forked from ukamnya/microdata_mirror
Compare commits
No commits in common. "master" and "master" have entirely different histories.
2
go.mod
2
go.mod
|
@ -1,4 +1,4 @@
|
|||
module git.ukamnya.ru/ukamnya/microdata
|
||||
module github.com/iand/microdata
|
||||
|
||||
go 1.21.4
|
||||
|
||||
|
|
40
microdata.go
40
microdata.go
|
@ -31,7 +31,7 @@ type Item struct {
|
|||
// NewItem creates a new microdata item
|
||||
func NewItem() *Item {
|
||||
return &Item{
|
||||
Properties: make(propertyMap),
|
||||
Properties: make(propertyMap, 0),
|
||||
Types: make([]string, 0),
|
||||
}
|
||||
}
|
||||
|
@ -104,7 +104,7 @@ func (p *Parser) Parse() (*Microdata, error) {
|
|||
}
|
||||
|
||||
topLevelItemNodes := make([]*html.Node, 0)
|
||||
p.identifiedNodes = make(map[string]*html.Node)
|
||||
p.identifiedNodes = make(map[string]*html.Node, 0)
|
||||
|
||||
walk(tree, func(n *html.Node) {
|
||||
if n.Type == html.ElementNode {
|
||||
|
@ -171,47 +171,39 @@ func (p *Parser) readItem(item *Item, node *html.Node) *Item {
|
|||
}
|
||||
}
|
||||
} else {
|
||||
var propertyValue *string
|
||||
var propertyValue string
|
||||
|
||||
switch node.DataAtom {
|
||||
case atom.Meta:
|
||||
if val, exists := getAttr("content", node); exists {
|
||||
propertyValue = &val
|
||||
propertyValue = val
|
||||
}
|
||||
case atom.Audio, atom.Embed, atom.Iframe, atom.Img, atom.Source, atom.Track, atom.Video:
|
||||
if urlValue, exists := getAttr("src", node); exists {
|
||||
if parsedURL, err := p.base.Parse(urlValue); err == nil {
|
||||
parsedStr := parsedURL.String()
|
||||
propertyValue = &parsedStr
|
||||
propertyValue = parsedURL.String()
|
||||
}
|
||||
}
|
||||
case atom.A, atom.Area, atom.Link:
|
||||
if urlValue, exists := getAttr("href", node); exists {
|
||||
if parsedURL, err := p.base.Parse(urlValue); err == nil {
|
||||
parsedStr := parsedURL.String()
|
||||
propertyValue = &parsedStr
|
||||
propertyValue = parsedURL.String()
|
||||
}
|
||||
}
|
||||
case atom.Object:
|
||||
if urlValue, exists := getAttr("data", node); exists {
|
||||
propertyValue = &urlValue
|
||||
propertyValue = urlValue
|
||||
}
|
||||
case atom.Data, atom.Meter:
|
||||
if urlValue, exists := getAttr("value", node); exists {
|
||||
propertyValue = &urlValue
|
||||
propertyValue = urlValue
|
||||
}
|
||||
case atom.Time:
|
||||
if urlValue, exists := getAttr("datetime", node); exists {
|
||||
propertyValue = &urlValue
|
||||
propertyValue = urlValue
|
||||
}
|
||||
|
||||
default:
|
||||
// The "content" attribute can be found on other tags besides the meta tag.
|
||||
if val, ok := getAttr("content", node); ok {
|
||||
propertyValue = &val
|
||||
break
|
||||
}
|
||||
|
||||
var text bytes.Buffer
|
||||
walk(node, func(n *html.Node) {
|
||||
if n.Type == html.TextNode {
|
||||
|
@ -219,22 +211,14 @@ func (p *Parser) readItem(item *Item, node *html.Node) *Item {
|
|||
}
|
||||
|
||||
})
|
||||
|
||||
val := text.String()
|
||||
propertyValue = &val
|
||||
propertyValue = text.String()
|
||||
}
|
||||
|
||||
if propertyValue == nil {
|
||||
if val, ok := getAttr("content", node); ok {
|
||||
propertyValue = &val
|
||||
}
|
||||
}
|
||||
|
||||
if propertyValue != nil {
|
||||
if len(propertyValue) > 0 {
|
||||
for _, propertyName := range strings.Split(strings.TrimSpace(itemprop), " ") {
|
||||
propertyName = strings.TrimSpace(propertyName)
|
||||
if propertyName != "" {
|
||||
item.AddString(propertyName, *propertyValue)
|
||||
item.AddString(propertyName, propertyValue)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
Loading…
Reference in New Issue