Refactor markup package (#32399)

To make the markup package easier to maintain:
1. Split some go files into small files
2. Use a shared util.NopCloser, remove duplicate code
3. Remove unused functions

modules/markup/render.go

@@ -0,0 +1,226 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package markup
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"net/url"
+	"path/filepath"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/gitrepo"
+	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/util"
+
+	"github.com/yuin/goldmark/ast"
+)
+
+type RenderMetaMode string
+
+const (
+	RenderMetaAsDetails RenderMetaMode = "details" // default
+	RenderMetaAsNone    RenderMetaMode = "none"
+	RenderMetaAsTable   RenderMetaMode = "table"
+)
+
+// RenderContext represents a render context
+type RenderContext struct {
+	Ctx              context.Context
+	RelativePath     string // relative path from tree root of the branch
+	Type             string
+	IsWiki           bool
+	Links            Links
+	Metas            map[string]string // user, repo, mode(comment/document)
+	DefaultLink      string
+	GitRepo          *git.Repository
+	Repo             gitrepo.Repository
+	ShaExistCache    map[string]bool
+	cancelFn         func()
+	SidebarTocNode   ast.Node
+	RenderMetaAs     RenderMetaMode
+	InStandalonePage bool // used by external render. the router "/org/repo/render/..." will output the rendered content in a standalone page
+}
+
+// Cancel runs any cleanup functions that have been registered for this Ctx
+func (ctx *RenderContext) Cancel() {
+	if ctx == nil {
+		return
+	}
+	ctx.ShaExistCache = map[string]bool{}
+	if ctx.cancelFn == nil {
+		return
+	}
+	ctx.cancelFn()
+}
+
+// AddCancel adds the provided fn as a Cleanup for this Ctx
+func (ctx *RenderContext) AddCancel(fn func()) {
+	if ctx == nil {
+		return
+	}
+	oldCancelFn := ctx.cancelFn
+	if oldCancelFn == nil {
+		ctx.cancelFn = fn
+		return
+	}
+	ctx.cancelFn = func() {
+		defer oldCancelFn()
+		fn()
+	}
+}
+
+// Render renders markup file to HTML with all specific handling stuff.
+func Render(ctx *RenderContext, input io.Reader, output io.Writer) error {
+	if ctx.Type != "" {
+		return renderByType(ctx, input, output)
+	} else if ctx.RelativePath != "" {
+		return renderFile(ctx, input, output)
+	}
+	return errors.New("render options both filename and type missing")
+}
+
+// RenderString renders Markup string to HTML with all specific handling stuff and return string
+func RenderString(ctx *RenderContext, content string) (string, error) {
+	var buf strings.Builder
+	if err := Render(ctx, strings.NewReader(content), &buf); err != nil {
+		return "", err
+	}
+	return buf.String(), nil
+}
+
+func renderIFrame(ctx *RenderContext, output io.Writer) error {
+	// set height="0" ahead, otherwise the scrollHeight would be max(150, realHeight)
+	// at the moment, only "allow-scripts" is allowed for sandbox mode.
+	// "allow-same-origin" should never be used, it leads to XSS attack, and it makes the JS in iframe can access parent window's config and CSRF token
+	// TODO: when using dark theme, if the rendered content doesn't have proper style, the default text color is black, which is not easy to read
+	_, err := io.WriteString(output, fmt.Sprintf(`
+<iframe src="%s/%s/%s/render/%s/%s"
+name="giteaExternalRender"
+onload="this.height=giteaExternalRender.document.documentElement.scrollHeight"
+width="100%%" height="0" scrolling="no" frameborder="0" style="overflow: hidden"
+sandbox="allow-scripts"
+></iframe>`,
+		setting.AppSubURL,
+		url.PathEscape(ctx.Metas["user"]),
+		url.PathEscape(ctx.Metas["repo"]),
+		ctx.Metas["BranchNameSubURL"],
+		url.PathEscape(ctx.RelativePath),
+	))
+	return err
+}
+
+func render(ctx *RenderContext, renderer Renderer, input io.Reader, output io.Writer) error {
+	var wg sync.WaitGroup
+	var err error
+	pr, pw := io.Pipe()
+	defer func() {
+		_ = pr.Close()
+		_ = pw.Close()
+	}()
+
+	var pr2 io.ReadCloser
+	var pw2 io.WriteCloser
+
+	var sanitizerDisabled bool
+	if r, ok := renderer.(ExternalRenderer); ok {
+		sanitizerDisabled = r.SanitizerDisabled()
+	}
+
+	if !sanitizerDisabled {
+		pr2, pw2 = io.Pipe()
+		defer func() {
+			_ = pr2.Close()
+			_ = pw2.Close()
+		}()
+
+		wg.Add(1)
+		go func() {
+			err = SanitizeReader(pr2, renderer.Name(), output)
+			_ = pr2.Close()
+			wg.Done()
+		}()
+	} else {
+		pw2 = util.NopCloser{Writer: output}
+	}
+
+	wg.Add(1)
+	go func() {
+		if r, ok := renderer.(PostProcessRenderer); ok && r.NeedPostProcess() {
+			err = PostProcess(ctx, pr, pw2)
+		} else {
+			_, err = io.Copy(pw2, pr)
+		}
+		_ = pr.Close()
+		_ = pw2.Close()
+		wg.Done()
+	}()
+
+	if err1 := renderer.Render(ctx, input, pw); err1 != nil {
+		return err1
+	}
+	_ = pw.Close()
+
+	wg.Wait()
+	return err
+}
+
+func renderByType(ctx *RenderContext, input io.Reader, output io.Writer) error {
+	if renderer, ok := renderers[ctx.Type]; ok {
+		return render(ctx, renderer, input, output)
+	}
+	return fmt.Errorf("unsupported render type: %s", ctx.Type)
+}
+
+// ErrUnsupportedRenderExtension represents the error when extension doesn't supported to render
+type ErrUnsupportedRenderExtension struct {
+	Extension string
+}
+
+func IsErrUnsupportedRenderExtension(err error) bool {
+	_, ok := err.(ErrUnsupportedRenderExtension)
+	return ok
+}
+
+func (err ErrUnsupportedRenderExtension) Error() string {
+	return fmt.Sprintf("Unsupported render extension: %s", err.Extension)
+}
+
+func renderFile(ctx *RenderContext, input io.Reader, output io.Writer) error {
+	extension := strings.ToLower(filepath.Ext(ctx.RelativePath))
+	if renderer, ok := extRenderers[extension]; ok {
+		if r, ok := renderer.(ExternalRenderer); ok && r.DisplayInIFrame() {
+			if !ctx.InStandalonePage {
+				// for an external render, it could only output its content in a standalone page
+				// otherwise, a <iframe> should be outputted to embed the external rendered page
+				return renderIFrame(ctx, output)
+			}
+		}
+		return render(ctx, renderer, input, output)
+	}
+	return ErrUnsupportedRenderExtension{extension}
+}
+
+// Init initializes the render global variables
+func Init(ph *ProcessorHelper) {
+	if ph != nil {
+		DefaultProcessorHelper = *ph
+	}
+
+	if len(setting.Markdown.CustomURLSchemes) > 0 {
+		CustomLinkURLSchemes(setting.Markdown.CustomURLSchemes)
+	}
+
+	// since setting maybe changed extensions, this will reload all renderer extensions mapping
+	extRenderers = make(map[string]Renderer)
+	for _, renderer := range renderers {
+		for _, ext := range renderer.Extensions() {
+			extRenderers[strings.ToLower(ext)] = renderer
+		}
+	}
+}