Java Code Examples for org.jsoup.nodes.Element#children()

private List<ArtifactBean> populate(Elements elements) {
	List<ArtifactBean> artifactList = Lists.newArrayList();
	for (Element dependency : elements) {
		
		ArtifactBean artifactBean = new ArtifactBean();
		for (Element child : dependency.children()) {
			if (child.nodeName().compareTo("groupid") == 0) {
				artifactBean.setGroupId(child.text());
			} else if (child.nodeName().compareTo("artifactid") == 0) {
				artifactBean.setArtifactId(child.text());
			} else if (child.nodeName().compareTo("type") == 0) {
				artifactBean.setType(child.text());
			}
		}
		
		if (StringUtils.hasText(artifactBean.getGroupId()) && StringUtils.hasText(artifactBean.getArtifactId())) {
			artifactBean.setId(artifactBean.getGroupId() + ":" + artifactBean.getArtifactId());
			artifactList.add(artifactBean);
		}
	}
	
	Collections.sort(artifactList);
	
	return artifactList;
}
 

private void parseTopic(Element classElement, Topic typeTopic) throws TopicMapException {
    System.out.println(classElement.text());
    
    Topic t = getOrCreateTopic(tm, null, classElement.text());
    
    if(typeTopic == null) typeTopic = wandoraClass;
    t.addType(typeTopic);
    
    // See if the next element is a list (of instances)
    Element listWrapper = classElement.nextElementSibling();
    if(listWrapper != null && !listWrapper.children().isEmpty()) {
        for(Element listCandidate: listWrapper.children()) {
            if(listCandidate.tagName().equals("ul")) {
                parseList(listCandidate, t);
            }
        }
    }
}
 

protected Node parseLi(Element liElement) {
    Elements children = liElement.children();
    if (children.size() == 1 && children.get(0).nodeName().equals("a")) {
        return parseA(children.get(0));
    }
    Node root = new Node();
    for (Element child : liElement.children()) {
        if (child.nodeName().equals("span")) {
            root = parseSpan(child);
        }
        if (child.nodeName().equals("ul")) {
            root.addAll(parseUL(child));
        }
    }
    return root;
}
 

private void parseTopic(Element classElement, Topic classTopic) throws TopicMapException {
    String name = classElement.text().trim();
    
    if(name.length() == 0) return;
    
    Topic t = getOrCreateTopic(tm, null , name);
    
    if(classTopic == null) classTopic = wandoraClass;
    makeSubclassOf(tm, t, classTopic);
    
    // See if the next element is a list (of instances)
    Element listWrapper = classElement.nextElementSibling();
    if(listWrapper != null && !listWrapper.children().isEmpty()) {
        for(Element listCandidate: listWrapper.children()){
            if(listCandidate.tagName().equals("ul"))
                parseList(listCandidate, t);
        }
    }
}
 

private static void testHtmlParser(String url) throws Exception {
    Document doc = Jsoup.connect(url).userAgent(USER_AGENT).cookie("auth", "token")
            .timeout(30000).get();
    Charset charset = doc.charset();
    System.out.println("charset = " + charset);
    System.out.println("location = " + doc.location());
    System.out.println("nodeName = " + doc.nodeName());
    Document.OutputSettings outputSettings = doc.outputSettings();
    System.out.println("charset = " + outputSettings.charset());
    System.out.println("indentAmount = " + outputSettings.indentAmount());
    System.out.println("syntax = " + outputSettings.syntax());
    System.out.println("escapeMode = " + outputSettings.escapeMode());
    System.out.println("prettyPrint = " + outputSettings.prettyPrint());
    System.out.println("outline = " + outputSettings.outline());

    System.out.println("title = " + doc.title());
    System.out.println("baseUri = " + doc.baseUri());

    Element head = doc.head();
    Elements children = head.children();
    for(Element child: children) {
        System.out.print(child.tag().getName() + " : ");
        System.out.println(child);
    }
    printElements(doc.body().children());
}
 

