fix(markup): resolve and proxy srcset links of <source> elements (#39708) (#39756)

Backport #39708

Co-authored-by: Basavaraj S m <basavaraj@digiscrypt.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>

modules/htmlutil/html_srcset.go

@@ -0,0 +1,105 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package htmlutil
+
+import (
+	"strings"
+)
+
+type SrcSetCandidate struct {
+	SrcLink string
+	SrcDesc []string
+}
+
+const (
+	asciiSpaces         = " \t\n\r\f"
+	asciiSpacesAndComma = " \t\n\r\f,"
+)
+
+func ParseSrcset(srcset string) (candidates []SrcSetCandidate) {
+	// ref: https://html.spec.whatwg.org/multipage/images.html#parse-a-srcset-attribute
+	// It can't simply use "split" because comma can appear in URLs and descriptors can be parenthesized and contain commas.
+	extractLink := func() (string, bool) {
+		srcset = strings.TrimLeft(srcset, asciiSpacesAndComma)
+		if srcset == "" {
+			return "", false
+		}
+		end := strings.IndexAny(srcset, asciiSpaces)
+		if end == -1 {
+			end = len(srcset)
+		}
+		link := srcset[:end]
+		srcset = srcset[end:]
+		if strings.HasSuffix(link, ",") {
+			return strings.TrimRight(link, ","), false
+		}
+		return link, true
+	}
+	extractDesc := func() (string, bool) {
+		srcset = strings.TrimLeft(srcset, asciiSpaces)
+		if srcset == "" {
+			return "", false
+		}
+		inParentheses := false
+		for i := range len(srcset) {
+			c := srcset[i]
+			if inParentheses {
+				inParentheses = c != ')'
+				continue
+			}
+			if c == '(' {
+				inParentheses = true
+				continue
+			}
+			if strings.IndexByte(asciiSpacesAndComma, c) != -1 {
+				desc := srcset[:i]
+				srcset = srcset[i+1:]
+				return desc, c != ','
+			}
+		}
+		desc := srcset
+		srcset = ""
+		return desc, false
+	}
+
+	for {
+		link, hasDesc := extractLink()
+		if link == "" {
+			break
+		}
+		var descriptors []string
+		if hasDesc {
+			for {
+				desc, hasMoreDesc := extractDesc()
+				if desc != "" {
+					descriptors = append(descriptors, desc)
+				}
+				if !hasMoreDesc {
+					break
+				}
+			}
+		}
+		candidates = append(candidates, SrcSetCandidate{SrcLink: link, SrcDesc: descriptors})
+	}
+	return candidates
+}
+
+func EncodeSrcset(candidates []SrcSetCandidate) string {
+	var sb strings.Builder
+	for i, c := range candidates {
+		if i != 0 {
+			sb.WriteString(", ")
+		}
+		link := c.SrcLink
+		if prefix, ok := strings.CutSuffix(link, ","); ok {
+			link = prefix + "%2C"
+		}
+		sb.WriteString(link)
+		for _, d := range c.SrcDesc {
+			sb.WriteByte(' ')
+			sb.WriteString(d)
+		}
+	}
+	return sb.String()
+}

modules/htmlutil/html_srcset_test.go

@@ -0,0 +1,81 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package htmlutil
+
+import (
+	"testing"
+
+	"gitea.dev/modules/util"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestSrcset(t *testing.T) {
+	for _, test := range []struct {
+		name       string
+		input      string
+		candidates []SrcSetCandidate
+		encoded    string
+	}{
+		{
+			name:       "bare URL",
+			input:      "image.png",
+			candidates: []SrcSetCandidate{{SrcLink: "image.png"}},
+		},
+		{
+			name:       "bare URLs",
+			input:      "a.png, b.png , c.png,",
+			candidates: []SrcSetCandidate{{SrcLink: "a.png"}, {SrcLink: "b.png"}, {SrcLink: "c.png"}},
+			encoded:    "a.png, b.png, c.png",
+		},
+		{
+			name:       "empty candidates",
+			input:      ", ,\ta.png 1x,, ,b.png 2x, ,",
+			candidates: []SrcSetCandidate{{SrcLink: "a.png", SrcDesc: []string{"1x"}}, {SrcLink: "b.png", SrcDesc: []string{"2x"}}},
+			encoded:    "a.png 1x, b.png 2x",
+		},
+		{
+			name:       "trailing commas in URL",
+			input:      "a.png,, b.png 2x",
+			candidates: []SrcSetCandidate{{SrcLink: "a.png"}, {SrcLink: "b.png", SrcDesc: []string{"2x"}}},
+			encoded:    "a.png, b.png 2x",
+		},
+		{
+			name:       "comma in URL",
+			input:      "comma,in,url.png",
+			candidates: []SrcSetCandidate{{SrcLink: "comma,in,url.png"}},
+		},
+		{
+			name:       "embedded data URL",
+			input:      "data:image/png;base64,abcd 1x, image.png 2x",
+			candidates: []SrcSetCandidate{{SrcLink: "data:image/png;base64,abcd", SrcDesc: []string{"1x"}}, {SrcLink: "image.png", SrcDesc: []string{"2x"}}},
+		},
+		{
+			name:       "parenthesized descriptor",
+			input:      "a.png (x,y) 1x, b.png 2x",
+			candidates: []SrcSetCandidate{{SrcLink: "a.png", SrcDesc: []string{"(x,y)", "1x"}}, {SrcLink: "b.png", SrcDesc: []string{"2x"}}},
+		},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			candidates := ParseSrcset(test.input)
+			assert.Equal(t, test.candidates, candidates)
+			encoded := EncodeSrcset(candidates)
+			assert.Equal(t, util.IfZero(test.encoded, test.input), encoded)
+		})
+	}
+
+	t.Run("encode trailing commas", func(t *testing.T) {
+		encoded := EncodeSrcset([]SrcSetCandidate{
+			{SrcLink: "image,"},
+			{SrcLink: "image,,", SrcDesc: []string{"2x"}},
+			{SrcLink: "image.svg?variant=light,", SrcDesc: []string{"3x"}},
+		})
+		assert.Equal(t, "image%2C, image,%2C 2x, image.svg?variant=light%2C 3x", encoded)
+		assert.Equal(t, []SrcSetCandidate{
+			{SrcLink: "image%2C"},
+			{SrcLink: "image,%2C", SrcDesc: []string{"2x"}},
+			{SrcLink: "image.svg?variant=light%2C", SrcDesc: []string{"3x"}},
+		}, ParseSrcset(encoded))
+	})
+}