com.gargoylesoftware.htmlunit.html.HTMLParser Java Examples

/**
 * Parses an HTML snippet.
 * @param valueAsString text that contains text and tags to be converted to a document fragment
 * @return a document fragment
 * @see <a href="https://developer.mozilla.org/en-US/docs/DOM/range.createContextualFragment">Mozilla
 * documentation</a>
 */
@JsxFunction
public Object createContextualFragment(final String valueAsString) {
    final SgmlPage page = startContainer_.getDomNodeOrDie().getPage();
    final DomDocumentFragment fragment = new DomDocumentFragment(page);
    try {
        HTMLParser.parseFragment(fragment, startContainer_.getDomNodeOrDie(), valueAsString);
    }
    catch (final Exception e) {
        LogFactory.getLog(Range.class).error("Unexpected exception occurred in createContextualFragment", e);
        throw Context.reportRuntimeError("Unexpected exception occurred in createContextualFragment: "
                + e.getMessage());
    }

    return fragment.getScriptableObject();
}
 

/**
 * Changes the number of options: removes options if the new length
 * is less than the current one else add new empty options to reach the
 * new length.
 * @param newLength the new length property value
 */
@JsxSetter
public void setLength(final int newLength) {
    if (newLength < 0) {
        if (getBrowserVersion().hasFeature(JS_SELECT_OPTIONS_IGNORE_NEGATIVE_LENGTH)) {
            return;
        }
        throw Context.reportRuntimeError("Length is negative");
    }

    final int currentLength = htmlSelect_.getOptionSize();
    if (currentLength > newLength) {
        htmlSelect_.setOptionSize(newLength);
    }
    else {
        for (int i = currentLength; i < newLength; i++) {
            final HtmlOption option = (HtmlOption) HTMLParser.getFactory(HtmlOption.TAG_NAME).createElement(
                    htmlSelect_.getPage(), HtmlOption.TAG_NAME, null);
            htmlSelect_.appendOption(option);
            if (!getBrowserVersion().hasFeature(JS_SELECT_OPTIONS_DONT_ADD_EMPTY_TEXT_CHILD_WHEN_EXPANDING)) {
                option.appendChild(new DomText(option.getPage(), ""));
            }
        }
    }
}
 

private void checkWebPageForExpectedEmptyResult(final AnalysisResult result) {
    try (WebClient webClient = createWebClient(false)) {
        WebResponse webResponse = webClient.getPage(result.getOwner(), DEFAULT_ENTRY_PATH).getWebResponse();
        HtmlPage htmlPage = HTMLParser.parseHtml(webResponse, webClient.getCurrentWindow());
        assertThat(getLinksWithGivenTargetName(htmlPage, DEFAULT_TAB_TO_INVESTIGATE)).isEmpty();
    }
    catch (IOException | SAXException e) {
        throw new AssertionError(e);
    }
}
 

/**
 * JavaScript constructor.
 */
@JsxConstructor({CHROME, FF, EDGE})
public void jsConstructor() {
    final SgmlPage page = (SgmlPage) getWindow().getWebWindow().getEnclosedPage();
    final DomElement fake =
            HTMLParser.getFactory(HtmlImage.TAG_NAME).createElement(page, HtmlImage.TAG_NAME, new AttributesImpl());
    setDomNode(fake);
}
 

/**
 * Create a Page object for the specified web response.
 *
 * @param webResponse the response from the server
 * @param webWindow the window that this page will be loaded into
 * @exception IOException if an IO problem occurs
 * @return the new page object
 */
@Override
public Page createPage(final WebResponse webResponse, final WebWindow webWindow) throws IOException {
    final String contentType = determineContentType(webResponse);

    final PageType pageType = determinePageType(contentType);
    switch (pageType) {
        case HTML:
            return createHtmlPage(webResponse, webWindow);

        case JAVASCRIPT:
            return createHtmlPage(webResponse, webWindow);

        case XML:
            final SgmlPage sgmlPage = createXmlPage(webResponse, webWindow);
            final DomElement doc = sgmlPage.getDocumentElement();
            if (doc != null && HTMLParser.XHTML_NAMESPACE.equals(doc.getNamespaceURI())) {
                return createXHtmlPage(webResponse, webWindow);
            }
            return sgmlPage;

        case TEXT:
            return createTextPage(webResponse, webWindow);

        default:
            return createUnexpectedPage(webResponse, webWindow);
    }
}
 

@Test
public void testRender() throws Exception {
    ScheduledTime end = new ScheduledTime("2015-09-03T13:17Z");
    ScheduledTime start = new ScheduledTime("2015-09-03T13:11Z");
    NavigableSet<ScheduledTime> tileTimes = new TreeSet<>(ImmutableSet.of(new ScheduledTime("2015-09-03T13:10Z"), new ScheduledTime("2015-09-03T13:15Z")));
    WorkflowID id = new WorkflowID("foo");
    List<WorkflowGroup> groups = ImmutableList.of(new WorkflowGroup("All workflows", ImmutableList.of(id)));
    WorkflowInfo workflowInfo = new WorkflowInfo(new URL("http://example.com"), ImmutableList.of());
    SlotState state1 = new SlotState(new SlotID(id, new ScheduledTime("2015-09-03T13:16Z")), SlotState.Status.FAILURE);
    SlotState state2 = new SlotState(new SlotID(id, new ScheduledTime("2015-09-03T13:12Z")), SlotState.Status.WAITING);
    List<SlotState> slotStates = ImmutableList.of(state1, state2);
    Map<WorkflowID, WorkflowStatus> statuses = ImmutableMap.of(id, new WorkflowStatus(workflowInfo, slotStates, false));
    UIConfiguration conf = new UIConfiguration(start, end, tileTimes, groups, statuses, new URL("http://example.com"));
    
