characters *including*
+	// '\n', '\r' and '\t'
+	t := runes.Remove(runes.Predicate(unicode.IsControl))
+	sanitized, _, err := transform.String(t, text)
+	if err != nil {
+		// transform.String should never return an error as our transformer doesn't returns one.
+		// Confirmed with fuzzing.
+		panic(err)
+	}
+
+	// trim extra new line not displayed in the github UI but still present in the data
+	return strings.TrimSpace(sanitized)
+}
+
+func CleanupOneLineArray(texts []string) []string {
+	for i := range texts {
+		texts[i] = CleanupOneLine(texts[i])
+	}
+	return texts
 }
diff --git a/util/text/validate.go b/util/text/validate.go
index 51e94fb4..4c3f7065 100644
--- a/util/text/validate.go
+++ b/util/text/validate.go
@@ -33,6 +33,18 @@ func Safe(s string) bool {
 	return true
 }
 
+// Safe will tell if a character in the string is considered unsafe
+// Currently trigger on all unicode control character
+func SafeOneLine(s string) bool {
+	for _, r := range s {
+		if unicode.IsControl(r) {
+			return false
+		}
+	}
+
+	return true
+}
+
 // ValidUrl will tell if the string contains what seems to be a valid URL
 func ValidUrl(s string) bool {
 	if strings.Contains(s, "\n") {