Fix markup heading parsing, fix emphasis parsing (go-gitea#36284)

Fixes go-gitea#36106, fix go-gitea#17958

---------

Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>

Author: heathdutton

modules/htmlutil/html.go

@@ -6,6 +6,7 @@ package htmlutil
 import (
 	"fmt"
 	"html/template"
+	"io"
 	"slices"
 	"strings"
 )
@@ -31,7 +32,7 @@ func ParseSizeAndClass(defaultSize int, defaultClass string, others ...any) (int
 	return size, class
 }
 
-func HTMLFormat(s template.HTML, rawArgs ...any) template.HTML {
+func htmlFormatArgs(s template.HTML, rawArgs []any) []any {
 	if !strings.Contains(string(s), "%") || len(rawArgs) == 0 {
 		panic("HTMLFormat requires one or more arguments")
 	}
@@ -50,5 +51,35 @@ func HTMLFormat(s template.HTML, rawArgs ...any) template.HTML {
 			args[i] = template.HTMLEscapeString(fmt.Sprint(v))
 		}
 	}
-	return template.HTML(fmt.Sprintf(string(s), args...))
+	return args
+}
+
+func HTMLFormat(s template.HTML, rawArgs ...any) template.HTML {
+	return template.HTML(fmt.Sprintf(string(s), htmlFormatArgs(s, rawArgs)...))
+}
+
+func HTMLPrintf(w io.Writer, s template.HTML, rawArgs ...any) (int, error) {
+	return fmt.Fprintf(w, string(s), htmlFormatArgs(s, rawArgs)...)
+}
+
+func HTMLPrint(w io.Writer, s template.HTML) (int, error) {
+	return io.WriteString(w, string(s))
+}
+
+func HTMLPrintTag(w io.Writer, tag template.HTML, attrs map[string]string) (written int, err error) {
+	n, err := io.WriteString(w, "<"+string(tag))
+	written += n
+	if err != nil {
+		return written, err
+	}
+	for k, v := range attrs {
+		n, err = fmt.Fprintf(w, ` %s="%s"`, template.HTMLEscapeString(k), template.HTMLEscapeString(v))
+		written += n
+		if err != nil {
+			return written, err
+		}
+	}
+	n, err = io.WriteString(w, ">")
+	written += n
+	return written, err
 }

modules/markup/common/footnote.go

@@ -405,9 +405,9 @@ func (r *FootnoteHTMLRenderer) renderFootnoteLink(w util.BufWriter, source []byt
 	if entering {
 		n := node.(*FootnoteLink)
 		is := strconv.Itoa(n.Index)
-		_, _ = w.WriteString(`<sup id="fnref:`)
+		_, _ = w.WriteString(`<sup id="fnref:user-content-`)
 		_, _ = w.Write(n.Name)
-		_, _ = w.WriteString(`"><a href="#fn:`)
+		_, _ = w.WriteString(`"><a href="#fn:user-content-`)
 		_, _ = w.Write(n.Name)
 		_, _ = w.WriteString(`" class="footnote-ref" role="doc-noteref">`) // FIXME: here and below, need to keep the classes
 		_, _ = w.WriteString(is)
@@ -419,7 +419,7 @@ func (r *FootnoteHTMLRenderer) renderFootnoteLink(w util.BufWriter, source []byt
 func (r *FootnoteHTMLRenderer) renderFootnoteBackLink(w util.BufWriter, source []byte, node ast.Node, entering bool) (ast.WalkStatus, error) {
 	if entering {
 		n := node.(*FootnoteBackLink)
-		_, _ = w.WriteString(` <a href="#fnref:`)
+		_, _ = w.WriteString(` <a href="#fnref:user-content-`)
 		_, _ = w.Write(n.Name)
 		_, _ = w.WriteString(`" class="footnote-backref" role="doc-backlink">`)
 		_, _ = w.WriteString("&#x21a9;&#xfe0e;")
@@ -431,7 +431,7 @@ func (r *FootnoteHTMLRenderer) renderFootnoteBackLink(w util.BufWriter, source [
 func (r *FootnoteHTMLRenderer) renderFootnote(w util.BufWriter, source []byte, node ast.Node, entering bool) (ast.WalkStatus, error) {
 	n := node.(*Footnote)
 	if entering {
-		_, _ = w.WriteString(`<li id="fn:`)
+		_, _ = w.WriteString(`<li id="fn:user-content-`)
 		_, _ = w.Write(n.Name)
 		_, _ = w.WriteString(`" role="doc-endnote"`)
 		if node.Attributes() != nil {

modules/markup/html.go

@@ -12,7 +12,9 @@ import (
 	"strings"
 	"sync"
 
