Hugo: XSS via text/html content files

@@ -74,6 +74,11 @@ var DefaultConfig = Config{
 			AllowChildProcess: []string{"tailwindcss"}, // detect-libc spawns getconf on some Linux setups.
 		},
 	},
+	// Content under /content is treated as untrusted. text/html bodies are
+	// emitted verbatim and are an XSS sink, so they are denied by default.
+	// Everything else is allowed because Whitelist treats a deny-only list as
+	// "allow anything not denied".
+	AllowContent: MustNewWhitelist("! ^text/html$"),
 }
 
 // Config is the top level security config.
@@ -92,6 +97,12 @@ type Config struct {
 	// Node holds Node.js security settings.
 	Node Node `json:"node"`
 
+	// AllowContent restricts which content media types may be used for
+	// pages under /content. Matched against the full MIME type (e.g.
+	// "text/html"). text/html is denied by default because Hugo emits the
+	// body verbatim.
+	AllowContent Whitelist `json:"allowContent"`
+
 	// Allow inline shortcodes
 	EnableInlineShortcodes bool `json:"enableInlineShortcodes"`
 }
@@ -200,6 +211,17 @@ func (c Config) CheckAllowedHTTPMethod(method string) error {
 	return nil
 }
 
+func (c Config) CheckAllowedContent(mediaType string) error {
+	if !c.AllowContent.Accept(mediaType) {
+		return &AccessDeniedError{
+			name:     mediaType,
+			path:     "security.allowContent",
+			policies: c.ToTOML(),
+		}
+	}
+	return nil
+}
+
 // ToSecurityMap converts c to a map with 'security' as the root key.
 func (c Config) ToSecurityMap() map[string]any {
 	// Take it to JSON and back to get proper casing etc.

@@ -135,7 +135,7 @@ func TestToTOML(t *testing.T) {
 	got := DefaultConfig.ToTOML()
 
 	c.Assert(got, qt.Equals,
-		"[security]\n  enableInlineShortcodes = false\n\n  [security.exec]\n    allow = ['^(dart-)?sass(-embedded)?$', '^go$', '^git$', '^node$', '^postcss$', '^tailwindcss$']\n    osEnv = ['(?i)^((HTTPS?|NO)_PROXY|PATH(EXT)?|APPDATA|TE?MP|TERM|GO\\w+|(XDG_CONFIG_)?HOME|USERPROFILE|SSH_AUTH_SOCK|DISPLAY|LANG|SYSTEMDRIVE|PROGRAMDATA)$']\n\n  [security.funcs]\n    getenv = ['^HUGO_', '^CI$']\n\n  [security.http]\n    methods = ['(?i)GET|POST']\n    urls = ['(?i)^https?://[a-z0-9]', '! ^https?://\\d+\\.', '! (?i)localhost', '! (?i)^https?://[^/?#]*@']\n\n  [security.node]\n    [security.node.permissions]\n      allowAddons = ['tailwindcss']\n      allowChildProcess = ['tailwindcss']\n      allowRead = ['.']\n      allowWorker = ['tailwindcss']\n      allowWrite = []\n      disable = false",
+		"[security]\n  allowContent = ['! ^text/html$']\n  enableInlineShortcodes = false\n\n  [security.exec]\n    allow = ['^(dart-)?sass(-embedded)?$', '^go$', '^git$', '^node$', '^postcss$', '^tailwindcss$']\n    osEnv = ['(?i)^((HTTPS?|NO)_PROXY|PATH(EXT)?|APPDATA|TE?MP|TERM|GO\\w+|(XDG_CONFIG_)?HOME|USERPROFILE|SSH_AUTH_SOCK|DISPLAY|LANG|SYSTEMDRIVE|PROGRAMDATA)$']\n\n  [security.funcs]\n    getenv = ['^HUGO_', '^CI$']\n\n  [security.http]\n    methods = ['(?i)GET|POST']\n    urls = ['(?i)^https?://[a-z0-9]', '! ^https?://\\d+\\.', '! (?i)localhost', '! (?i)^https?://[^/?#]*@']\n\n  [security.node]\n    [security.node.permissions]\n      allowAddons = ['tailwindcss']\n      allowChildProcess = ['tailwindcss']\n      allowRead = ['.']\n      allowWorker = ['tailwindcss']\n      allowWrite = []\n      disable = false",
 	)
 }
 
