Java HtmlPage类代码示例

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
public static List<String> searchDuck (String keyword) {
    List<String> searchResults = new ArrayList<>();
    try{
        WebClient webClient = new WebClient(BrowserVersion.CHROME);
        HtmlPage page = webClient.getPage("https://duckduckgo.com/html/?q=" + keyword);
        List<HtmlAnchor> l = page.getByXPath("//a[@class='result__url']");
        for(HtmlAnchor a: l) {
            searchResults.add(a.getHrefAttribute());
        }


    }
    catch(Exception e){
        System.err.println(e);
    }
    return searchResults;
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
private boolean isAlreadyLoadedByAncestor(
                                           final URL url,
                                           HtmlPage page ) {

    WebWindow window = page.getEnclosingWindow();
    while (window != null) {
        if (url.sameFile(window.getEnclosedPage().getWebResponse().getWebRequest().getUrl())) {
            return true;
        }
        if (window == window.getParentWindow()) {
            window = null;
        } else {
            window = window.getParentWindow();
        }
    }
    return false;
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
public String mockLogin(URL url, String username, String password) {
	try {
		HtmlPage page1 = webClient.getPage(url.getURLValue());  
		// find the login form
        HtmlForm form = page1.getForms().get(0);  
        
        // fill in the input
        HtmlInput hi = form.getInputByName("commit");
        HtmlTextInput textField = form.getInputByName("login");  
        HtmlPasswordInput pass = form.getInputByName("password");  
        
        textField.click();
        textField.setValueAttribute(username);  
        pass.click();
        pass.setValueAttribute(password);  
      
        // push the button
        HtmlPage page2 = hi.click();  
        return page2.asXml();

	} catch(IOException ioe)  {
		ioe.printStackTrace();
		return null;
	}
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
/**
	 * Scraping base experiment. To be used as starting point.
	 *
	 * First things first:
	 * <ol>
	 * <li>
	 * Make sure you're able to retrieve the page. Try again with JS disabled if
	 * the client crashes.
	 * </li>
	 * <li>
	 * Make sure you're actually getting what you want, and that there are no
	 * redirect-, proxy-, or login shenanigans or what not. Have a look at the
	 * retrieved text.
	 * </li>
	 * <li>
	 * Also consider disabling CSS to speed things up a bit.
	 * </li>
	 * </ol>
	 */
	@Test
	public void scrapingBaseExperiment() {
		assertOutputDirectoryExists();

		String url = "http://www.google.com";

		try (WebClient client = new WebClient(BrowserVersion.CHROME)) {
//			client.getOptions().setCssEnabled(false);
//			client.getOptions().setJavaScriptEnabled(false);

			try {
				HtmlPage page = client.getPage(url);
				System.out.println("HtmlPage:");
				System.out.println(page.asText());
				System.out.print("\n");
			} catch (IOException ex) {
				System.err.println("WARNING: failed to visit: " + url);
				ex.printStackTrace(System.err);
			}
		}
	}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
private HtmlPage allowAccess(WebClient webClient, HtmlPage allowAccessPage) throws IOException {
	HtmlButton allowAccessButton = (HtmlButton) allowAccessPage.getElementById("submit_approve_access");
	if (allowAccessButton == null) {
		throw new RuntimeException("Cannot find allow access button in html page :\n" + allowAccessPage.asXml());
	}
	webClient.waitForBackgroundJavaScriptStartingBefore(WAIT_DELAY_MS);
	// allowAccessButton.click() does not work because
	// allowAccessButton.isVisible() is false
	// for some reason (click() was working with htmlunit 2.23)
	HtmlPage tokenPage = clickButtonIgnoringVisibility(allowAccessButton);
	return tokenPage;
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
@Test
public void neverDispose() throws Throwable {
    final IOException error = new IOException("to be thrown");

    Disposable disposable = spy(new ThrowDisposable(error));

    @SuppressWarnings("deprecation")
    AsyncResourceDisposer.WorkItem item = disposer.disposeAndWait(disposable).get();

