Java Code Examples for com.gargoylesoftware.htmlunit.Page#isHtmlPage()

private URL makeUrlForOpenWindow(final String urlString) {
    if (urlString.isEmpty()) {
        return WebClient.URL_ABOUT_BLANK;
    }

    try {
        final Page page = getWebWindow().getEnclosedPage();
        if (page != null && page.isHtmlPage()) {
            return ((HtmlPage) page).getFullyQualifiedUrl(urlString);
        }
        return new URL(urlString);
    }
    catch (final MalformedURLException e) {
        if (LOG.isWarnEnabled()) {
            LOG.error("Unable to create URL for openWindow: relativeUrl=[" + urlString + "]", e);
        }
        return null;
    }
}
 

/**
 * Override of default clickAction that makes this radio button the selected
 * one when it is clicked.
 * {@inheritDoc}
 *
 * @throws IOException if an IO error occurred
 */
@Override
protected boolean doClickStateUpdate(final boolean shiftKey, final boolean ctrlKey) throws IOException {
    final HtmlForm form = getEnclosingForm();
    final boolean changed = !isChecked();

    final Page page = getPage();
    if (form != null) {
        form.setCheckedRadioButton(this);
    }
    else if (page != null && page.isHtmlPage()) {
        setCheckedForPage((HtmlPage) page);
    }
    super.doClickStateUpdate(shiftKey, ctrlKey);
    return changed;
}
 

/**
 * Sets the {@code checked} attribute.
 *
 * @param isChecked true if this element is to be selected
 * @return the page that occupies this window after setting checked status
 * It may be the same window or it may be a freshly loaded one.
 */
@Override
public Page setChecked(final boolean isChecked) {
    Page page = getPage();

    final boolean changed = isChecked() != isChecked;
    checkedState_ = isChecked;
    if (isChecked) {
        final HtmlForm form = getEnclosingForm();
        if (form != null) {
            form.setCheckedRadioButton(this);
        }
        else if (page != null && page.isHtmlPage()) {
            setCheckedForPage((HtmlPage) page);
        }
    }

    if (changed) {
        final ScriptResult scriptResult = fireEvent(Event.TYPE_CHANGE);
        if (scriptResult != null) {
            page = scriptResult.getNewPage();
        }
    }
    return page;
}
 

/**
 * Override of default clickAction that makes this radio button the selected
 * one when it is clicked.
 * {@inheritDoc}
 *
 * @throws IOException if an IO error occurred
 */
@Override
protected boolean doClickStateUpdate(final boolean shiftKey, final boolean ctrlKey) throws IOException {
    final HtmlForm form = getEnclosingForm();
    final boolean changed = !isChecked();

    final Page page = getPage();
    if (form != null) {
        form.setCheckedRadioButton(this);
    }
    else if (page != null && page.isHtmlPage()) {
        setCheckedForPage((HtmlPage) page);
    }
    super.doClickStateUpdate(shiftKey, ctrlKey);
    return changed;
}
 

private URL makeUrlForOpenWindow(final String urlString) {
    if (urlString.isEmpty()) {
        return WebClient.URL_ABOUT_BLANK;
    }

    try {
        final Page page = getWebWindow().getEnclosedPage();
        if (page != null && page.isHtmlPage()) {
            return ((HtmlPage) page).getFullyQualifiedUrl(urlString);
        }
        return new URL(urlString);
    }
    catch (final MalformedURLException e) {
        LOG.error("Unable to create URL for openWindow: relativeUrl=[" + urlString + "]", e);
        return null;
    }
}
 

/**
 * Deregister frames that are no longer in use.
 */
public void deregisterFramesIfNeeded() {
    for (final WebWindow window : getFrames()) {
        getWebClient().deregisterWebWindow(window);
        final Page page = window.getEnclosedPage();
        if (page != null && page.isHtmlPage()) {
            // seems quite silly, but for instance if the src attribute of an iframe is not
            // set, the error only occurs when leaving the page
            ((HtmlPage) page).deregisterFramesIfNeeded();
        }
    }
}
 

/**
 * Adds a body element to the current page, if necessary. Strictly speaking, this should
 * probably be done by NekoHTML. See the bug linked below. If and when that bug is fixed,
 * we may be able to get rid of this code.
 *
 * http://sourceforge.net/p/nekohtml/bugs/15/
 * @param page
 * @param originalCall
 * @param checkInsideFrameOnly true if the original page had body that was removed by JavaScript
 */
private static void addBodyToPageIfNecessary(
        final HtmlPage page, final boolean originalCall, final boolean checkInsideFrameOnly) {
    // IE waits for the whole page to load before initializing bodies for frames.
    final boolean waitToLoad = page.hasFeature(PAGE_WAIT_LOAD_BEFORE_BODY);
    if (page.getEnclosingWindow() instanceof FrameWindow && originalCall && waitToLoad) {
        return;
    }

