Support external renderers to PDF (#17635)

docs/content/doc/advanced/external-renderers.en-us.md

@@ -158,6 +158,35 @@ RENDER_COMMAND = "jupyter-nbconvert --stdin --stdout --to html --template basic"
 ALLOW_DATA_URI_IMAGES = true
 ```
 
+### Example: Office PPTX
+
+Convert Office PPTX files to PDF using
+[LibreOffice CLI](https://help.libreoffice.org/latest/en-US/text/shared/guide/start_parameters.html):
+
+```ini
+[markup.pptx]
+ENABLED = true
+FILE_EXTENSIONS = .pptx
+IS_INPUT_FILE = true
+RENDER_COMMAND = ./convert-pptx.sh
+RENDER_CONTENT_MODE = pdf
+```
+
+The script `convert-pptx.sh`:
+
+```sh
+#!/usr/bin/env sh
+set -eu
+file="$1"
+dir=`mktemp -d`
+libreoffice --convert-to pdf "$file" --outdir "$dir"
+cat "$dir/$(basename $file .pptx).pdf"
+rm -rf "$dir"
+```
+
+Using `RENDER_CONTENT_MODE = pdf` makes Gitea to embed files into a PDF viewer.
+It is mutually exclusive with post-processing and sanitization.
+
 ## Customizing CSS
 
 The external renderer is specified in the .ini in the format `[markup.XXXXX]` and the HTML supplied by your external renderer will be wrapped in a `<div>` with classes `markup` and `XXXXX`. The `markup` class provides out of the box styling (as does `markdown` if `XXXXX` is `markdown`). Otherwise you can use these classes to specifically target the contents of your rendered HTML.

modules/markup/external/external.go

@@ -61,14 +61,20 @@ func (p *Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
 
 // SanitizerDisabled disabled sanitize if return true
 func (p *Renderer) SanitizerDisabled() bool {
-	return p.RenderContentMode == setting.RenderContentModeNoSanitizer || p.RenderContentMode == setting.RenderContentModeIframe
+	return p.RenderContentMode == setting.RenderContentModeNoSanitizer ||
+		p.RenderContentMode == setting.RenderContentModeIframe ||
+		p.RenderContentMode == setting.RenderContentModePDF
 }
 
 // DisplayInIFrame represents whether render the content with an iframe
 func (p *Renderer) DisplayInIFrame() bool {
 	return p.RenderContentMode == setting.RenderContentModeIframe
 }
 
+func (p *Renderer) DisplayAsPDF() bool {
+	return p.RenderContentMode == setting.RenderContentModePDF
+}
+
 func envMark(envName string) string {
 	if runtime.GOOS == "windows" {
 		return "%" + envName + "%"

modules/markup/renderer.go

@@ -106,6 +106,9 @@ type ExternalRenderer interface {
 
 	// DisplayInIFrame represents whether render the content with an iframe
 	DisplayInIFrame() bool
+
+	// DisplayAsPDF represents whether to the renderer output should be viewed as PDF.
+	DisplayAsPDF() bool
 }
 
 // RendererContentDetector detects if the content can be rendered
@@ -177,23 +180,38 @@ type nopCloser struct {
 
 func (nopCloser) Close() error { return nil }
 
+func getRenderURL(ctx *RenderContext) string {
+	return fmt.Sprintf("%s/%s/%s/render/%s/%s",
+		setting.AppSubURL,
+		url.PathEscape(ctx.Metas["user"]),
+		url.PathEscape(ctx.Metas["repo"]),
+		ctx.Metas["BranchNameSubURL"],
+		url.PathEscape(ctx.RelativePath),
+	)
+}
+
 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"
+<iframe src="%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),
+		getRenderURL(ctx),
+	))
+	return err
+}
+
+func renderPDFViewer(ctx *RenderContext, output io.Writer) error {
+	_, err := io.WriteString(output, fmt.Sprintf(`
+<iframe width="100%%" height="600px" src="%s/vendor/plugins/pdfjs/web/viewer.html?file=%s"></iframe>`,
+		setting.StaticURLPrefix+"/assets",
+		getRenderURL(ctx),
 	))
 	return err
 }
@@ -281,11 +299,13 @@ func (err ErrUnsupportedRenderExtension) Error() string {
 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