    Set<AsyncResourceDisposer.WorkItem> remaining = disposer.getBacklog();
    assertEquals(1, remaining.size());
    assertThat(remaining.iterator().next(), equalTo(item));
    assertEquals(error, ((Disposable.State.Thrown) item.getLastState()).getCause());

    verify(disposable).dispose();
    assertThat(disposer.getBacklog(), not(IsEmptyCollection.<AsyncResourceDisposer.WorkItem>empty()));

    int itemId = item.getId();
    HtmlPage page = j.createWebClient().goTo(disposer.getUrl());
    page = page.getFormByName("stop-tracking-" + itemId).getInputByName("submit").click();
    assertThat(page.getWebResponse().getContentAsString(), containsString(disposer.getDisplayName())); // Redirected back

    assertThat(disposer.getBacklog(), emptyCollectionOf(AsyncResourceDisposer.WorkItem.class));
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
@Test
public void shouldShowPetIndexPage() throws Exception {
  HtmlPage ownerIndexPage = getPage("/pets");
  assertEquals(13.0, ownerIndexPage.getByXPath("count(//div[@class='card'])").get(0));

  String content = ownerIndexPage.asText();
  assertTrue(content.contains("Leo"));
  assertTrue(content.contains("Owned by: George Franklin"));
  assertTrue(content.contains("Birthday: 2010-09-07"));
  assertTrue(content.contains("Type: cat"));


  List viewEditDeleteLinks = ownerIndexPage.getByXPath("//div[@class='card'][1]//@href");
  assertEquals("/owners/1", ((DomAttr) viewEditDeleteLinks.get(0)).getValue());
  assertEquals("/pets/1", ((DomAttr) viewEditDeleteLinks.get(1)).getValue());
  assertEquals("/pets/1/edit", ((DomAttr) viewEditDeleteLinks.get(2)).getValue());
  assertEquals("/pets/1/delete", ((DomAttr) viewEditDeleteLinks.get(3)).getValue());
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
public void loginIfNecessary() {
	if (isSessionOut()) {
		try {
			RespWithTokenJSON respWithTokenJSON = loginBySpecialPost();
			String token = null;
			if (respWithTokenJSON == null || (token = respWithTokenJSON.getToken()) == null) {
				throw new LoginFailedException("Failed to login " + MP_WEIXIN_LOGIN
						+ "; No expected json with token responsed.");
			}
			HtmlPage homePage = this.webClient.getPage(MP_WEIXIN_BASE + respWithTokenJSON.getRedirect_url());
			setCurrentPage(homePage);
			updateIdentityUnit(token);
		} catch (LoginFailedException e) {
			throw e;
		} catch (Exception ex) {
			LOGGER.error("Request Login Page Error.", ex);
		}
	}
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
@Test
public void testConversation() throws Exception {
    HtmlPage page;
    Iterator<HtmlElement> it;

    page = webClient.getPage(webUrl + "resources/converse/start");
    it = page.getDocumentElement().getHtmlElementsByTagName("p").iterator();
    final String secret = it.next().asText();
    System.out.println(secret);

    page = webClient.getPage(webUrl + "resources/converse/" +
        page.getDocumentElement().getHtmlElementsByTagName("a")
            .iterator().next().getAttribute("href"));
    it = page.getDocumentElement().getHtmlElementsByTagName("p").iterator();
    assertTrue(secret.equals(it.next().asText()));

    page = webClient.getPage(webUrl + "resources/converse/stop");
    it = page.getDocumentElement().getHtmlElementsByTagName("p").iterator();
    assertFalse(secret.equals(it.next().asText()));
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
/**
 * Retrieve a form and submit it making sure the CSRF hidden field is present
 *
 * @throws Exception an error occurs or validation fails.
 */
@Test
public void testFormOk() throws Exception {
    HtmlPage page1 = webClient.getPage(webUrl + "resources/csrf");
    HtmlForm form = (HtmlForm) page1.getDocumentElement().getHtmlElementsByTagName("form").get(0);

    // Check hidden input field
    HtmlElement input = form.getHtmlElementsByTagName("input").get(1);
    assertTrue(input.getAttribute("type").equals("hidden"));
    assertTrue(input.getAttribute("name").equals(CSRF_PARAM));
    assertTrue(input.hasAttribute("value"));        // token

