Java Code Examples for com.gargoylesoftware.htmlunit.WebClient#URL_ABOUT_BLANK

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;
    }
}
 

/**
 * Returns the domain name of the server that served the document, or {@code null} if the server
 * cannot be identified by a domain name.
 * @return the domain name of the server that served the document
 * @see <a href="http://www.w3.org/TR/2000/WD-DOM-Level-1-20000929/level-one-html.html#ID-2250147">
 * W3C documentation</a>
 */
@JsxGetter({CHROME, IE})
public String getDomain() {
    if (domain_ == null && getPage().getWebResponse() != null) {
        URL url = getPage().getUrl();
        if (url == WebClient.URL_ABOUT_BLANK) {
            final WebWindow w = getWindow().getWebWindow();
            if (w instanceof FrameWindow) {
                url = ((FrameWindow) w).getEnclosingPage().getUrl();
            }
            else {
                return null;
            }
        }
        domain_ = url.getHost().toLowerCase(Locale.ROOT);
    }

    return domain_;
}
 

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;
    }
}
 

/**
 * Returns the domain name of the server that served the document, or {@code null} if the server
 * cannot be identified by a domain name.
 * @return the domain name of the server that served the document
 * @see <a href="http://www.w3.org/TR/2000/WD-DOM-Level-1-20000929/level-one-html.html#ID-2250147">
 * W3C documentation</a>
 */
@JsxGetter({CHROME, IE})
public String getDomain() {
    if (domain_ == null && getPage().getWebResponse() != null) {
        URL url = getPage().getUrl();
        if (url == WebClient.URL_ABOUT_BLANK) {
            final WebWindow w = getWindow().getWebWindow();
            if (w instanceof FrameWindow) {
                url = ((FrameWindow) w).getEnclosingPage().getUrl();
            }
            else {
                return null;
            }
        }
        domain_ = url.getHost().toLowerCase(Locale.ROOT);
    }

    return domain_;
}
 

@Override
URL toUrlUnsafeClassic(final String url) throws MalformedURLException {
    final String protocol = StringUtils.substringBefore(url, ":")
            .toLowerCase(Locale.ROOT);

    if (protocol.isEmpty() || UrlUtils.isNormalUrlProtocol(protocol)) {
        return toNormalUrl(url);
    }
    else if (JavaScriptURLConnection.JAVASCRIPT_PREFIX.equals(protocol + ":")) {
        return new URL(null, url, JS_HANDLER);
    }
    else if ("about".equals(protocol)) {
        if (WebClient.URL_ABOUT_BLANK != null
                && StringUtils.equalsIgnoreCase(WebClient.URL_ABOUT_BLANK.toExternalForm(), url)) {
            return WebClient.URL_ABOUT_BLANK;
        }
        return new URL(null, url, ABOUT_HANDLER);
    }
    else if ("data".equals(protocol)) {
        return new URL(null, url, DATA_HANDLER);
    }
    else {
        return new URL(null, url, AnyHandler.INSTANCE);
    }
}
 

@Override
URL toUrlUnsafeClassic(final String url) throws MalformedURLException {
    if (WebClient.URL_ABOUT_BLANK != null
            && StringUtils.equalsIgnoreCase(WebClient.URL_ABOUT_BLANK.toExternalForm(), url)) {
        return WebClient.URL_ABOUT_BLANK;
    }
    else if (StringUtils.startsWithIgnoreCase(url, JavaScriptURLConnection.JAVASCRIPT_PREFIX)) {
        return new URL(PREFIX + url.replaceFirst(":", "/"));
    }
    else if (StringUtils.startsWithIgnoreCase(url, WebClient.ABOUT_SCHEME)) {
        return new URL(PREFIX + url.replaceFirst(":", "/"));
    }
    else if (StringUtils.startsWithIgnoreCase(url, "data:")) {
        return new URL(PREFIX + url.replaceFirst(":", "/"));
    }
    else {
        return toNormalUrl(url);
    }
}
 

