private PhoneResult mineAgentList(SearchResultItem item) {
if (!isMinableAgentList(item)) return null;
String html = getHtml(item.getRealUrl());
if (TextUtils.isEmpty(html) || !html.contains(item.getPhone())) return null;
List<String> list = TextUtils.getList(html, "<div id='list_", "<div class=\"clear\">");
if (list != null && list.size() > 0) {
for (String l : list) {
if (l.contains(item.getPhone())) {
String agent_href = TextUtils.getSubString(l, "<a href='", "'");
if (!TextUtils.isEmpty(agent_href) && agent_href.startsWith("/")) {
String domain = TextUtils.getMatchGroup(PATTERN_AGENT_LIST, item.getRealUrl());
if (!TextUtils.isEmpty(domain)) {
String aLink = domain + agent_href;
return mineAgent(aLink);
}
}
break;
}
}
}
return null;
}
private boolean isMinableAgentList(SearchResultItem item) {
return TextUtils.isMatchReg(PATTERN_AGENT_LIST, item.getRealUrl());
}
private PhoneResult mineAgent(String url) {
if (!isMinableAgent(url)) return null;
PhoneResult phoneResult = new PhoneResult();
String html = getHtml(url);
if (TextUtils.isEmpty(html)) return null;
String floatHtml = TextUtils.getSubString(html, "<div class=\"Floating\">", "</div>");
if (!TextUtils.isEmpty(floatHtml)) {
String TAG_COM = "<dd style=\"padding-top:2px\\9;*padding-top:3px;\" >";
String company = TextUtils.getSubString(floatHtml, TAG_COM, "</dd>");
phoneResult.setJigou(company);
String name = TextUtils.getSubString(floatHtml, "<dd>", " ");
if (!TextUtils.isEmpty(name)) phoneResult.setChenghu(TextUtils.appendJob(name, "房产经纪"));
String rz = TextUtils.getSubString(html, "<div class=\"rzren\">", "</div>");
if (!TextUtils.isEmpty(rz)) {
String avartar = TextUtils.getSubString(rz, "<img src=\"", "\"");
phoneResult.setImage(avartar);
}
String address = TextUtils.getSubString(html, "<li>地<span class=\"pl24\">址</span>：", "</li>");
phoneResult.setAddress(address);
List<String> hangyeList = new ArrayList<String>();
hangyeList.add("房屋中介");
phoneResult.setHangyeList(hangyeList);
}
// 如果没有人头像，就试着用公司logo
if (TextUtils.isEmpty(phoneResult.getImage())) {
String componyLogoHtml = TextUtils.getSubString(html, "<li class=\"companylogo\">", "</li>");
String componyLogo = TextUtils.getSubString(componyLogoHtml, "<img src=\"", "\"");
phoneResult.setImage(componyLogo);
}
// 如果公司头像也没有，可能网页是另外一种格式的，如：
if (TextUtils.isEmpty(phoneResult.getImage())) {
String photoHtml = TextUtils.getSubString(html, "<div class=\"photo\">", "</div>");
if (!TextUtils.isEmpty(photoHtml)) {
String avartar = TextUtils.getSubString(photoHtml, "<img src=\"", "\"");
phoneResult.setImage(avartar);
}
}
if (phoneResult.isFound()) return phoneResult;
else return null;
}
@Override
public boolean isMinableDomain(SearchResultItem item) {
return TextUtils.isMatchReg(PATTERN_DOMAIN, item.getDomain());
}
private boolean isMinableFang(SearchResultItem item) {
return TextUtils.isMatchReg(PATTERN_FANG, item.getRealUrl());
}
private PhoneResult mineFang(SearchResultItem item) {
String html = getHtml(item);
if (TextUtils.isEmpty(html)) return null;
String link = TextUtils.getMatchGroup(PATTERN_AGENT_LINK, html);
System.out.println("link=" + link);
if (!TextUtils.isEmpty(link)) {
return mineAgent(link);
}
if (html.contains("100%个人房源")) {
PhoneResult phoneResult = new PhoneResult();
String chenghu =
TextUtils.getSubString(html, "<span class=\"name floatl\" id=\"Span2\">", "</span>");
if (!TextUtils.isEmpty(chenghu)) {
chenghu = chenghu.trim();
chenghu = TextUtils.clearHuanhang(chenghu);
if (!TextUtils.isEmpty(chenghu)) phoneResult.setChenghu(TextUtils.appendJob(chenghu, "房东"));
List<String> hangyeList = new ArrayList<String>();
hangyeList.add("房屋出租或出售");
phoneResult.setHangyeList(hangyeList);
}
// 房产图片
String imageHtml =
TextUtils.getSubString(html, "<div class=\"slider\" id=\"thumbnail\">", "</div>");
if (!TextUtils.isEmpty(imageHtml)) {
String avartar = TextUtils.getSubString(imageHtml, "<img src=\"", "\"");
phoneResult.setImage(avartar);
}
if (phoneResult.isFound()) return phoneResult;
}
return null;
}
private boolean isMinableAgent(String url) {
return TextUtils.isMatchReg(PATTERN_AGENT, url)
|| TextUtils.isMatchReg(PATTERN_AGENT_SUBPAGE, url);
}