Example 6 with Document
use of org.jsoup.nodes.Document in project jsoup by jhy.
the class ParseTest method testNewsHomepage.
@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"));
Element hs = doc.select("a[href*=naughty-corners-are-a-bad-idea]").first();
assertEquals("http://www.heraldsun.com.au/news/naughty-corners-are-a-bad-idea-for-kids/story-e6frf7jo-1225817899003", hs.attr("href"));
assertEquals(hs.attr("href"), hs.attr("abs:href"));
}Example 7 with Document
use of org.jsoup.nodes.Document in project jsoup by jhy.
the class ParseTest method testBinary.
@Test
public void testBinary() throws IOException {
File in = getFile("/htmltests/thumb.jpg");
Document doc = Jsoup.parse(in, "UTF-8");
// nothing useful, but did not blow up
assertTrue(doc.text().contains("gd-jpeg"));
}Example 8 with Document
use of org.jsoup.nodes.Document in project jsoup by jhy.
the class ParseTest method testNytArticle.
@Test
public void testNytArticle() throws IOException {
// has tags like <nyt_text>
File in = getFile("/htmltests/nyt-article-1.html");
Document doc = Jsoup.parse(in, null, "http://www.nytimes.com/2010/07/26/business/global/26bp.html?hp");
Element headline = doc.select("nyt_headline[version=1.0]").first();
assertEquals("As BP Lays Out Future, It Will Not Include Hayward", headline.text());
}Example 9 with Document
use of org.jsoup.nodes.Document in project jsoup by jhy.
the class ParseTest method testBrokenHtml5CharsetWithASingleDoubleQuote.
@Test
public void testBrokenHtml5CharsetWithASingleDoubleQuote() throws IOException {
InputStream in = inputStreamFrom("<html>\n" + "<head><meta charset=UTF-8\"></head>\n" + "<body></body>\n" + "</html>");
Document doc = Jsoup.parse(in, null, "http://example.com/");
assertEquals("UTF-8", doc.outputSettings().charset().displayName());
}Example 10 with Document
use of org.jsoup.nodes.Document in project jsoup by jhy.
the class ParseTest method testGoogleSearchIpod.
@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"));
assertEquals("http://www.apple.com/itunes/", results.get(1).attr("href"));
}Aggregations
Document (org.jsoup.nodes.Document)391
Test (org.junit.Test)194
Element (org.jsoup.nodes.Element)153
IOException (java.io.IOException)100
File (java.io.File)81
Elements (org.jsoup.select.Elements)70
ElementHandlerImpl (org.asqatasun.ruleimplementation.ElementHandlerImpl)51
Connection (org.jsoup.Connection)37
ArrayList (java.util.ArrayList)36
URL (java.net.URL)24
HashMap (java.util.HashMap)16
InputStream (java.io.InputStream)13
List (java.util.List)9
MalformedURLException (java.net.MalformedURLException)8
Matcher (java.util.regex.Matcher)7
Logger (org.slf4j.Logger)7
Pattern (java.util.regex.Pattern)6
HttpGet (org.apache.http.client.methods.HttpGet)6
Jsoup (org.jsoup.Jsoup)6
LoggerFactory (org.slf4j.LoggerFactory)6
