ensure itemscope always starts a new item

microdata.go

@@ -121,8 +121,18 @@ func (p *Parser) Parse() (*Microdata, error) {
 	})
 
 	for _, node := range topLevelItemNodes {
-		item := NewItem()
-		p.data.Items = append(p.data.Items, item)
+		p.data.Items = append(p.data.Items, p.readItem(nil, node))
+	}
+
+	return p.data, nil
+}
+
+func (p *Parser) readItem(item *Item, node *html.Node) *Item {
+	var parent *Item
+
+	if _, exists := getAttr("itemscope", node); exists {
+		parent, item = item, NewItem()
+
 		if itemtypes, exists := getAttr("itemtype", node); exists {
 			for _, itemtype := range strings.Split(strings.TrimSpace(itemtypes), " ") {
 				itemtype = strings.TrimSpace(itemtype)
@@ -143,53 +153,24 @@ func (p *Parser) Parse() (*Microdata, error) {
 				itemref = strings.TrimSpace(itemref)
 
 				if refnode, exists := p.identifiedNodes[itemref]; exists {
+					if refnode != node {
 						p.readItem(item, refnode)
 					}
 				}
 			}
-
-		for child := node.FirstChild; child != nil; {
-			p.readItem(item, child)
-			child = child.NextSibling
 		}
 	}
 
-	return p.data, nil
-}
-
-func (p *Parser) readItem(item *Item, node *html.Node) {
 	if itemprop, exists := getAttr("itemprop", node); exists {
-		if _, exists := getAttr("itemscope", node); exists {
-			subitem := NewItem()
-
-			if itemrefs, exists := getAttr("itemref", node); exists {
-				for _, itemref := range strings.Split(strings.TrimSpace(itemrefs), " ") {
-					itemref = strings.TrimSpace(itemref)
-
-					if refnode, exists := p.identifiedNodes[itemref]; exists {
-						if refnode != node {
-							p.readItem(subitem, refnode)
-						}
-					}
-				}
-			}
-
-			for child := node.FirstChild; child != nil; {
-				p.readItem(subitem, child)
-				child = child.NextSibling
-			}
-
+		if parent != nil {
+			// an itemprop on an itemscope has value of the item created by the itemscope
 			for _, propertyName := range strings.Split(strings.TrimSpace(itemprop), " ") {
 				propertyName = strings.TrimSpace(propertyName)
 				if propertyName != "" {
-					item.AddItem(propertyName, subitem)
+					parent.AddItem(propertyName, item)
 				}
 			}
-
-			return
-
-		}
-
+		} else {
 			var propertyValue string
 
 			switch node.DataAtom {
@@ -241,7 +222,7 @@ func (p *Parser) readItem(item *Item, node *html.Node) {
 					}
 				}
 			}
-
+		}
 	}
 
 	for child := node.FirstChild; child != nil; {
@@ -249,6 +230,7 @@ func (p *Parser) readItem(item *Item, node *html.Node) {
 		child = child.NextSibling
 	}
 
+	return item
 }
 
 func getAttr(name string, node *html.Node) (string, bool) {

microdata_test.go

@@ -583,8 +583,9 @@ func TestSkipSelfReferencingItemref(t *testing.T) {
 	actual := ParseData(html, t)
 
 	child := NewItem()
-	child.AddString("title", "Foo")
+	child.AddType("http://data-vocabulary.org/Breadcrumb")
 	child.AddString("url", "http://example.com/foo/bar")
+	child.AddString("title", "Foo")
 
 	item := NewItem()
 	item.AddType("http://schema.org/WebPage")
@@ -609,11 +610,9 @@ func TestPropertiesInContainedItem(t *testing.T) {
 			<div itemscope itemtype="http://schema.org/Person">
 				<meta itemprop="bar" content="bar value">
 			</div>
-
 			<div itemscope itemtype="http://schema.org/Person" itemprop="author">
 				<meta itemprop="baz" content="baz value">
 			</div>
-
 		</body>`
 
 	actual := ParseData(html, t)