    // Find out if the document already has a body element (or frameset).
    final Element doc = page.getDocumentElement();
    boolean hasBody = false;
    for (Node child = doc.getFirstChild(); child != null; child = child.getNextSibling()) {
        if (child instanceof HtmlBody || child instanceof HtmlFrameSet) {
            hasBody = true;
            break;
        }
    }

    // If the document does not have a body, add it.
    if (!hasBody && !checkInsideFrameOnly) {
        final HtmlBody body = new HtmlBody("body", page, null, false);
        doc.appendChild(body);
    }

    // If this is IE, we need to initialize the bodies of any frames, as well.
    // This will already have been done when emulating FF (see above).
    if (waitToLoad) {
        for (final FrameWindow frame : page.getFrames()) {
            final Page containedPage = frame.getEnclosedPage();
            if (containedPage != null && containedPage.isHtmlPage()) {
                addBodyToPageIfNecessary((HtmlPage) containedPage, false, false);
            }
        }
    }
}
 

private static String getFileExtension(final Page enclosedPage) {
    if (enclosedPage != null) {
        if (enclosedPage.isHtmlPage()) {
            return "html";
        }

        final URL url = enclosedPage.getUrl();
        if (url.getPath().contains(".")) {
            return StringUtils.substringAfterLast(url.getPath(), ".");
        }
    }

    return ".unknown";
}
 

private Map<String, DomAttr> getAttributesFor(final BaseFrameElement frame) throws IOException {
    final Map<String, DomAttr> map = createAttributesCopyWithClonedAttribute(frame, "src");
    final DomAttr srcAttr = map.get("src");
    if (srcAttr == null) {
        return map;
    }

    final Page enclosedPage = frame.getEnclosedPage();
    final String suffix = getFileExtension(enclosedPage);
    final File file = createFile(srcAttr.getValue(), "." + suffix);

    if (enclosedPage != null) {
        if (enclosedPage.isHtmlPage()) {
            file.delete(); // TODO: refactor as it is stupid to create empty file at one place
            // and then to complain that it already exists
            ((HtmlPage) enclosedPage).save(file);
        }
        else {
            try (InputStream is = enclosedPage.getWebResponse().getContentAsStream()) {
                try (FileOutputStream fos = new FileOutputStream(file)) {
                    IOUtils.copyLarge(is, fos);
                }
            }
        }
    }

    srcAttr.setValue(file.getParentFile().getName() + FILE_SEPARATOR + file.getName());
    return map;
}
 

/**
 * Initialize the object.
 * @param enclosedPage the page containing the JavaScript
 */
public void initialize(final Page enclosedPage) {
    if (enclosedPage != null && enclosedPage.isHtmlPage()) {
        final HtmlPage htmlPage = (HtmlPage) enclosedPage;

        // Windows don't have corresponding DomNodes so set the domNode
        // variable to be the page. If this isn't set then SimpleScriptable.get()
        // won't work properly
        setDomNode(htmlPage);
        clearEventListenersContainer();

        WebAssert.notNull("document_", document_);
        document_.setDomNode(htmlPage);
    }
}
 

/**
 * <span style="color:red">INTERNAL API - SUBJECT TO CHANGE AT ANY TIME - USE AT YOUR OWN RISK.</span><br>
 *
 * Called after the node for the {@code frame} or {@code iframe} has been added to the containing page.
 * The node needs to be added first to allow JavaScript in the frame to see the frame in the parent.
 * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
 *      {@link com.gargoylesoftware.htmlunit.WebClientOptions#setThrowExceptionOnFailingStatusCode(boolean)} is
 *      set to true
 */

public void loadInnerPage() throws FailingHttpStatusCodeException {
    String source = getSrcAttribute();
    if (source.isEmpty() || StringUtils.startsWithIgnoreCase(source, WebClient.ABOUT_SCHEME)) {
        source = WebClient.ABOUT_BLANK;
    }

    loadInnerPageIfPossible(source);

    final Page enclosedPage = getEnclosedPage();
    if (enclosedPage != null && enclosedPage.isHtmlPage()) {
        final HtmlPage htmlPage = (HtmlPage) enclosedPage;

        final AbstractJavaScriptEngine<?> jsEngine = getPage().getWebClient().getJavaScriptEngine();
        if (jsEngine != null && jsEngine.isScriptRunning()) {
            final PostponedAction action = new PostponedAction(getPage()) {
                @Override
                public void execute() throws Exception {
                    htmlPage.setReadyState(READY_STATE_COMPLETE);
                }
            };
            jsEngine.addPostponedAction(action);
        }
        else {
            htmlPage.setReadyState(READY_STATE_COMPLETE);
        }
    }
}
 

