First commit

4 files changed, 556 insertions, 0 deletions

diff --git a/reader/sanitizer/sanitizer.go b/reader/sanitizer/sanitizer.go
new file mode 100644
index 0000000..6af034c
--- /dev/null
+++ b/reader/sanitizer/sanitizer.go
@@ -0,0 +1,360 @@
+// Copyright 2017 Frédéric Guillot. All rights reserved.
+// Use of this source code is governed by the Apache 2.0
+// license that can be found in the LICENSE file.
+
+package sanitizer
+
+import (
+	"bytes"
+	"fmt"
+	"github.com/miniflux/miniflux2/reader/url"
+	"io"
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+// Sanitize returns safe HTML.
+func Sanitize(baseURL, input string) string {
+	tokenizer := html.NewTokenizer(bytes.NewBufferString(input))
+	var buffer bytes.Buffer
+	var tagStack []string
+
+	for {
+		if tokenizer.Next() == html.ErrorToken {
+			err := tokenizer.Err()
+			if err == io.EOF {
+				return buffer.String()
+			}
+
+			return ""
+		}
+
+		token := tokenizer.Token()
+		switch token.Type {
+		case html.TextToken:
+			buffer.WriteString(token.Data)
+		case html.StartTagToken:
+			tagName := token.DataAtom.String()
+
+			if !isPixelTracker(tagName, token.Attr) && isValidTag(tagName) {
+				attrNames, htmlAttributes := sanitizeAttributes(baseURL, tagName, token.Attr)
+
+				if hasRequiredAttributes(tagName, attrNames) {
+					if len(attrNames) > 0 {
+						buffer.WriteString("<" + tagName + " " + htmlAttributes + ">")
+					} else {
+						buffer.WriteString("<" + tagName + ">")
+					}
+
+					tagStack = append(tagStack, tagName)
+				}
+			}
+		case html.EndTagToken:
+			tagName := token.DataAtom.String()
+			if isValidTag(tagName) && inList(tagName, tagStack) {
+				buffer.WriteString(fmt.Sprintf("</%s>", tagName))
+			}
+		case html.SelfClosingTagToken:
+			tagName := token.DataAtom.String()
+			if !isPixelTracker(tagName, token.Attr) && isValidTag(tagName) {
+				attrNames, htmlAttributes := sanitizeAttributes(baseURL, tagName, token.Attr)
+
+				if hasRequiredAttributes(tagName, attrNames) {
+					if len(attrNames) > 0 {
+						buffer.WriteString("<" + tagName + " " + htmlAttributes + "/>")
+					} else {
+						buffer.WriteString("<" + tagName + "/>")
+					}
+				}
+			}
+		}
+	}
+}
+
+func sanitizeAttributes(baseURL, tagName string, attributes []html.Attribute) (attrNames []string, html string) {
+	var htmlAttrs []string
+	var err error
+
+	for _, attribute := range attributes {
+		value := attribute.Val
+
+		if !isValidAttribute(tagName, attribute.Key) {
+			continue
+		}
+
+		if isExternalResourceAttribute(attribute.Key) {
+			if tagName == "iframe" && !isValidIframeSource(attribute.Val) {
+				continue
+			} else {
+				value, err = url.GetAbsoluteURL(baseURL, value)
+				if err != nil {
+					continue
+				}
+
+				if !hasValidScheme(value) || isBlacklistedResource(value) {
+					continue
+				}
+			}
+		}
+
+		attrNames = append(attrNames, attribute.Key)
+		htmlAttrs = append(htmlAttrs, fmt.Sprintf(`%s="%s"`, attribute.Key, value))
+	}
+
+	extraAttrNames, extraHTMLAttributes := getExtraAttributes(tagName)
+	if len(extraAttrNames) > 0 {
+		attrNames = append(attrNames, extraAttrNames...)
+		htmlAttrs = append(htmlAttrs, extraHTMLAttributes...)
