private void findRecommendations(
@NotNull Document doc, @NotNull BilingualQueryResultBuilder resultBuilder) {
// Determine all candidate nodes:
Elements alternativeNodes = doc.select("div.cc > p > *");
Language currentLanguage = null;
for (Element node : alternativeNodes) {
// If the next node is a flagicon, try to determine the language for the next entries from the
// class name
if (node.tagName().equals("span") && node.hasClass("flagicon")) {
Set<String> classNames = node.classNames();
classNames.remove("flagicon");
for (String className : classNames) {
Language candidate = Language.getExistingLanguageById(className);
if (candidate != null) {
currentLanguage = candidate;
break;
}
}
} else if (node.tagName().equals("a")) {
String recommendationText = node.text();
DictionaryObjectBuilder objectBuilder = ImmutableDictionaryObject.builder();
objectBuilder.setLanguage(currentLanguage).setGeneralForm(recommendationText);
resultBuilder.addSimilarRecommendation(objectBuilder.build());
}
}
}
public List<AreaVO> parseMessage(String text, int pid) {
Document doc = Jsoup.parse(text);
Element body = doc.body();
List<AreaVO> areas = new ArrayList<AreaVO>();
Elements divs = body.getElementsByClass("subarea");
if (divs.size() > 0) {
Element div = divs.get(0);
Elements childs = div.children();
String letter = "";
for (int i = 1; i < childs.size(); i++) {
Element child = childs.get(i);
if ("b".equals(child.tagName())) {
letter = child.text();
continue;
}
if ("a".equals(child.tagName())) {
AreaVO area = new AreaVO();
area.setLetter(letter);
area.setName(child.text());
area.setOrderIdx(index);
area.setPid(pid);
String href = child.attr("href");
String pinyin = href.substring(7, href.lastIndexOf("/"));
area.setPinyin(pinyin);
index++;
System.out.println(area.toString());
areas.add(area);
}
}
}
return areas;
}
public static void main(String[] args) throws IOException {
// Validate.isTrue(args.length == 1, "usage: supply url to fetch");
// String url = args[0];
// String url = "http://www.hao123.com";
String url = "http://www.iteye.com/login";
print("Fetching %s...", url);
Document doc = Jsoup.connect(url).get();
Elements links = doc.select("a[href]");
Elements media = doc.select("[src]");
Elements imports = doc.select("link[href]");
print("\nMedia: (%d)", media.size());
for (Element src : media) {
if (src.tagName().equals("img"))
print(
" * %s: <%s> %sx%s (%s)",
src.tagName(),
src.attr("abs:src"),
src.attr("width"),
src.attr("height"),
trim(src.attr("alt"), 20));
else print(" * %s: <%s>", src.tagName(), src.attr("abs:src"));
}
print("\nImports: (%d)", imports.size());
for (Element link : imports) {
print(" * %s <%s> (%s)", link.tagName(), link.attr("abs:href"), link.attr("rel"));
}
print("\nLinks: (%d)", links.size());
for (Element link : links) {
print(" * a: <%s> (%s)", link.attr("abs:href"), trim(link.text(), 35));
}
}
/** 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();
}
/**
* Reads an Item from a design and inserts it into the data source. Recursively handles any
* children of the item as well.
*
* @since 7.5.0
* @param node an element representing the item (tree node).
* @param selected A set accumulating selected items. If the item that is read is marked as
* selected, its item id should be added to this set.
* @param context the DesignContext instance used in parsing
* @return the item id of the new item
* @throws DesignException if the tag name of the {@code node} element is not {@code node}.
