@SuppressLint("DefaultLocale")
private String improveHtml(final String html) {
final Document document = Jsoup.parse(html);
for (final Element e : document.getAllElements()) {
if (e.hasAttr("style")) {
for (final Attribute a : e.attributes()) {
if (a.getKey().compareTo("style") == 0) {
final String[] items = a.getValue().trim().split(";");
String newValue = "";
for (final String item : items) {
if (!item.toLowerCase(Locale.ENGLISH).contains("font-family:")
&& !item.toLowerCase(Locale.ENGLISH).contains("font-size:")) {
newValue = newValue.concat(item).concat(";");
}
}
a.setValue(newValue);
}
}
}
}
return document.body().html();
}
/**
* @param attributes A list of attribs
* @return Returns a mutable map parsed out of the attribute list
*/
public static Map<String, String> parseAttribs(Attributes attributes) {
Map<String, String> attrs = new LinkedHashMap<String, String>(attributes.size() + 4);
for (Attribute a : attributes.asList())
if (!SKIP_ATTR.contains(a.getKey())) attrs.put(a.getKey(), a.getValue());
return attrs;
}
private static String cleanHtml(final Node node) {
if (node instanceof Element) {
Element element = ((Element) node);
StringBuilder accum = new StringBuilder();
accum.append("<").append(element.tagName());
for (Attribute attribute : element.attributes()) {
if (!(attribute.getKey().startsWith("_"))) {
accum.append(" ");
accum.append(attribute.getKey());
accum.append("=\"");
accum.append(attribute.getValue());
accum.append('"');
}
}
if (element.childNodes().isEmpty() && element.tag().isEmpty()) {
accum.append(" />");
} else {
accum.append(">");
for (Node child : element.childNodes()) accum.append(cleanHtml(child));
accum.append("</").append(element.tagName()).append(">");
}
return accum.toString();
} else if (node instanceof TextNode) {
return ((TextNode) node).getWholeText();
} else if (node instanceof XmlDeclaration) {
// HACK
if (node.childNodes().isEmpty()) {
return "";
}
return node.outerHtml();
} else if (node instanceof Comment) {
// HACK: elide comments for now.
return "";
} else if (node instanceof DataNode && node.childNodes().isEmpty()) {
// No child nodes are defined but we have to handle content if such exists, example
// <script language="JavaScript">var a = { name: "${user.name}"}</script>
String content = node.attr("data");
if (Strings.empty(content)) {
return "";
}
return content;
} else {
return node.outerHtml();
}
}
private boolean testValidProtocol(Element el, Attribute attr, Set<Protocol> protocols) {
// try to resolve relative urls to abs, and optionally update the attribute so output html has
// abs.
// rels without a baseuri get removed
String value = el.absUrl(attr.getKey());
if (value.length() == 0)
value = attr.getValue(); // if it could not be made abs, run as-is to allow custom unknown
// protocols
if (!preserveRelativeLinks) attr.setValue(value);
for (Protocol protocol : protocols) {
String prot = protocol.toString() + ":";
if (value.toLowerCase().startsWith(prot)) {
return true;
}
}
return false;
}