+	}
+
+	return attrNames, strings.Join(htmlAttrs, " ")
+}
+
+func getExtraAttributes(tagName string) ([]string, []string) {
+	if tagName == "a" {
+		return []string{"rel", "target", "referrerpolicy"}, []string{`rel="noopener noreferrer"`, `target="_blank"`, `referrerpolicy="no-referrer"`}
+	}
+
+	if tagName == "video" || tagName == "audio" {
+		return []string{"controls"}, []string{"controls"}
+	}
+
+	return nil, nil
+}
+
+func isValidTag(tagName string) bool {
+	for element := range getTagWhitelist() {
+		if tagName == element {
+			return true
+		}
+	}
+
+	return false
+}
+
+func isValidAttribute(tagName, attributeName string) bool {
+	for element, attributes := range getTagWhitelist() {
+		if tagName == element {
+			if inList(attributeName, attributes) {
+				return true
+			}
+		}
+	}
+
+	return false
+}
+
+func isExternalResourceAttribute(attribute string) bool {
+	switch attribute {
+	case "src", "href", "poster", "cite":
+		return true
+	default:
+		return false
+	}
+}
+
+func isPixelTracker(tagName string, attributes []html.Attribute) bool {
+	if tagName == "img" {
+		hasHeight := false
+		hasWidth := false
+
+		for _, attribute := range attributes {
+			if attribute.Key == "height" && attribute.Val == "1" {
+				hasHeight = true
+			}
+
+			if attribute.Key == "width" && attribute.Val == "1" {
+				hasWidth = true
+			}
+		}
+
+		return hasHeight && hasWidth
+	}
+
+	return false
+}
+
+func hasRequiredAttributes(tagName string, attributes []string) bool {
+	elements := make(map[string][]string)
+	elements["a"] = []string{"href"}
+	elements["iframe"] = []string{"src"}
+	elements["img"] = []string{"src"}
+	elements["source"] = []string{"src"}
+
+	for element, attrs := range elements {
+		if tagName == element {
+			for _, attribute := range attributes {
+				for _, attr := range attrs {
+					if attr == attribute {
+						return true
+					}
+				}
+			}
+
+			return false
+		}
+	}
+
+	return true
+}
+
+func hasValidScheme(src string) bool {
+	// See https://www.iana.org/assignments/uri-schemes/uri-schemes.xhtml
+	whitelist := []string{
+		"apt://",
+		"bitcoin://",
+		"callto://",
+		"ed2k://",
+		"facetime://",
+		"feed://",
+		"ftp://",
+		"geo://",
+		"gopher://",
+		"git://",
+		"http://",
+		"https://",
+		"irc://",
+		"irc6://",
+		"ircs://",
+		"itms://",
+		"jabber://",
+		"magnet://",
+		"mailto://",
+		"maps://",
+		"news://",
+		"nfs://",
+		"nntp://",
+		"rtmp://",
+		"sip://",
+		"sips://",
+		"skype://",
+		"smb://",
+		"sms://",
+		"spotify://",
+		"ssh://",
+		"sftp://",
+		"steam://",
+		"svn://",
+		"tel://",
+		"webcal://",
+		"xmpp://",
+	}
+
+	for _, prefix := range whitelist {
+		if strings.HasPrefix(src, prefix) {
+			return true
+		}
+	}
+
+	return false
+}
+
+func isBlacklistedResource(src string) bool {
+	blacklist := []string{
+		"feedsportal.com",
+		"api.flattr.com",
+		"stats.wordpress.com",
+		"plus.google.com/share",
+		"twitter.com/share",
+		"feeds.feedburner.com",
+	}
+
+	for _, element := range blacklist {
+		if strings.Contains(src, element) {
+			return true
+		}
