Refactor markup render system (#32533)

Remove unmaintainable sanitizer rules. No need to add special "class"
regexp rules anymore, use RenderInternal.SafeAttr instead, more details
(and examples) are in the tests

Author: wxiaoguang

modules/html/html.go

@@ -0,0 +1,48 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package htmlutil
+
+import (
+	"fmt"
+	"html/template"
+	"slices"
+)
+
+// ParseSizeAndClass get size and class from string with default values
+// If present, "others" expects the new size first and then the classes to use
+func ParseSizeAndClass(defaultSize int, defaultClass string, others ...any) (int, string) {
+	size := defaultSize
+	if len(others) >= 1 {
+		if v, ok := others[0].(int); ok && v != 0 {
+			size = v
+		}
+	}
+	class := defaultClass
+	if len(others) >= 2 {
+		if v, ok := others[1].(string); ok && v != "" {
+			if class != "" {
+				class += " "
+			}
+			class += v
+		}
+	}
+	return size, class
+}
+
+func HTMLFormat(s string, rawArgs ...any) template.HTML {
+	args := slices.Clone(rawArgs)
+	for i, v := range args {
+		switch v := v.(type) {
+		case nil, bool, int, int8, int16, int32, int64, uint, uint8, uint16, uint32, uint64, float32, float64, template.HTML:
+			// for most basic types (including template.HTML which is safe), just do nothing and use it
+		case string:
+			args[i] = template.HTMLEscapeString(v)
+		case fmt.Stringer:
+			args[i] = template.HTMLEscapeString(v.String())
+		default:
+			args[i] = template.HTMLEscapeString(fmt.Sprint(v))
+		}
+	}
+	return template.HTML(fmt.Sprintf(s, args...))
+}

modules/htmlutil/html.go

@@ -0,0 +1,15 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package htmlutil
+
+import (
+	"html/template"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestHTMLFormat(t *testing.T) {
+	assert.Equal(t, template.HTML("<a>&lt; < 1</a>"), HTMLFormat("<a>%s %s %d</a>", "<", template.HTML("<"), 1))
+}

modules/htmlutil/html_test.go

@@ -7,7 +7,6 @@ import (
 	"fmt"
 	"io"
 	"net/url"
-	"regexp"
 
 	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/setting"
@@ -38,10 +37,7 @@ const (
 
 // SanitizerRules implements markup.Renderer
 func (Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
-	return []setting.MarkupSanitizerRule{
-		{Element: "div", AllowAttr: "class", Regexp: regexp.MustCompile(playerClassName)},
-		{Element: "div", AllowAttr: playerSrcAttr},
-	}
+	return []setting.MarkupSanitizerRule{{Element: "div", AllowAttr: playerSrcAttr}}
 }
 
 // Render implements markup.Renderer
@@ -53,12 +49,5 @@ func (Renderer) Render(ctx *markup.RenderContext, _ io.Reader, output io.Writer)
 		ctx.Metas["BranchNameSubURL"],
 		url.PathEscape(ctx.RelativePath),
 	)
-
-	_, err := io.WriteString(output, fmt.Sprintf(
-		`<div class="%s" %s="%s"></div>`,
-		playerClassName,
-		playerSrcAttr,
-		rawURL,
-	))
-	return err
+	return ctx.RenderInternal.FormatWithSafeAttrs(output, `<div class="%s" %s="%s"></div>`, playerClassName, playerSrcAttr, rawURL)
 }

modules/markup/asciicast/asciicast.go

@@ -9,15 +9,27 @@ package common
 import (
 	"bytes"
 	"regexp"
+	"sync"
 
 	"github.com/yuin/goldmark"
 	"github.com/yuin/goldmark/ast"
 	"github.com/yuin/goldmark/parser"
 	"github.com/yuin/goldmark/text"
 	"github.com/yuin/goldmark/util"
+	"mvdan.cc/xurls/v2"
 )
 
-var wwwURLRegxp = regexp.MustCompile(`^www\.[-a-zA-Z0-9@:%._\+~#=]{2,256}\.[a-z]{2,6}((?:/|[#?])[-a-zA-Z0-9@:%_\+.~#!?&//=\(\);,'">\^{}\[\]` + "`" + `]*)?`)
+type GlobalVarsType struct {
+	wwwURLRegxp *regexp.Regexp
+	LinkRegex   *regexp.Regexp // fast matching a URL link, no any extra validation.
+}
+
+var GlobalVars = sync.OnceValue[*GlobalVarsType](func() *GlobalVarsType {
+	v := &GlobalVarsType{}
+	v.wwwURLRegxp = regexp.MustCompile(`^www\.[-a-zA-Z0-9@:%._\+~#=]{2,256}\.[a-z]{2,6}((?:/|[#?])[-a-zA-Z0-9@:%_\+.~#!?&//=\(\);,'">\^{}\[\]` + "`" + `]*)?`)
+	v.LinkRegex, _ = xurls.StrictMatchingScheme("https?://")
+	return v
+})
 
