Examples with Document org.jsoup.nodes.Document used on opensource projects

Example 11 with Document

use of org.jsoup.nodes.Document in project jsoup by jhy.

the class ParseTest method testBaiduVariant.

@Test
public void testBaiduVariant() throws IOException {
    // tests <meta charset> when preceded by another <meta>
    File in = getFile("/htmltests/baidu-variant.html");
    Document doc = Jsoup.parse(in, null, // http charset is gb2312, but NOT specifying it, to test http-equiv parse
    "http://www.baidu.com/");
    // check auto-detect from meta
    assertEquals("GB2312", doc.outputSettings().charset().displayName());
    assertEquals("<title>百度一下，你就知道</title>", doc.select("title").outerHtml());
}

Example 12 with Document

use of org.jsoup.nodes.Document in project jsoup by jhy.

the class UrlConnectTest method sendHeadRequest.

@Test
public void sendHeadRequest() throws IOException {
    String url = "http://direct.infohound.net/tools/parse-xml.xml";
    Connection con = Jsoup.connect(url).method(Connection.Method.HEAD);
    final Connection.Response response = con.execute();
    assertEquals("text/xml", response.header("Content-Type"));
    // head ought to have no body
    assertEquals("", response.body());
    Document doc = response.parse();
    assertEquals("", doc.text());
}

Example 13 with Document

use of org.jsoup.nodes.Document in project jsoup by jhy.

the class UrlConnectTest method ignores500tExceptionIfSoConfigured.

@Test
public void ignores500tExceptionIfSoConfigured() throws IOException {
    Connection con = Jsoup.connect("http://direct.infohound.net/tools/500.pl").ignoreHttpErrors(true);
    Connection.Response res = con.execute();
    Document doc = res.parse();
    assertEquals(500, res.statusCode());
    assertEquals("Application Error", res.statusMessage());
    assertEquals("Woops", doc.select("h1").first().text());
}

Example 14 with Document

use of org.jsoup.nodes.Document in project jsoup by jhy.

the class UrlConnectTest method fetchToW3c.

@Test
public void fetchToW3c() throws IOException {
    String url = "https://jsoup.org";
    Document doc = Jsoup.connect(url).get();
    W3CDom dom = new W3CDom();
    org.w3c.dom.Document wDoc = dom.fromJsoup(doc);
    assertEquals(url, wDoc.getDocumentURI());
    String html = dom.asString(wDoc);
    assertTrue(html.contains("jsoup"));
}

Example 15 with Document

use of org.jsoup.nodes.Document in project jsoup by jhy.

the class UrlConnectTest method baseHrefCorrectAfterHttpEquiv.

@Test
public void baseHrefCorrectAfterHttpEquiv() throws IOException {
    // https://github.com/jhy/jsoup/issues/440
    Connection.Response res = Jsoup.connect("http://direct.infohound.net/tools/charset-base.html").execute();
    Document doc = res.parse();
    assertEquals("http://example.com/foo.jpg", doc.select("img").first().absUrl("src"));
}