public void InsertTest()
{
Job param = new Job(); // TODO: 初始化为适当的值
param.title = "test";
param.poster_email = "*****@*****.**";
JobMap.Insert(param);
Assert.Inconclusive("无法验证不返回值的方法。");
}
public Job GetJobInfoParser(string url)
{
Job jobinfo = new Job();
string title = string.Empty;
string description = string.Empty;
DateTime dt = DateTime.Now;
string email = string.Empty;
Parser parser = new Parser(new HttpProtocol(new Uri(url)));
NodeFilter detail = new HasAttributeFilter("class", "d_left");
NodeList nodeDetail = parser.ExtractAllNodesThatMatch(detail);
if (nodeDetail == null || nodeDetail.Count == 0)
{
return jobinfo;
}
description = GetDetailString(nodeDetail);
Match m = Regex.Match(description, @"发布时间：(?<date>\d\d\d\d-\d{1,2}\-\d{1,2} \d{1,2}\:\d{1,2})");
dt = DateTime.Now;
if (m.Success && m.Groups["date"].Success && DateTime.TryParse(m.Groups["date"].Value, out dt)) { }
Match emailMatch = Regex.Match(description, @"([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)");
if (emailMatch.Success)
{
email = emailMatch.Value;
}
Match telMatch = Regex.Match(description, @"(1[3|5|8][0-9]|15[0|3|6|7|8|9]|18[8|9])\d{8}");
if (telMatch.Success)
{
jobinfo.tel = telMatch.Value;
}
jobinfo.category_id = Catalog.id;
jobinfo.title = title;
jobinfo.description = description;
jobinfo.created_on = dt;
jobinfo.is_active = true;
jobinfo.city_id = Catalog.city.id;
jobinfo.sp1010url = url;
jobinfo.poster_email = email;
return jobinfo;
}
public void GetContartInfo(NodeList list, ref Job info)
{
string miaoshu = list[0].ToString();
if (string.IsNullOrEmpty(miaoshu))
{
return;
}
miaoshu = Regex.Replace(miaoshu,@"(\\t|\s)","");
Match company = Regex.Match(miaoshu, @"Txt\(4903\[108\,12\]\,4935\[110\,16\]\)\:\\n(?<company>\w*)\\n...End", RegexOptions.Multiline);
if (company.Success)
{
info.company = company.Value;
}
}
public Job GetDetail(string url)
{
Job info = new Job();
Parser parser = ParserHelp.GetParser(url);
NodeFilter miaoShu = new HasAttributeFilter("id", "miaoshu");
NodeFilter mainBox = new HasAttributeFilter("class", "mainBox");
NodeFilter orfilter = new OrFilter(miaoShu, mainBox);
NodeList list = new NodeList();
list = parser.Parse(orfilter);
if (list == null || list.Count < 2)
{
return info;
}
GetMiaoShu(list, ref info);
GetContartInfo(list, ref info);
return info;
}
public void GetMiaoShu(NodeList list, ref Job info)
{
string miaoshu = list[1].ToPlainTextString();
miaoshu = Regex.Replace(miaoshu, @"^[\s|\t]{1,}", "", RegexOptions.Multiline);
info.description = miaoshu;
}
public static void Insert(Job param)
{
MapperHelper.Instance().Insert("JobMap.InsertJob", param);
}
private void InsertJobInfo(Job jobinfo)
{
if (string.IsNullOrEmpty(jobinfo.title) || string.IsNullOrEmpty(jobinfo.description))
{
return;
}
bool sign = true;
while (sign)
{
try
{
JobMap.Insert(jobinfo);
sign = false;
}
catch (Exception ex)
{
SpiderEventLog.WriteWarningLog("HtmlParse InsertJobInfo error:" + ex.ToString());
Thread.Sleep(3000);
sign=true;
}
}
}