/**
 * Creates an {@link HTMLDocument}.
 * @see <a href="https://developer.mozilla.org/en-US/docs/Web/API/DOMImplementation/createHTMLDocument">
 *   createHTMLDocument (MDN)</a>
 *
 * @param titleObj the document title
 * @return the newly created {@link HTMLDocument}
 */
@JsxFunction
public HTMLDocument createHTMLDocument(final Object titleObj) {
    if (Undefined.isUndefined(titleObj)
            && getBrowserVersion().hasFeature(JS_DOMIMPLEMENTATION_CREATE_HTMLDOCOMENT_REQUIRES_TITLE)) {
        throw Context.reportRuntimeError("Title is required");
    }

    final HTMLDocument document = new HTMLDocument();
    document.setParentScope(getParentScope());
    document.setPrototype(getPrototype(document.getClass()));

    // According to spec and behavior of function in browsers new document
    // has no location object and is not connected with any window
    final WebResponse resp = new StringWebResponse("", WebClient.URL_ABOUT_BLANK);
    final HtmlPage page = new HtmlPage(resp, getWindow().getWebWindow());
    page.setEnclosingWindow(null);
    document.setDomNode(page);

    final HTMLHtmlElement html = (HTMLHtmlElement) document.createElement("html");
    page.appendChild(html.getDomNodeOrDie());

    final HTMLHeadElement head = (HTMLHeadElement) document.createElement("head");
    html.appendChild(head);

    if (!Undefined.isUndefined(titleObj)) {
        final HTMLTitleElement title = (HTMLTitleElement) document.createElement("title");
        head.appendChild(title);
        title.setTextContent(Context.toString(titleObj));
    }

    final HTMLBodyElement body = (HTMLBodyElement) document.createElement("body");
    html.appendChild(body);

    return document;
}
 

/**
 * Sets the domain of this document.
 *
 * Domains can only be set to suffixes of the existing domain
 * with the exception of setting the domain to itself.
 * <p>
 * The domain will be set according to the following rules:
 * <ol>
 * <li>If the newDomain.equalsIgnoreCase(currentDomain) the method returns with no error.</li>
 * <li>If the browser version is netscape, the newDomain is downshifted.</li>
 * <li>The change will take place if and only if the suffixes of the
 *       current domain and the new domain match AND there are at least
 *       two domain qualifiers e.g. the following transformations are legal
 *       d1.d2.d3.gargoylesoftware.com may be transformed to itself or:
 *          d2.d3.gargoylesoftware.com
 *             d3.gargoylesoftware.com
 *                gargoylesoftware.com
 *
 *        transformation to:        com
 *        will fail
 * </li>
 * </ol>
 * <p>
 * TODO This code could be modified to understand country domain suffixes.
 * The domain www.bbc.co.uk should be trimmable only down to bbc.co.uk
 * trimming to co.uk should not be possible.
 * @param newDomain the new domain to set
 */
@JsxSetter({CHROME, IE})
public void setDomain(String newDomain) {
    final BrowserVersion browserVersion = getBrowserVersion();

    // IE (at least 6) doesn't allow to set domain of about:blank
    if (WebClient.URL_ABOUT_BLANK == getPage().getUrl()
        && browserVersion.hasFeature(JS_DOCUMENT_SETTING_DOMAIN_THROWS_FOR_ABOUT_BLANK)) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from \""
                + WebClient.URL_ABOUT_BLANK + "\" to: \""
                + newDomain + "\".");
    }

    newDomain = newDomain.toLowerCase(Locale.ROOT);

    final String currentDomain = getDomain();
    if (currentDomain.equalsIgnoreCase(newDomain)) {
        return;
    }

    if (newDomain.indexOf('.') == -1) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from: \""
                + currentDomain + "\" to: \"" + newDomain + "\" (new domain has to contain a dot).");
    }

    if (currentDomain.indexOf('.') > -1
            && !currentDomain.toLowerCase(Locale.ROOT).endsWith("." + newDomain.toLowerCase(Locale.ROOT))) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from: \""
                + currentDomain + "\" to: \"" + newDomain + "\"");
    }

    domain_ = newDomain;
}
 

/**
 * Returns the pathname portion of the location URL.
 * @return the pathname portion of the location URL
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms534332.aspx">MSDN Documentation</a>
 */
@JsxGetter
public String getPathname() {
    if (WebClient.URL_ABOUT_BLANK == getUrl()) {
        if (getBrowserVersion().hasFeature(URL_ABOUT_BLANK_HAS_BLANK_PATH)) {
            return "blank";
        }
        return "/blank";
    }
    return getUrl().getPath();
}
 

