@Test
public void testHashAndEqualsAndValue() {
// .equals and hashcode are identity. value is content.
String doc1 =
"<div id=1><p class=one>One</p><p class=one>One</p><p class=one>Two</p><p class=two>One</p></div>"
+ "<div id=2><p class=one>One</p><p class=one>One</p><p class=one>Two</p><p class=two>One</p></div>";
Document doc = Jsoup.parse(doc1);
Elements els = doc.select("p");
/*
for (Element el : els) {
System.out.println(el.hashCode() + " - " + el.outerHtml());
}
0 1534787905 - <p class="one">One</p>
1 1534787905 - <p class="one">One</p>
2 1539683239 - <p class="one">Two</p>
3 1535455211 - <p class="two">One</p>
4 1534787905 - <p class="one">One</p>
5 1534787905 - <p class="one">One</p>
6 1539683239 - <p class="one">Two</p>
7 1535455211 - <p class="two">One</p>
*/
assertEquals(8, els.size());
Element e0 = els.get(0);
Element e1 = els.get(1);
Element e2 = els.get(2);
Element e3 = els.get(3);
Element e4 = els.get(4);
Element e5 = els.get(5);
Element e6 = els.get(6);
Element e7 = els.get(7);
assertEquals(e0, e0);
assertTrue(e0.hasSameValue(e1));
assertTrue(e0.hasSameValue(e4));
assertTrue(e0.hasSameValue(e5));
assertFalse(e0.equals(e2));
assertFalse(e0.hasSameValue(e2));
assertFalse(e0.hasSameValue(e3));
assertFalse(e0.hasSameValue(e6));
assertFalse(e0.hasSameValue(e7));
assertEquals(e0.hashCode(), e0.hashCode());
assertFalse(e0.hashCode() == (e2.hashCode()));
assertFalse(e0.hashCode() == (e3).hashCode());
assertFalse(e0.hashCode() == (e6).hashCode());
assertFalse(e0.hashCode() == (e7).hashCode());
}
private void getChildElement(Element parentElement, Integer level) {
parentElement.html(deleteComent(parentElement.html()));
// System.out.println("key:"+(level+","+parentElement.hashCode())+",value:"+parentElement.html());
if (parentElement.children().size() > 0) {
level += 1;
for (int i = 0; i < parentElement.children().size(); i++) {
if (("ul".equals(parentElement.tagName().toLowerCase()))
|| ("table".equals(parentElement.tagName().toLowerCase()))) { // 整体标签
String html = parentElement.html().replaceAll(" ", "").replaceAll(" ", ""); // 去中英文空格
if (html.contains("首页") || parentElement.id().contains("nav")) {
//
// System.out.println("----------------------首页Start-----------------------------");
Elements links = parentElement.select("a");
for (Element ele : links) {
if (topMenumap.get(level + "," + ele.hashCode()) == null) {
topMenumap.put(level + "," + ele.hashCode(), ele);
// System.out.println(level + "," + ele.hashCode() + ",---------------" +
// ele.html());
// System.out.println("a:" +
// ele.attr("abs:href") + ",文本:" + ele.text());
}
}
// System.out.println("----------------------首页End-----------------------------");
} else {
map.put(level + "," + parentElement.hashCode(), parentElement);
}
} else {
getChildElement(parentElement.child(i), level);
}
}
} else {
if ("script".equals(parentElement.tagName().toLowerCase())) {
return;
}
if (StringUtils.isNotEmpty(parentElement.html())) {
level += 1;
map.put(level + "," + parentElement.hashCode(), parentElement);
}
}
}