+		if r, ok := renderer.(ExternalRenderer); ok && !ctx.InStandalonePage {
+			// for an external render, it could only output its content in a standalone page
+			// otherwise, the output may need to be embedded into an <iframe> or a viewer
+			if r.DisplayInIFrame() {
 				return renderIFrame(ctx, output)
+			} else if r.DisplayAsPDF() {
+				return renderPDFViewer(ctx, output)
 			}
 		}
 		return render(ctx, renderer, input, output)

modules/setting/markup.go

@@ -9,7 +9,6 @@ import (
 	"strings"
 
 	"code.gitea.io/gitea/modules/log"
-
 	"gopkg.in/ini.v1"
 )
 
@@ -24,6 +23,7 @@ const (
 	RenderContentModeSanitized   = "sanitized"
 	RenderContentModeNoSanitizer = "no-sanitizer"
 	RenderContentModeIframe      = "iframe"
+	RenderContentModePDF         = "pdf"
 )
 
 // MarkupRenderer defines the external parser configured in ini
@@ -160,18 +160,27 @@ func newMarkupRenderer(name string, sec *ini.Section) {
 	}
 	if renderContentMode != RenderContentModeSanitized &&
 		renderContentMode != RenderContentModeNoSanitizer &&
-		renderContentMode != RenderContentModeIframe {
+		renderContentMode != RenderContentModeIframe &&
+		renderContentMode != RenderContentModePDF {
 		log.Error("invalid RENDER_CONTENT_MODE: %q, default to %q", renderContentMode, RenderContentModeSanitized)
 		renderContentMode = RenderContentModeSanitized
 	}
 
+	needPostProcessDefault := renderContentMode != RenderContentModePDF
+	needPostProcess := sec.Key("NEED_POSTPROCESS").MustBool(needPostProcessDefault)
+	if needPostProcess && renderContentMode == RenderContentModePDF {
+		log.Error("NEED_POSTPROCESS: %q is incompatible with RENDER_CONTENT_MODE: %q, default to %q",
+			needPostProcess, renderContentMode, false)
+		needPostProcess = false
+	}
+
 	ExternalMarkupRenderers = append(ExternalMarkupRenderers, &MarkupRenderer{
 		Enabled:           sec.Key("ENABLED").MustBool(false),
 		MarkupName:        name,
 		FileExtensions:    exts,
 		Command:           command,
 		IsInputFile:       sec.Key("IS_INPUT_FILE").MustBool(false),
-		NeedPostProcess:   sec.Key("NEED_POSTPROCESS").MustBool(true),
+		NeedPostProcess:   needPostProcess,
 		RenderContentMode: renderContentMode,
 	})
 }

routers/web/repo/render.go

@@ -44,7 +44,10 @@ func RenderFile(ctx *context.Context) {
 	st := typesniffer.DetectContentType(buf)
 	isTextFile := st.IsText()
 
-	rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc))
+	rd := io.MultiReader(bytes.NewReader(buf), dataRc)
+	if isTextFile {
+		rd = charset.ToUTF8WithFallbackReader(rd)
+	}
 
 	if markupType := markup.Type(blob.Name()); markupType == "" {
 		if isTextFile {

routers/web/repo/view.go

@@ -462,6 +462,9 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 		ctx.Data["EditFileTooltip"] = ctx.Tr("repo.editor.cannot_edit_non_text_files")
 	}
 
+	metas := ctx.Repo.Repository.ComposeDocumentMetas()
+	metas["BranchNameSubURL"] = ctx.Repo.BranchNameSubURL()
+
 	switch {
 	case isRepresentableAsText:
 		if st.IsSvgImage() {
@@ -498,8 +501,6 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 			if !detected {
 				markupType = ""
 			}
-			metas := ctx.Repo.Repository.ComposeDocumentMetas()
-			metas["BranchNameSubURL"] = ctx.Repo.BranchNameSubURL()
 			ctx.Data["EscapeStatus"], ctx.Data["FileContent"], err = markupRender(ctx, &markup.RenderContext{
 				Ctx:          ctx,
 				Type:         markupType,
@@ -613,7 +614,7 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 				Ctx:          ctx,
 				RelativePath: ctx.Repo.TreePath,
 				URLPrefix:    path.Dir(treeLink),
-				Metas:        ctx.Repo.Repository.ComposeDocumentMetas(),
+				Metas:        metas,
 				GitRepo:      ctx.Repo.GitRepo,
 			}, rd)
 			if err != nil {