/// <summary>
/// 企业负责人信息
/// </summary>
/// <param name="info"></param>
/// <param name="html"></param>
public void AddCorpLeader(CorpInfo info, string html)
{
Parser parser = new Parser(new Lexer(html));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "fzrxx")));
if (nodeList != null && nodeList.Count > 0)
{
parser = new Parser(new Lexer(nodeList.ToHtml().Replace("th", "td")));
NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (tableNode != null && tableNode.Count > 0)
{
TableTag table = tableNode[0] as TableTag;
for (int i = 1; i < table.RowCount; i++)
{
TableRow tr = table.Rows[i];
if (tr.Columns[0].ToPlainTextString().Contains("没有显示结果"))
{
break;
}
string LeaderName = string.Empty, LeaderDuty = string.Empty, LeaderType = string.Empty, htlCtx = string.Empty, phone = string.Empty;
LeaderName = tr.Columns[1].ToNodePlainString();
LeaderDuty = tr.Columns[2].ToNodePlainString();
phone = tr.Columns[3].ToNodePlainString();
LeaderType = tr.Columns[4].ToNodePlainString();
CorpLeader corpLeader = ToolDb.GenCorpLeader(info.Id, LeaderName, LeaderDuty, LeaderType, info.Url, phone);
ToolDb.SaveEntity(corpLeader, string.Empty);
}
}
}
}
protected void AddCorpLeader(CorpInfo info, string enterpriseId)
{
string url = "http://202.104.65.182:8081/G2/webdrive/web-enterprise-leader.do?enterpriseId=" + enterpriseId + "&data&filter_params_=rowNum,leaderId,name,title,safetyLicenseCode,safetyLicenseIssuer,safetyLicenseValidEnd&defined_operations_=&nocheck_operations_=&";
string gridSearch = "false";
string nd = ToolHtml.GetDateTimeLong(DateTime.Now).ToString();
string PAGESIZE = "1000";
string PAGE = "1";
string sortField = "";
string sortDirection = "asc";
NameValueCollection nvc = ToolWeb.GetNameValueCollection(new string[] {
"gridSearch", "nd", "PAGESIZE", "PAGE", "sortField", "sortDirection"
}, new string[] {
gridSearch, nd, PAGESIZE, PAGE, sortField, sortDirection
});
string strJson = string.Empty;
try
{
strJson = ToolWeb.GetHtmlByUrl(url, nvc);
}
catch { return; }
JavaScriptSerializer serializer = new JavaScriptSerializer();
Dictionary <string, object> smsTypeJson = (Dictionary <string, object>)serializer.DeserializeObject(strJson);
object[] objList = smsTypeJson["data"] as object[];
if (objList != null)
{
foreach (object obj in objList)
{
Dictionary <string, object> dic = obj as Dictionary <string, object>;
string LeaderName = string.Empty, LeaderDuty = string.Empty, LeaderType = string.Empty, htlCtx = string.Empty, phone = string.Empty;
LeaderName = Convert.ToString(dic["name"]);
LeaderType = Convert.ToString(dic["title"]);
CorpLeader corpLeader = ToolDb.GenCorpLeader(info.Id, LeaderName, LeaderDuty, LeaderType, info.Url, phone);
ToolDb.SaveEntity(corpLeader, string.Empty);
}
}
}
protected override IList ExecuteCrawl(bool crawlAll)
{
string html = string.Empty;
string cookiestr = string.Empty;
string viewState = string.Empty;
int pageInt = 1;
string eventValidation = string.Empty;
try
{
html = ToolWeb.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8, ref cookiestr);
}
catch { return(null); }
Parser parser = new Parser(new Lexer(html));
NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "ContentPlaceHolder1_AspNetPager1")), true), new TagNameFilter("a")));
if (pageNode != null && pageNode.Count > 0)
{
try
{
string temp = pageNode[pageNode.Count - 1].GetATagHref().Replace("'", "").Replace(")", "kdxx").Replace(",", "xxdk");
pageInt = int.Parse(temp.GetRegexBegEnd("xxdk", "kdxx"));
}