+	"code.gitea.io/gitea/modules/htmlutil"
 	"code.gitea.io/gitea/modules/markup/common"
+	"code.gitea.io/gitea/modules/translation"
 
 	"golang.org/x/net/html"
 	"golang.org/x/net/html/atom"
@@ -234,6 +236,49 @@ func postProcessString(ctx *RenderContext, procs []processor, content string) (s
 	return buf.String(), nil
 }
 
+func RenderTocHeadingItems(ctx *RenderContext, nodeDetailsAttrs map[string]string, out io.Writer) {
+	locale, ok := ctx.Value(translation.ContextKey).(translation.Locale)
+	if !ok {
+		locale = translation.NewLocale("")
+	}
+	_, _ = htmlutil.HTMLPrintTag(out, "details", nodeDetailsAttrs)
+	_, _ = htmlutil.HTMLPrintf(out, "<summary>%s</summary>\n", locale.TrString("toc"))
+
+	baseLevel := 6
+	for _, header := range ctx.TocHeadingItems {
+		if header.HeadingLevel < baseLevel {
+			baseLevel = header.HeadingLevel
+		}
+	}
+
+	currentLevel := baseLevel
+	indent := []byte{' ', ' '}
+	_, _ = htmlutil.HTMLPrint(out, "<ul>\n")
+	for _, header := range ctx.TocHeadingItems {
+		for currentLevel < header.HeadingLevel {
+			_, _ = out.Write(indent)
+			_, _ = htmlutil.HTMLPrint(out, "<ul>\n")
+			indent = append(indent, ' ', ' ')
+			currentLevel++
+		}
+		for currentLevel > header.HeadingLevel {
+			indent = indent[:len(indent)-2]
+			_, _ = out.Write(indent)
+			_, _ = htmlutil.HTMLPrint(out, "</ul>\n")
+			currentLevel--
+		}
+		_, _ = out.Write(indent)
+		_, _ = htmlutil.HTMLPrintf(out, "<li><a href=\"#%s\">%s</a></li>\n", header.AnchorID, header.InnerText)
+	}
+	for currentLevel > baseLevel {
+		indent = indent[:len(indent)-2]
+		_, _ = out.Write(indent)
+		_, _ = htmlutil.HTMLPrint(out, "</ul>\n")
+		currentLevel--
+	}
+	_, _ = htmlutil.HTMLPrint(out, "</ul>\n</details>\n")
+}
+
 func postProcess(ctx *RenderContext, procs []processor, input io.Reader, output io.Writer) error {
 	if !ctx.usedByRender && ctx.RenderHelper != nil {
 		defer ctx.RenderHelper.CleanUp()
@@ -284,6 +329,9 @@ func postProcess(ctx *RenderContext, procs []processor, input io.Reader, output
 	}
 
 	// Render everything to buf.
+	if ctx.TocShowInSection == TocShowInMain && len(ctx.TocHeadingItems) > 0 {
+		RenderTocHeadingItems(ctx, nil, output)
+	}
 	for _, node := range newNodes {
 		if err := html.Render(output, node); err != nil {
 			return fmt.Errorf("markup.postProcess: html.Render: %w", err)
@@ -314,7 +362,7 @@ func visitNode(ctx *RenderContext, procs []processor, node *html.Node) *html.Nod
 		return node.NextSibling
 	}
 
-	processNodeAttrID(ctx, node)
+	processNodeHeadingAndID(ctx, node)
 	processFootnoteNode(ctx, node) // FIXME: the footnote processing should be done in the "footnote.go" renderer directly
 
 	if isEmojiNode(node) {

modules/markup/html_node.go

@@ -14,7 +14,7 @@ import (
 func isAnchorIDUserContent(s string) bool {
 	// blackfridayExtRegex is for blackfriday extensions create IDs like fn:user-content-footnote
 	// old logic: blackfridayExtRegex = regexp.MustCompile(`[^:]*:user-content-`)
-	return strings.HasPrefix(s, "user-content-") || strings.Contains(s, ":user-content-")
+	return strings.HasPrefix(s, "user-content-") || strings.Contains(s, ":user-content-") || isAnchorIDFootnote(s)
 }
 
 func isAnchorIDFootnote(s string) bool {
@@ -34,7 +34,10 @@ func isHeadingTag(node *html.Node) bool {
 }
 
