/*
* Read current tag with attributes.
* @param tk *html.Tokenizer - tokenizer instance
* @return *Tag - reference to read tag
*/
func readTag(tk *html.Tokenizer) *Tag {
// we are only interested in certain tags
tag, _ := tk.TagName()
name := string(tag)
switch name {
//-----------------------------------------------------
// external script files
//-----------------------------------------------------
case "script":
attrs := getAttrs(tk)
if attrs != nil {
if _, ok := attrs["src"]; ok {
// add external reference to script file
return NewTag("script", attrs)
}
}
//-----------------------------------------------------
// external image
//-----------------------------------------------------
case "img":
attrs := getAttrs(tk)
if attrs != nil {
return NewTag("img", attrs)
}
//-----------------------------------------------------
// external links (style sheets)
//-----------------------------------------------------
case "link":
attrs := getAttrs(tk)
if attrs != nil {
if _, ok := attrs["href"]; ok {
// add external reference to link
return NewTag("link", attrs)
}
}
//-----------------------------------------------------
// input fields
//-----------------------------------------------------
case "input":
attrs := getAttrs(tk)
if attrs != nil {
if _, ok := attrs["type"]; ok {
// add external reference to link
return NewTag("input", attrs)
}
}
}
//-----------------------------------------------------
// ignore all other tags (no tag processed).
//-----------------------------------------------------
return nil
}
func traverse_html_tokenizer(z *html.Tokenizer) {
for {
if z.Next() == html.ErrorToken {
return
}
text_b := z.Text()
tag_name_b, hasAttri := z.TagName()
tag_attr_key_b, tag_attr_value_b, _ := z.TagAttr()
text := string(text_b)
tag_name := string(tag_name_b)
tag_attr_key := string(tag_attr_key_b)
tag_attr_value := string(tag_attr_value_b)
fmt.Printf("|Tokenizer.Text:%-10s|Tokenizer.TagName:%-10s|hasAttri:%-10t|tag_attr_key:%-10s|tag_attr_value:%-10s|\n", text, tag_name, hasAttri, tag_attr_key, tag_attr_value)
}
}
func burnTokensUntilEndTag(firewood *html.Tokenizer, tagName string) {
rawTagName := []byte(tagName)
for {
token := firewood.Next()
switch token {
case html.ErrorToken:
return
case html.EndTagToken:
name, _ := firewood.TagName()
// log.Println("Struck token " + string(name))
if bytes.Equal(name, rawTagName) {
// log.Println("Extinguishing token fire.")
return
}
}
}
}
func textUpToEndTag(tokenizer *html.Tokenizer, tagName string) []byte {
var textBuffer bytes.Buffer
rawTagName := []byte(tagName)
for done := false; !done; {
token := tokenizer.Next()
switch token {
case html.TextToken:
textBuffer.Write(tokenizer.Text())
case html.EndTagToken:
name, _ := tokenizer.TagName()
if bytes.Equal(rawTagName, name) {
done = true
}
case html.ErrorToken:
done = true
}
}
return textBuffer.Bytes()
}
// getTagName gets a tagName from tokenizer.
func getTagName(tokenizer *html.Tokenizer) string {
tagName, _ := tokenizer.TagName()
return string(tagName)
}