public void AddDbAddress(Entity.AddressBase address)
{
AddressBase result = Utility.EFContextFactory.GetCurrentDbContext().Set <AddressBase>().Find(address.Code);
if (result == null)
{
DataRepository.DB.Set <AddressBase>().Add(address);
}
}
public int AddAddress(Entity.AddressBase address)
{
AddressBase result = Utility.EFContextFactory.GetCurrentDbContext().Set <AddressBase>().Find(address.Code);
if (result == null)
{
return(DataRepository.Add <AddressBase>(address));
}
return(-1);
}
private static int GetAddress(string url)
{
string html = HttpHelper.Get(url);
if (html == "err")
{
System.Threading.Thread.Sleep(1000);
//GetHtml(url);
html = HttpHelper.Get(url);
if (html == "err")
{
throw new Exception();
}
}
if (url.IndexOf("Html/Address") > -1)
{
url = "http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2016" + url.Substring(url.LastIndexOf("/"));
}
Regex reg = new Regex("<tr class='(.*?)'>(.*?)</tr>", RegexOptions.IgnoreCase);
MatchCollection matches = reg.Matches(html);
foreach (Match match in matches)
{
string entityCode = url.Substring(url.LastIndexOf('/') + 1, url.Length - url.LastIndexOf('/') - 6).PadRight(12, '0');
Regex hrefReg = new Regex("<a href='(.*?)'>");
Regex phrefReg = new Regex("<a href='(.*?)'>(.*?)</a>");
MatchCollection hrefs = phrefReg.Matches(match.Value);
if (match.Value.IndexOf(EnumEntity.AddressType.provincetr.ToString()) > -1)
{
foreach (Match hrefm in hrefs)
{
//(obj as List<string>).Add(hrefm.Value);
AddressBase address = new Entity.AddressBase();
string[] hrefarr = hrefm.Value.Replace("<a href='", "").Replace("'>", ";").Replace("<br/></a>", "").Split(';');
address.Code = hrefarr[0].Substring(hrefarr[0].LastIndexOf('/') + 1, hrefarr[0].Length - hrefarr[0].LastIndexOf('/') - 6).PadRight(12, '0');
address.Text = hrefarr[1];
address.Type = "provincetr";
addressLogic.AddDbAddress(address);
if (!string.IsNullOrWhiteSpace(hrefm.Value))
{
GetAddress(url.Substring(0, url.LastIndexOf('/') + 1) + hrefarr[0].Substring(0, 2) + ".html");
}
}
}
else
{
if (hrefs.Count != 0)
{
string type = match.Value.Substring("<tr class='".Length, match.Value.IndexOf("'>") - "<tr class='".Length);
string[] hrefarr = new string[2];
hrefarr[0] = hrefs[0].Value.Replace("<a href='", "").Replace("'>", ";").Replace("</a>", "").Split(';')[1];
hrefarr[1] = hrefs[1].Value.Replace("<a href='", "").Replace("'>", ";").Replace("</a>", "").Split(';')[1];
Match hmatch = hrefReg.Match(match.Value);
AddressBase address = new Entity.AddressBase();
address.Code = hrefarr[0];
address.Text = hrefarr[1];
address.Type = type;
addressLogic.AddDbAddress(address);
if (!string.IsNullOrWhiteSpace(hmatch.Value) && type != EnumEntity.AddressType.towntr.ToString())
{
GetAddress(url.Substring(0, url.LastIndexOf('/') + 1) + hmatch.Value.Replace("<a href='", "").ToString().Replace("'>", ""));
}
}
}
}
int result = addressLogic.SaveDbAddress();
count += result;
Log4net.LogHelper.WriteLog("地址统计:" + count.ToString(), url + ";共计:" + result.ToString());
return(count);
}
