protected synchronized boolean isHavingUrl(String url) {
boolean isHaving = false;
// 判断该URL是否这次任务中被抓取
if (URLCache.findUrlInCache(url)) // 存在为true
return true;
// 判断该URL是否曾已被抓取
if (storage.getSinfoDAO().isCrawlUrl(url)) return true;
Statement stmt = null;
String sql = "select top 1 url from spider_resource where url = '" + url + "'";
ResultSet rs = null;
try {
stmt = daoHelp.getConnection().createStatement();
rs = stmt.executeQuery(sql);
while (rs.next()) {
isHaving = true;
}
daoHelp.safeClose(rs);
daoHelp.safeClose(stmt);
} catch (SQLException e) {
log.error("SQLException -> ", e);
daoHelp.safeClose(rs);
daoHelp.safeClose(stmt);
} finally {
daoHelp.safeClose(rs);
daoHelp.safeClose(stmt);
}
return isHaving;
}
protected Resource createResourceFromRecord(ResultSet rs) throws SQLException {
Resource res = this.createResource(rs);
ResInfo resInfo = storage.getResourceInfoDAO().findById(res.getRId());
Folder folder = storage.getFolderDAO().findById(res.getRId());
Sinfo sinfo = storage.getSinfoDAO().findById(res.getRId());
res.setFolder(folder);
res.setResInfo(resInfo);
res.setSinfo(sinfo);
return res;
}