Examples of org.jsoup.nodes.Document.title()

org.jsoup.nodes.Document.title()
Get the string contents of the document's {@code title} element. @return Trimed title, or empty string if none set.

              recordContent = recordContent.replaceAll(replace.getMatcher(),
                  replace.getReplacement());
            }
          }
          String sourceId = source.getId().toString();
          String title = document.title();
          return new Record(title, url, recordContent, sourceId);
        }
      }
      EntityUtils.consume(entity);
      return null;

View Full Code Here

    public void testSmhBizArticle() throws IOException {
        File in = getFile("/htmltests/smh-biz-article-1.html");
        Document doc = Jsoup.parse(in, "UTF-8",
            "http://www.smh.com.au/business/the-boards-next-fear-the-female-quota-20100106-lteq.html");
        assertEquals("The board’s next fear: the female quota",
            doc.title()); // note that the apos in the source is a literal ’ (8217), not escaped or '
        assertEquals("en", doc.select("html").attr("xml:lang"));


        Elements articleBody = doc.select(".articleBody > *");
        assertEquals(17, articleBody.size());
        // todo: more tests!

View Full Code Here


    @Test
    public void testNewsHomepage() throws IOException {
        File in = getFile("/htmltests/news-com-au-home.html");
        Document doc = Jsoup.parse(in, "UTF-8", "http://www.news.com.au/");
        assertEquals("News.com.au | News from Australia and around the world online | NewsComAu", doc.title());
        assertEquals("Brace yourself for Metro meltdown", doc.select(".id1225817868581 h4").text().trim());


        Element a = doc.select("a[href=/entertainment/horoscopes]").first();
        assertEquals("/entertainment/horoscopes", a.attr("href"));
        assertEquals("http://www.news.com.au/entertainment/horoscopes", a.attr("abs:href"));

View Full Code Here


    @Test
    public void testGoogleSearchIpod() throws IOException {
        File in = getFile("/htmltests/google-ipod.html");
        Document doc = Jsoup.parse(in, "UTF-8", "http://www.google.com/search?hl=en&q=ipod&aq=f&oq=&aqi=g10");
        assertEquals("ipod - Google Search", doc.title());
        Elements results = doc.select("h3.r > a");
        assertEquals(12, results.size());
        assertEquals(
            "http://news.google.com/news?hl=en&q=ipod&um=1&ie=UTF-8&ei=uYlKS4SbBoGg6gPf-5XXCw&sa=X&oi=news_group&ct=title&resnum=1&ved=0CCIQsQQwAA",
            results.get(0).attr("href"));

View Full Code Here


    @Test
    public void testYahooJp() throws IOException {
        File in = getFile("/htmltests/yahoo-jp.html");
        Document doc = Jsoup.parse(in, "UTF-8", "http://www.yahoo.co.jp/index.html"); // http charset is utf-8.
        assertEquals("Yahoo! JAPAN", doc.title());
        Element a = doc.select("a[href=t/2322m2]").first();
        assertEquals("http://www.yahoo.co.jp/_ylh=X3oDMTB0NWxnaGxsBF9TAzIwNzcyOTYyNjUEdGlkAzEyBHRtcGwDZ2Ex/t/2322m2",
            a.attr("abs:href")); // session put into <base>
        assertEquals("全国、人気の駅ランキング", a.text());
    }

View Full Code Here


    @Test
    public void fetchURl() throws IOException {
        String url = "http://www.google.com"; // no trailing / to force redir
        Document doc = Jsoup.parse(new URL(url), 10*1000);
        assertTrue(doc.title().contains("Google"));
    }


    @Test
    public void fetchBaidu() throws IOException {
        Connection.Response res = Jsoup.connect("http://www.baidu.com/").timeout(10*1000).execute();

View Full Code Here

    }


    @Test
    public void ignoresContentTypeIfSoConfigured() throws IOException {
        Document doc = Jsoup.connect("http://jsoup.org/rez/osi_logo.png").ignoreContentType(true).get();
        assertEquals("", doc.title()); // this will cause an ugly parse tree
    }


    @Test
    public void doesPost() throws IOException {
        Document doc = Jsoup.connect(echoURL)

View Full Code Here


    @Test
    public void followsTempRedirect() throws IOException {
        Connection con = Jsoup.connect("http://direct.infohound.net/tools/302.pl"); // http://jsoup.org
        Document doc = con.get();
        assertTrue(doc.title().contains("jsoup"));
    }


    @Test
    public void postRedirectsFetchWithGet() throws IOException {
        Connection con = Jsoup.connect("http://direct.infohound.net/tools/302.pl")

View Full Code Here

    @Test
    public void followsRedirectToHttps() throws IOException {
        Connection con = Jsoup.connect("http://direct.infohound.net/tools/302-secure.pl"); // https://www.google.com
        con.data("id", "5");
        Document doc = con.get();
        assertTrue(doc.title().contains("Google"));
    }


    @Test
    public void followsRelativeRedirect() throws IOException {
        Connection con = Jsoup.connect("http://direct.infohound.net/tools/302-rel.pl"); // to ./ - /tools/

View Full Code Here


    @Test
    public void followsRelativeRedirect() throws IOException {
        Connection con = Jsoup.connect("http://direct.infohound.net/tools/302-rel.pl"); // to ./ - /tools/
        Document doc = con.post();
        assertTrue(doc.title().contains("HTML Tidy Online"));
    }


    @Test
    public void throwsExceptionOnError() {
        String url = "http://direct.infohound.net/tools/404";

View Full Code Here

0 1 2

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.