catch { }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
if (i == 2)
{
viewState = ToolWeb.GetAspNetViewState(html);
eventValidation = ToolWeb.GetAspNetEventValidation(html);
}
NameValueCollection nvc = ToolWeb.GetNameValueCollection(
new string[] {
"ctl00$ContentPlaceHolder1$ScriptManager1",
"ctl00$ContentPlaceHolder1$txtORGNAME",
"ctl00$ContentPlaceHolder1$txtORGCODE",
"ctl00$ContentPlaceHolder1$txtPNAME",
"ctl00$ContentPlaceHolder1$txtIDNUM",
"ctl00$ContentPlaceHolder1$txtHIREERORGNAME",
"ctl00$ContentPlaceHolder1$txtHIREERORGCODE",
"ctl00$ContentPlaceHolder1$ddlRegType",
"ctl00$ContentPlaceHolder1$ddlTitle",
"ctl00$ContentPlaceHolder1$ddlABC",
"ctl00$ContentPlaceHolder1$ddlCert",
"__VIEWSTATE",
"__EVENTTARGET",
"__EVENTARGUMENT",
"__EVENTVALIDATION",
"__ASYNCPOST"
},
new string[] {
"ctl00$ContentPlaceHolder1$UpdatePanel1|ctl00$ContentPlaceHolder1$AspNetPager1",
"", "", "", "", "", "", "", "", "", "",
viewState,
"ctl00$ContentPlaceHolder1$AspNetPager1",
i.ToString(),
eventValidation,
"true"
}
);
html = ToolWeb.GetHtmlByUrl(SiteUrl, nvc, Encoding.UTF8, ref cookiestr);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "data-grid")));
if (nodeList != null && nodeList.Count > 0)
{
TableTag table = nodeList[0] as TableTag;
for (int j = 1; j < table.RowCount; j++)
{
string CorpName = string.Empty, CorpCode = string.Empty, CorpAddress = string.Empty,
RegDate = string.Empty, RegFund = string.Empty, BusinessCode = string.Empty,
BusinessType = string.Empty, LinkMan = string.Empty, LinkPhone = string.Empty,
Fax = string.Empty, Email = string.Empty, CorpSite = string.Empty, cUrl = string.Empty, CorpType,
ISOQualNum = string.Empty, ISOEnvironNum = string.Empty, OffAdr = string.Empty, Cert = string.Empty;
TableRow tr = table.Rows[j];
CorpName = tr.Columns[0].ToNodePlainString();
LinkMan = tr.Columns[1].ToNodePlainString();
cUrl = tr.Columns[0].GetATagValue("onclick").Replace("OpenWin('", "");
if (cUrl.IndexOf("'") > 0)
{
cUrl = "http://113.108.219.40/intogd/" + cUrl.Remove(cUrl.IndexOf("'"));
}
string htmldtl = string.Empty;
try
{
htmldtl = ToolWeb.GetHtmlByUrl(cUrl, Encoding.UTF8);
}
catch { continue; }
parser = new Parser(new Lexer(htmldtl));
NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "data-table")));
if (dtlNode != null && dtlNode.Count > 0)
{
string ctx = string.Empty;
TableTag dtlTable = dtlNode[0] as TableTag;
for (int k = 0; k < dtlTable.RowCount; k++)
{
for (int d = 0; d < dtlTable.Rows[k].ColumnCount; d++)
{
TableColumn col = dtlTable.Rows[k].Columns[d];
if (col.GetAttribute("class") == "td-left")
{
ctx += col.ToNodePlainString() + "：";
}
else
{
ctx += col.ToNodePlainString() + "\r\n";
}
}
}
RegDate = ctx.GetRegex("成立时间,注册时间").GetDateRegex();
RegFund = ctx.GetRegex("注册资本");
BusinessCode = ctx.GetRegex("营业执照注册号");
CorpType = "外地进粤企业";
CorpAddress = ctx.GetRegex("注册详细地址");
if (!string.IsNullOrEmpty(RegFund) && !RegFund.Contains("万"))
{
RegFund += "万";
}
CorpInfo corp = ToolDb.GenCorpInfo(CorpName, CorpCode, CorpAddress, RegDate, RegFund, BusinessCode, BusinessType, LinkMan, LinkPhone, Fax, Email, CorpSite, CorpType, "广东省", "广东地区", "广东省住房和城乡建设厅", cUrl, ISOQualNum, ISOEnvironNum, OffAdr);
string strSql = string.Format("select Id from CorpInfo where CorpName='{0}' and CorpType='{1}'", corp.CorpName, corp.CorpType);
DataTable dt = ToolCoreDb.GetDbData(strSql);
if (dt != null && dt.Rows.Count > 0)
{
string id = dt.Rows[0]["Id"].ToString();
StringBuilder delCorpQual = new System.Text.StringBuilder();