    StringWebResponse response = new StringWebResponse(UIServlet.render(conf), new URL("http://example.com"));
    WebClient webClient = new WebClient();
    webClient.setThrowExceptionOnFailingStatusCode(false);
    HtmlPage page = HTMLParser.parse(response, new TopLevelWindow("top", webClient));
    
    // Some basic sanity checking
    
    List<HtmlTableDataCell> slotCells = (List<HtmlTableDataCell>) page.getByXPath("//td[contains(@class, 'slot')]");
    Assert.assertEquals("fail", slotCells.get(0).getTextContent());
    Assert.assertEquals("wait", slotCells.get(1).getTextContent());
    
    List<HtmlTableDataCell> hourCells = (List<HtmlTableDataCell>) page.getByXPath("//td[contains(@class, 'hour')]");
    Assert.assertEquals("1315", hourCells.get(0).getTextContent());
    Assert.assertEquals("1310", hourCells.get(1).getTextContent());
    
    List<HtmlTableDataCell> workflowCells = (List<HtmlTableDataCell>) page.getByXPath("//td[@class='workflow']");
    Assert.assertEquals("foo", workflowCells.get(0).getTextContent());
    
    System.out.println(response.getContentAsString());
}
 

private static DomNode createFrom(final SgmlPage page, final Node source, final boolean handleXHTMLAsHTML,
        final Map<Integer, List<String>> attributesOrderMap) {
    if (source.getNodeType() == Node.TEXT_NODE) {
        return new DomText(page, source.getNodeValue());
    }
    if (source.getNodeType() == Node.PROCESSING_INSTRUCTION_NODE) {
        return new DomProcessingInstruction(page, source.getNodeName(), source.getNodeValue());
    }
    if (source.getNodeType() == Node.COMMENT_NODE) {
        return new DomComment(page, source.getNodeValue());
    }
    if (source.getNodeType() == Node.DOCUMENT_TYPE_NODE) {
        final DocumentType documentType = (DocumentType) source;
        return new DomDocumentType(page, documentType.getName(), documentType.getPublicId(),
                documentType.getSystemId());
    }
    final String ns = source.getNamespaceURI();
    String localName = source.getLocalName();
    if (handleXHTMLAsHTML && HTMLParser.XHTML_NAMESPACE.equals(ns)) {
        final ElementFactory factory = HTMLParser.getFactory(localName);
        return factory.createElementNS(page, ns, localName,
                namedNodeMapToSaxAttributes(source.getAttributes(), attributesOrderMap, source));
    }
    final NamedNodeMap nodeAttributes = source.getAttributes();
    if (page != null && page.isHtmlPage()) {
        localName = localName.toUpperCase(Locale.ROOT);
    }
    final String qualifiedName;
    if (source.getPrefix() == null) {
        qualifiedName = localName;
    }
    else {
        qualifiedName = source.getPrefix() + ':' + localName;
    }

    final String namespaceURI = source.getNamespaceURI();
    if (HTMLParser.SVG_NAMESPACE.equals(namespaceURI)) {
        return HTMLParser.SVG_FACTORY.createElementNS(page, namespaceURI, qualifiedName,
                namedNodeMapToSaxAttributes(nodeAttributes, attributesOrderMap, source));
    }

    final Map<String, DomAttr> attributes = new LinkedHashMap<>();
    for (int i = 0; i < nodeAttributes.getLength(); i++) {
        final int orderedIndex = getIndex(nodeAttributes, attributesOrderMap, source, i);
        final Attr attribute = (Attr) nodeAttributes.item(orderedIndex);
        final String attributeNamespaceURI = attribute.getNamespaceURI();
        final String attributeQualifiedName;
        if (attribute.getPrefix() != null) {
            attributeQualifiedName = attribute.getPrefix() + ':' + attribute.getLocalName();
        }
        else {
            attributeQualifiedName = attribute.getLocalName();
        }
        final String value = attribute.getNodeValue();
        final boolean specified = attribute.getSpecified();
        final DomAttr xmlAttribute =
                new DomAttr(page, attributeNamespaceURI, attributeQualifiedName, value, specified);
        attributes.put(attribute.getNodeName(), xmlAttribute);
    }
    return new DomElement(namespaceURI, qualifiedName, page, attributes);
}
 

/**
 * Creates an HtmlPage for this WebResponse.
 *
 * @param webResponse the page's source
 * @param webWindow the WebWindow to place the HtmlPage in
 * @return the newly created HtmlPage
 * @throws IOException if the page could not be created
 */
protected HtmlPage createHtmlPage(final WebResponse webResponse, final WebWindow webWindow) throws IOException {
    return HTMLParser.parseHtml(webResponse, webWindow);
}
 

/**
 * Creates an XHtmlPage for this WebResponse.
 *
 * @param webResponse the page's source
 * @param webWindow the WebWindow to place the HtmlPage in
 * @return the newly created XHtmlPage
 * @throws IOException if the page could not be created
 */
protected XHtmlPage createXHtmlPage(final WebResponse webResponse, final WebWindow webWindow) throws IOException {
    return HTMLParser.parseXHtml(webResponse, webWindow);
}