From f9d3983de28abe7cce306083986fe8527c44b82b Mon Sep 17 00:00:00 2001
From: Gregorius Bima Kharisma Wicaksana
 <51526537+bimakw@users.noreply.github.com>
Date: Tue, 6 Jan 2026 12:09:44 +0700
Subject: [PATCH] fix: generate IDs for HTML headings without id attribute
 (#36233)

This PR fixes #27383 where HTML headings like `<h1>Title</h1>` in
markdown files would have empty permalink anchors

---------

Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
---
 models/renderhelper/repo_file.go |   2 +-
 models/renderhelper/repo_wiki.go |   2 +-
 modules/markup/html.go           |   2 +-
 modules/markup/html_node.go      |  45 ++++++++++++-
 modules/markup/html_node_test.go | 104 +++++++++++++++++++++++++++++++
 modules/markup/render.go         |   9 +++
 6 files changed, 160 insertions(+), 4 deletions(-)
 create mode 100644 modules/markup/html_node_test.go

diff --git a/models/renderhelper/repo_file.go b/models/renderhelper/repo_file.go
index e0375ed280..f1df8e89e0 100644
--- a/models/renderhelper/repo_file.go
+++ b/models/renderhelper/repo_file.go
@@ -70,6 +70,6 @@ func NewRenderContextRepoFile(ctx context.Context, repo *repo_model.Repository,
 			"repo": helper.opts.DeprecatedRepoName,
 		})
 	}
-	rctx = rctx.WithHelper(helper)
+	rctx = rctx.WithHelper(helper).WithEnableHeadingIDGeneration(true)
 	return rctx
 }
diff --git a/models/renderhelper/repo_wiki.go b/models/renderhelper/repo_wiki.go
index b75f1b9701..218b1e4a67 100644
--- a/models/renderhelper/repo_wiki.go
+++ b/models/renderhelper/repo_wiki.go
@@ -71,7 +71,7 @@ func NewRenderContextRepoWiki(ctx context.Context, repo *repo_model.Repository,
 			"markupAllowShortIssuePattern": "true",
 		})
 	}
-	rctx = rctx.WithHelper(helper)
+	rctx = rctx.WithHelper(helper).WithEnableHeadingIDGeneration(true)
 	helper.ctx = rctx
 	return rctx
 }
diff --git a/modules/markup/html.go b/modules/markup/html.go
index 51afd4be00..9676241b54 100644
--- a/modules/markup/html.go
+++ b/modules/markup/html.go
@@ -314,7 +314,7 @@ func visitNode(ctx *RenderContext, procs []processor, node *html.Node) *html.Nod
 		return node.NextSibling
 	}
 
-	processNodeAttrID(node)
+	processNodeAttrID(ctx, node)
 	processFootnoteNode(ctx, node) // FIXME: the footnote processing should be done in the "footnote.go" renderer directly
 
 	if isEmojiNode(node) {
diff --git a/modules/markup/html_node.go b/modules/markup/html_node.go
index 4eb78fdd2b..066ee9711d 100644
--- a/modules/markup/html_node.go
+++ b/modules/markup/html_node.go
@@ -6,6 +6,8 @@ package markup
 import (
 	"strings"
 
+	"code.gitea.io/gitea/modules/markup/common"
+
 	"golang.org/x/net/html"
 )
 
@@ -23,16 +25,57 @@ func isAnchorHrefFootnote(s string) bool {
 	return strings.HasPrefix(s, "#fnref:user-content-") || strings.HasPrefix(s, "#fn:user-content-")
 }
 
-func processNodeAttrID(node *html.Node) {
+// isHeadingTag returns true if the node is a heading tag (h1-h6)
+func isHeadingTag(node *html.Node) bool {
+	return node.Type == html.ElementNode &&
+		len(node.Data) == 2 &&
+		node.Data[0] == 'h' &&
+		node.Data[1] >= '1' && node.Data[1] <= '6'
+}
+
+// getNodeText extracts the text content from a node and its children
+func getNodeText(node *html.Node) string {
+	var text strings.Builder
+	var extractText func(*html.Node)
+	extractText = func(n *html.Node) {
+		if n.Type == html.TextNode {
+			text.WriteString(n.Data)
+		}
+		for c := n.FirstChild; c != nil; c = c.NextSibling {
+			extractText(c)
+		}
+	}
+	extractText(node)
+	return text.String()
+}
+
+func processNodeAttrID(ctx *RenderContext, node *html.Node) {
 	// 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
 	for idx, attr := range node.Attr {
 		if attr.Key == "id" {
+			hasID = true
 			if !isAnchorIDUserContent(attr.Val) {
 				node.Attr[idx].Val = "user-content-" + attr.Val
 			}
 		}
 	}
+
+	// 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 != "" {
+			// 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})
+			}
+		}
+	}
 }
 