StringBuilder delCorpLeader = new System.Text.StringBuilder();
StringBuilder delCorpSecLicStaff = new System.Text.StringBuilder();
StringBuilder delCorpInstitution = new StringBuilder();
delCorpInstitution.AppendFormat("delete from CorpInstitution where CorpId='{0}'", id);
delCorpQual.AppendFormat("delete from CorpQual where CorpId='{0}'", id);
delCorpLeader.AppendFormat("delete from CorpLeader where CorpId='{0}'", id);
delCorpSecLicStaff.AppendFormat("delete from CorpTecStaff where CorpId='{0}'", id);
ToolCoreDb.ExecuteSql(delCorpInstitution.ToString());
ToolCoreDb.ExecuteSql(delCorpQual.ToString());
ToolCoreDb.ExecuteSql(delCorpLeader.ToString());
ToolCoreDb.ExecuteSql(delCorpSecLicStaff.ToString());
string corpSql = string.Format("delete from CorpInfo where Id='{0}'", id);
ToolCoreDb.ExecuteSql(corpSql);
}
if (ToolDb.SaveEntity(corp, this.ExistCompareFields))
{
if (!string.IsNullOrEmpty(LinkMan))
{
CorpLeader leader = ToolDb.GenCorpLeader(corp.Id, LinkMan, "", "企业法定代表人", cUrl);
ToolDb.SaveEntity(leader, "");
}
if (!string.IsNullOrEmpty(tr.Columns[2].ToNodePlainString()))
{
CorpLeader leader = ToolDb.GenCorpLeader(corp.Id, tr.Columns[2].ToNodePlainString(), "", "技术负责人", cUrl);
ToolDb.SaveEntity(leader, "");
}
if (!string.IsNullOrEmpty(tr.Columns[3].ToNodePlainString()))
{
CorpLeader leader = ToolDb.GenCorpLeader(corp.Id, tr.Columns[3].ToNodePlainString(), "", "驻粤负责人", cUrl);
ToolDb.SaveEntity(leader, "");
}
AddCorpQual(corp, htmldtl);
AddCorpTecStaff(corp, htmldtl);
GetOffAddress(htmldtl, cUrl, corp);
}
}
}
}
}
return(null);
}
/// <summary>
/// 企业负责人信息
/// </summary>
/// <param name="info"></param>
/// <param name="html"></param>
public void AddCorpLeader(CorpInfo info, string param, string corpType)
{
string url = "http://portal.szjs.gov.cn:8888/publicShow/queryPrincipal.html";
string[] postParams = new string[] { "param", "corpType", "orgCode", "page" };
string[] postValues = new string[] { param, corpType, info.CorpCode, "1" };
NameValueCollection nvc = ToolWeb.GetNameValueCollection(postParams, postValues);
string html = string.Empty;
try
{
html = ToolWeb.GetHtmlByUrl(url, nvc, Encoding.UTF8);
}
catch
{
Thread.Sleep(12 * 60 * 1000);
try
{
html = ToolWeb.GetHtmlByUrl(url, nvc, Encoding.UTF8);
}
catch
{
Thread.Sleep(8 * 60 * 1000);
return;
}
}
JavaScriptSerializer java = new JavaScriptSerializer();
Dictionary <string, object> jsonResults = (Dictionary <string, object>)java.DeserializeObject(html);
int pageInt = 1;
try
{
pageInt = (int)jsonResults["totalPage"];
}
catch { }
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
postValues = new string[] { param, corpType, info.CorpCode, i.ToString() };
nvc = ToolWeb.GetNameValueCollection(postParams, postValues);
try
{
html = ToolWeb.GetHtmlByUrl(url, nvc, Encoding.UTF8);
}
catch
{
Thread.Sleep(12 * 60 * 1000);
try
{
html = ToolWeb.GetHtmlByUrl(url, nvc, Encoding.UTF8);
}
catch
{
Thread.Sleep(8 * 60 * 1000);
continue;
}
}
jsonResults = (Dictionary <string, object>)java.DeserializeObject(html);
}
object[] dicRecords = (object[])jsonResults["records"];
foreach (object dicRecord in dicRecords)
{
Dictionary <string, object> dic = (Dictionary <string, object>)dicRecord;
string LeaderName = string.Empty, LeaderDuty = string.Empty, LeaderType = string.Empty, htlCtx = string.Empty, phone = string.Empty;
LeaderName = Convert.ToString(dic["name"]);
LeaderDuty = Convert.ToString(dic["duty"]);
phone = Convert.ToString(dic["tel"]);
LeaderType = Convert.ToString(dic["emptype"]);