@Override
public boolean matches(Element root, Element element) {
	final Element p = element.parent();
	if (p==null || p instanceof Document) return false;
	
	int pos = 0;
      	Elements family = p.children();
      	for (int i = 0; i < family.size(); i++) {
      		if (family.get(i).tag().equals(element.tag())) pos++;
      	}
      	return pos == 1;
}
 

public static Prefix extractPrefix(Element element){
    Prefix prefix = new Prefix();
    List<Element> tds = element.children();
    if(tds==null || tds.size()!=3){
        return prefix;
    }
    String p = tds.get(0).text().trim();
    if(!p.endsWith("-")){
        return prefix;
    }
    String des = tds.get(1).text();
    return new Prefix(p, des);
}
 

private static String printNode(Element root, int indentation) {
    StringBuilder sb = new StringBuilder(indentation);
    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();
}
 

public Map<String, String> parse(Document doc) {
	Elements elements = doc.select("table[id=conf]").select("tbody").select("tr");
	Iterator<Element> iter = elements.iterator();
	Map<String, String> configs = new HashMap<String, String>();
	while(iter.hasNext()) {
		Element element = iter.next();
		Elements tds = element.children();
		String key = tds.get(0).text();
		String value = tds.get(1).text();
		configs.put(key, value);
	}
	return configs;
}
 

@Test
public void examplesTraversing() {
    Elements sections = doc.select("section");

    Element firstSection = sections.first();
    Element lastSection = sections.last();
    Element secondSection = sections.get(2);
    Elements allParents = firstSection.parents();
    Element parent = firstSection.parent();
    Elements children = firstSection.children();
    Elements siblings = firstSection.siblingElements();

    sections.forEach(el -> System.out.println("section: " + el));
}
 

private static String printNode(Element root, int indentation) {
    StringBuilder sb = new StringBuilder(indentation);
    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();
}
 

private static Element findElement(Element e, ElementPredicate f) {
    if (f.check(e)) {
        return e;
    }
    for (Element c : e.children()) {
        Element elem = findElement(c, f);
        if (elem != null) {
            return elem;
        }
    }
    return null;
}
 

private void parseList(Element list, Topic typeTopic) throws TopicMapException {
    Elements listElements = list.children();
    for(Element outerElement: listElements){
        if(outerElement.children().isEmpty()){
            parseTopic(outerElement, typeTopic);
        }
    }
}
 

private void parseList(Element list, Topic classTopic) throws TopicMapException {
    
    Elements listElements = list.children();
    
    for(Element outerElement: listElements){
        if(outerElement.children().isEmpty()){
            parseTopic(outerElement, classTopic);
        }
    }
    
}
 

private static RDisplay htmlDisplay(Element body, String imageWidth) {
  String div = "";
  for (Element element : body.children()) {
    String eHtml = element.html();
    String eOuterHtml = element.outerHtml();

    eOuterHtml = eOuterHtml.replace("“%html " , "").replace("”", "");

    Matcher matcher = pattern.matcher(eHtml);
    if (matcher.matches()) {
      eOuterHtml = eOuterHtml.replace(matcher.group(), "");
    }

    div = div + eOuterHtml;
  }

  String content =  div
    .replaceAll("src=\"//", "src=\"http://")
    .replaceAll("href=\"//", "href=\"http://");

  body.html(content);

  for (Element image : body.getElementsByTag("img")) {
    image.attr("width", imageWidth);
  }

  return new RDisplay(body.html(), Type.HTML, Code.SUCCESS);
}
 

private void parseMicroformats(Element elem, URI baseUri, JsonList items) {
    if (hasRootClass(elem)) {
        JsonDict itemDict = parseMicroformat(elem, baseUri);
        items.add(itemDict);
    }
    else {
        for (Element child : elem.children()) {
            parseMicroformats(child, baseUri, items);
        }
    }
}
 

public Content toContent(@Nonnull String url) {
    Content result = new Content();

    result.setSite(Site.HENTAI2READ);
    if (url.isEmpty()) return result.setStatus(StatusContent.IGNORED);

    result.setUrl(url.replace(Site.HENTAI2READ.getUrl(), ""));
    result.setCoverImageUrl(coverUrl);
    if (!title.isEmpty()) {
        String titleStr = title.get(title.size() - 1).text();
        result.setTitle(!titleStr.isEmpty() ? Helper.removeNonPrintableChars(titleStr) : "");
    } else result.setTitle("<no title>");
    result.setUniqueSiteId(uniqueId);