/**
 * <p>Constructs a URL instance based on the specified URL string, taking into account the fact that the
 * specified URL string may represent an <tt>"about:..."</tt> URL, a <tt>"javascript:..."</tt> URL, or
 * a <tt>data:...</tt> URL.</p>
 *
 * <p>Unlike {@link #toUrlSafe(String)}, the caller need not be sure that URL strings passed to this
 * method will parse correctly as URLs.</p>
 *
 * @param url the URL string to convert into a URL instance
 * @return the constructed URL instance
 * @throws MalformedURLException if the URL string cannot be converted to a URL instance
 */
public static URL toUrlUnsafe(final String url) throws MalformedURLException {
    WebAssert.notNull("url", url);

    final String protocol = org.apache.commons.lang3.StringUtils.substringBefore(url, ":").toLowerCase(Locale.ROOT);

    if (protocol.isEmpty() || UrlUtils.isNormalUrlProtocol(protocol)) {
        final URL response = new URL(url);
        if (response.getProtocol().startsWith("http")
                && org.apache.commons.lang3.StringUtils.isEmpty(response.getHost())) {
            throw new MalformedURLException("Missing host name in url: " + url);
        }
        return response;
    }

    if (JavaScriptURLConnection.JAVASCRIPT_PREFIX.equals(protocol + ":")) {
        return new URL(null, url, JS_HANDLER);
    }

    if ("about".equals(protocol)) {
        if (WebClient.URL_ABOUT_BLANK != null
                && org.apache.commons.lang3.StringUtils.
                    equalsIgnoreCase(WebClient.URL_ABOUT_BLANK.toExternalForm(), url)) {
            return WebClient.URL_ABOUT_BLANK;
        }
        return new URL(null, url, ABOUT_HANDLER);
    }

    if ("data".equals(protocol)) {
        return new URL(null, url, DATA_HANDLER);
    }

    return new URL(null, url, AnyHandler.INSTANCE);
}
 

/**
 * Loads the content of the contained frames. This is done after the page is completely loaded, to allow script
 * contained in the frames to reference elements from the page located after the closing </frame> tag.
 * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
 *         {@link WebClient#setThrowExceptionOnFailingStatusCode(boolean)} is set to {@code true}
 */
void loadFrames() throws FailingHttpStatusCodeException {
    for (final FrameWindow w : getFrames()) {
        final BaseFrameElement frame = w.getFrameElement();
        // test if the frame should really be loaded:
        // if a script has already changed its content, it should be skipped
        // use == and not equals(...) to identify initial content (versus URL set to "about:blank")
        if (frame.getEnclosedWindow() != null
                && WebClient.URL_ABOUT_BLANK == frame.getEnclosedPage().getUrl()
                && !frame.isContentLoaded()) {
            frame.loadInnerPage();
        }
    }
}
 

/**
 * Creates an {@link HTMLDocument}.
 * @see <a href="https://developer.mozilla.org/en-US/docs/Web/API/DOMImplementation/createHTMLDocument">
 *   createHTMLDocument (MDN)</a>
 *
 * @param titleObj the document title
 * @return the newly created {@link HTMLDocument}
 */
@JsxFunction
public HTMLDocument createHTMLDocument(final Object titleObj) {
    if (titleObj == Undefined.instance
            && getBrowserVersion().hasFeature(JS_DOMIMPLEMENTATION_CREATE_HTMLDOCOMENT_REQUIRES_TITLE)) {
        throw Context.reportRuntimeError("Title is required");
    }

    final HTMLDocument document = new HTMLDocument();
    document.setParentScope(getParentScope());
    document.setPrototype(getPrototype(document.getClass()));

    // According to spec and behavior of function in browsers new document
    // has no location object and is not connected with any window
    final WebResponse resp = new StringWebResponse("", WebClient.URL_ABOUT_BLANK);
    final HtmlPage page = new HtmlPage(resp, getWindow().getWebWindow());
    page.setEnclosingWindow(null);
    document.setDomNode(page);

    final HTMLHtmlElement html = (HTMLHtmlElement) document.createElement("html");
    page.appendChild(html.getDomNodeOrDie());

    final HTMLHeadElement head = (HTMLHeadElement) document.createElement("head");
    html.appendChild(head);

    if (titleObj != Undefined.instance) {
        final HTMLTitleElement title = (HTMLTitleElement) document.createElement("title");
        head.appendChild(title);
        title.setTextContent(Context.toString(titleObj));
    }