+	}
+
+	return false
+}
+
+func isValidIframeSource(src string) bool {
+	whitelist := []string{
+		"http://www.youtube.com",
+		"https://www.youtube.com",
+		"http://player.vimeo.com",
+		"https://player.vimeo.com",
+		"http://www.dailymotion.com",
+		"https://www.dailymotion.com",
+		"http://vk.com",
+		"https://vk.com",
+	}
+
+	for _, prefix := range whitelist {
+		if strings.HasPrefix(src, prefix) {
+			return true
+		}
+	}
+
+	return false
+}
+
+func getTagWhitelist() map[string][]string {
+	whitelist := make(map[string][]string)
+	whitelist["img"] = []string{"alt", "title", "src"}
+	whitelist["audio"] = []string{"src"}
+	whitelist["video"] = []string{"poster", "height", "width", "src"}
+	whitelist["source"] = []string{"src", "type"}
+	whitelist["dt"] = []string{}
+	whitelist["dd"] = []string{}
+	whitelist["dl"] = []string{}
+	whitelist["table"] = []string{}
+	whitelist["caption"] = []string{}
+	whitelist["thead"] = []string{}
+	whitelist["tfooter"] = []string{}
+	whitelist["tr"] = []string{}
+	whitelist["td"] = []string{"rowspan", "colspan"}
+	whitelist["th"] = []string{"rowspan", "colspan"}
+	whitelist["h1"] = []string{}
+	whitelist["h2"] = []string{}
+	whitelist["h3"] = []string{}
+	whitelist["h4"] = []string{}
+	whitelist["h5"] = []string{}
+	whitelist["h6"] = []string{}
+	whitelist["strong"] = []string{}
+	whitelist["em"] = []string{}
+	whitelist["code"] = []string{}
+	whitelist["pre"] = []string{}
+	whitelist["blockquote"] = []string{}
+	whitelist["q"] = []string{"cite"}
+	whitelist["p"] = []string{}
+	whitelist["ul"] = []string{}
+	whitelist["li"] = []string{}
+	whitelist["ol"] = []string{}
+	whitelist["br"] = []string{}
+	whitelist["del"] = []string{}
+	whitelist["a"] = []string{"href", "title"}
+	whitelist["figure"] = []string{}
+	whitelist["figcaption"] = []string{}
+	whitelist["cite"] = []string{}
+	whitelist["time"] = []string{"datetime"}
+	whitelist["abbr"] = []string{"title"}
+	whitelist["acronym"] = []string{"title"}
+	whitelist["wbr"] = []string{}
+	whitelist["dfn"] = []string{}
+	whitelist["sub"] = []string{}
+	whitelist["sup"] = []string{}
+	whitelist["var"] = []string{}
+	whitelist["samp"] = []string{}
+	whitelist["s"] = []string{}
+	whitelist["del"] = []string{}
+	whitelist["ins"] = []string{}
+	whitelist["kbd"] = []string{}
+	whitelist["rp"] = []string{}
+	whitelist["rt"] = []string{}
+	whitelist["rtc"] = []string{}
+	whitelist["ruby"] = []string{}
+	whitelist["iframe"] = []string{"width", "height", "frameborder", "src", "allowfullscreen"}
+	return whitelist
+}
+
+func inList(needle string, haystack []string) bool {
+	for _, element := range haystack {
+		if element == needle {
+			return true
+		}
+	}
+
+	return false
+}
diff --git a/reader/sanitizer/sanitizer_test.go b/reader/sanitizer/sanitizer_test.go
new file mode 100644
index 0000000..73862d3
--- /dev/null
+++ b/reader/sanitizer/sanitizer_test.go
@@ -0,0 +1,144 @@
+// Copyright 2017 Frédéric Guillot. All rights reserved.
+// Use of this source code is governed by the Apache 2.0
+// license that can be found in the LICENSE file.