/**
 * Deregister frames that are no longer in use.
 */
public void deregisterFramesIfNeeded() {
    for (final WebWindow window : getFrames()) {
        getWebClient().deregisterWebWindow(window);
        final Page page = window.getEnclosedPage();
        if (page != null && page.isHtmlPage()) {
            // seems quite silly, but for instance if the src attribute of an iframe is not
            // set, the error only occurs when leaving the page
            ((HtmlPage) page).deregisterFramesIfNeeded();
        }
    }
}
 

@Override
public HtmlPage makeObject(TvMaoObjectKey key) throws Exception {
	Page page = WebCrawler.crawl(key.url);
	if (page.isHtmlPage()) {
		return (HtmlPage) page;
	}
	throw new MyTvException("invalid web page which url: " + key.url);
}
 

/**
 * <span style="color:red">INTERNAL API - SUBJECT TO CHANGE AT ANY TIME - USE AT YOUR OWN RISK.</span><br>
 *
 * Called after the node for the {@code frame} or {@code iframe} has been added to the containing page.
 * The node needs to be added first to allow JavaScript in the frame to see the frame in the parent.
 * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
 *      {@link com.gargoylesoftware.htmlunit.WebClientOptions#setThrowExceptionOnFailingStatusCode(boolean)} is
 *      set to true
 */

public void loadInnerPage() throws FailingHttpStatusCodeException {
    String source = getSrcAttribute();
    if (source.isEmpty() || StringUtils.startsWithIgnoreCase(source, WebClient.ABOUT_SCHEME)) {
        source = WebClient.ABOUT_BLANK;
    }

    loadInnerPageIfPossible(source);

    final Page enclosedPage = getEnclosedPage();
    if (enclosedPage != null && enclosedPage.isHtmlPage()) {
        final HtmlPage htmlPage = (HtmlPage) enclosedPage;
        final AbstractJavaScriptEngine<?> jsEngine = getPage().getWebClient().getJavaScriptEngine();
        if (jsEngine.isScriptRunning()) {
            final PostponedAction action = new PostponedAction(getPage()) {
                @Override
                public void execute() throws Exception {
                    htmlPage.setReadyState(READY_STATE_COMPLETE);
                }
            };
            jsEngine.addPostponedAction(action);
        }
        else {
            htmlPage.setReadyState(READY_STATE_COMPLETE);
        }
    }
}
 

/**
 * Adds a body element to the current page, if necessary. Strictly speaking, this should
 * probably be done by NekoHTML. See the bug linked below. If and when that bug is fixed,
 * we may be able to get rid of this code.
 *
 * http://sourceforge.net/p/nekohtml/bugs/15/
 * @param page
 * @param originalCall
 * @param checkInsideFrameOnly true if the original page had body that was removed by JavaScript
 */
private void addBodyToPageIfNecessary(
        final HtmlPage page, final boolean originalCall, final boolean checkInsideFrameOnly) {
    // IE waits for the whole page to load before initializing bodies for frames.
    final boolean waitToLoad = page.hasFeature(PAGE_WAIT_LOAD_BEFORE_BODY);
    if (page.getEnclosingWindow() instanceof FrameWindow && originalCall && waitToLoad) {
        return;
    }

    // Find out if the document already has a body element (or frameset).
    final Element doc = page.getDocumentElement();
    boolean hasBody = false;
    for (Node child = doc.getFirstChild(); child != null; child = child.getNextSibling()) {
        if (child instanceof HtmlBody || child instanceof HtmlFrameSet) {
            hasBody = true;
            break;
        }
    }

    // If the document does not have a body, add it.
    if (!hasBody && !checkInsideFrameOnly) {
        final DomElement body = getFactory("body").createElement(page, "body", null);
        doc.appendChild(body);
    }

    // If this is IE, we need to initialize the bodies of any frames, as well.
    // This will already have been done when emulating FF (see above).
    if (waitToLoad) {
        for (final FrameWindow frame : page.getFrames()) {
            final Page containedPage = frame.getEnclosedPage();
            if (containedPage != null && containedPage.isHtmlPage()) {
                addBodyToPageIfNecessary((HtmlPage) containedPage, false, false);
            }
        }
    }
}
 

private static String getFileExtension(final Page enclosedPage) {
    if (enclosedPage != null) {
        if (enclosedPage.isHtmlPage()) {
            return "html";
        }

        final URL url = enclosedPage.getUrl();
        if (url.getPath().contains(".")) {
            return StringUtils.substringAfterLast(url.getPath(), ".");
        }
    }

