@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();
}
/** Produce predictable html (attributes in alphabetical order), always include close tags */
private String elementToHtml(Element producedElem, StringBuilder sb) {
ArrayList<String> names = new ArrayList<String>();
for (Attribute a : producedElem.attributes().asList()) {
names.add(a.getKey());
}
Collections.sort(names);
sb.append("<" + producedElem.tagName() + "");
for (String attrName : names) {
sb.append(" ")
.append(attrName)
.append("=")
.append("\'")
.append(producedElem.attr(attrName))
.append("\'");
}
sb.append(">");
for (Node child : producedElem.childNodes()) {
if (child instanceof Element) {
elementToHtml((Element) child, sb);
} else if (child instanceof TextNode) {
String text = ((TextNode) child).text();
sb.append(text.trim());
}
}
sb.append("</").append(producedElem.tagName()).append(">");
return sb.toString();
}
/**
* @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;
}
/**
* Test if the supplied attribute is allowed by this whitelist for this tag
*
* @param tagName tag to consider allowing the attribute in
* @param el element under test, to confirm protocol
* @param attr attribute under test
* @return true if allowed
*/
protected boolean isSafeAttribute(String tagName, Element el, Attribute attr) {
TagName tag = TagName.valueOf(tagName);
AttributeKey key = AttributeKey.valueOf(attr.getKey());
if (attributes.containsKey(tag)) {
if (attributes.get(tag).contains(key)) {
if (protocols.containsKey(tag)) {
Map<AttributeKey, Set<Protocol>> attrProts = protocols.get(tag);
// ok if not defined protocol; otherwise test
return !attrProts.containsKey(key) || testValidProtocol(el, attr, attrProts.get(key));
} else { // attribute found, no protocols defined, so OK
return true;
}
}
}
// no attributes defined for tag, try :all tag
return !tagName.equals(":all") && isSafeAttribute(":all", el, attr);
}
public static String ConvertHtmlToEnml(String html) // , Note note)
{
String[] prohibitedArray =
new String[] {
"applet",
"base",
"basefont",
"bgsound",
"blink",
"body",
"button",
"dir",
"embed",
"fieldset",
"form",
"frame",
"frameset",
"head",
"html",
"iframe",
"ilayer",
"input",
"isindex",
"label",
"layer",
"legend",
"link",
"marquee",
"menu",
"meta",
"noframes",
"noscript",
"object",
"optgroup",
"option",
"param",
"plaintext",
"script",
"select",
"style",
"textarea",
"xml",
"image"
};
String[] disableAttributesArray =
new String[] {"id", "class", "accesskey", "data", "dynsrc", "tabindex", "sizset"};
List<String> prohibited = Arrays.asList(prohibitedArray);
List<String> disableAttributes = Arrays.asList(disableAttributesArray);
Document doc = Jsoup.parse(html);
// var imgs = new ImageRecordDbPersistence().LoadFromFile().ToList();
ImageRecordDalHelper helper = new ImageRecordDalHelper();
Elements nodes = doc.getAllElements();
int total = nodes.size() - 1;
for (int j = total; j >= 0; j--) {
// remove all prohibited node
if (prohibited.contains(nodes.get(j).tagName())) {
if (!(nodes.get(j).childNodeSize() > 0)) nodes.get(j).remove();
else {
for (Element child : nodes.get(j).children()) nodes.get(j).parent().appendChild(child);
nodes.get(j).remove();
}
}
// remove disabled attribute
if (nodes.get(j).attributes().size() > 0) {
int count = nodes.get(j).attributes().size() - 1;
List<Attribute> attributes = nodes.get(j).attributes().asList();
// int count = disableAttributes.size();
for (int i = count; i >= 0; i--) {
if (disableAttributes.contains(attributes.get(i).getKey())) {
nodes.get(j).removeAttr(attributes.get(i).getKey());
continue;
}
// deal with on*
if (attributes.get(i).getKey().startsWith("on")) {
nodes.get(j).removeAttr(attributes.get(i).getKey());
continue;
}
if (attributes.get(i).getKey().startsWith("sizcache")) {
nodes.get(j).removeAttr(attributes.get(i).getKey());
continue;
}
if (attributes.get(i).getKey().startsWith("f-size")) {
nodes.get(j).removeAttr(attributes.get(i).getKey());
continue;
}
}
}
// deal with relative href
if (nodes.get(j).tagName().equals("a")) {
if (nodes.get(j).attributes().size() > 0 && nodes.get(j).hasAttr("href")) {
String href = nodes.get(j).attr("href");
if (!href.startsWith("http") || !href.startsWith("https") || !href.startsWith("www")) {
nodes.get(j).removeAttr("href");
}
}
}
// deal with cached img, replace with online src
if (nodes.get(j).tagName().equals("img") && nodes.get(j).attributes().size() > 0) {
if (nodes.get(j).hasAttr("xsrc")) {
if (nodes.get(j).attr("xsrc").startsWith("mnt:")) {
String src = nodes.get(j).attr("xsrc");
for (Attribute attr : nodes.get(j).attributes().asList()) {
nodes.get(j).removeAttr(attr.getKey());
}
// nodes.get(j).attr("src", imgs.First(r => src.Contains(r.StoredName)).OriginUrl);
nodes.get(j).attr("src", helper.GetImageRecordEntityByStoreName(src).OriginUrl);
} else {
String xsrc = nodes.get(j).attr("xsrc");
;
for (Attribute attr : nodes.get(j).attributes().asList()) {
nodes.get(j).removeAttr(attr.getKey());
}
nodes.get(j).attr("src", xsrc);
}
} else {
for (Attribute attr : nodes.get(j).attributes().asList()) {
nodes.get(j).removeAttr(attr.getKey());
}
}
// better reading experience in mobile client and web client
nodes.get(j).attr("style", "max-height:100%; max-width:100%;");
}
if (nodes.get(j).tagName().equals("a") && nodes.get(j).attributes().size() > 0) {
if (nodes.get(j).hasAttr("href")) {
String href = nodes.get(j).attr("href");
for (Attribute attr : nodes.get(j).attributes().asList()) {
nodes.get(j).removeAttr(attr.getKey());
}
nodes.get(j).attr("href", href);
} else {
for (Attribute attr : nodes.get(j).attributes().asList()) {
nodes.get(j).removeAttr(attr.getKey());
}
}
}
}
char[] xmlChar = doc.html().toCharArray();
for (int i = 0; i < xmlChar.length; ++i) {
if (xmlChar[i] > 0xFFFD) {
// 或者直接替换掉0xb
xmlChar[i] = ' '; // 用空格替换
} else if (xmlChar[i] < 0x20 && xmlChar[i] != 't' & xmlChar[i] != 'n' & xmlChar[i] != 'r') {
// 或者直接替换掉0xb
xmlChar[i] = ' '; // 用空格替换
}
}
helper.Close();
return new String(xmlChar).replace("<?xml version=\"1.0\" encoding=\"utf-8\"?>", "");
}
