Turn message part viewers into plugins

5 files changed, 283 insertions, 0 deletions

diff --git a/plugins/viewhtml/plugin.go b/plugins/viewhtml/plugin.go
new file mode 100644
index 0000000..f22364e
--- /dev/null
+++ b/plugins/viewhtml/plugin.go
@@ -0,0 +1,10 @@
+package koushinviewhtml
+
+import (
+	"git.sr.ht/~emersion/koushin"
+)
+
+func init() {
+	p := koushin.GoPlugin{Name: "viewhtml"}
+	koushin.RegisterPluginLoader(p.Loader())
+}
diff --git a/plugins/viewhtml/public/assets/script.js b/plugins/viewhtml/public/assets/script.js
new file mode 100644
index 0000000..fd7a1d5
--- /dev/null
+++ b/plugins/viewhtml/public/assets/script.js
@@ -0,0 +1,18 @@
+var emailFrame = document.getElementById("email-frame");
+if (emailFrame) {
+	// Resize the frame with its content
+	var resizeFrame = function() {
+		emailFrame.style.height = emailFrame.contentWindow.document.documentElement.scrollHeight + "px";
+	};
+	emailFrame.addEventListener("load", resizeFrame);
+	emailFrame.contentWindow.addEventListener("resize", resizeFrame);
+
+	// Polyfill in case the srcdoc attribute isn't supported
+	if (!emailFrame.srcdoc) {
+		var srcdoc = emailFrame.getAttribute("srcdoc");
+		var doc = emailFrame.contentWindow.document;
+		doc.open();
+		doc.write(srcdoc);
+		doc.close();
+	}
+}
diff --git a/plugins/viewhtml/public/assets/style.css b/plugins/viewhtml/public/assets/style.css
new file mode 100644
index 0000000..4f91f63
--- /dev/null
+++ b/plugins/viewhtml/public/assets/style.css
@@ -0,0 +1,5 @@
+iframe {
+  width: 100%;
+  height: 400px;
+  border: 0;
+}
diff --git a/plugins/viewhtml/sanitize.go b/plugins/viewhtml/sanitize.go
new file mode 100644
index 0000000..8cee481
--- /dev/null
+++ b/plugins/viewhtml/sanitize.go
@@ -0,0 +1,193 @@
+package koushinviewhtml
+
+import (
+	"bytes"
+	"fmt"
+	"regexp"
+	"strings"
+
+	"github.com/aymerick/douceur/css"
+	cssparser "github.com/chris-ramon/douceur/parser"
+	"github.com/microcosm-cc/bluemonday"
+	"golang.org/x/net/html"
+)
+
+// TODO: this doesn't accomodate for quoting
+var (
+	cssURLRegexp  = regexp.MustCompile(`url\([^)]*\)`)
+	cssExprRegexp = regexp.MustCompile(`expression\([^)]*\)`)
+)
+
+var allowedStyles = map[string]bool{
+	"direction":       true,
+	"font":            true,
+	"font-family":     true,
+	"font-style":      true,
+	"font-variant":    true,
+	"font-size":       true,
+	"font-weight":     true,
+	"letter-spacing":  true,
+	"line-height":     true,
+	"text-align":      true,
+	"text-decoration": true,
+	"text-indent":     true,
+	"text-overflow":   true,
+	"text-shadow":     true,
+	"text-transform":  true,
+	"white-space":     true,
+	"word-spacing":    true,
+	"word-wrap":       true,
+	"vertical-align":  true,
+
+	"color":             true,
+	"background":        true,
+	"background-color":  true,
+	"background-image":  true,
+	"background-repeat": true,
+
+	"border":        true,
+	"border-color":  true,
+	"border-radius": true,
+	"height":        true,
+	"margin":        true,
+	"padding":       true,
+	"width":         true,
+	"max-width":     true,
+	"min-width":     true,
+
+	"clear": true,
+	"float": true,
+
+	"border-collapse": true,
+	"border-spacing":  true,
+	"caption-side":    true,
+	"empty-cells":     true,
+	"table-layout":    true,
+
+	"list-style-type":     true,
+	"list-style-position": true,
+}
+
+func sanitizeCSSDecls(decls []*css.Declaration) []*css.Declaration {
+	sanitized := make([]*css.Declaration, 0, len(decls))
+	for _, decl := range decls {
+		if !allowedStyles[decl.Property] {
+			continue
+		}
+		if cssExprRegexp.FindStringIndex(decl.Value) != nil {
+			continue
+		}
+
+		// TODO: more robust CSS declaration parsing
+		decl.Value = cssURLRegexp.ReplaceAllString(decl.Value, "url(about:blank)")
+
+		sanitized = append(sanitized, decl)
+	}
+	return sanitized
+}
+
+func sanitizeCSSRule(rule *css.Rule) {
+	// Disallow @import
