fix: coalesce duplicate class attributes in CodeBlock and Heading

html/coalesce_test.go

@@ -0,0 +1,64 @@
+package html
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestCoalesceClassAttrs(t *testing.T) {
+	tests := []struct {
+		name string
+		in   []string
+		want []string
+	}{
+		{
+			name: "no class attrs",
+			in:   []string{`id="foo"`},
+			want: []string{`id="foo"`},
+		},
+		{
+			name: "single class attr unchanged",
+			in:   []string{`class="language-go"`},
+			want: []string{`class="language-go"`},
+		},
+		{
+			name: "two class attrs merged",
+			in:   []string{`class="language-yml"`, `class="my-class"`},
+			want: []string{`class="language-yml my-class"`},
+		},
+		{
+			name: "class attrs with other attrs preserved",
+			in:   []string{`class="language-go"`, `id="code1"`, `class="highlight"`},
+			want: []string{`class="language-go highlight"`, `id="code1"`},
+		},
+		{
+			name: "empty input",
+			in:   nil,
+			want: nil,
+		},
+		{
+			name: "no attrs at all",
+			in:   []string{},
+			want: []string{},
+		},
+		{
+			name: "empty class value stripped",
+			in:   []string{`class=""`, `class="my-class"`},
+			want: []string{`class="my-class"`},
+		},
+		{
+			name: "multi-word class values preserved",
+			in:   []string{`class="foo bar"`, `class="baz"`},
+			want: []string{`class="foo bar baz"`},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := coalesceClassAttrs(tt.in)
+			if !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("coalesceClassAttrs(%v) = %v, want %v", tt.in, got, tt.want)
+			}
+		})
+	}
+}

html/renderer.go

@@ -659,7 +659,6 @@ func (r *Renderer) MakeUniqueHeadingID(hdr *ast.Heading) string {
 func (r *Renderer) HeadingEnter(w io.Writer, hdr *ast.Heading) {
 	var attrs []string
 	var class string
-	// TODO(miek): add helper functions for coalescing these classes.
 	if hdr.IsTitleblock {
 		class = "title"
 	}
@@ -680,6 +679,7 @@ func (r *Renderer) HeadingEnter(w io.Writer, hdr *ast.Heading) {
 		attrs = append(attrs, attrID)
 	}
 	attrs = append(attrs, BlockAttrs(hdr)...)
+	attrs = coalesceClassAttrs(attrs)
 	r.CR(w)
 	r.OutTag(w, HeadingOpenTagFromLevel(hdr.Level), attrs)
 }
@@ -864,10 +864,9 @@ Parse:
 // CodeBlock writes ast.CodeBlock node
 func (r *Renderer) CodeBlock(w io.Writer, codeBlock *ast.CodeBlock) {
 	var attrs []string
-	// TODO(miek): this can add multiple class= attribute, they should be coalesced into one.
-	// This is probably true for some other elements as well
 	attrs = appendLanguageAttr(attrs, codeBlock.Info)
 	attrs = append(attrs, BlockAttrs(codeBlock)...)
+	attrs = coalesceClassAttrs(attrs)
 	r.CR(w)
 
 	r.Outs(w, "<pre>")
@@ -1338,6 +1337,44 @@ func BlockAttrs(node ast.Node) []string {
 	return s
 }
 
+// coalesceClassAttrs merges multiple class="..." attributes into a single one.
+// For example, class="language-go" and class="my-class" become class="language-go my-class".
+func coalesceClassAttrs(attrs []string) []string {
+	const prefix = `class="`
+	var classes []string
+	firstClassIdx := -1
+	for i, a := range attrs {
+		if strings.HasPrefix(a, prefix) && strings.HasSuffix(a, `"`) {
+			if firstClassIdx == -1 {
+				firstClassIdx = i
+			}
+			// extract the value between class=" and trailing "
+			val := a[len(prefix) : len(a)-1]
+			if val != "" {
+				classes = append(classes, val)
+			}
+		}
+	}
+	if firstClassIdx == -1 {
+		return attrs
+	}
+
+	merged := fmt.Sprintf(`class="%s"`, strings.Join(classes, " "))
+	out := make([]string, 0, len(attrs))
+	seenClass := false
+	for i, a := range attrs {
+		if strings.HasPrefix(a, prefix) && strings.HasSuffix(a, `"`) {
+			if !seenClass && i == firstClassIdx {
+				out = append(out, merged)
+				seenClass = true
+			}
+			continue
+		}
+		out = append(out, a)
+	}
+	return out
+}
+
 // TagWithAttributes creates a HTML tag with a given name and attributes
 func TagWithAttributes(name string, attrs []string) string {
 	s := name

html_renderer_test.go

@@ -70,6 +70,40 @@ func TestTagParagraphCode(t *testing.T) {
 	doTestsParam(t, tests, params)
 }
 
+// TestCodeBlockClassCoalescing verifies that when a code block has both
+// a language annotation and a custom class attribute, they are merged into
+// a single class attribute. See https://github.com/gomarkdown/markdown/issues/209.
+func TestCodeBlockClassCoalescing(t *testing.T) {
+	input := "``` yml\ntext: something\n```\n"
+
+	p := parser.NewWithExtensions(parser.FencedCode)
+	doc := p.Parse([]byte(input))
+
+	// Walk the AST and add a custom class to the CodeBlock node.
+	ast.WalkFunc(doc, func(node ast.Node, entering bool) ast.WalkStatus {
+		if cb, ok := node.(*ast.CodeBlock); ok {
+			cb.Attribute = &ast.Attribute{
+				Classes: [][]byte{[]byte("my-class")},
+			}
+		}
+		return ast.GoToNext
+	})
+
+	renderer := html.NewRenderer(html.RendererOptions{})
+	got := string(Render(doc, renderer))
+
+	// Before the fix, this produced two separate class= attributes:
+	//   <code class="language-yml" class="my-class">
+	// After the fix, they should be coalesced:
+	//   <code class="language-yml my-class">
+	expected := `<pre><code class="language-yml my-class">text: something
+</code></pre>
+`
+	if got != expected {
+		t.Errorf("CodeBlock class coalescing failed.\nExpected:\n%s\nGot:\n%s", expected, got)
+	}
+}
+
 func TestRenderNodeHookLinkAttrs(t *testing.T) {
 	tests := []string{
 		`[Click Me](gopher://foo.bar "Click Me")`,

testdata/mmark.test

@@ -140,7 +140,7 @@ A> This is an aside
 {.myclass3}
 .# Preface section
 +++
-<h1 class="special" class="myclass3">Preface section</h1>
+<h1 class="special myclass3">Preface section</h1>
 +++
 {.myclass4}
 A> hello