    return ".unknown";
}
 

private Map<String, DomAttr> getAttributesFor(final BaseFrameElement frame) throws IOException {
    final Map<String, DomAttr> map = createAttributesCopyWithClonedAttribute(frame, DomElement.SRC_ATTRIBUTE);
    final DomAttr srcAttr = map.get(DomElement.SRC_ATTRIBUTE);
    if (srcAttr == null) {
        return map;
    }

    final Page enclosedPage = frame.getEnclosedPage();
    final String suffix = getFileExtension(enclosedPage);
    final File file = createFile(srcAttr.getValue(), "." + suffix);

    if (enclosedPage != null) {
        if (enclosedPage.isHtmlPage()) {
            file.delete(); // TODO: refactor as it is stupid to create empty file at one place
            // and then to complain that it already exists
            ((HtmlPage) enclosedPage).save(file);
        }
        else {
            try (InputStream is = enclosedPage.getWebResponse().getContentAsStream()) {
                try (OutputStream fos = Files.newOutputStream(file.toPath())) {
                    IOUtils.copyLarge(is, fos);
                }
            }
        }
    }

    srcAttr.setValue(file.getParentFile().getName() + FILE_SEPARATOR + file.getName());
    return map;
}
 

/**
 * Initialize the object.
 * @param enclosedPage the page containing the JavaScript
 */
public void initialize(final Page enclosedPage) {
    if (enclosedPage != null && enclosedPage.isHtmlPage()) {
        final HtmlPage htmlPage = (HtmlPage) enclosedPage;

        // Windows don't have corresponding DomNodes so set the domNode
        // variable to be the page. If this isn't set then SimpleScriptable.get()
        // won't work properly
        setDomNode(htmlPage);
        clearEventListenersContainer();

        WebAssert.notNull("document_", document_);
        document_.setDomNode(htmlPage);
    }
}
 

/** {@inheritDoc} */
@Override
public void run() {
    final WebWindow w = window_.get();
    if (w == null) {
        // The window has been garbage collected! No need to execute, obviously.
        return;
    }

    if (LOG.isDebugEnabled()) {
        LOG.debug("Executing " + this + ".");
    }

    try {
        // Verify that the window is still open
        if (w.isClosed()) {
            if (LOG.isDebugEnabled()) {
                LOG.debug("Enclosing window is now closed. Execution cancelled.");
            }
            return;
        }
        if (!w.getWebClient().containsWebWindow(w)) {
            if (LOG.isDebugEnabled()) {
                LOG.debug("Enclosing window is now closed. Execution cancelled.");
            }
            return;
        }

        // Verify that the current page is still available and a html page
        final Page enclosedPage = w.getEnclosedPage();
        if (enclosedPage == null || !enclosedPage.isHtmlPage()) {
            if (enclosedPage == null) {
                if (LOG.isDebugEnabled()) {
                    LOG.debug("The page that originated this job doesn't exist anymore. Execution cancelled.");
                }
                return;
            }
            if (LOG.isDebugEnabled()) {
                LOG.debug("The page that originated this job is no html page ("
                            + enclosedPage.getClass().getName() + "). Execution cancelled.");
            }
            return;
        }

        runJavaScript((HtmlPage) enclosedPage);
    }
    finally {
        if (LOG.isDebugEnabled()) {
            LOG.debug("Finished executing " + this + ".");
        }
    }
}
 

/** {@inheritDoc} */
@Override
public void run() {
    final WebWindow w = window_.get();
    if (w == null) {
        // The window has been garbage collected! No need to execute, obviously.
        return;
    }

    if (LOG.isDebugEnabled()) {
        LOG.debug("Executing " + this + ".");
    }

    try {
        // Verify that the window is still open
        if (w.isClosed()) {
            if (LOG.isDebugEnabled()) {
                LOG.debug("Enclosing window is now closed. Execution cancelled.");
            }
            return;
        }
        if (!w.getWebClient().containsWebWindow(w)) {
            if (LOG.isDebugEnabled()) {
                LOG.debug("Enclosing window is now closed. Execution cancelled.");
            }
            return;
        }

        // Verify that the current page is still available and a html page
        final Page enclosedPage = w.getEnclosedPage();
        if (enclosedPage == null || !enclosedPage.isHtmlPage()) {
            if (enclosedPage == null) {
                if (LOG.isDebugEnabled()) {
                    LOG.debug("The page that originated this job doesn't exist anymore. Execution cancelled.");
                }
                return;
            }
            if (LOG.isDebugEnabled()) {
                LOG.debug("The page that originated this job is no html page ("
                            + enclosedPage.getClass().getName() + "). Execution cancelled.");
            }
            return;
        }

        runJavaScript((HtmlPage) enclosedPage);
    }
    finally {
        if (LOG.isDebugEnabled()) {
            LOG.debug("Finished executing " + this + ".");
        }
    }
}