1 package us.codecraft.webmagic.utils;
2
3 /**
4 * @author hooy
5 */
6 public class BaseSelectorUtils {
7
8 /**
9 * Jsoup/HtmlCleaner could not parse "tr" or "td" tag directly
10 * https://stackoverflow.com/questions/63607740/jsoup-couldnt-parse-tr-tag
11 *
12 * @param text - the html string
13 * @return text
14 */
15 public static String preParse(String text) {
16 if (((text.startsWith("<tr>") || text.startsWith("<tr ")) && text.endsWith("</tr>"))
17 || ((text.startsWith("<td>") || text.startsWith("<td ")) && text.endsWith("</td>"))) {
18 text = "<table>" + text + "</table>";
19 }
20 return text;
21 }
22
23 }