 type linkifyParser struct{}
 
@@ -60,10 +72,10 @@ func (s *linkifyParser) Parse(parent ast.Node, block text.Reader, pc parser.Cont
 	var protocol []byte
 	typ := ast.AutoLinkURL
 	if bytes.HasPrefix(line, protoHTTP) || bytes.HasPrefix(line, protoHTTPS) || bytes.HasPrefix(line, protoFTP) {
-		m = LinkRegex.FindSubmatchIndex(line)
+		m = GlobalVars().LinkRegex.FindSubmatchIndex(line)
 	}
 	if m == nil && bytes.HasPrefix(line, domainWWW) {
-		m = wwwURLRegxp.FindSubmatchIndex(line)
+		m = GlobalVars().wwwURLRegxp.FindSubmatchIndex(line)
 		protocol = []byte("http")
 	}
 	if m != nil {

modules/markup/common/html.go

@@ -6,8 +6,7 @@ package console
 import (
 	"bytes"
 	"io"
-	"path/filepath"
-	"regexp"
+	"path"
 
 	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/setting"
@@ -36,7 +35,7 @@ func (Renderer) Extensions() []string {
 // SanitizerRules implements markup.Renderer
 func (Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
 	return []setting.MarkupSanitizerRule{
-		{Element: "span", AllowAttr: "class", Regexp: regexp.MustCompile(`^term-((fg[ix]?|bg)\d+|container)$`)},
+		{Element: "span", AllowAttr: "class", Regexp: `^term-((fg[ix]?|bg)\d+|container)$`},
 	}
 }
 
@@ -46,7 +45,7 @@ func (Renderer) CanRender(filename string, input io.Reader) bool {
 	if err != nil {
 		return false
 	}
-	if enry.GetLanguage(filepath.Base(filename), buf) != enry.OtherLanguage {
+	if enry.GetLanguage(path.Base(filename), buf) != enry.OtherLanguage {
 		return false
 	}
 	return bytes.ContainsRune(buf, '\x1b')

modules/markup/common/linkify.go

@@ -7,7 +7,6 @@ import (
 	"bufio"
 	"html"
 	"io"
-	"regexp"
 	"strconv"
 
 	"code.gitea.io/gitea/modules/csv"
@@ -37,9 +36,9 @@ func (Renderer) Extensions() []string {
 // SanitizerRules implements markup.Renderer
 func (Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
 	return []setting.MarkupSanitizerRule{
-		{Element: "table", AllowAttr: "class", Regexp: regexp.MustCompile(`data-table`)},
-		{Element: "th", AllowAttr: "class", Regexp: regexp.MustCompile(`line-num`)},
-		{Element: "td", AllowAttr: "class", Regexp: regexp.MustCompile(`line-num`)},
+		{Element: "table", AllowAttr: "class", Regexp: `^data-table$`},
+		{Element: "th", AllowAttr: "class", Regexp: `^line-num$`},
+		{Element: "td", AllowAttr: "class", Regexp: `^line-num$`},
 	}
 }
 
@@ -51,13 +50,13 @@ func writeField(w io.Writer, element, class, field string) error {
 		return err
 	}
 	if len(class) > 0 {
-		if _, err := io.WriteString(w, " class=\""); err != nil {
+		if _, err := io.WriteString(w, ` class="`); err != nil {
 			return err
 		}
 		if _, err := io.WriteString(w, class); err != nil {
 			return err
 		}
-		if _, err := io.WriteString(w, "\""); err != nil {
+		if _, err := io.WriteString(w, `"`); err != nil {
 			return err
 		}
 	}

modules/markup/console/console.go

@@ -102,7 +102,7 @@ func (p *Renderer) Render(ctx *markup.RenderContext, input io.Reader, output io.
 
 		_, err = io.Copy(f, input)
 		if err != nil {
-			f.Close()
+			_ = f.Close()
 			return fmt.Errorf("%s write data to temp file when rendering %s failed: %w", p.Name(), p.Command, err)
 		}
 
@@ -113,10 +113,9 @@ func (p *Renderer) Render(ctx *markup.RenderContext, input io.Reader, output io.
 		args = append(args, f.Name())
 	}
 
-	if ctx == nil || ctx.Ctx == nil {
-		if ctx == nil {
-			log.Warn("RenderContext not provided defaulting to empty ctx")
-			ctx = &markup.RenderContext{}
+	if ctx.Ctx == nil {
+		if !setting.IsProd || setting.IsInTesting {
+			panic("RenderContext did not provide context")
 		}
 		log.Warn("RenderContext did not provide context, defaulting to Shutdown context")
 		ctx.Ctx = graceful.GetManager().ShutdownContext()