v3.0.0 - Config API redone to support custom elements and attributes

.codeclimate.yml

@@ -0,0 +1,15 @@
+engines:
+  fixme:
+    enabled: true
+  gofmt:
+    enabled: true
+  golint:
+    enabled: true
+  govet:
+    enabled: true
+  markdownlint:
+    enabled: true
+ratings:
+  paths:
+    - "**.go"
+    - "**.md"

.travis.yml

@@ -0,0 +1,16 @@
+language: go
+sudo: false
+go:
+  - 1.7
+  - tip
+before_install:
+  - go get github.com/mattn/goveralls
+script:
+  - $HOME/gopath/bin/goveralls -service=travis-ci
+notifications:
+  webhooks:
+    urls:
+      - https://webhooks.gitter.im/e/b7a046aa93ac3d8e5262
+    on_success: change
+    on_failure: always
+    on_start: never

README.md

@@ -1,4 +1,7 @@
 # HTML Cleaner
 
+[![Build Status](https://travis-ci.org/BenLubar/htmlcleaner.svg?branch=master)](https://travis-ci.org/BenLubar/htmlcleaner)
+[![Join the chat at https://gitter.im/BenLubar/htmlcleaner](https://badges.gitter.im/BenLubar/webscale.svg)](https://gitter.im/BenLubar/htmlcleaner)
 [![Go Report Card](https://goreportcard.com/badge/github.com/BenLubar/htmlcleaner)](https://goreportcard.com/report/github.com/BenLubar/htmlcleaner)
 [![GoDoc](https://godoc.org/github.com/BenLubar/htmlcleaner?status.svg)](https://godoc.org/github.com/BenLubar/htmlcleaner)
+[![Coverage Status](https://coveralls.io/repos/github/BenLubar/htmlcleaner/badge.svg?branch=master)](https://coveralls.io/github/BenLubar/htmlcleaner?branch=master)

cleaner.go

@@ -50,8 +50,18 @@ func Preprocess(config *Config, fragment string) string {
 		case html.StartTagToken, html.EndTagToken, html.SelfClosingTagToken:
 			raw := string(t.Raw())
 			tagName, _ := t.TagName()
-			tag := atom.Lookup(tagName)
-			if _, ok := config.Elem[tag]; !ok {
+			allowed := false
+			if tag := atom.Lookup(tagName); tag != 0 {
+				if _, ok := config.elem[tag]; ok {
+					allowed = true
+				}
+			}
+			if !allowed {
+				if _, ok := config.elemCustom[string(tagName)]; ok {
+					allowed = true
+				}
+			}
+			if !allowed {
 				raw = html.EscapeString(raw)
 			}
 			write(raw)
@@ -239,7 +249,9 @@ func filterNode(c *Config, n *html.Node) *html.Node {
 }
 
 func cleanNode(c *Config, n *html.Node) *html.Node {
-	if allowedAttr, ok := c.Elem[n.DataAtom]; ok {
+	allowedAttr, ok1 := c.elem[n.DataAtom]
+	customAttr, ok2 := c.elemCustom[n.Data]
+	if ok1 || ok2 {
 		// copy the node
 		tmp := *n
 		n = &tmp
@@ -252,15 +264,24 @@ func cleanNode(c *Config, n *html.Node) *html.Node {
 		n.Attr = make([]html.Attribute, 0, len(attrs))
 		for _, attr := range attrs {
 			a := atom.Lookup([]byte(attr.Key))
-			if attr.Namespace != "" || (!allowedAttr[a] && !c.Attr[a]) {
+
+			re1, ok1 := allowedAttr[a]
+			re2, ok2 := customAttr[attr.Key]
+			_, ok3 := c.attr[a]
+			_, ok4 := c.attrCustom[attr.Key]
+
+			if attr.Namespace != "" || (!ok1 && !ok2 && !ok3 && !ok4) {
 				continue
 			}
 
-			if !c.AllowJavascriptURL && !cleanURL(c, a, &attr) {
+			if !cleanURL(c, a, &attr) {
 				continue
 			}
 
-			if re, ok := c.AttrMatch[n.DataAtom][a]; ok && !re.MatchString(attr.Val) {
+			if re1 != nil && !re1.MatchString(attr.Val) {
+				continue
+			}
+			if re2 != nil && !re2.MatchString(attr.Val) {
 				continue
 			}
 
@@ -287,6 +308,12 @@ var allowedURLSchemes = map[string]bool{
 	"":       true,
 }
 