    // Submit form
    HtmlSubmitInput button = (HtmlSubmitInput) form.getHtmlElementsByTagName("input").get(0);
    HtmlPage page2 = button.click();
    Iterator<HtmlElement> it = page2.getDocumentElement().getHtmlElementsByTagName("h1").iterator();
    assertTrue(it.next().asText().contains("CSRF Protection OK"));
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
/**
 * Retrieves a form, removes CSRF hidden field and attempts to submit. Should
 * result in a 403 error.
 *
 * @throws Exception an error occurs or validation fails.
 */
@Test
public void testFormFail() throws Exception {
    HtmlPage page1 = webClient.getPage(webUrl + "resources/csrf");
    HtmlForm form = (HtmlForm) page1.getDocumentElement().getHtmlElementsByTagName("form").get(0);

    // Remove hidden input field to cause a CSRF validation failure
    HtmlElement input = form.getHtmlElementsByTagName("input").get(1);
    form.removeChild(input);

    // Submit form - should fail
    HtmlSubmitInput button = (HtmlSubmitInput) form.getHtmlElementsByTagName("input").get(0);
    try {
        button.click();
        fail("CSRF validation should have failed!");
    } catch (FailingHttpStatusCodeException e) {
        // falls through
    }
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
/**
 * Checks that CSRF validation works if token sent as header instead of
 * form field.
 *
 * @throws Exception an error occurs or validation fails.
 */
@Test
public void testFormHeaderOk() throws Exception {
    HtmlPage page1 = webClient.getPage(webUrl + "resources/csrf");

    // Check response and CSRF header
    WebResponse res = page1.getWebResponse();
    assertEquals(Response.Status.OK.getStatusCode(), res.getStatusCode());
    assertNotNull(res.getResponseHeaderValue(CSRF_HEADER));

    WebRequest req = new WebRequest(new URL(webUrl + "resources/csrf"));
    req.setHttpMethod(HttpMethod.POST);
    req.setAdditionalHeader(CSRF_HEADER, res.getResponseHeaderValue(CSRF_HEADER));
    res = webClient.loadWebResponse(req);
    assertEquals(Response.Status.OK.getStatusCode(), res.getStatusCode());
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
public static boolean setAttribute(HtmlPage htmlPage, String xpath,
		String key, String value) {
	boolean flag = true;
	try {
		HtmlElement element = htmlPage.getFirstByXPath(xpath);
		if (element != null) {
			element.click();
			element.setAttribute(key, "");
			element.setAttribute(key, value);
		} else {
			logger.error("setAttribute element is null xpath " + xpath);
		}

	} catch (Exception e) {
		logger.error(e.getMessage(), e);
		flag = false;
	}
	return flag;
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
private Map<String, String> getEverything(final Account a, final HtmlPage page) {
    Map<String, String> values = new HashMap<>();

    try {
        AccountXpaths xpaths = a.getAccountXpaths();

        values.put("messages", xpathUtils.checkXpathListForString(xpaths.getMessages(), page));
        values.put("alerts", xpathUtils.checkXpathListForString(xpaths.getAlerts(), page));
        values.put("posts", xpathUtils.checkXpathListForString(xpaths.getPosts(), page));
        values.put("ratings", xpathUtils.checkXpathListForString(xpaths.getRatings(), page));

        return values;
    } catch (Exception e) {
        e.printStackTrace();

    }
    return values;
}
 

import com.gargoylesoftware.htmlunit.html.HtmlPage; //导入依赖的package包/类
@Test
public void testGetWithCorrectCredentials() throws Exception {
	logger.info("start get request with correct credentials");
	loginForm.getInputByName("j_username").setValueAttribute("u1");
	loginForm.getInputByName("j_password").setValueAttribute("p1");
	HtmlSubmitInput submitButton = loginForm.getInputByName("submitButton");
	HtmlPage page2 = submitButton.click();
	assertEquals("Form-based Security - Success", page2.getTitleText());
}