    final HTMLBodyElement body = (HTMLBodyElement) document.createElement("body");
    html.appendChild(body);

    return document;
}
 

/**
 * Sets the domain of this document.
 *
 * Domains can only be set to suffixes of the existing domain
 * with the exception of setting the domain to itself.
 * <p>
 * The domain will be set according to the following rules:
 * <ol>
 * <li>If the newDomain.equalsIgnoreCase(currentDomain) the method returns with no error.</li>
 * <li>If the browser version is netscape, the newDomain is downshifted.</li>
 * <li>The change will take place if and only if the suffixes of the
 *       current domain and the new domain match AND there are at least
 *       two domain qualifiers e.g. the following transformations are legal
 *       d1.d2.d3.gargoylesoftware.com may be transformed to itself or:
 *          d2.d3.gargoylesoftware.com
 *             d3.gargoylesoftware.com
 *                gargoylesoftware.com
 *
 *        transformation to:        com
 *        will fail
 * </li>
 * </ol>
 * <p>
 * TODO This code could be modified to understand country domain suffixes.
 * The domain www.bbc.co.uk should be trimmable only down to bbc.co.uk
 * trimming to co.uk should not be possible.
 * @param newDomain the new domain to set
 */
@JsxSetter({CHROME, IE})
public void setDomain(String newDomain) {
    final BrowserVersion browserVersion = getBrowserVersion();

    // IE (at least 6) doesn't allow to set domain of about:blank
    if (WebClient.URL_ABOUT_BLANK == getPage().getUrl()
        && browserVersion.hasFeature(JS_DOCUMENT_SETTING_DOMAIN_THROWS_FOR_ABOUT_BLANK)) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from \""
                + WebClient.URL_ABOUT_BLANK + "\" to: \""
                + newDomain + "\".");
    }

    newDomain = newDomain.toLowerCase(Locale.ROOT);

    final String currentDomain = getDomain();
    if (currentDomain.equalsIgnoreCase(newDomain)) {
        return;
    }

    if (newDomain.indexOf('.') == -1) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from: \""
                + currentDomain + "\" to: \"" + newDomain + "\" (new domain has to contain a dot).");
    }

    if (currentDomain.indexOf('.') > -1
            && !currentDomain.toLowerCase(Locale.ROOT).endsWith("." + newDomain.toLowerCase(Locale.ROOT))) {
        throw Context.reportRuntimeError("Illegal domain value, cannot set domain from: \""
                + currentDomain + "\" to: \"" + newDomain + "\"");
    }

    domain_ = newDomain;
}
 

/**
 * Returns the pathname portion of the location URL.
 * @return the pathname portion of the location URL
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms534332.aspx">MSDN Documentation</a>
 */
@JsxGetter
public String getPathname() {
    if (WebClient.URL_ABOUT_BLANK == getUrl()) {
        if (getBrowserVersion().hasFeature(URL_ABOUT_BLANK_HAS_EMPTY_PATH)) {
            return "";
        }
        if (getBrowserVersion().hasFeature(URL_ABOUT_BLANK_HAS_BLANK_PATH)) {
            return "blank";
        }
        return "/blank";
    }
    return getUrl().getPath();
}
 

/**
 * Loads the content of the contained frames. This is done after the page is completely loaded, to allow script
 * contained in the frames to reference elements from the page located after the closing </frame> tag.
 * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
 *         {@link WebClient#setThrowExceptionOnFailingStatusCode(boolean)} is set to {@code true}
 */
void loadFrames() throws FailingHttpStatusCodeException {
    for (final FrameWindow w : getFrames()) {
        final BaseFrameElement frame = w.getFrameElement();
        // test if the frame should really be loaded:
        // if a script has already changed its content, it should be skipped
        // use == and not equals(...) to identify initial content (versus URL set to "about:blank")
        if (frame.getEnclosedWindow() != null
                && WebClient.URL_ABOUT_BLANK == frame.getEnclosedPage().getUrl()
                && !frame.isContentLoaded()) {
            frame.loadInnerPage();
        }
    }
}
 