+// SafeURLScheme returns true if u.Scheme is http, https, mailto, data, or an
+// empty string.
+func SafeURLScheme(u *url.URL) bool {
+	return allowedURLSchemes[u.Scheme]
+}
+
 func cleanURL(c *Config, a atom.Atom, attr *html.Attribute) bool {
 	if a != atom.Href && a != atom.Src && a != atom.Poster {
 		return true
@@ -296,9 +323,6 @@ func cleanURL(c *Config, a atom.Atom, attr *html.Attribute) bool {
 	if err != nil {
 		return false
 	}
-	if !allowedURLSchemes[u.Scheme] {
-		return false
-	}
 	if c.ValidateURL != nil && !c.ValidateURL(u) {
 		return false
 	}

cleaner_test.go

@@ -29,6 +29,14 @@ func doTableTest(f func(*Config, string) string, t *testing.T, table []testTable
 	}
 }
 
+var wrapConfig = func() *Config {
+	c := *DefaultConfig
+
+	c.WrapText = true
+
+	return &c
+}()
+
 var testTableClean = []testTable{
 	{"Empty", ``, ``, nil},
 	{"PlainText", `a`, `a`, nil},
@@ -46,9 +54,9 @@ var testTableClean = []testTable{
 	{"Ampersand", `&`, `&`, nil},
 	{"AmpersandEntity", `&`, `&`, nil},
 	{"InvalidTagEntity", `<invalidtag>&#34;</invalidtag>`, `&lt;invalidtag&gt;&#34;&lt;/invalidtag&gt;`, nil},
-	{"StrayListItem", `<li>`, `<ul><li></li></ul>`, &Config{Elem: map[atom.Atom]map[atom.Atom]bool{atom.Ul: nil, atom.Li: nil}}},
+	{"StrayListItem", `<li>`, `<ul><li></li></ul>`, (&Config{}).ElemAtom(atom.Ul, atom.Li)},
 	{"LinkPercent", `<a href="https://www.%google.com">google</a>`, `<a>google</a>`, nil},
-	{"LinkPercentWrap", `<a href="https://www.%google.com">google</a>`, `<p><a>google</a></p>`, &Config{Elem: DefaultConfig.Elem, WrapText: true}},
+	{"LinkPercentWrap", `<a href="https://www.%google.com">google</a>`, `<p><a>google</a></p>`, wrapConfig},
 	{"GreaterThanInfix", `foo>bar`, `foo&gt;bar`, nil},
 	{"GreaterThanPrefix", `>bar`, `&gt;bar`, nil},
 	{"GreaterThanSuffix", `foo>`, `foo&gt;`, nil},
@@ -63,9 +71,9 @@ var testTableClean = []testTable{
 	{"PHP", `<?php echo mysql_real_escape_string('foo'); ?>`, `<!--?php echo mysql_real_escape_string('foo'); ?-->`, nil},
 	{"PHPEscaped", `<?php echo mysql_real_escape_string('foo'); ?>`, `&lt;!--?php echo mysql_real_escape_string(&#39;foo&#39;); ?--&gt;`, &Config{EscapeComments: true}},
 	{"Small250", strings.Repeat(`<small>a `, 250), strings.Repeat(`<small>a `, 99) + "<small>[omitted]" + strings.Repeat(`</small>`, 100), nil},
-	{"WrapUnclosed", `hello <em>world`, `<p>hello <em>world</em></p>`, &Config{Elem: DefaultConfig.Elem, WrapText: true}},
-	{"WrapStraySpace", `<p>hello</p> <p>world</p>`, `<p>hello</p> <p>world</p>`, &Config{Elem: DefaultConfig.Elem, WrapText: true}},
-	{"WrapInvalidNesting", `<em>hello <p>world</p>`, `<p><em>hello </em></p><p><em>world</em></p><p></p>`, &Config{Elem: DefaultConfig.Elem, WrapText: true}},
+	{"WrapUnclosed", `hello <em>world`, `<p>hello <em>world</em></p>`, wrapConfig},
+	{"WrapStraySpace", `<p>hello</p> <p>world</p>`, `<p>hello</p> <p>world</p>`, wrapConfig},
+	{"WrapInvalidNesting", `<em>hello <p>world</p>`, `<p><em>hello </em></p><p><em>world</em></p><p></p>`, wrapConfig},
 }
 
 func TestClean(t *testing.T) {