+	if rule.Kind == css.AtRule && strings.EqualFold(rule.Name, "@import") {
+		rule.Prelude = "url(about:blank)"
+	}
+
+	rule.Declarations = sanitizeCSSDecls(rule.Declarations)
+
+	for _, child := range rule.Rules {
+		sanitizeCSSRule(child)
+	}
+}
+
+func sanitizeNode(n *html.Node) {
+	if n.Type == html.ElementNode {
+		if strings.EqualFold(n.Data, "img") {
+			for i := range n.Attr {
+				attr := &n.Attr[i]
+				if strings.EqualFold(attr.Key, "src") {
+					attr.Val = "about:blank"
+				}
+			}
+		} else if strings.EqualFold(n.Data, "style") {
+			var s string
+			c := n.FirstChild
+			for c != nil {
+				if c.Type == html.TextNode {
+					s += c.Data
+				}
+
+				next := c.NextSibling
+				n.RemoveChild(c)
+				c = next
+			}
+
+			stylesheet, err := cssparser.Parse(s)
+			if err != nil {
+				s = ""
+			} else {
+				for _, rule := range stylesheet.Rules {
+					sanitizeCSSRule(rule)
+				}
+
+				s = stylesheet.String()
+			}
+
+			n.AppendChild(&html.Node{
+				Type: html.TextNode,
+				Data: s,
+			})
+		}
+
+		for i := range n.Attr {
+			// Don't use `i, attr := range n.Attr` since `attr` would be a copy
+			attr := &n.Attr[i]
+
+			if strings.EqualFold(attr.Key, "style") {
+				decls, err := cssparser.ParseDeclarations(attr.Val)
+				if err != nil {
+					attr.Val = ""
+					continue
+				}
+
+				decls = sanitizeCSSDecls(decls)
+
+				attr.Val = ""
+				for _, d := range decls {
+					attr.Val += d.String()
+				}
+			}
+		}
+	}
+
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		sanitizeNode(c)
+	}
+}
+
+func sanitizeHTML(b []byte) ([]byte, error) {
+	doc, err := html.Parse(bytes.NewReader(b))
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse HTML: %v", err)
+	}
+
+	sanitizeNode(doc)
+
+	var buf bytes.Buffer
+	if err := html.Render(&buf, doc); err != nil {
+		return nil, fmt.Errorf("failed to render HTML: %v", err)
+	}
+	b = buf.Bytes()
+
+	// bluemonday must always be run last
+	p := bluemonday.UGCPolicy()
+
+	// TODO: use bluemonday's AllowStyles once it's released and
+	// supports <style>
+	p.AllowElements("style")
+	p.AllowAttrs("style").Globally()
+
+	p.AddTargetBlankToFullyQualifiedLinks(true)
+	p.RequireNoFollowOnLinks(true)
+
+	return p.SanitizeBytes(b), nil
+}
diff --git a/plugins/viewhtml/viewer.go b/plugins/viewhtml/viewer.go
new file mode 100644
index 0000000..9734c37
--- /dev/null
+++ b/plugins/viewhtml/viewer.go
@@ -0,0 +1,57 @@
+package koushinviewhtml
+
+import (
+	"bytes"
+	"fmt"
+	"html/template"
+	"io/ioutil"
+	"strings"
+
+	"git.sr.ht/~emersion/koushin"
+	koushinbase "git.sr.ht/~emersion/koushin/plugins/base"
+	"github.com/emersion/go-message"
+)
+
+const tpl = `
+<!-- allow-same-origin is required to resize the frame with its content -->
+<!-- allow-popups is required for target="_blank" links -->
+<iframe id="email-frame" srcdoc="{{.}}" sandbox="allow-same-origin allow-popups"></iframe>
+<script src="/plugins/viewhtml/assets/script.js"></script>
+<link rel="stylesheet" href="/plugins/viewhtml/assets/style.css">
+`
+
+type viewer struct{}
+
+func (viewer) ViewMessagePart(ctx *koushin.Context, msg *koushinbase.IMAPMessage, part *message.Entity) (interface{}, error) {
+	mimeType, _, err := part.Header.ContentType()
+	if err != nil {
+		return nil, err
+	}
+	if !strings.EqualFold(mimeType, "text/html") {
+		return nil, koushinbase.ErrViewUnsupported
+	}
+
+	body, err := ioutil.ReadAll(part.Body)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read part body: %v", err)
+	}
+
+	body, err = sanitizeHTML(body)
+	if err != nil {
+		return nil, fmt.Errorf("failed to sanitize HTML part: %v", err)
+	}
+
+	t := template.Must(template.New("view-html.html").Parse(tpl))
+
+	var buf bytes.Buffer
+	err = t.Execute(&buf, string(body))
+	if err != nil {
+		return nil, err
+	}
+
+	return template.HTML(buf.String()), nil
+}
+
+func init() {
+	koushinbase.RegisterViewer(viewer{})
+}