func main() {
// Find ourself some html
resp, err := http.Get("https://github.com/Wessie/unhtml/commits/master")
if err != nil {
log.Fatal(err)
}
// Parse the HTML, accepts an io.Reader
d, err := unhtml.NewDecoder(resp.Body)
if err != nil {
log.Fatal(err)
}
// Unmarshal into a struct with xpath tags for extracting things
result := GithubCommits{}
if err := d.Unmarshal(&result); err != nil {
log.Fatal(err)
}
// Convert to JSON for pretty printing the result
b, err := json.MarshalIndent(result, "", " ")
if err != nil {
log.Fatal(err)
}
fmt.Println(string(b))
}
func Parse(r io.Reader) ([]Item, error) {
for i, v := range xml.HTMLAutoClose {
if v == "link" {
xml.HTMLAutoClose[i] = "bogus"
break
}
}
d, err := unhtml.NewDecoder(r)
if err != nil {
return nil, err
}
var result []Item
var startPath = "rss/channel/item"
if err := d.UnmarshalRelative(startPath, &result); err != nil {
return nil, err
}
return result, nil
}