    AttributeMap attributes = new AttributeMap();
    String currentProperty = "";
    for (Element e : properties) {
        for (Element child : e.children()) {
            if (child.nodeName().equals("b"))
                currentProperty = child.text().toLowerCase().trim();
            else if (child.nodeName().equals("a")) {
                switch (currentProperty) {
                    case "page":
                        String qtyPages = child.text().substring(0, child.text().indexOf(" page"));
                        result.setQtyPages(Integer.parseInt(qtyPages));
                        break;
                    case "parody":
                        ParseHelper.parseAttribute(attributes, AttributeType.SERIE, child, true, Site.HENTAI2READ);
                        break;
                    case "artist":
                        ParseHelper.parseAttribute(attributes, AttributeType.ARTIST, child, true, Site.HENTAI2READ);
                        break;
                    case "language":
                        ParseHelper.parseAttribute(attributes, AttributeType.LANGUAGE, child, true, Site.HENTAI2READ);
                        break;
                    case "character":
                        ParseHelper.parseAttribute(attributes, AttributeType.CHARACTER, child, true, Site.HENTAI2READ);
                        break;
                    case "content":
                    case "category":
                        ParseHelper.parseAttribute(attributes, AttributeType.TAG, child, true, Site.HENTAI2READ);
                        break;
                    default:
                        // Other cases aren't interesting
                }
            }
        }
    }
    result.addAttributes(attributes);

    return result;
}
 

private String parseImpliedName(Element elem) {
    if (("img".equals(elem.tagName()) || ("area".equals(elem.tagName())) && elem.hasAttr("alt"))) {
        return elem.attr("alt");
    }
    if ("abbr".equals(elem.tagName()) && elem.hasAttr("title")) {
        return elem.attr("title");
    }

    Elements children = elem.children();
    if (children.size() == 1) {
        Element child = children.first();
        // else if .h-x>img:only-child[alt]:not[.h-*] then use that img alt for name
        // else if .h-x>area:only-child[alt]:not[.h-*] then use that area alt for name
        if (!hasRootClass(child)
                && ("img".equals(child.tagName()) || "area".equals(child.tagName()))
                && child.hasAttr("alt")) {
            return child.attr("alt");
        }
        // else if .h-x>abbr:only-child[title] then use that abbr title for name
        if ("abbr".equals(child.tagName()) && child.hasAttr("title")) {
            return child.attr("title");
        }

        Elements grandChildren = child.children();
        if (grandChildren.size() == 1) {
            Element grandChild = grandChildren.first();
            // else if .h-x>:only-child>img:only-child[alt]:not[.h-*] then use that img alt for name
            // else if .h-x>:only-child>area:only-child[alt]:not[.h-*] then use that area alt for name
            if (!hasRootClass(grandChild)
                    && ("img".equals(grandChild.tagName()) || "area".equals(grandChild.tagName()))
                    && grandChild.hasAttr("alt")) {
                return grandChild.attr("alt");
            }
            // else if .h-x>:only-child>abbr:only-child[title] use that abbr title for name
            if ("abbr".equals(grandChild.tagName()) && grandChild.hasAttr("c")) {
                return grandChild.attr("title");
            }
        }
    }

    // else use the textContent of the .h-x for name
    // drop leading & trailing white-space from name, including nbsp
    return elem.text().trim();
}
 

private void parseText(Elements elements, int startId) {
	Log.d(TAG, "Start ID:" + startId);
	boolean isSkip = true;
	if (startId == 0)
		isSkip = false;

	for (int idx = 0; idx < elements.size(); idx++) {
		Element el = elements.get(idx);
		if (el.hasAttr("id") && isSkip) {
			try {
				int id = Integer.parseInt(el.attr("id"));
				if (id >= startId)
					isSkip = false;
			} catch (Exception ex) {
				Log.e(TAG, ex.getMessage());
			}
		}
		if (isSkip)
			continue;
		if (el.parent().hasClass("editsection"))
			continue;
		if (isWhiteSpace(el.tagName())) {
			SpeakValue s = new SpeakValue();
			s.Val = SILENCE;
			s.ID = null;
			queue.add(s);
		}

		SpeakValue val = new SpeakValue();
		// check if have children element for formatting
		boolean hasFormattingChild = false;
		for (Element child : el.children()) {
			if (FORMATTING_ELEMENTS.contains(child.tagName().toLowerCase())) {
				hasFormattingChild = true;
				break;
			}
		}

		if (hasFormattingChild) {
			Log.d(TAG, "Got formatting text: " + el.html());
			val.Val = el.text();
			Log.d(TAG, "Use text: " + el.text());
			removeAllChildren(el, elements);
			idx--;
		} else {
			val.Val = el.ownText();
		}

		if (el.hasAttr("id"))
			val.ID = el.attr("id");
		else
			val.ID = null;

		queue.add(val);
	}
}
 

/**
 * 直接子节点
 *
 * @param e
 * @return
 */
public Elements child(Element e) {
    return e.children();
}