/**
* Carries out preprocessing that makes JEuclid handle the document better.
*
* @param doc Document
*/
static void preprocessForJEuclid(Document doc) {
// underbrace and overbrace
NodeList list = doc.getElementsByTagName("mo");
for (int i = 0; i < list.getLength(); i++) {
Element mo = (Element) list.item(i);
String parentName = ((Element) mo.getParentNode()).getTagName();
if (parentName == null) {
continue;
}
if (parentName.equals("munder") && isTextChild(mo, "\ufe38")) {
mo.setAttribute("stretchy", "true");
mo.removeChild(mo.getFirstChild());
mo.appendChild(doc.createTextNode("\u23df"));
} else if (parentName.equals("mover") && isTextChild(mo, "\ufe37")) {
mo.setAttribute("stretchy", "true");
mo.removeChild(mo.getFirstChild());
mo.appendChild(doc.createTextNode("\u23de"));
}
}
// menclose for long division doesn't allow enough top padding. Oh, and
// <mpadded> isn't implemented. And there isn't enough padding to left of
// the bar either. Solve by adding an <mover> with just an <mspace> over#
// the longdiv, contained within an mrow that adds a <mspace> before it.
list = doc.getElementsByTagName("menclose");
for (int i = 0; i < list.getLength(); i++) {
Element menclose = (Element) list.item(i);
// Only for longdiv
if (!"longdiv".equals(menclose.getAttribute("notation"))) {
continue;
}
Element mrow = doc.createElementNS(WebMathsService.NS, "mrow");
Element mover = doc.createElementNS(WebMathsService.NS, "mover");
Element mspace = doc.createElementNS(WebMathsService.NS, "mspace");
Element mspaceW = doc.createElementNS(WebMathsService.NS, "mspace");
boolean previousElement = false;
for (Node previous = menclose.getPreviousSibling();
previous != null;
previous = previous.getPreviousSibling()) {
if (previous.getNodeType() == Node.ELEMENT_NODE) {
previousElement = true;
break;
}
}
if (previousElement) {
mspaceW.setAttribute("width", "4px");
}
menclose.getParentNode().insertBefore(mrow, menclose);
menclose.getParentNode().removeChild(menclose);
mrow.appendChild(mspaceW);
mrow.appendChild(mover);
mover.appendChild(menclose);
mover.appendChild(mspace);
}
}
private static boolean isTextChild(Node parent, String text) {
NodeList list = parent.getChildNodes();
if (list.getLength() != 1) {
return false;
}
Node child = list.item(0);
if (child.getNodeType() != Node.TEXT_NODE) {
return false;
}
return child.getNodeValue().equals(text);
}
private static NodeList getIFrames(NodeList node_list) {
return node_list.extractAllNodesThatMatch(new TagNameFilter("iframe"), true);
}
private static NodeList getRemarks(NodeList node_list) {
return node_list.extractAllNodesThatMatch(new NodeClassFilter(RemarkNode.class), true);
}
private static NodeList getPhotosetPosts(NodeList node_list) {
return node_list.extractAllNodesThatMatch(
new HasAttributeFilter("class", "my_photoset_post"), true);
}
private static boolean handleURL(String address) {
Main.status(String.format("Processing page \"%s\".", address));
try {
NodeList posts = getPosts(address);
if (posts.toNodeArray().length == 0) {
return false;
}
for (Node post_node : posts.toNodeArray()) {
if (post_node instanceof TagNode) {
TagNode post = (TagNode) post_node;
Post new_post = new Post(Long.parseLong(post.getAttribute("id").substring(5)));
if (!Main.post_post_hash.containsKey(new_post)) {
NodeList photo_posts = getPhotoPosts(post.getChildren());
NodeList remarks = getRemarks(photo_posts);
for (Node node : remarks.toNodeArray()) {
Matcher matcher = lores.matcher(node.getText());
String media_url = "";
if (matcher.find()) {
media_url = matcher.group();
media_url = media_url.substring(17, media_url.length() - 1);
}
String thumb =
media_url.replace(
media_url.substring(media_url.lastIndexOf("_"), media_url.lastIndexOf(".")),
"_75sq");
URL thumb_url = new URL(thumb);
new_post.pictures.add(new Picture(new URL(media_url), thumb_url));
}
NodeList photoset_posts = getPhotosetPosts(post.getChildren());
NodeList iframes = getIFrames(photoset_posts);
for (Node node : iframes.toNodeArray()) {
if (node instanceof TagNode) {
String iframe_url = ((TagNode) node).getAttribute("src");
Parser parser2 = new Parser(iframe_url);
NodeList a_list = parser2.extractAllNodesThatMatch(new TagNameFilter("a"));
Node[] a_array = a_list.toNodeArray();
Node[] img_array =
a_list.extractAllNodesThatMatch(new TagNameFilter("img"), true).toNodeArray();
String media_url;
for (int i = 0; i < a_array.length; i++) {
media_url = ((TagNode) img_array[i]).getAttribute("src");
String thumb =
media_url.replace(
media_url.substring(
media_url.lastIndexOf("_"), media_url.lastIndexOf(".")),
"_75sq");
URL thumb_url = new URL(thumb);
new_post.pictures.add(new Picture(new URL(media_url), thumb_url));
}
}
}
Main.handlePost(new_post);
} else {
new_post = post_post_hash.get(new_post);
handleNonDownloadPost(new_post);
}
}
}
} catch (Exception ex) {
ex.printStackTrace();
Main.status("Error handling post.");
}
return true;
}