*/
@Override
protected String readItem(Element node, Set<String> selected, DesignContext context) {
if (!"node".equals(node.tagName())) {
throw new DesignException(
"Unrecognized child element in " + getClass().getSimpleName() + ": " + node.tagName());
}
String itemId = node.attr("text");
addItem(itemId);
if (node.hasAttr("icon")) {
Resource icon =
DesignAttributeHandler.readAttribute("icon", node.attributes(), Resource.class);
setItemIcon(itemId, icon);
}
if (node.hasAttr("selected")) {
selected.add(itemId);
}
for (Element child : node.children()) {
String childItemId = readItem(child, selected, context);
setParent(childItemId, itemId);
}
return itemId;
}
private static Node toNode(Element aInElement) {
int i = aInElement.siblingIndex();
Node lNode = aInElement.parent().childNode(i);
if (!lNode.nodeName().equals(aInElement.tagName())) {
throw new RuntimeException(lNode.nodeName() + " != " + aInElement.tagName());
}
return lNode;
}
@Test
public void parsesSimpleDocument() {
String html =
"<html><head><title>First!</title></head><body><p>First post! <img src=\"foo.png\" /></p></body></html>";
Document doc = Jsoup.parse(html);
// need a better way to verify these:
Element p = doc.child(1).child(0);
assertEquals("p", p.tagName());
Element img = p.child(0);
assertEquals("foo.png", img.attr("src"));
assertEquals("img", img.tagName());
}
@Test
public void testGetElementById() {
Document doc = Jsoup.parse(reference);
Element div = doc.getElementById("div1");
assertEquals("div1", div.id());
assertNull(doc.getElementById("none"));
Document doc2 =
Jsoup.parse("<div id=1><div id=2><p>Hello <span id=2>world!</span></p></div></div>");
Element div2 = doc2.getElementById("2");
assertEquals("div", div2.tagName()); // not the span
Element span = div2.child(0).getElementById("2"); // called from <p> context should be span
assertEquals("span", span.tagName());
}
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();
}
}
public void download(Connection aInConnection, Collection<Image> images) throws IOException {
aInConnection.url(url);
Document lDocument = aInConnection.get();
Element lMain = lDocument.getElementById("main");
Elements lContents = lMain.getElementsByClass("content");
if (lContents.size() == 1) {
StringBuilder sb = new StringBuilder();
Element lContent = lContents.first();
collectImages(lContent, images);
Elements lLightboxElements = lContent.getElementsByClass("lightbox");
for (Element lLightboxElement : lLightboxElements) {
Collection<Node> lImageNodes = extractImageNodes(lLightboxElement);
Element lParent = lLightboxElement.parent();
int i = lLightboxElement.siblingIndex();
lParent.insertChildren(i, lImageNodes);
lLightboxElement.remove();
}
Elements lChildElements = lContent.children();
for (Element lChildElement : lChildElements) {
if (lChildElement.hasClass("clear")) {
// no more post content
break;
}
if (title == null && lChildElement.tagName().equals("h1")) {
// the first h1 header is the title
title = lChildElement.html();
} else {
if (excerpt == null && lChildElement.tagName().equals("p")) {
excerpt = lChildElement.text();
}
String lStr = lChildElement.toString();
sb.append(lStr);
}
}
content = sb.toString();
Elements lDateElements = lContent.getElementsByClass("date");
String lHunDate = lDateElements.first().html();
date = new PostDate(lHunDate);
} else {
System.out.println("More than one content in main section of post page " + toString());
}
}
public static Pupil getSelectedPupil(Document doc) throws ParseException {
boolean found = false;
Pupil p, selectedP = null;
Elements pupilSelectors =
doc.getElementsByAttributeValue("id", "ctl00_topMenu_pupil_drdPupils");
for (Element pupilSelector : pupilSelectors) {
Elements pupils = pupilSelector.getAllElements();
for (Element pupil : pupils) {
if (pupil.tagName().equals("option")) {
String value = pupil.attr("value");
found = true;
if ((p = Pupil.getByFormId(value)) == null) {
p = new Pupil(pupil.text(), value);
long rowId = p.insert();
if (BuildConfig.DEBUG)
Log.d("GshisHTMLParser", TS.get() + " Pupil.insert() = " + rowId);
}
if (pupil.hasAttr("selected") && pupil.attr("selected").equals("selected")) {
selectedP = p;
}
}
}
}
if (!found) {
if (BuildConfig.DEBUG) Log.d("GshisParser", TS.get() + " Alternative fields found!");
Element userName = doc.getElementsByClass("user-name").first();
Element userId = doc.getElementsByAttributeValue("id", "ctl00_topMenu_tbUserId").first();
String name = userName.text();
String id = userId.attr("value");
if (BuildConfig.DEBUG) Log.d("GshisParser", TS.get() + " name=" + name + " id=" + id);
if ((p = Pupil.getByFormId(id)) == null) {
p = new Pupil(name, id);
long rowId = p.insert();
if (BuildConfig.DEBUG) Log.d("GshisParser", TS.get() + " Pupil.insert() = " + rowId);
}
selectedP = p;
}
if (selectedP == null) throw new ParseException("Pupils not found", 0);
return selectedP;
}
/**
* @param theLinks
* @throws IOException
*/
public void getSecondLinks(ArrayList<String> theLinks) throws IOException {
String temp = null;
Document doc = null;
boolean flag;
for (String sLink : theLinks) {
if (sLink.endsWith(".asx") == true || sLink.endsWith(".swf") == true) {
stationLinks2.add(sLink);
print("Written to file: %s", sLink);
} else {
doc = parseUrl(sLink, 0);
if (doc != null) {
Elements media = doc.select("[src]");
print("Fetching %s --> ", sLink);
flag = false;
for (Element src : media) {
if (src.tagName().equals("embed") == true) {
flag = true;
temp = src.attr("abs:src");
stationLinks2.add(temp);
break; // link found, load next url
}
} // end nested for
if (flag == false) { // the code has no embed tag
stationLinks2.add(sLink);
}
}
}
} // end outer for
writeLinksToFile(links2FileName, stationLinks2);
print("Written %s to file, second links.", stationLinks2.size());
} // end method
private static void accumulateParents(Element el, Elements parents) {
Element parent = el.parent();
if (parent != null && !parent.tagName().equals("#root")) {
parents.add(parent);
accumulateParents(parent, parents);
}
}
private void recurse(Element element) {
ElementAction action = classifyElement(element);
if (action == ElementAction.Whitespace || action == ElementAction.Sentence) {
appendSpace();
}
for (Node childNode : element.childNodes()) {
// n.b., cdata not possible if we are coming from TagSoup. If we also handle
// real xhtml by directly parsing it, then we have another story on our hands.