 // getNodeText extracts the text content from a node and its children
-func getNodeText(node *html.Node) string {
+func getNodeText(node *html.Node, cached **string) string {
+	if *cached != nil {
+		return **cached
+	}
 	var text strings.Builder
 	var extractText func(*html.Node)
 	extractText = func(n *html.Node) {
@@ -46,36 +49,56 @@ func getNodeText(node *html.Node) string {
 		}
 	}
 	extractText(node)
-	return text.String()
+	textStr := text.String()
+	*cached = &textStr
+	return textStr
 }
 
-func processNodeAttrID(ctx *RenderContext, node *html.Node) {
+func processNodeHeadingAndID(ctx *RenderContext, node *html.Node) {
+	// TODO: handle duplicate IDs, need to track existing IDs in the document
 	// Add user-content- to IDs and "#" links if they don't already have them,
 	// and convert the link href to a relative link to the host root
-	hasID := false
+	attrIDVal := ""
 	for idx, attr := range node.Attr {
 		if attr.Key == "id" {
-			hasID = true
-			if !isAnchorIDUserContent(attr.Val) {
-				node.Attr[idx].Val = "user-content-" + attr.Val
+			attrIDVal = attr.Val
+			if !isAnchorIDUserContent(attrIDVal) {
+				attrIDVal = "user-content-" + attrIDVal
+				node.Attr[idx].Val = attrIDVal
 			}
 		}
 	}
 
+	if !isHeadingTag(node) || !ctx.RenderOptions.EnableHeadingIDGeneration {
+		return
+	}
+
 	// For heading tags (h1-h6) without an id attribute, generate one from the text content.
 	// This ensures HTML headings like <h1>Title</h1> get proper permalink anchors
 	// matching the behavior of Markdown headings.
 	// Only enabled for repository files and wiki pages via EnableHeadingIDGeneration option.
-	if !hasID && isHeadingTag(node) && ctx.RenderOptions.EnableHeadingIDGeneration {
-		text := getNodeText(node)
-		if text != "" {
+	var nodeTextCached *string
+	if attrIDVal == "" {
+		nodeText := getNodeText(node, &nodeTextCached)
+		if nodeText != "" {
 			// Use the same CleanValue function used by Markdown heading ID generation
-			cleanedID := string(common.CleanValue([]byte(text)))
-			if cleanedID != "" {
-				node.Attr = append(node.Attr, html.Attribute{Key: "id", Val: "user-content-" + cleanedID})
+			attrIDVal = string(common.CleanValue([]byte(nodeText)))
+			if attrIDVal != "" {
+				attrIDVal = "user-content-" + attrIDVal
+				node.Attr = append(node.Attr, html.Attribute{Key: "id", Val: attrIDVal})
 			}
 		}
 	}
+	if ctx.TocShowInSection != "" {
+		nodeText := getNodeText(node, &nodeTextCached)
+		if nodeText != "" && attrIDVal != "" {
+			ctx.TocHeadingItems = append(ctx.TocHeadingItems, &TocHeadingItem{
+				HeadingLevel: int(node.Data[1] - '0'),
+				AnchorID:     attrIDVal,
+				InnerText:    nodeText,
+			})
+		}
+	}
 }
 
