|
|
@@ -49,23 +49,27 @@ public class DetailCleaner {
|
|
|
*/
|
|
|
public static String clean(String html) {
|
|
|
Document doc = Jsoup.parse(html);
|
|
|
- doc.select("meta, script").remove();
|
|
|
- doc.select("#hidden-box").remove();
|
|
|
- doc.select("div.hidden-box").remove();
|
|
|
- for (Element container : doc.select("div.container")) {
|
|
|
- Elements children = new Elements();
|
|
|
- container.children().forEach(children::add);
|
|
|
- for (Element child : children) {
|
|
|
- boolean keep = child.is("div")
|
|
|
- && (child.hasClass("article-title")
|
|
|
- || child.hasClass("not-exist-media-leader")
|
|
|
- || child.hasClass("article-content"));
|
|
|
- if (!keep) {
|
|
|
- child.remove();
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- return doc.html();
|
|
|
+
|
|
|
+ html = """
|
|
|
+ <!DOCTYPE html>
|
|
|
+ <html lang="en" style="font-size: 18px;">
|
|
|
+ <head></head>
|
|
|
+ <body>
|
|
|
+ </body>
|
|
|
+ </html>
|
|
|
+ """;
|
|
|
+ Document result = Jsoup.parse(html);
|
|
|
+
|
|
|
+ Elements style = doc.select("link[rel='stylesheet']");
|
|
|
+ Element head = result.head();
|
|
|
+ head.insertChildren(0, style);
|
|
|
+
|
|
|
+ Element container = doc.select("div.container").first();
|
|
|
+ Elements keep = container.children().select("div.article-title, div.not-exist-media-leader, div.article-content");
|
|
|
+ keep.select("video").remove();
|
|
|
+ result.body().appendChildren(keep);
|
|
|
+
|
|
|
+ return result.html();
|
|
|
}
|
|
|
|
|
|
/**
|
