Advertisement
Guest User

Untitled

a guest
Feb 12th, 2016
64
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.19 KB | None | 0 0
  1. private String cleanHTML(String str) {
  2. Element element = Jsoup.parse(str);
  3.  
  4. final StringBuilder buffer = new StringBuilder();
  5.  
  6. new NodeTraversor(new NodeVisitor() {
  7. boolean isNewline = true;
  8.  
  9. @Override
  10. public void head(Node node, int depth) {
  11. }
  12.  
  13. @Override
  14. public void tail(Node node, int depth) {
  15. if (node instanceof TextNode) {
  16. TextNode textNode = (TextNode) node;
  17. String text = textNode.text();//.replace('\u00A0', ' ').trim();
  18. if(!text.isEmpty()) {
  19. buffer.append(text);
  20. isNewline = false;
  21. }
  22. } else if (node instanceof Element) {
  23. Element element = (Element) node;
  24. if (!isNewline) {
  25. if((element.isBlock() || element.tagName().equals("br"))) {
  26. buffer.append("\n");
  27. isNewline = true;
  28. }
  29. }
  30. }
  31. }
  32. }).traverse(element);
  33.  
  34. return buffer.toString();
  35. }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement