Java Code Examples for org.jsoup.nodes.Document#head()

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());
}
 

@Test
public void headHasMetaTags() throws Exception {
    initUI(testUI, createVaadinRequest());

    Document page = pageBuilder.getBootstrapPage(new BootstrapContext(
            request, null, session, testUI, this::contextRootRelativePath));

    Element head = page.head();
    Elements metas = head.getElementsByTag("meta");

    Assert.assertEquals(3, metas.size());
    Element meta = metas.get(0);
    assertEquals("Content-Type", meta.attr("http-equiv"));
    assertEquals("text/html; charset=utf-8", meta.attr("content"));

    meta = metas.get(1);
    assertEquals("X-UA-Compatible", meta.attr("http-equiv"));
    assertEquals("IE=edge", meta.attr("content"));

    meta = metas.get(2);
    assertEquals(BootstrapHandler.VIEWPORT, meta.attr("name"));
    assertEquals(Viewport.DEFAULT,
            meta.attr(BootstrapHandler.CONTENT_ATTRIBUTE));
}
 

private void bootstrapPage_productionModeTest(boolean productionMode) {
    mocks.setProductionMode(productionMode);
    TestUI anotherUI = new TestUI();
    initUI(testUI);

    anotherUI.getInternals().setSession(session);
    VaadinRequest vaadinRequest = createVaadinRequest();
    anotherUI.doInit(vaadinRequest, 0);
    anotherUI.getRouter().initializeUI(anotherUI, request);
    BootstrapContext bootstrapContext = new BootstrapContext(vaadinRequest,
            null, session, anotherUI, this::contextRootRelativePath);
    anotherUI.getInternals()
            .setContextRoot(contextRootRelativePath(request));

    Document page = pageBuilder.getBootstrapPage(bootstrapContext);

    Element head = page.head();
    Assert.assertTrue(
            head.outerHtml().contains("mode = " + productionMode));
}
 

@Test public void createsDocumentStructure() {
    String html = "<meta name=keywords /><link rel=stylesheet /><title>jsoup</title><p>Hello world</p>";
    Document doc = Jsoup.parse(html);
    Element head = doc.head();
    Element body = doc.body();

    assertEquals(1, doc.children().size()); // root node: contains html node
    assertEquals(2, doc.child(0).children().size()); // html node: head and body
    assertEquals(3, head.children().size());
    assertEquals(1, body.children().size());

    assertEquals("keywords", head.getElementsByTag("meta").get(0).attr("name"));
    assertEquals(0, body.getElementsByTag("meta").size());
    assertEquals("jsoup", doc.title());
    assertEquals("Hello world", body.text());
    assertEquals("Hello world", body.children().get(0).text());
}
 

@Test public void createsDocumentStructure() {
    String html = "<meta name=keywords /><link rel=stylesheet /><title>jsoup</title><p>Hello world</p>";
    Document doc = Jsoup.parse(html);
    Element head = doc.head();
    Element body = doc.body();

    assertEquals(1, doc.children().size()); // root node: contains html node
    assertEquals(2, doc.child(0).children().size()); // html node: head and body
    assertEquals(3, head.children().size());
    assertEquals(1, body.children().size());

    assertEquals("keywords", head.getElementsByTag("meta").get(0).attr("name"));
    assertEquals(0, body.getElementsByTag("meta").size());
    assertEquals("jsoup", doc.title());
    assertEquals("Hello world", body.text());
    assertEquals("Hello world", body.children().get(0).text());
}
 

private String getArticleTitle(Document sourceDoc) {
    String title = "";
    if(sourceDoc.head() != null &&
            StringUtils.isNotEmpty(sourceDoc.head().attr(WxCrawlerConstant.BackupArticle.ARTICLE_TITLE))) {
        title = sourceDoc.head().attr(WxCrawlerConstant.BackupArticle.ARTICLE_TITLE);
    } else if (sourceDoc.select(WxCrawlerConstant.HTMLElementSelector.TITLE).first() != null) {
        title = sourceDoc.select(WxCrawlerConstant.HTMLElementSelector.TITLE).first().text();
    } else {
        title = sourceDoc.title();
    }
    return title;
}
 

public static ArticleTransferVO convert2ArticleTransferVO(ArticleTransferVO articleTransferVO, Document doc){
    Element header = doc.head();
    articleTransferVO.setPublishDate(header.attr(WxCrawlerConstant.BackupArticle.PUBLISH_DATE));
    articleTransferVO.setAuthor(header.attr(WxCrawlerConstant.BackupArticle.AUTHOR));
    articleTransferVO.setAccountId(header.attr(WxCrawlerConstant.BackupArticle.ACCOUNT_ID));
    articleTransferVO.setAccountName(header.attr(WxCrawlerConstant.BackupArticle.ACCOUNT_NAME));
    articleTransferVO.setDigest(header.attr(WxCrawlerConstant.BackupArticle.DIGEST));
    articleTransferVO.setOssCover(header.attr(WxCrawlerConstant.BackupArticle.COVER));
    articleTransferVO.setArticleType(header.attr(WxCrawlerConstant.BackupArticle.ARTICLE_TYPE));
    articleTransferVO.setTitle(header.attr(WxCrawlerConstant.BackupArticle.ARTICLE_TITLE));
    return articleTransferVO;
}
 

public Map<String, Object> doParse(Object... arg) {
    mArguments = arg;
    if (mValues == null) {
        mValues = new HashMap<>();
    }
    Document doc = Jsoup.parse(mHtml);
    mHeader = doc.head();
    mBody = doc.body();
    parse(doc, mHeader, mBody, mValues);
    return mValues;
}
 

@Test
public void bootstrapPage_configJsonPatternIsReplacedBeforeInitialUidl() {
    TestUI anotherUI = new TestUI();
    initUI(testUI);

    SystemMessages messages = Mockito.mock(SystemMessages.class);
    service.setSystemMessagesProvider(info -> messages);
    Mockito.when(messages.isSessionExpiredNotificationEnabled())
            .thenReturn(true);
    Mockito.when(session.getSession())
            .thenReturn(Mockito.mock(WrappedSession.class));

    String url = "http://{{CONFIG_JSON}}/file";
    Mockito.when(messages.getSessionExpiredURL()).thenReturn(url);

    anotherUI.getInternals().setSession(session);
    VaadinRequest vaadinRequest = createVaadinRequest();
    anotherUI.doInit(vaadinRequest, 0);
    anotherUI.getRouter().initializeUI(anotherUI, request);
    BootstrapContext bootstrapContext = new BootstrapContext(vaadinRequest,
            null, session, anotherUI, this::contextRootRelativePath);
    anotherUI.getInternals()
            .setContextRoot(contextRootRelativePath(request));

    Document page = pageBuilder.getBootstrapPage(bootstrapContext);
    Element head = page.head();
    Assert.assertTrue(head.outerHtml().contains(url));
}
 

@Test
public void addMultiMetaTagViaMetaAnnotation_MetaSizeCorrect_ContentCorrect()
        throws InvalidRouteConfigurationException {
    initUI(testUI, createVaadinRequest(),
            Collections.singleton(MetaAnnotations.class));

    Document page = pageBuilder.getBootstrapPage(new BootstrapContext(
            request, null, session, testUI, this::contextRootRelativePath));

    Element head = page.head();
    Elements metas = head.getElementsByTag("meta");

    Assert.assertEquals(5, metas.size());
    Element meta = metas.get(0);
    assertEquals("Content-Type", meta.attr("http-equiv"));
    assertEquals("text/html; charset=utf-8", meta.attr("content"));

    meta = metas.get(1);
    assertEquals("X-UA-Compatible", meta.attr("http-equiv"));
    assertEquals("IE=edge", meta.attr("content"));

    meta = metas.get(2);
    assertEquals(BootstrapHandler.VIEWPORT, meta.attr("name"));
    assertEquals(Viewport.DEFAULT,
            meta.attr(BootstrapHandler.CONTENT_ATTRIBUTE));

    meta = metas.get(3);
    assertEquals("apple-mobile-web-app-status-bar-style",
            meta.attr("name"));
    assertEquals("black", meta.attr(BootstrapHandler.CONTENT_ATTRIBUTE));

    meta = metas.get(4);
    assertEquals("apple-mobile-web-app-capable", meta.attr("name"));
    assertEquals("yes", meta.attr(BootstrapHandler.CONTENT_ATTRIBUTE));
}
 

@Test
public void defaultViewport() {
    initUI(testUI);
    Document page = pageBuilder.getBootstrapPage(context);
    Element head = page.head();
    Elements viewports = head.getElementsByAttributeValue("name",
            BootstrapHandler.VIEWPORT);
    Assert.assertEquals(1, viewports.size());
    Element viewport = viewports.get(0);
    Assert.assertEquals(Viewport.DEFAULT,
            viewport.attr(BootstrapHandler.CONTENT_ATTRIBUTE));

}
 

@Test
public void viewportAnnotationOverridesDefault() throws Exception {
    initUI(testUI, createVaadinRequest(),
            Collections.singleton(RouteWithViewport.class));
    Document page = pageBuilder.getBootstrapPage(context);
    Element head = page.head();
    Elements viewports = head.getElementsByAttributeValue("name",
            BootstrapHandler.VIEWPORT);
    Assert.assertEquals(1, viewports.size());
    Element viewport = viewports.get(0);
    Assert.assertEquals("viewport-annotation-value",
            viewport.attr(BootstrapHandler.CONTENT_ATTRIBUTE));

}
 

@Override
protected void doProcess(final JCas jCas) throws AnalysisEngineProcessException {
  final File f = getFileName(jCas);
  final DocumentAnnotation da = getDocumentAnnotation(jCas);

  final Document doc =
      Jsoup.parse("<!DOCTYPE html>\n<html lang=\"" + da.getLanguage() + "\"></html>");
  doc.outputSettings(new Document.OutputSettings().prettyPrint(false));
  final Element head = doc.head();

  if (!Strings.isNullOrEmpty(css)) {
    final Element cssLink = head.appendElement("link");
    cssLink.attr("rel", "stylesheet");
    cssLink.attr("href", css);
  }

  final Element charset = head.appendElement("meta");
  charset.attr("charset", "utf-8");

  appendMeta(head, "document.type", da.getDocType());
  appendMeta(head, "document.sourceUri", da.getSourceUri());
  appendMeta(head, "externalId", da.getHash());

  appendMeta(head, "document.classification", da.getDocumentClassification());
  appendMeta(
      head,
      "document.caveats",
      String.join(",", UimaTypesUtils.toArray(da.getDocumentCaveats())));
  appendMeta(
      head,
      "document.releasability",
      String.join(",", UimaTypesUtils.toArray(da.getDocumentReleasability())));

  String title = null;
  for (final Metadata md : JCasUtil.select(jCas, Metadata.class)) {
    appendMeta(head, md.getKey(), md.getValue());
    if ("documentTitle".equalsIgnoreCase(md.getKey())) {
      title = md.getValue();
    }
  }

  if (!Strings.isNullOrEmpty(title)) {
    doc.title(title);
  }

  final Element body = doc.body();

  writeBody(jCas, body);

  try {
    FileUtils.writeStringToFile(f, doc.html(), Charset.defaultCharset());
  } catch (final IOException e) {
    throw new AnalysisEngineProcessException(e);
  }
}
 

private static void setupPwa(Document document, PwaRegistry registry) {
    if (registry == null) {
        return;
    }

    PwaConfiguration config = registry.getPwaConfiguration();

    if (config.isEnabled()) {
        // Add header injections
        Element head = document.head();

        // Describe PWA capability for iOS devices
        head.appendElement(META_TAG)
                .attr("name", "apple-mobile-web-app-capable")
                .attr(CONTENT_ATTRIBUTE, "yes");

        // Theme color
        head.appendElement(META_TAG).attr("name", "theme-color")
                .attr(CONTENT_ATTRIBUTE, config.getThemeColor());
        head.appendElement(META_TAG)
                .attr("name", "apple-mobile-web-app-status-bar-style")
                .attr(CONTENT_ATTRIBUTE, config.getThemeColor());

        // Add manifest
        head.appendElement("link").attr("rel", "manifest").attr("href",
                config.getManifestPath());

        // Add icons
        for (PwaIcon icon : registry.getHeaderIcons()) {
            head.appendChild(icon.asElement());
        }

        // Add service worker initialization
        head.appendElement(SCRIPT_TAG)
                .text("if ('serviceWorker' in navigator) {\n"
                        + "  window.addEventListener('load', function() {\n"
                        + "    navigator.serviceWorker.register('"
                        + config.getServiceWorkerPath() + "');\n"
                        + "  });\n" + "}");
    }
}