Add sanitizer rules per renderer (go-gitea#16110)

* Added sanitizer rules per renderer.

* Updated documentation.

Co-authored-by: techknowlogick <techknowlogick@gitea.io>

docs/content/doc/advanced/config-cheat-sheet.en-us.md

@@ -907,13 +907,17 @@ Gitea supports customizing the sanitization policy for rendered HTML. The exampl
 ELEMENT = span
 ALLOW_ATTR = class
 REGEXP = ^\s*((math(\s+|$)|inline(\s+|$)|display(\s+|$)))+
+ALLOW_DATA_URI_IMAGES = true
 ```
 
  - `ELEMENT`: The element this policy applies to. Must be non-empty.
  - `ALLOW_ATTR`: The attribute this policy allows. Must be non-empty.
  - `REGEXP`: A regex to match the contents of the attribute against. Must be present but may be empty for unconditional whitelisting of this attribute.
+ - `ALLOW_DATA_URI_IMAGES`: **false** Allow data uri images (`<img src="data:image/png;base64,..."/>`).
 
 Multiple sanitisation rules can be defined by adding unique subsections, e.g. `[markup.sanitizer.TeX-2]`.
+To apply a sanitisation rules only for a specify external renderer they must use the renderer name, e.g. `[markup.sanitizer.asciidoc.rule-1]`.
+If the rule is defined above the renderer ini section or the name does not match a renderer it is applied to every renderer.
 
 ## Time (`time`)
 

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

@@ -64,8 +64,8 @@ IS_INPUT_FILE = false
 [markup.jupyter]
 ENABLED = true
 FILE_EXTENSIONS = .ipynb
-RENDER_COMMAND = "jupyter nbconvert --stdout --to html --template basic "
-IS_INPUT_FILE = true
+RENDER_COMMAND = "jupyter nbconvert --stdin --stdout --to html --template basic"
+IS_INPUT_FILE = false
 
 [markup.restructuredtext]
 ENABLED = true
@@ -90,15 +90,50 @@ FILE_EXTENSIONS = .md,.markdown
 RENDER_COMMAND  = pandoc -f markdown -t html --katex
 ```
 
-You must define `ELEMENT`, `ALLOW_ATTR`, and `REGEXP` in each section.
+You must define `ELEMENT` and `ALLOW_ATTR` in each section.
 
 To define multiple entries, add a unique alphanumeric suffix (e.g., `[markup.sanitizer.1]` and `[markup.sanitizer.something]`).
 
+To apply a sanitisation rules only for a specify external renderer they must use the renderer name, e.g. `[markup.sanitizer.asciidoc.rule-1]`, `[markup.sanitizer.<renderer>.rule-1]`.
+
+**Note**: If the rule is defined above the renderer ini section or the name does not match a renderer it is applied to every renderer.
+
 Once your configuration changes have been made, restart Gitea to have changes take effect.
 
 **Note**: Prior to Gitea 1.12 there was a single `markup.sanitiser` section with keys that were redefined for multiple rules, however,
 there were significant problems with this method of configuration necessitating configuration through multiple sections.
 
+### Example: Office DOCX
+
+Display Office DOCX files with [`pandoc`](https://pandoc.org/):
+```ini
+[markup.docx]
+ENABLED = true
+FILE_EXTENSIONS = .docx
+RENDER_COMMAND = "pandoc --from docx --to html --self-contained --template /path/to/basic.html"
+
+[markup.sanitizer.docx.img]
+ALLOW_DATA_URI_IMAGES = true
+```
+
+The template file has the following content:
+```
+$body$
+```
+
+### Example: Jupyter Notebook
+
+Display Jupyter Notebook files with [`nbconvert`](https://github.com/jupyter/nbconvert):
+```ini
+[markup.jupyter]
+ENABLED = true
+FILE_EXTENSIONS = .ipynb
+RENDER_COMMAND = "jupyter-nbconvert --stdin --stdout --to html --template basic"
+
+[markup.sanitizer.jupyter.img]
+ALLOW_DATA_URI_IMAGES = true
+```
+
 ## 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/csv/csv.go

@@ -10,6 +10,7 @@ import (
 	"html"
 	"io"
 	"io/ioutil"
+	"regexp"
 	"strconv"
 
 	"code.gitea.io/gitea/modules/csv"
@@ -38,6 +39,15 @@ func (Renderer) Extensions() []string {
 	return []string{".csv", ".tsv"}
 }
 
+// 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`)},
+	}
+}
+
 func writeField(w io.Writer, element, class, field string) error {
 	if _, err := io.WriteString(w, "<"); err != nil {
 		return err

modules/markup/external/external.go

@@ -30,7 +30,7 @@ func RegisterRenderers() {
 
 // Renderer implements markup.Renderer for external tools
 type Renderer struct {
-	setting.MarkupRenderer
+	*setting.MarkupRenderer
 }
 
 // Name returns the external tool name
@@ -48,6 +48,11 @@ func (p *Renderer) Extensions() []string {
 	return p.FileExtensions
 }
 
