Merge pull request #612 from sergioperezcheco/fix/class-ascii-token-boundaries

Fix class helpers splitting non-ASCII whitespace class names

property.go

@@ -108,7 +108,7 @@ func (s *Selection) Html() (ret string, e error) {
 // AddClass adds the given class(es) to each element in the set of matched elements.
 // Multiple class names can be specified, separated by a space or via multiple arguments.
 func (s *Selection) AddClass(class ...string) *Selection {
-	classStr := strings.TrimSpace(strings.Join(class, " "))
+	classStr := strings.Trim(strings.Join(class, " "), " \t\r\n\f")
 
 	if classStr == "" {
 		return s
@@ -157,7 +157,7 @@ func (s *Selection) HasClass(class string) bool {
 func (s *Selection) RemoveClass(class ...string) *Selection {
 	var rclasses []string
 
-	classStr := strings.TrimSpace(strings.Join(class, " "))
+	classStr := strings.Trim(strings.Join(class, " "), " \t\r\n\f")
 	remove := classStr == ""
 
 	if !remove {
@@ -183,7 +183,7 @@ func (s *Selection) RemoveClass(class ...string) *Selection {
 // ToggleClass adds or removes the given class(es) for each element in the set of matched elements.
 // Multiple class names can be specified, separated by a space or via multiple arguments.
 func (s *Selection) ToggleClass(class ...string) *Selection {
-	classStr := strings.TrimSpace(strings.Join(class, " "))
+	classStr := strings.Trim(strings.Join(class, " "), " \t\r\n\f")
 
 	if classStr == "" {
 		return s
@@ -258,7 +258,10 @@ func getClassesAndAttr(n *html.Node) (classes string, attr *html.Attribute) {
 }
 
 func getClassesSlice(classes string) []string {
-	return strings.Fields(classes)
+	// HTML class tokens are separated by ASCII whitespace, not Unicode whitespace.
+	return strings.FieldsFunc(classes, func(r rune) bool {
+		return r == ' ' || r == '\t' || r == '\r' || r == '\n' || r == '\f'
+	})
 }
 
 func removeAttr(n *html.Node, attrName string) {
@@ -272,7 +275,7 @@ func removeAttr(n *html.Node, attrName string) {
 }
 
 func setClasses(n *html.Node, attr *html.Attribute, classes string) {
-	classes = strings.TrimSpace(classes)
+	classes = strings.Trim(classes, " \t\r\n\f")
 	if classes == "" {
 		removeAttr(n, "class")
 		return

property_unicode_class_test.go

@@ -0,0 +1,75 @@
+package goquery
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestClassOperationsPreserveNonASCIIWhitespace(t *testing.T) {
+	for _, whitespace := range []string{"\u00a0", "\u2003", "\u202f", "\v"} {
+		for _, class := range []string{"a" + whitespace + "b", whitespace + "a", "a" + whitespace, whitespace} {
+			for _, op := range []struct {
+				name string
+				run  func(*Selection, string) *Selection
+				want string
+			}{
+				{"add", func(s *Selection, c string) *Selection { return s.AddClass(c) }, class + " keep"},
+				{"remove", func(s *Selection, c string) *Selection { return s.RemoveClass(c) }, "keep"},
+				{"toggle", func(s *Selection, c string) *Selection { return s.ToggleClass(c) }, "keep"},
+			} {
+				t.Run(op.name+"/"+class, func(t *testing.T) {
+					doc, err := NewDocumentFromReader(strings.NewReader("<div></div>"))
+					if err != nil {
+						t.Fatal(err)
+					}
+					sel := doc.Find("div").SetAttr("class", class+" keep")
+					if !sel.HasClass(class) {
+						t.Fatal("existing class not recognized")
+					}
+					if op.run(sel, "\t"+class+"\r\n") != sel {
+						t.Fatal("lost chaining")
+					}
+					if got := sel.AttrOr("class", ""); got != op.want {
+						t.Fatalf("class = %q, want %q", got, op.want)
+					}
+					markup, err := OuterHtml(sel)
+					if err != nil {
+						t.Fatal(err)
+					}
+					reopened, err := NewDocumentFromReader(strings.NewReader(markup))
+					if err != nil {
+						t.Fatal(err)
+					}
+					if got := reopened.Find("div").AttrOr("class", ""); got != op.want {
+						t.Fatalf("round trip = %q, want %q", got, op.want)
+					}
+				})
+			}
+		}
+	}
+}
+
+func TestClassTokensASCIIWhitespace(t *testing.T) {
+	for _, sep := range []string{" ", "\t", "\r", "\n", "\f"} {
+		got := getClassesSlice(sep + "a\u00a0b" + sep + sep + "keep" + sep)
+		if len(got) != 2 || got[0] != "a\u00a0b" || got[1] != "keep" {
+			t.Errorf("separator %q: %q", sep, got)
+		}
+	}
+}
+
+func TestClassOperationsKeepUnicodeAtAttributeEdges(t *testing.T) {
+	doc, err := NewDocumentFromReader(strings.NewReader("<div></div>"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	sel := doc.Find("div")
+	sel.AddClass("\u00a0a\u00a0").AddClass("keep").RemoveClass("keep")
+	if got := sel.AttrOr("class", ""); got != "\u00a0a\u00a0" {
+		t.Fatalf("class = %q", got)
+	}
+	sel.RemoveClass()
+	if _, exists := sel.Attr("class"); exists {
+		t.Fatal("RemoveClass() did not clear attribute")
+	}
+}