 func processFootnoteNode(ctx *RenderContext, node *html.Node) {
diff --git a/modules/markup/html_node_test.go b/modules/markup/html_node_test.go
new file mode 100644
index 0000000000..007e3c2a12
--- /dev/null
+++ b/modules/markup/html_node_test.go
@@ -0,0 +1,104 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package markup
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestProcessNodeAttrID_HTMLHeadingWithoutID(t *testing.T) {
+	// Test that HTML headings without id get an auto-generated id from their text content
+	// when EnableHeadingIDGeneration is true (for repo files and wiki pages)
+	testCases := []struct {
+		name     string
+		input    string
+		expected string
+	}{
+		{
+			name:     "h1 without id",
+			input:    `<h1>Heading without ID</h1>`,
+			expected: `<h1 id="user-content-heading-without-id">Heading without ID</h1>`,
+		},
+		{
+			name:     "h2 without id",
+			input:    `<h2>Another Heading</h2>`,
+			expected: `<h2 id="user-content-another-heading">Another Heading</h2>`,
+		},
+		{
+			name:     "h3 without id",
+			input:    `<h3>Third Level</h3>`,
+			expected: `<h3 id="user-content-third-level">Third Level</h3>`,
+		},
+		{
+			name:     "h1 with existing id should keep it",
+			input:    `<h1 id="my-custom-id">Heading with ID</h1>`,
+			expected: `<h1 id="user-content-my-custom-id">Heading with ID</h1>`,
+		},
+		{
+			name:     "h1 with user-content prefix should not double prefix",
+			input:    `<h1 id="user-content-already-prefixed">Already Prefixed</h1>`,
+			expected: `<h1 id="user-content-already-prefixed">Already Prefixed</h1>`,
+		},
+		{
+			name:     "heading with special characters",
+			input:    `<h1>What is Wine Staging?</h1>`,
+			expected: `<h1 id="user-content-what-is-wine-staging">What is Wine Staging?</h1>`,
+		},
+		{
+			name:     "heading with nested elements",
+			input:    `<h2><strong>Bold</strong> and <em>Italic</em></h2>`,
+			expected: `<h2 id="user-content-bold-and-italic"><strong>Bold</strong> and <em>Italic</em></h2>`,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			var result strings.Builder
+			ctx := NewTestRenderContext().WithEnableHeadingIDGeneration(true)
+			err := PostProcessDefault(ctx, strings.NewReader(tc.input), &result)
+			assert.NoError(t, err)
+			assert.Equal(t, tc.expected, strings.TrimSpace(result.String()))
+		})
+	}
+}
+
+func TestProcessNodeAttrID_SkipHeadingIDForComments(t *testing.T) {
+	// Test that HTML headings in comment-like contexts (issue comments)
+	// do NOT get auto-generated IDs to avoid duplicate IDs on pages with multiple documents.
+	// This is controlled by EnableHeadingIDGeneration which defaults to false.
+	testCases := []struct {
+		name     string
+		input    string
+		expected string
+	}{
+		{
+			name:     "h1 without id in comment context",
+			input:    `<h1>Heading without ID</h1>`,
+			expected: `<h1>Heading without ID</h1>`,
+		},
+		{
+			name:     "h2 without id in comment context",
+			input:    `<h2>Another Heading</h2>`,
+			expected: `<h2>Another Heading</h2>`,
+		},
+		{
+			name:     "h1 with existing id should still be prefixed",
+			input:    `<h1 id="my-custom-id">Heading with ID</h1>`,
+			expected: `<h1 id="user-content-my-custom-id">Heading with ID</h1>`,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			var result strings.Builder
+			// Default context without EnableHeadingIDGeneration (simulates comment rendering)
+			err := PostProcessDefault(NewTestRenderContext(), strings.NewReader(tc.input), &result)
+			assert.NoError(t, err)
+			assert.Equal(t, tc.expected, strings.TrimSpace(result.String()))
+		})
+	}
+}
diff --git a/modules/markup/render.go b/modules/markup/render.go
index c645749065..12f002b0c6 100644
--- a/modules/markup/render.go
+++ b/modules/markup/render.go
@@ -54,6 +54,10 @@ type RenderOptions struct {
 
 	// used by external render. the router "/org/repo/render/..." will output the rendered content in a standalone page
 	InStandalonePage bool
+
+	// EnableHeadingIDGeneration controls whether to auto-generate IDs for HTML headings without id attribute.
+	// This should be enabled for repository files and wiki pages, but disabled for comments to avoid duplicate IDs.
+	EnableHeadingIDGeneration bool
 }
 
 // RenderContext represents a render context
@@ -112,6 +116,11 @@ func (ctx *RenderContext) WithInStandalonePage(v bool) *RenderContext {
 	return ctx
 }
 
+func (ctx *RenderContext) WithEnableHeadingIDGeneration(v bool) *RenderContext {
+	ctx.RenderOptions.EnableHeadingIDGeneration = v
+	return ctx
+}
+
 func (ctx *RenderContext) WithUseAbsoluteLink(v bool) *RenderContext {
 	ctx.RenderOptions.UseAbsoluteLink = v
 	return ctx