+
+package sanitizer
+
+import "testing"
+
+func TestValidInput(t *testing.T) {
+	input := `<p>This is a <strong>text</strong> with an image: <img src="http://example.org/" alt="Test">.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if input != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
+	}
+}
+
+func TestSelfClosingTags(t *testing.T) {
+	input := `<p>This <br> is a <strong>text</strong> <br/>with an image: <img src="http://example.org/" alt="Test"/>.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if input != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
+	}
+}
+
+func TestTable(t *testing.T) {
+	input := `<table><tr><th>A</th><th colspan="2">B</th></tr><tr><td>C</td><td>D</td><td>E</td></tr></table>`
+	output := Sanitize("http://example.org/", input)
+
+	if input != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
+	}
+}
+
+func TestRelativeURL(t *testing.T) {
+	input := `This <a href="/test.html">link is relative</a> and this image: <img src="../folder/image.png"/>`
+	expected := `This <a href="http://example.org/test.html" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">link is relative</a> and this image: <img src="http://example.org/folder/image.png"/>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestProtocolRelativeURL(t *testing.T) {
+	input := `This <a href="//static.example.org/index.html">link is relative</a>.`
+	expected := `This <a href="https://static.example.org/index.html" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">link is relative</a>.`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestInvalidTag(t *testing.T) {
+	input := `<p>My invalid <b>tag</b>.</p>`
+	expected := `<p>My invalid tag.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestVideoTag(t *testing.T) {
+	input := `<p>My valid <video src="videofile.webm" autoplay poster="posterimage.jpg">fallback</video>.</p>`
+	expected := `<p>My valid <video src="http://example.org/videofile.webm" poster="http://example.org/posterimage.jpg" controls>fallback</video>.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestAudioAndSourceTag(t *testing.T) {
+	input := `<p>My music <audio controls="controls"><source src="foo.wav" type="audio/wav"></audio>.</p>`
+	expected := `<p>My music <audio controls><source src="http://example.org/foo.wav" type="audio/wav"></audio>.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestUnknownTag(t *testing.T) {
+	input := `<p>My invalid <unknown>tag</unknown>.</p>`
+	expected := `<p>My invalid tag.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestInvalidNestedTag(t *testing.T) {
+	input := `<p>My invalid <b>tag with some <em>valid</em> tag</b>.</p>`
+	expected := `<p>My invalid tag with some <em>valid</em> tag.</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestInvalidIFrame(t *testing.T) {
+	input := `<iframe src="http://example.org/"></iframe>`
+	expected := ``
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestInvalidURLScheme(t *testing.T) {
+	input := `<p>This link is <a src="file:///etc/passwd">not valid</a></p>`
+	expected := `<p>This link is not valid</p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestBlacklistedLink(t *testing.T) {
+	input := `<p>This image is not valid <img src="https://stats.wordpress.com/some-tracker"></p>`
+	expected := `<p>This image is not valid </p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
+
+func TestPixelTracker(t *testing.T) {
+	input := `<p><img src="https://tracker1.example.org/" height="1" width="1"> and <img src="https://tracker2.example.org/" height="1" width="1"/></p>`
+	expected := `<p> and </p>`
+	output := Sanitize("http://example.org/", input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}
diff --git a/reader/sanitizer/strip_tags.go b/reader/sanitizer/strip_tags.go
new file mode 100644
index 0000000..f3cecd1
--- /dev/null
+++ b/reader/sanitizer/strip_tags.go
@@ -0,0 +1,35 @@
+// Copyright 2017 Frédéric Guillot. All rights reserved.
+// Use of this source code is governed by the Apache 2.0
+// license that can be found in the LICENSE file.
+
+package sanitizer
+
+import (
+	"bytes"
+	"io"
+
+	"golang.org/x/net/html"
+)
+
+// StripTags removes all HTML/XML tags from the input string.
+func StripTags(input string) string {
+	tokenizer := html.NewTokenizer(bytes.NewBufferString(input))
+	var buffer bytes.Buffer
+
+	for {
+		if tokenizer.Next() == html.ErrorToken {
+			err := tokenizer.Err()
+			if err == io.EOF {
+				return buffer.String()
+			}
+
+			return ""
+		}
+
+		token := tokenizer.Token()
+		switch token.Type {
+		case html.TextToken:
+			buffer.WriteString(token.Data)
+		}
+	}
+}
diff --git a/reader/sanitizer/strip_tags_test.go b/reader/sanitizer/strip_tags_test.go
new file mode 100644
index 0000000..763ccc9
--- /dev/null
+++ b/reader/sanitizer/strip_tags_test.go
@@ -0,0 +1,17 @@
+// Copyright 2017 Frédéric Guillot. All rights reserved.
+// Use of this source code is governed by the Apache 2.0
+// license that can be found in the LICENSE file.
+
+package sanitizer
+
+import "testing"
+
+func TestStripTags(t *testing.T) {
+	input := `This <a href="/test.html">link is relative</a> and <strong>this</strong> image: <img src="../folder/image.png"/>`
+	expected := `This link is relative and this image: `
+	output := StripTags(input)
+
+	if expected != output {
+		t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
+	}
+}