/**
 * <span style="color:red">INTERNAL API - SUBJECT TO CHANGE AT ANY TIME - USE AT YOUR OWN RISK.</span><br>
 *
 * Gets the request for a submission of this form with the specified SubmittableElement.
 * @param submitElement the element that caused the submit to occur
 * @return the request
 */
public WebRequest getWebRequest(final SubmittableElement submitElement) {
    final HtmlPage htmlPage = (HtmlPage) getPage();
    final List<NameValuePair> parameters = getParameterListForSubmit(submitElement);
    final HttpMethod method;
    final String methodAttribute = getMethodAttribute();
    if ("post".equalsIgnoreCase(methodAttribute)) {
        method = HttpMethod.POST;
    }
    else {
        if (!"get".equalsIgnoreCase(methodAttribute) && StringUtils.isNotBlank(methodAttribute)) {
            notifyIncorrectness("Incorrect submit method >" + getMethodAttribute() + "<. Using >GET<.");
        }
        method = HttpMethod.GET;
    }

    final BrowserVersion browser = getPage().getWebClient().getBrowserVersion();
    String actionUrl = getActionAttribute();
    String anchor = null;
    String queryFromFields = "";
    if (HttpMethod.GET == method) {
        if (actionUrl.contains("#")) {
            anchor = StringUtils.substringAfter(actionUrl, "#");
        }
        final Charset enc = getPage().getCharset();
        queryFromFields =
            URLEncodedUtils.format(Arrays.asList(NameValuePair.toHttpClient(parameters)), enc);

        // action may already contain some query parameters: they have to be removed
        actionUrl = StringUtils.substringBefore(actionUrl, "#");
        actionUrl = StringUtils.substringBefore(actionUrl, "?");
        parameters.clear(); // parameters have been added to query
    }
    URL url;
    try {
        if (actionUrl.isEmpty()) {
            url = WebClient.expandUrl(htmlPage.getUrl(), actionUrl);
        }
        else {
            url = htmlPage.getFullyQualifiedUrl(actionUrl);
        }

        if (!queryFromFields.isEmpty()) {
            url = UrlUtils.getUrlWithNewQuery(url, queryFromFields);
        }

        if (HttpMethod.GET == method && browser.hasFeature(FORM_SUBMISSION_URL_WITHOUT_HASH)
                && WebClient.URL_ABOUT_BLANK != url) {
            url = UrlUtils.getUrlWithNewRef(url, null);
        }
        else if (HttpMethod.POST == method
                && browser.hasFeature(FORM_SUBMISSION_URL_WITHOUT_HASH)
                && WebClient.URL_ABOUT_BLANK != url
                && StringUtils.isEmpty(actionUrl)) {
            url = UrlUtils.getUrlWithNewRef(url, null);
        }
        else if (anchor != null
                && WebClient.URL_ABOUT_BLANK != url) {
            url = UrlUtils.getUrlWithNewRef(url, anchor);
        }
    }
    catch (final MalformedURLException e) {
        throw new IllegalArgumentException("Not a valid url: " + actionUrl);
    }

    final WebRequest request = new WebRequest(url, method);
    request.setAdditionalHeader(HttpHeader.ACCEPT, browser.getHtmlAcceptHeader());
    request.setAdditionalHeader(HttpHeader.ACCEPT_ENCODING, "gzip, deflate");
    request.setRequestParameters(parameters);
    if (HttpMethod.POST == method) {
        request.setEncodingType(FormEncodingType.getInstance(getEnctypeAttribute()));
    }
    request.setCharset(getSubmitCharset());

    String referer = htmlPage.getUrl().toExternalForm();
    request.setAdditionalHeader(HttpHeader.REFERER, referer);

    if (HttpMethod.POST == method
            && browser.hasFeature(FORM_SUBMISSION_HEADER_ORIGIN)) {
        referer = StringUtils.stripEnd(referer, "/");
        request.setAdditionalHeader(HttpHeader.ORIGIN, referer);
    }
    if (HttpMethod.POST == method
            && browser.hasFeature(FORM_SUBMISSION_HEADER_CACHE_CONTROL_MAX_AGE)) {
        request.setAdditionalHeader(HttpHeader.CACHE_CONTROL, "max-age=0");
    }
    if (browser.hasFeature(FORM_SUBMISSION_HEADER_CACHE_CONTROL_NO_CACHE)) {
        request.setAdditionalHeader(HttpHeader.CACHE_CONTROL, "no-cache");
    }

    return request;
}