// though we could use canonical XML to get rid of them.
if (childNode instanceof TextNode && action != ElementAction.Banned) {
TextNode textContent = (TextNode) childNode;
String textString = textContent.text();
append(textContent, textString);
} else if (childNode instanceof Element) {
recurse((Element) childNode);
}
}
if (action == ElementAction.Whitespace) {
appendSpace();
} else if (action == ElementAction.Sentence) {
appendPeriod();
} else if (action == ElementAction.Mark) {
Mark mark = new Mark();
mark.setOffset(pcDataOffset);
mark.setTag(element.tagName());
}
}
public void head(Node source, int depth) {
if (source instanceof Element) {
Element sourceEl = (Element) source;
if (whitelist.isSafeTag(sourceEl.tagName())) { // safe, clone and copy safe attrs
ElementMeta meta = createSafeElement(sourceEl);
Element destChild = meta.el;
destination.appendChild(destChild);
numDiscarded += meta.numAttribsDiscarded;
destination = destChild;
} else if (source
!= root) { // not a safe tag, so don't add. don't count root against discarded.
numDiscarded++;
}
} else if (source instanceof TextNode) {
TextNode sourceText = (TextNode) source;
TextNode destText = new TextNode(sourceText.getWholeText(), source.baseUri());
destination.appendChild(destText);
} else if (source instanceof DataNode && whitelist.isSafeTag(source.parent().nodeName())) {
DataNode sourceData = (DataNode) source;
DataNode destData = new DataNode(sourceData.getWholeData(), source.baseUri());
destination.appendChild(destData);
} else { // else, we don't care about comments, xml proc instructions, etc
numDiscarded++;
}
}
private boolean isBlock(Node n) {
boolean block = false;
if (n != null && n instanceof Element) {
Element el = (Element) n;
block = el.isBlock() || el.tagName().equals("br");
}
return block;
}
private static void parseTitle(Topic.Builder topicBuilder, Element ele) {
ele = ele.select(".item_title > a").get(0);
Preconditions.checkState(ele.tagName().equals("a"));
String url = ele.attr("href");
topicBuilder.setId(Topic.getIdFromUrl(url));
topicBuilder.setTitle(ele.html());
}
@Test
public void parsesUnterminatedAttribute() {
String h1 = "<p id=\"foo";
Document doc = Jsoup.parse(h1);
Element p = doc.getElementById("foo");
assertNotNull(p);
assertEquals("p", p.tagName());
}
static void parseMember(Topic.Builder builder, Element ele) {
final Member.Builder memberBuilder = new Member.Builder();
// get member url
ele = ele.child(0);
Preconditions.checkState(ele.tagName().equals("a"));
final String url = ele.attr("href");
memberBuilder.setUsername(Member.getNameFromUrl(url));
// get member avatar
final Avatar.Builder avatarBuilder = new Avatar.Builder();
ele = ele.child(0);
Preconditions.checkState(ele.tagName().equals("img"));
avatarBuilder.setUrl(ele.attr("src"));
memberBuilder.setAvatar(avatarBuilder.createAvatar());
builder.setMember(memberBuilder.createMember());
}
public static Week getSelectedWeek(Document doc, Schedule s) throws ParseException {
boolean found = false;
Week selectedW = null;
SimpleDateFormat f = new SimpleDateFormat("yyyy dd.MM", Locale.ENGLISH);
f.setTimeZone(TimeZone.getTimeZone("Europe/Moscow"));
Elements weekSelectors = doc.getElementsByAttributeValue("id", "ctl00_body_week_drdWeeks");
for (Element weekSelector : weekSelectors) {
Elements weeks = weekSelector.getAllElements();
for (Element week : weeks) {
if (week.tagName().equals("option")) {
String value = week.text();
Week w;
found = true;
if ((w = s.getWeek(week.attr("value"))) == null) {
w = new Week();
String wBegin = value.substring(0, value.indexOf("-") - 1);
String wMonth = wBegin.substring(wBegin.indexOf(".") + 1, wBegin.length());
String year;
if (Integer.parseInt(wMonth) > 7) {
year = s.getFormText().substring(0, s.getFormText().indexOf("-") - 1);
} else {
year =
s.getFormText()
.substring(s.getFormText().indexOf("-") + 2, s.getFormText().length());
}
