public ContentModel view(String url) {
ContentModel model = new ContentModel();
try {
NodeFilter filter = new TagNameFilter("html");
Parser parser = new Parser();
parser.setURL(SearchHelper.decrypt(url));
parser.setEncoding(parser.getEncoding());
// parser.setEncoding("gb2312");
NodeList list = parser.extractAllNodesThatMatch(filter);
for (int i = 0; i < list.size(); i++) {
String s = list.elementAt(i).toHtml();
model.setContent(s);
}
} catch (Exception e) {
e.printStackTrace();
}
return model;
}
public ContentModel listHtml(String param, String type) {
ContentModel model = new ContentModel();
StringBuffer html = new StringBuffer();
try {
NodeFilter filter = new TagNameFilter("body");
Parser parser = new Parser();
parser.setURL(SearchHelper.SEARCH_URL_BAIDU + param);
parser.setEncoding(parser.getEncoding());
NodeList list = parser.extractAllNodesThatMatch(filter);
String body = list.toHtml();
Parser content = new Parser();
content.setInputHTML(body);
content.setEncoding(parser.getEncoding());
NodeFilter content_filter = new TagNameFilter("table");
NodeList content_list = content.extractAllNodesThatMatch(content_filter);
for (int i = 0; i < content_list.size(); i++) {
String s = content_list.elementAt(i).toHtml();
if (s.indexOf("div") != -1) {
continue;
}
if (s.indexOf("相关搜索") != -1) {
html.append("<div id=\"rs\">" + s + "</div>");
continue;
}
html.append("<div class=\"content\">");
for (Node n : extractHtml(content_list.elementAt(i), type)) {
if (n instanceof LinkTag) {
if (n.toPlainTextString().equals("百度快照")) {
continue;
}
html.append("<h3 class=\"t\">" + n.toHtml() + "</h3>");
} else {
html.append(n.toHtml());
}
}
html.append("<br/></div><br>");
}
/** 获取分页数据 */
Parser page = new Parser();
page.setInputHTML(body);
page.setEncoding(parser.getEncoding());
NodeFilter page_filter = new TagNameFilter("p");
NodeList page_list = page.extractAllNodesThatMatch(page_filter);
for (int i = 0; i < page_list.size(); i++) {
String s = page_list.elementAt(i).toHtml();
if (s.indexOf("page") == -1) {
continue;
}
html.append("<p id=\"page\">" + page_list.elementAt(i).toHtml() + "</div>");
}
} catch (Exception e) {
e.printStackTrace();
}
model.setContent(html.toString());
return model;
}
