Exemple #1
0
  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;
  }
Exemple #2
0
  public static void main(String[] args) {
    SearchResultItem item = new SearchResultItem();
    item.setRealUrl("http://esf.wuhan.fang.com/a/long878158?s=BingXML");
    item.setPhone("18707151870");
    item.setDomain("esf.wuhan.fang.com");

    DomainFang domainFang = new DomainFang();

    System.out.println(domainFang.mine(item));
  }
Exemple #3
0
 @Override
 public PhoneResult mine(SearchResultItem item) {
   String url = item.getRealUrl();
   if (isMinableAgent(item)) {
     return mineAgent(url);
   } else if (isMinableFang(item)) {
     return mineFang(item);
   } else if (isMinableAgentList(item)) {
     return mineAgentList(item);
   }
   return null;
 }
Exemple #4
0
 private boolean isMinableAgentList(SearchResultItem item) {
   return TextUtils.isMatchReg(PATTERN_AGENT_LIST, item.getRealUrl());
 }
Exemple #5
0
 private boolean isMinableAgent(SearchResultItem item) {
   return isMinableAgent(item.getRealUrl());
 }
Exemple #6
0
 @Override
 public boolean isMinableDomain(SearchResultItem item) {
   return TextUtils.isMatchReg(PATTERN_DOMAIN, item.getDomain());
 }
Exemple #7
0
 private boolean isMinableFang(SearchResultItem item) {
   return TextUtils.isMatchReg(PATTERN_FANG, item.getRealUrl());
 }