w.setStart(f.parse(year + " " + wBegin));
w.setFormText(week.text());
w.setFormId(week.attr("value"));
s.addWeek(w);
}
if (week.hasAttr("selected") && week.attr("selected").equals("selected")) {
selectedW = w;
long u = w.setLoaded().update();
if (BuildConfig.DEBUG) Log.d("GshisHTMLParser", TS.get() + " Week.update() = " + u);
}
}
}
}
if (!found) throw new ParseException("Weeks not found", 0);
return selectedW;
}
// recursively processes the element to replace <br>'s with \n
private void fixLineBreaks(Element el) {
for (final Element e : el.children()) {
if (e.tagName().equals("br")) {
e.before("\n");
e.remove();
} else {
fixLineBreaks(e);
}
}
}
/**
* Complement of HtmlTemplateCompiler#lexicalClimb(). This method pops off the stack of lexical
* scopes when we're done processing a sitebricks widget.
*/
private void lexicalDescend(PageCompilingContext pc, Element element, boolean shouldPopScope) {
// pop form
if ("form".equals(element.tagName())) pc.form = null;
// pop compiler if the scope ends
if (shouldPopScope) {
pc.lexicalScopes.pop();
}
}
/**
* Determines whether an element is not focusable using keys.
*
* <p>TODO(jharty): We need to determine which elements are generally focusable (where focus is
* disabled by setting tabindex to a negative number) and which ones generally are NOT focusable,
* but may become focusable by setting a positive tabindex. TODO(jharty): This could move to
* OnClickIsFocusable?
*
* @param element the element to test
* @return true when the element is not focusable, else false.
*/
static boolean notFocusable(Element element) {
List<String> focusable =
Arrays.asList(new String[] {"a", "button", "input", "select", "textarea"});
if (focusable.contains(element.tagName())) {
if (element.hasAttr(TAB_INDEX) && -1 >= Integer.parseInt(element.attr(TAB_INDEX))) {
return true;
}
}
return false;
}
@Test
public void parsesRoughAttributes() {
String html =
"<html><head><title>First!</title></head><body><p class=\"foo > bar\">First post! <img src=\"foo.png\" /></p></body></html>";
Document doc = Jsoup.parse(html);
// need a better way to verify these:
Element p = doc.child(1).child(0);
assertEquals("p", p.tagName());
assertEquals("foo > bar", p.attr("class"));
}
public Map doProcess() {
if (!isInit) {
init();
}
Iterator<String> it = map.keySet().iterator();
while (it.hasNext()) {
String key = it.next();
Element val = map.get(key);
// 如果是ul或者table，用相应的组件渲染
if (("ul".equals(val.tagName().toLowerCase()))
|| ("table".equals(val.tagName().toLowerCase()))) {
System.out.println("------------------列表开始-----------------------------");
Elements links = val.select("a");
for (Element ele : links) {
System.out.println("a:" + ele.attr("abs:href") + ",文本:" + ele.text());
}
System.out.println("------------------列表结束-----------------------------");
} else {
System.out.println("------------------非列表-----------------------------");
if ("a".equals(val.tagName().toLowerCase())) {
System.out.println("a:" + val.attr("abs:href") + ",文本:" + val.text());
} else {
if ("span".equals(val.tagName().toLowerCase())) {
System.out.println("父容器:" + val.parent().tagName());
}
System.out.println("标签:" + val.tagName() + ",html:" + val.html());
}
}
}
System.out.println("---------------top menu---------------------");
it = topMenumap.keySet().iterator();
while (it.hasNext()) {
String key = (String) it.next();
Element val = topMenumap.get(key);
Elements links = val.select("a");
for (Element ele : links) {
System.out.println("a:" + ele.attr("abs:href") + ",文本:" + ele.text());
}
}
return map;
}
private void getChildElement(Element parentElement, Integer level) {
parentElement.html(deleteComent(parentElement.html()));
// System.out.println("key:"+(level+","+parentElement.hashCode())+",value:"+parentElement.html());
if (parentElement.children().size() > 0) {
level += 1;
for (int i = 0; i < parentElement.children().size(); i++) {
if (("ul".equals(parentElement.tagName().toLowerCase()))
