Apply the document's parse options to HTML added to it later

manipulation.go

@@ -388,7 +388,7 @@ func (s *Selection) WrapHtml(htmlStr string) *Selection {
 		if parent == nil {
 			parent = &html.Node{Type: html.ElementNode}
 		}
-		nodes := cachedParseHtmlWithContext(cache, htmlStr, parent)
+		nodes := cachedParseHtmlWithContext(cache, htmlStr, parent, s.parseOpts())
 		newSingleSelection(n, s.document).wrapAllNodes(cloneNodes(nodes)...)
 	}
 	return s
@@ -449,9 +449,9 @@ func (s *Selection) WrapAllHtml(htmlStr string) *Selection {
 	if len(s.Nodes) > 0 {
 		context = s.Nodes[0]
 		if context.Parent != nil {
-			nodes = parseHtmlWithContext(htmlStr, context)
+			nodes = parseHtmlWithContext(htmlStr, context, s.parseOpts())
 		} else {
-			nodes = parseHtml(htmlStr)
+			nodes = parseHtml(htmlStr, s.parseOpts())
 		}
 	}
 	return s.wrapAllNodes(nodes...)
@@ -526,7 +526,7 @@ func (s *Selection) WrapInnerSelection(sel *Selection) *Selection {
 func (s *Selection) WrapInnerHtml(htmlStr string) *Selection {
 	cache := make(map[string][]*html.Node)
 	for _, n := range s.Nodes {
-		nodes := cachedParseHtmlWithContext(cache, htmlStr, n)
+		nodes := cachedParseHtmlWithContext(cache, htmlStr, n, s.parseOpts())
 		newSingleSelection(n, s.document).wrapInnerNodes(cloneNodes(nodes)...)
 	}
 	return s
@@ -559,14 +559,23 @@ func (s *Selection) wrapInnerNodes(ns ...*html.Node) *Selection {
 	return s
 }
 
-func parseHtml(h string) []*html.Node {
+// parseOpts returns the parse options of the document this selection belongs to,
-	return parseHtmlWithContext(h, &html.Node{Type: html.ElementNode})
+// so an HTML string added later is parsed the same way the document was.
+func (s *Selection) parseOpts() []html.ParseOption {
+	if s == nil || s.document == nil {
+		return nil
+	}
+	return s.document.parseOptions
 }
 
-func parseHtmlWithContext(h string, context *html.Node) []*html.Node {
+func parseHtml(h string, opts []html.ParseOption) []*html.Node {
+	return parseHtmlWithContext(h, &html.Node{Type: html.ElementNode}, opts)
+}
+
+func parseHtmlWithContext(h string, context *html.Node, opts []html.ParseOption) []*html.Node {
 	// Errors are only returned when the io.Reader returns any error besides
 	// EOF, but strings.Reader never will
-	nodes, err := html.ParseFragment(strings.NewReader(h), context)
+	nodes, err := html.ParseFragmentWithOptions(strings.NewReader(h), context, opts...)
 	if err != nil {
 		panic("goquery: failed to parse HTML: " + err.Error())
 	}
@@ -660,7 +669,7 @@ func (s *Selection) eachNodeHtml(htmlStr string, isParent bool, mergeFn func(n *
 		if context == nil {
 			continue
 		}
-		nodes := cachedParseHtmlWithContext(cache, htmlStr, context)
+		nodes := cachedParseHtmlWithContext(cache, htmlStr, context, s.parseOpts())
 		mergeFn(n, cloneNodes(nodes))
 	}
 	return s
@@ -669,12 +678,12 @@ func (s *Selection) eachNodeHtml(htmlStr string, isParent bool, mergeFn func(n *
 // cachedParseHtmlWithContext returns parseHtmlWithContext(htmlStr, context), reusing a prior
 // result when context's nodeName has already been seen. Callers pass their own
 // cache map so the cache lifetime matches the caller's loop scope.
-func cachedParseHtmlWithContext(cache map[string][]*html.Node, htmlStr string, context *html.Node) []*html.Node {
+func cachedParseHtmlWithContext(cache map[string][]*html.Node, htmlStr string, context *html.Node, opts []html.ParseOption) []*html.Node {
 	key := nodeName(context)
 	if nodes, ok := cache[key]; ok {
 		return nodes
 	}
-	nodes := parseHtmlWithContext(htmlStr, context)
+	nodes := parseHtmlWithContext(htmlStr, context, opts)
 	cache[key] = nodes
 	return nodes
 }

type.go

@@ -19,6 +19,9 @@ type Document struct {
 	*Selection
 	Url      *url.URL
 	rootNode *html.Node
+	// parse options this document was built with, reused whenever an HTML
+	// string is parsed into it later on
+	parseOptions []html.ParseOption
 }
 
 // NewDocumentFromNode is a Document constructor that takes a root html Node
@@ -71,7 +74,7 @@ func NewDocumentFromReaderWithOptions(r io.Reader, opts ...html.ParseOption) (*D
 	if e != nil {
 		return nil, e
 	}
-	return newDocument(root, nil), nil
+	return newDocument(root, nil, opts...), nil
 }
 
 // NewDocumentFromResponse is another Document constructor that takes an http response as argument.
@@ -100,13 +103,13 @@ func NewDocumentFromResponse(res *http.Response) (*Document, error) {
 
 // CloneDocument creates a deep-clone of a document.
 func CloneDocument(doc *Document) *Document {
-	return newDocument(cloneNode(doc.rootNode), doc.Url)
+	return newDocument(cloneNode(doc.rootNode), doc.Url, doc.parseOptions...)
 }
 
 // Private constructor, make sure all fields are correctly filled.
-func newDocument(root *html.Node, url *url.URL) *Document {
+func newDocument(root *html.Node, url *url.URL, opts ...html.ParseOption) *Document {
 	// Create and fill the document
-	d := &Document{nil, url, root}
+	d := &Document{Url: url, rootNode: root, parseOptions: opts}
 	d.Selection = newSingleSelection(root, d)
 	return d
 }

type_test.go

@@ -290,3 +290,29 @@ func TestNewDocumentFromReaderWithOptionsNoOptions(t *testing.T) {
 		t.Errorf("got %q, want %q", got, "hi")
 	}
 }
+
+func TestNewDocumentFromReaderWithOptionsAppliesToAddedHtml(t *testing.T) {
+	const doc = `<html><body><p>before</p></body></html>`
+	const frag = `<noscript><a href="http://example.org">click this link</a></noscript>`
+
+	// The options given to the constructor hold for HTML added later, so the
+	// fragment is parsed the same way the document was.
+	d, e := NewDocumentFromReaderWithOptions(strings.NewReader(doc), html.ParseOptionEnableScripting(false))
+	if e != nil {
+		t.Fatal(e)
+	}
+	d.Find("body").AppendHtml(frag)
+	if _, ok := d.Find("noscript a").Attr("href"); !ok {
+		t.Error("expected the added noscript to be parsed with scripting disabled")
+	}
+
+	// Without the options the added fragment keeps the default behaviour.
+	d, e = NewDocumentFromReader(strings.NewReader(doc))
+	if e != nil {
+		t.Fatal(e)
+	}
+	d.Find("body").AppendHtml(frag)
+	if _, ok := d.Find("noscript a").Attr("href"); ok {
+		t.Error("expected no match inside the added noscript with scripting enabled")
+	}
+}