Fix edge case in truncate function allowing too long slugs (#74)

Fix #72

* Add more TestSlugMakeSmartTruncate cases

Co-authored-by: Romain Poirot <romain.poirot@gitguardian.com>
Co-authored-by: Dobrosław Żybort <matrixik@gmail.com>

slug.go

@@ -160,25 +160,19 @@ func SubstituteRune(s string, sub map[rune]string) string {
 }
 
 func smartTruncate(text string) string {
-	if len(text) < MaxLength {
+	if len(text) <= MaxLength {
 		return text
 	}
 
-	var truncated string
-	words := strings.SplitAfter(text, "-")
-	// If MaxLength is smaller than length of the first word return word
-	// truncated after MaxLength.
-	if len(words[0]) > MaxLength {
-		return words[0][:MaxLength]
-	}
-	for _, word := range words {
-		if len(truncated)+len(word)-1 <= MaxLength {
-			truncated = truncated + word
-		} else {
-			break
+	// If slug is too long, we need to find the last '-' before MaxLength, and
+	// we cut there.
+	// If we don't find any, we have only one word, and we cut at MaxLength.
+	for i := MaxLength; i >= 0; i-- {
+		if text[i] == '-' {
+			return text[:i]
 		}
 	}
-	return strings.Trim(truncated, "-")
+	return text[:MaxLength]
 }
 
 // IsSlug returns True if provided text does not contain white characters,

slug_test.go

@@ -283,12 +283,39 @@ func TestSlugMakeSmartTruncate(t *testing.T) {
 		want          string
 		smartTruncate bool
 	}{
-		{"DOBROSLAWZYBORT", 100, "dobroslawzybort", true},
-		{"Dobroslaw Zybort", 100, "dobroslaw-zybort", true},
+		{"Dobrosław Żybort", 5, "dobro", true},
+		{"Dobroslaw Zybort", 9, "dobroslaw", true},
 		{"Dobroslaw Zybort", 12, "dobroslaw", true},
+		{"Dobroslaw Zybort", 15, "dobroslaw", true},
+		{"Dobroslaw Zybort", 16, "dobroslaw-zybort", true},
+		{"Dobroslaw Zybort", 17, "dobroslaw-zybort", true},
+		{"Dobroslaw Zybort", 100, "dobroslaw-zybort", true},
+		{"abc", 2, "ab", true},
+		{"-abc", 2, "ab", true},
+		{"abc-", 2, "ab", true},
+		{"abc", 3, "abc", true},
+		{"-abc", 3, "abc", true},
+		{"abc", 4, "abc", true},
+		{"abc-", 4, "abc", true},
+		{"-abc-", 4, "abc", true},
+		{"----abc----", 4, "abc", true},
+		{"abc-de", 4, "abc", true},
+		{"abc-de", 5, "abc", true},
+		{"abc-de", 6, "abc-de", true},
+		{"abc-de", 7, "abc-de", true},
+		{"abc-de-fg", 6, "abc-de", true},
+		{"abc-de-fg", 7, "abc-de", true},
+		{"abc-de-fg", 8, "abc-de", true},
+		{"abc-de-fg", 9, "abc-de-fg", true},
+		{"abc-de-fg", 10, "abc-de-fg", true},
+
+		{"DOBROSLAWZYBORT", 9, "dobroslaw", true},
+		{"DOBROSLAWZYBORT", 15, "dobroslawzybort", true},
+		{"DOBROSLAWZYBORT", 100, "dobroslawzybort", true},
 		{"  Dobroslaw     Zybort  ?", 12, "dobroslaw", true},
 		{"Ala ma 6 kotów.", 10, "ala-ma-6", true},
-		{"Dobrosław Żybort", 5, "dobro", true},
+
+		// No smart truncate
 		{"Long branch-name", 14, "long-branch-na", false},
 		{"Long branch-name", 12, "long-branch", false},
 	}