 func processFootnoteNode(ctx *RenderContext, node *html.Node) {

modules/markup/html_toc_test.go

@@ -0,0 +1,60 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package markup_test
+
+import (
+	"regexp"
+	"testing"
+
+	"code.gitea.io/gitea/modules/markup"
+	"code.gitea.io/gitea/modules/markup/markdown"
+	"code.gitea.io/gitea/modules/test"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestToCWithHTML(t *testing.T) {
+	defer test.MockVariableValue(&markup.RenderBehaviorForTesting.DisableAdditionalAttributes, true)()
+
+	t1 := `tag <a href="link">link</a> and <b>Bold</b>`
+	t2 := "code block `<a>`"
+	t3 := "markdown **bold**"
+	input := `---
+include_toc: true
+---
+
+# ` + t1 + `
+## ` + t2 + `
+#### ` + t3 + `
+## last
+`
+
+	renderCtx := markup.NewTestRenderContext().WithEnableHeadingIDGeneration(true)
+	resultHTML, err := markdown.RenderString(renderCtx, input)
+	assert.NoError(t, err)
+	result := string(resultHTML)
+	re := regexp.MustCompile(`(?s)<details class="frontmatter-content">.*?</details>`)
+	result = re.ReplaceAllString(result, "\n")
+	expected := `<details><summary>toc</summary>
+<ul>
+  <li><a href="#user-content-tag-link-and-bold" rel="nofollow">tag link and Bold</a></li>
+  <ul>
+    <li><a href="#user-content-code-block-a" rel="nofollow">code block &lt;a&gt;</a></li>
+    <ul>
+      <ul>
+        <li><a href="#user-content-markdown-bold" rel="nofollow">markdown bold</a></li>
+      </ul>
+    </ul>
+    <li><a href="#user-content-last" rel="nofollow">last</a></li>
+  </ul>
+</ul>
+</details>
+
+<h1 id="user-content-tag-link-and-bold">tag <a href="/link" rel="nofollow">link</a> and <b>Bold</b></h1>
+<h2 id="user-content-code-block-a">code block <code>&lt;a&gt;</code></h2>
+<h4 id="user-content-markdown-bold">markdown <strong>bold</strong></h4>
+<h2 id="user-content-last">last</h2>
+`
+	assert.Equal(t, expected, result)
+}

modules/markup/markdown/goldmark.go

@@ -41,11 +41,10 @@ func (g *ASTTransformer) applyElementDir(n ast.Node) {
 // Transform transforms the given AST tree.
 func (g *ASTTransformer) Transform(node *ast.Document, reader text.Reader, pc parser.Context) {
 	firstChild := node.FirstChild()
-	tocMode := ""
 	ctx := pc.Get(renderContextKey).(*markup.RenderContext)
 	rc := pc.Get(renderConfigKey).(*RenderConfig)
 
-	tocList := make([]Header, 0, 20)
+	tocMode := ""
 	if rc.yamlNode != nil {
 		metaNode := rc.toMetaNode(g)
 		if metaNode != nil {
@@ -60,8 +59,6 @@ func (g *ASTTransformer) Transform(node *ast.Document, reader text.Reader, pc pa
 		}
 
 		switch v := n.(type) {
-		case *ast.Heading:
-			g.transformHeading(ctx, v, reader, &tocList)
 		case *ast.Paragraph:
 			g.applyElementDir(v)
 		case *ast.List:
@@ -79,19 +76,18 @@ func (g *ASTTransformer) Transform(node *ast.Document, reader text.Reader, pc pa
 		return ast.WalkContinue, nil
 	})
 
-	showTocInMain := tocMode == "true" /* old behavior, in main view */ || tocMode == "main"
-	showTocInSidebar := !showTocInMain && tocMode != "false" // not hidden, not main, then show it in sidebar
-	if len(tocList) > 0 && (showTocInMain || showTocInSidebar) {
-		if showTocInMain {
-			tocNode := createTOCNode(tocList, rc.Lang, nil)
-			node.InsertBefore(node, firstChild, tocNode)
-		} else {
-			tocNode := createTOCNode(tocList, rc.Lang, map[string]string{"open": "open"})
-			ctx.SidebarTocNode = tocNode
+	if ctx.RenderOptions.EnableHeadingIDGeneration {
+		showTocInMain := tocMode == "true" /* old behavior, in main view */ || tocMode == "main"
+		showTocInSidebar := !showTocInMain && tocMode != "false" // not hidden, not main, then show it in sidebar
+		switch {
+		case showTocInMain:
+			ctx.TocShowInSection = markup.TocShowInMain
+		case showTocInSidebar:
+			ctx.TocShowInSection = markup.TocShowInSidebar
 		}
 	}
 
-	if len(rc.Lang) > 0 {
+	if rc.Lang != "" {
 		node.SetAttributeString("lang", []byte(rc.Lang))
 	}
 }