public ArrayList<JuickMessage> parseWebMessageListPure(String htmlStr) {
ArrayList<JuickMessage> retval = new ArrayList<JuickMessage>();
Document parsed = Jsoup.parse(htmlStr);
Elements posts = parsed.select("div");
ISimpleDateFormat sdf;
ISimpleDateFormat sdf2;
sdf = DevJuickComMessages.sdftz.createSDF("yyyy dd MMM HH:mm", "en", "US", "UTC");
sdf2 = DevJuickComMessages.sdftz.createSDF("yyyy dd MMM HH:mm", "ru", "RU", "UTC");
Calendar cal = Calendar.getInstance();
int currentYear = cal.get(Calendar.YEAR);
for (Element post : posts) {
String postClass = post.attr("class");
if (postClass.equals("post") || postClass.startsWith("post ")) {
PointMessage message = new PointMessage();
message.User = new JuickUser();
message.User.UName = post.select("div[class=info] > a > img").attr("alt");
if (message.User.UName.length() == 0) {
message.User.UName = post.select("div[class=author] > a").text();
}
String dataId = post.attr("data-id");
String dataCommentId = post.attr("data-comment-id");
String dataToCommentId = post.attr("data-to-comment-id");
message.setMID(new PointMessageID(message.User.UName, dataId, 0));
if (dataCommentId.length() > 0) {
message.setRID(Integer.parseInt(dataCommentId));
}
if (dataToCommentId.length() > 0) {
message.setReplyTo(Integer.parseInt(dataToCommentId));
}
message.tags = new Vector<String>();
for (Element el : post.select("a[class=tag]")) {
message.tags.add(el.text());
}
message.microBlogCode = PointMessageID.CODE;
StringBuilder dt = new StringBuilder();
for (Element el : post.select("div[class=created]")) {
dt.append(" ");
dt.append(el.text());
}
try {
message.Timestamp = new Date(sdf.parse(currentYear + " " + dt.toString().trim()));
} catch (IllegalArgumentException e) {
try {
message.Timestamp = new Date(sdf2.parse(currentYear + " " + dt.toString().trim()));
} catch (IllegalArgumentException e1) {
continue;
}
}
Date mt = message.Timestamp;
if (mt.getTime() > System.currentTimeMillis() + 50 * 24 * 60 * 60 * 1000L) {
Calendar cal2 = Calendar.getInstance();
cal2.setTime(mt);
cal2.set(Calendar.YEAR, cal2.get(Calendar.YEAR) - 1);
message.Timestamp = cal2.getTime();
}
Elements postEls = post.select("div[class=text-content]");
if (postEls.size() < 1) {
postEls = post.select("div[class=text]");
}
String referencedImages = "";
Elements postimg = post.select("a[class=postimg]");
for (Element as : postimg) {
Elements imgs = as.select("img");
for (Element img : imgs) {
String src = img.attr("src");
referencedImages += " " + src;
}
}
message.csrf_token = post.select("input[name=csrf_token]").attr("value");
// last part
if (postEls.size() < 1) {
message.Text = "Error parsing text ;-(";
} else {
Element elem = cleanupElement(postEls.get(0));
postEls.get(0).appendChild(elem); // add to document
Document.OutputSettings os = elem.ownerDocument().outputSettings();
os.prettyPrint(false);
String text = Utils.replace(elem.toString(), "\n", " ");
text = Utils.replace(text, "&", "&"); // this was improperly done in cleanupElement
while (true) {
long olds = text.length();
text = Utils.replace(text, " ", " ");
long news = text.length();
if (news == olds) break;
}
try {
message.replies = Integer.parseInt(post.select("span[class=cn]").text());
} catch (Exception ex) {
}
text += referencedImages;
message.Text = unwebMessageTextPoint(text);
}
retval.add(message);
}
}
for (JuickMessage juickMessage : retval) {
if (juickMessage.getRID() != 0
&& juickMessage.getReplyTo() != 0
&& !juickMessage.Text.startsWith("@")) {
String uzur = null;
for (JuickMessage scan : retval) {
if (scan.getRID() == juickMessage.getReplyTo()) {
uzur = scan.User.UName;
break;
}
}
if (uzur != null) {
juickMessage.Text = "@" + uzur + " " + juickMessage.Text;
}
}
}
return retval;
}