您的位置：首页 > 编程语言 > Java开发

Java Code Examples for org.jsoup.select.Elements

2014-06-09 17:20 525 查看

http://www.programcreek.com/java-api-examples/index.php?api=org.jsoup.select.Elements

Code Example 1:

1

From project jsoup, under directory /src/test/java/org/jsoup/integration/.

Source ParseTest.java

@Test
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!

}

Code Example 2:

1

From project jsoup, under directory /src/test/java/org/jsoup/integration/.

Source ParseTest.java

@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"));
}

Code Example 3:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source AttributeParseTest.java

@Test public void strictAttributeUnescapes() {
String html = "<a id=1 href='?foo=bar&mid<=true'>One</a> <a id=2 href='?foo=bar<qux&lg=1'>Two</a>";
Elements els = Jsoup.parse(html).select("a");
assertEquals("?foo=bar&mid<=true", els.first().attr("href"));
assertEquals("?foo=bar<qux&lg=1", els.last().attr("href"));
}

Code Example 4:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void parsesUnterminatedOption() {
// bit weird this -- browsers and spec get stuck in select until there's a </select>
Document doc = Jsoup.parse("<body><p><select><option>One<option>Two</p><p>Three</p>");
Elements options = doc.select("option");
assertEquals(2, options.size());
assertEquals("One", options.first().text());
assertEquals("TwoThree", options.last().text());
}

Code Example 5:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void handlesTextArea() {
Document doc = Jsoup.parse("<textarea>Hello</textarea>");
Elements els = doc.select("textarea");
assertEquals("Hello", els.text());
assertEquals("Hello", els.val());
}

Code Example 6:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void doesNotCreateImplicitLists() {
// old jsoup used to wrap this in <ul>, but that's not to spec
String h = "<li>Point one<li>Point two";
Document doc = Jsoup.parse(h);
Elements ol = doc.select("ul"); // should NOT have created a default ul.
assertEquals(0, ol.size());
Elements lis = doc.select("li");
assertEquals(2, lis.size());
assertEquals("body", lis.first().parent().tagName());

// no fiddling with non-implicit lists
String h2 = "<ol><li><p>Point the first<li><p>Point the second";
Document doc2 = Jsoup.parse(h2);

assertEquals(0, doc2.select("ul").size());
assertEquals(1, doc2.select("ol").size());
assertEquals(2, doc2.select("ol li").size());
assertEquals(2, doc2.select("ol li p").size());
assertEquals(1, doc2.select("ol li").get(0).children().size()); // one p in first li
}

Code Example 7:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void handlesBaseTags() {
// only listen to the first base href
String h = "<a href=1>#</a><base href='/2/'><a href='3'>#</a><base href='http://bar'><a href=/4>#</a>";
Document doc = Jsoup.parse(h, "http://foo/");
assertEquals("http://foo/2/", doc.baseUri()); // gets set once, so doc and descendants have first only

Elements anchors = doc.getElementsByTag("a");
assertEquals(3, anchors.size());

assertEquals("http://foo/2/", anchors.get(0).baseUri());
assertEquals("http://foo/2/", anchors.get(1).baseUri());
assertEquals("http://foo/2/", anchors.get(2).baseUri());

assertEquals("http://foo/2/1", anchors.get(0).absUrl("href"));
assertEquals("http://foo/2/3", anchors.get(1).absUrl("href"));
assertEquals("http://foo/4", anchors.get(2).absUrl("href"));
}

Code Example 8:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void handlesUnknownTags() {
String h = "<div><foo title=bar>Hello<foo title=qux>there</foo></div>";
Document doc = Jsoup.parse(h);
Elements foos = doc.select("foo");
assertEquals(2, foos.size());
assertEquals("bar", foos.first().attr("title"));
assertEquals("qux", foos.last().attr("title"));
assertEquals("there", foos.last().text());
}

Code Example 9:

1

From project jsoup, under directory /src/test/java/org/jsoup/parser/.

Source HtmlParserTest.java

@Test public void handlesUnclosedDefinitionLists() {
// jsoup used to create a <dl>, but that's not to spec
String h = "<dt>Foo<dd>Bar<dt>Qux<dd>Zug";
Document doc = Jsoup.parse(h);
assertEquals(0, doc.select("dl").size()); // no auto dl
assertEquals(4, doc.select("dt, dd").size());
Elements dts = doc.select("dt");
assertEquals(2, dts.size());
assertEquals("Zug", dts.get(1).nextElementSibling().text());
}

Code Example 10:

1