|| ("table".equals(parentElement.tagName().toLowerCase()))) { // 整体标签
String html = parentElement.html().replaceAll(" ", "").replaceAll(" ", ""); // 去中英文空格
if (html.contains("首页") || parentElement.id().contains("nav")) {
//
// System.out.println("----------------------首页Start-----------------------------");
Elements links = parentElement.select("a");
for (Element ele : links) {
if (topMenumap.get(level + "," + ele.hashCode()) == null) {
topMenumap.put(level + "," + ele.hashCode(), ele);
// System.out.println(level + "," + ele.hashCode() + ",---------------" +
// ele.html());
// System.out.println("a:" +
// ele.attr("abs:href") + ",文本:" + ele.text());
}
}
// System.out.println("----------------------首页End-----------------------------");
} else {
map.put(level + "," + parentElement.hashCode(), parentElement);
}
} else {
getChildElement(parentElement.child(i), level);
}
}
} else {
if ("script".equals(parentElement.tagName().toLowerCase())) {
return;
}
if (StringUtils.isNotEmpty(parentElement.html())) {
level += 1;
map.put(level + "," + parentElement.hashCode(), parentElement);
}
}
}
public static String printNode(Element root, int indentation) {
StringBuilder sb = new StringBuilder();
for (int i = 0; i < indentation; i++) {
sb.append(' ');
}
sb.append(root.tagName());
sb.append(":");
sb.append(root.ownText());
sb.append("\n");
for (Element el : root.children()) {
sb.append(printNode(el, indentation + 1));
sb.append("\n");
}
return sb.toString();
}
void appendTextSkipHidden(Element e, StringBuilder accum) {
for (Node child : e.childNodes()) {
if (unlikely(child)) continue;
if (child instanceof TextNode) {
TextNode textNode = (TextNode) child;
String txt = textNode.text();
accum.append(txt);
} else if (child instanceof Element) {
Element element = (Element) child;
if (accum.length() > 0 && element.isBlock() && !lastCharIsWhitespace(accum))
accum.append(" ");
else if (element.tagName().equals("br")) accum.append(" ");
appendTextSkipHidden(element, accum);
}
}
}
private ElementMeta createSafeElement(Element sourceEl) {
String sourceTag = sourceEl.tagName();
Attributes destAttrs = new Attributes();
Element dest = new Element(Tag.valueOf(sourceTag), sourceEl.baseUri(), destAttrs);
int numDiscarded = 0;
Attributes sourceAttrs = sourceEl.attributes();
for (Attribute sourceAttr : sourceAttrs) {
if (whitelist.isSafeAttribute(sourceTag, sourceEl, sourceAttr)) destAttrs.put(sourceAttr);
else numDiscarded++;
}
Attributes enforcedAttrs = whitelist.getEnforcedAttributes(sourceTag);
destAttrs.addAll(enforcedAttrs);
return new ElementMeta(dest, numDiscarded);
}
public static GradeSemester getActiveGradeSemester(Document doc, Schedule sch)
throws ParseException {
boolean found = false;
GradeSemester selG = null;
SimpleDateFormat fmt = new SimpleDateFormat("dd.MM.yyyy", Locale.ENGLISH);
fmt.setTimeZone(TimeZone.getTimeZone("Europe/Moscow"));
Elements semesterSelectors = doc.getElementsByAttributeValue("id", "ctl00_body_drdTerms");
for (Element semesterSelector : semesterSelectors) {
Elements semesters = semesterSelector.getAllElements();
for (Element semester : semesters) {
if (semester.tagName().equals("option")) {
String value = semester.text();
GradeSemester sem;
found = true;
if ((sem = sch.getSemester(semester.attr("value"))) == null) {
sem = new GradeSemester();
sem.setStart(fmt.parse(value.substring(12, value.indexOf("-") - 1)));
sem.setStop(fmt.parse(value.substring(value.indexOf("-") + 2, value.length() - 2)));
sem.setFormText(semester.text());
sem.setFormId(semester.attr("value"));
sch.addSemester(sem);
}
if (semester.hasAttr("selected") && semester.attr("selected").equals("selected")) {
long u = sem.setLoaded().update();
selG = sem;
if (BuildConfig.DEBUG) Log.d("GshisHTMLParser", TS.get() + " Semester.update() = " + u);
}
}
}
}
if (!found) throw new ParseException("Semesters not found", 0);
return selG;
}