@@ -298,6 +298,48 @@ func TestCheckAllowedHTTPURLDigitHostnameIssue14837(t *testing.T) {
 	}
 }
 
+func TestCheckAllowedContent(t *testing.T) {
+	t.Parallel()
+	c := qt.New(t)
+
+	c.Run("text/html denied by default", func(c *qt.C) {
+		c.Parallel()
+		pc, err := DecodeConfig(config.New())
+		c.Assert(err, qt.IsNil)
+		err = pc.CheckAllowedContent("text/html")
+		c.Assert(err, qt.IsNotNil)
+		c.Assert(err, qt.ErrorMatches, `(?s).*"text/html" is not whitelisted in policy "security\.allowContent".*`)
+	})
+
+	c.Run("Other content types allowed by default", func(c *qt.C) {
+		c.Parallel()
+		pc, err := DecodeConfig(config.New())
+		c.Assert(err, qt.IsNil)
+		for _, mt := range []string{
+			"text/markdown",
+			"text/asciidoc",
+			"text/x-org",
+			"text/rst",
+			"text/pandoc",
+		} {
+			c.Assert(pc.CheckAllowedContent(mt), qt.IsNil, qt.Commentf(mt))
+		}
+	})
+
+	c.Run("User can opt in to HTML", func(c *qt.C) {
+		c.Parallel()
+		tomlConfig := `
+[security]
+allowContent = ['.*']
+`
+		cfg, err := config.FromConfigString(tomlConfig, "toml")
+		c.Assert(err, qt.IsNil)
+		pc, err := DecodeConfig(cfg)
+		c.Assert(err, qt.IsNil)
+		c.Assert(pc.CheckAllowedContent("text/html"), qt.IsNil)
+	})
+}
+
 func TestDecodeConfigNodePermissions(t *testing.T) {
 	c := qt.New(t)
 

@@ -22,6 +22,8 @@ import (
 func TestPageBundlerBasic(t *testing.T) {
 	files := `
 -- hugo.toml --
+[security]
+allowContent = ['.*']
 -- content/mybundle/index.md --
 ---
 title: "My Bundle"
@@ -630,6 +632,8 @@ func TestHTMLFilesIsue11999(t *testing.T) {
 disableKinds = ["taxonomy", "term", "rss", "sitemap", "robotsTXT", "404"]
 [permalinks]
 posts = "/myposts/:slugorcontentbasename"
+[security]
+allowContent = ['.*']
 -- content/posts/markdown-without-frontmatter.md --
 -- content/posts/html-without-frontmatter.html --
 <html>hello</html>
@@ -705,6 +709,8 @@ func TestBundleDuplicatePagesAndResources(t *testing.T) {
 -- hugo.toml --
 baseURL = "https://example.com"
 disableKinds = ["taxonomy", "term"]
+[security]
+allowContent = ['.*']
 -- content/mysection/mybundle/index.md --
 -- content/mysection/mybundle/index.html --
 -- content/mysection/mybundle/p1.md --

@@ -807,7 +807,8 @@ func (ps *pageState) getContentConverter() converter.Converter {
 		markup := ps.m.pageConfigSource.ContentMediaType.SubType
 
 		if markup == "html" {
-			// Only used for shortcode inner content.
+			// Only reachable for shortcode inner content rendering; file-based
+			// HTML pages are gated at initFrontMatter via security.allowContent.
 			markup = "markdown"
 		}
 		ps.contentConverter, err = ps.m.newContentConverter(ps, markup)

@@ -238,6 +238,17 @@ func (m *pageMetaSource) initFrontMatter(h *HugoSites) error {
 		return err
 	}
 
+	// Gate the content format against the security policy. The body of a
+	// content file is treated as untrusted; text/html is denied by default
+	// because Hugo emits it verbatim and that is an XSS sink. This applies
+	// to pages emitted by content adapters too -- the adapter is trusted
+	// but the data it pulls in may not be.
+	if m.f != nil && !m.pageConfigSource.ContentMediaType.IsZero() {
+		if err := h.Deps.ExecHelper.Sec().CheckAllowedContent(m.pageConfigSource.ContentMediaType.Type); err != nil {
+			return err
+		}
+	}
+
 	return nil
 }
 