+// SanitizerRules implements markup.Renderer
+func (p *Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
+	return p.MarkupSanitizerRules
+}
+
 func envMark(envName string) string {
 	if runtime.GOOS == "windows" {
 		return "%" + envName + "%"

modules/markup/html_test.go

@@ -112,7 +112,7 @@ func TestRender_links(t *testing.T) {
 
 	defaultCustom := setting.Markdown.CustomURLSchemes
 	setting.Markdown.CustomURLSchemes = []string{"ftp", "magnet"}
-	ReplaceSanitizer()
+	InitializeSanitizer()
 	CustomLinkURLSchemes(setting.Markdown.CustomURLSchemes)
 
 	test(
@@ -192,7 +192,7 @@ func TestRender_links(t *testing.T) {
 
 	// Restore previous settings
 	setting.Markdown.CustomURLSchemes = defaultCustom
-	ReplaceSanitizer()
+	InitializeSanitizer()
 	CustomLinkURLSchemes(setting.Markdown.CustomURLSchemes)
 }
 

modules/markup/markdown/markdown.go

@@ -199,7 +199,7 @@ func actualRender(ctx *markup.RenderContext, input io.Reader, output io.Writer)
 		}
 		_ = lw.Close()
 	}()
-	buf := markup.SanitizeReader(rd)
+	buf := markup.SanitizeReader(rd, "")
 	_, err := io.Copy(output, buf)
 	return err
 }
@@ -215,7 +215,7 @@ func render(ctx *markup.RenderContext, input io.Reader, output io.Writer) error
 		if log.IsDebug() {
 			log.Debug("Panic in markdown: %v\n%s", err, string(log.Stack(2)))
 		}
-		ret := markup.SanitizeReader(input)
+		ret := markup.SanitizeReader(input, "")
 		_, err = io.Copy(output, ret)
 		if err != nil {
 			log.Error("SanitizeReader failed: %v", err)
@@ -249,6 +249,11 @@ func (Renderer) Extensions() []string {
 	return setting.Markdown.FileExtensions
 }
 
+// SanitizerRules implements markup.Renderer
+func (Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
+	return []setting.MarkupSanitizerRule{}
+}
+
 // Render implements markup.Renderer
 func (Renderer) Render(ctx *markup.RenderContext, input io.Reader, output io.Writer) error {
 	return render(ctx, input, output)

modules/markup/orgmode/orgmode.go

@@ -13,6 +13,7 @@ import (
 
 	"code.gitea.io/gitea/modules/highlight"
 	"code.gitea.io/gitea/modules/markup"
+	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/util"
 
 	"github.com/alecthomas/chroma"
@@ -41,6 +42,11 @@ func (Renderer) Extensions() []string {
 	return []string{".org"}
 }
 
+// SanitizerRules implements markup.Renderer
+func (Renderer) SanitizerRules() []setting.MarkupSanitizerRule {
+	return []setting.MarkupSanitizerRule{}
+}
+
 // Render renders orgmode rawbytes to HTML
 func Render(ctx *markup.RenderContext, input io.Reader, output io.Writer) error {
 	htmlWriter := org.NewHTMLWriter()

modules/markup/renderer.go

@@ -81,6 +81,7 @@ type Renderer interface {
 	Name() string // markup format name
 	Extensions() []string
 	NeedPostProcess() bool
+	SanitizerRules() []setting.MarkupSanitizerRule
 	Render(ctx *RenderContext, input io.Reader, output io.Writer) error
 }
 
@@ -136,37 +137,32 @@ func render(ctx *RenderContext, renderer Renderer, input io.Reader, output io.Wr
 		_ = pw.Close()
 	}()
 
-	if renderer.NeedPostProcess() {
-		pr2, pw2 := io.Pipe()
-		defer func() {
-			_ = pr2.Close()
-			_ = pw2.Close()
-		}()
-
-		wg.Add(1)
-		go func() {
-			buf := SanitizeReader(pr2)
-			_, err = io.Copy(output, buf)
-			_ = pr2.Close()
-			wg.Done()
-		}()
-
-		wg.Add(1)
-		go func() {
+	pr2, pw2 := io.Pipe()
+	defer func() {
+		_ = pr2.Close()
+		_ = pw2.Close()
+	}()
+
+	wg.Add(1)
+	go func() {
+		buf := SanitizeReader(pr2, renderer.Name())
+		_, err = io.Copy(output, buf)
+		_ = pr2.Close()
+		wg.Done()
+	}()
+
+	wg.Add(1)
+	go func() {
+		if renderer.NeedPostProcess() {
 			err = PostProcess(ctx, pr, pw2)
-			_ = pr.Close()
-			_ = pw2.Close()
-			wg.Done()
-		}()
-	} else {
-		wg.Add(1)
-		go func() {
-			buf := SanitizeReader(pr)
-			_, err = io.Copy(output, buf)
-			_ = pr.Close()
-			wg.Done()
-		}()
-	}
+		} else {
+			_, err = io.Copy(pw2, pr)
+		}
+		_ = pr.Close()
+		_ = pw2.Close()
+		wg.Done()
+	}()
+
 	if err1 := renderer.Render(ctx, input, pw); err1 != nil {
 		return err1
 	}