@@ -31,6 +31,8 @@ const filesPagesFromDataTempleBasic = `
 disableKinds = ["taxonomy", "term", "rss", "sitemap"]
 baseURL = "https://example.com"
 disableLiveReload = true
+[security]
+allowContent = ['.*']
 -- assets/a/pixel.png --
 iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNkYPhfDwAChwGA60e6kgAAAABJRU5ErkJggg==
 -- assets/mydata.yaml --
@@ -589,6 +591,8 @@ func TestPagesFromGoTmplShortcodeNoPreceddingCharacterIssue12544(t *testing.T) {
 	files := `
 -- hugo.toml --
 disableKinds = ['home','rss','section','sitemap','taxonomy','term']
+[security]
+allowContent = ['.*']
 -- content/_content.gotmpl --
 {{ $content := dict "mediaType" "text/html" "value" "x{{< sc >}}" }}
 {{ .AddPage (dict "content" $content "path" "a") }}

@@ -741,6 +741,8 @@ func TestSummaryManualSplitHTML(t *testing.T) {
 	t.Parallel()
 	Test(t, `
 -- hugo.toml --
+[security]
+allowContent = ['.*']
 -- content/simple.html --
 ---
 title: Simple

@@ -247,6 +247,8 @@ func TestRenderShortcodesNestedPageContextIssue12356(t *testing.T) {
 	files := `
 -- hugo.toml --
 disableKinds = ["taxonomy", "term", "rss", "sitemap", "robotsTXT", "404"]
+[security]
+allowContent = ['.*']
 -- layouts/_markup/render-image.html --
 {{- with .PageInner.Resources.Get .Destination -}}Image: {{ .RelPermalink }}|{{- end -}}
 -- layouts/_markup/render-link.html --

@@ -30,6 +30,74 @@ import (
 func TestSecurityPolicies(t *testing.T) {
 	c := qt.New(t)
 
+	c.Run("HTML content, denied by default", func(c *qt.C) {
+		c.Parallel()
+		files := `
+-- hugo.toml --
+baseURL = "https://example.org"
+-- content/page.html --
+---
+title: "Untrusted"
+---
+<script>alert(1)</script>
+-- layouts/single.html --
+{{ .Content }}
+`
+		_, err := TestE(c, files)
+		c.Assert(err, qt.IsNotNil)
+		c.Assert(err, qt.ErrorMatches, `(?s).*"text/html" is not whitelisted in policy "security\.allowContent".*`)
+	})
+
+	c.Run("HTML content, allowed via override", func(c *qt.C) {
+		c.Parallel()
+		files := `
+-- hugo.toml --
+baseURL = "https://example.org"
+[security]
+allowContent = ['.*']
+-- content/page.html --
+---
+title: "Trusted"
+---
+<p>hello</p>
+-- layouts/single.html --
+{{ .Content }}
+`
+		b := Test(c, files)
+		b.AssertFileContent("public/page/index.html", "<p>hello</p>")
+	})
+
+	c.Run("HTML content from content adapter, denied by default", func(c *qt.C) {
+		c.Parallel()
+		files := `
+-- hugo.toml --
+baseURL = "https://example.org"
+-- content/_content.gotmpl --
+{{ .AddPage (dict "path" "p1" "title" "Untrusted" "content" (dict "value" "<script>alert(1)</script>" "mediaType" "text/html")) }}
+-- layouts/single.html --
+{{ .Content }}
+`
+		_, err := TestE(c, files)
+		c.Assert(err, qt.IsNotNil)
+		c.Assert(err, qt.ErrorMatches, `(?s).*"text/html" is not whitelisted in policy "security\.allowContent".*`)
+	})
+
+	c.Run("HTML content from content adapter, allowed via override", func(c *qt.C) {
+		c.Parallel()
+		files := `
+-- hugo.toml --
+baseURL = "https://example.org"
+[security]
+allowContent = ['.*']
+-- content/_content.gotmpl --
+{{ .AddPage (dict "path" "p1" "title" "Trusted" "content" (dict "value" "<p>hello</p>" "mediaType" "text/html")) }}
+-- layouts/single.html --
+{{ .Content }}
+`
+		b := Test(c, files)
+		b.AssertFileContent("public/p1/index.html", "<p>hello</p>")
+	})
+
 	c.Run("os.GetEnv, denied", func(c *qt.C) {
 		c.Parallel()
 		files := `