CorpLeader corpLeader = ToolDb.GenCorpLeader(info.Id, LeaderName, LeaderDuty, LeaderType, info.Url, phone);
ToolDb.SaveEntity(corpLeader, string.Empty);
}
}
}
protected override IList ExecuteCrawl(bool crawlAll)
{
string html = string.Empty;
string cookiestr = string.Empty;
string viewState = string.Empty;
int pageInt = 1;
string eventValidation = string.Empty;
try
{
html = ToolWeb.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr);
}
catch
{
return(null);
}
string opValue = string.Empty, leveVlaue = string.Empty;
string[] levelNode = new string[] {
"特级", "特级(旧标准)", "一级", "一级(旧标准)", "二级", "二级(旧标准)", "三级", "三级(旧标准)", "暂定三级(旧标准)", "不分等级"
};
Parser parser = new Parser(new Lexer(html));
NodeList typeNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("id", "ctl00_ContentPlaceHolder1_ddlENT_SORT_ID")), true), new TagNameFilter("option")));
if (typeNode != null && typeNode.Count > 0)
{
for (int t = 1; t < typeNode.Count; t++)
{
for (int l = 1; l < levelNode.Length; l++)
{
leveVlaue = levelNode[l];
OptionTag opTag = typeNode[t] as OptionTag;
opValue = opTag.GetAttribute("value");
parser = new Parser(new Lexer(html));
NodeList inputNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("input"), new HasAttributeFilter("id", "ctl00_ContentPlaceHolder1_ValidateCode1_txtRanNum")));
string valiCode = string.Empty;
if (inputNode != null && inputNode.Count > 0)
{
valiCode = (inputNode[0] as InputTag).GetAttribute("value");
}
viewState = ToolWeb.GetAspNetViewState(html);
NameValueCollection typeNvc = ToolWeb.GetNameValueCollection(
new string[] {
"ctl00_ContentPlaceHolder1_toolkitScriptManager1_HiddenField",
"__EVENTTARGET",
"__EVENTARGUMENT",
"__LASTFOCUS",
"__VIEWSTATE",
"ctl00$ContentPlaceHolder1$ddlENT_SORT_ID",
"ctl00$ContentPlaceHolder1$ddlRank",
"ctl00$ContentPlaceHolder1$txtEnt_name",
"ctl00$ContentPlaceHolder1$ValidateCode1$txtValidateCode",
"ctl00$ContentPlaceHolder1$ValidateCode1$txtRanNum",
"ctl00$ContentPlaceHolder1$btnsearch"
},
new string[] {
"", "", "", "",
viewState, opValue, leveVlaue, "", valiCode, valiCode,
"搜 索"
});
try
{
html = ToolWeb.GetHtmlByUrl(SiteUrl, typeNvc, Encoding.UTF8, ref cookiestr);
}
catch { continue; }
parser = new Parser(new Lexer(html));
NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "ctl00_ContentPlaceHolder1_AspNetPager1")));
if (pageList != null && pageList.Count > 0)
{
try
{
string temp = pageList[0].ToPlainTextString().GetRegexBegEnd("共", "条");
int page = int.Parse(temp);
int result = page / 15;
if (page % 15 != 0)
{
pageInt = result + 1;
}
else
{
pageInt = result;
}
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 20)
{
break;
}
if (i > 1)
{
try
{
parser = new Parser(new Lexer(html));
NodeList pageInputNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("input"), new HasAttributeFilter("id", "ctl00_ContentPlaceHolder1_ValidateCode1_txtRanNum")));
string pageValiCode = string.Empty;
if (pageInputNode != null && pageInputNode.Count > 0)
{
pageValiCode = (pageInputNode[0] as InputTag).GetAttribute("value");
}
viewState = ToolWeb.GetAspNetViewState(html);
NameValueCollection nvc = ToolWeb.GetNameValueCollection(
new string[] {
"ctl00$ContentPlaceHolder1$ddlENT_SORT_ID",
"ctl00$ContentPlaceHolder1$ddlRank",
"ctl00$ContentPlaceHolder1$txtEnt_name",
"ctl00$ContentPlaceHolder1$ValidateCode1$txtRanNum",
"ctl00$ContentPlaceHolder1$ValidateCode1$txtValidateCode",
"ctl00_ContentPlaceHolder1_toolkitScriptManager1_HiddenField",
"__EVENTARGUMENT",
"__EVENTTARGET",
"__LASTFOCUS",
"__VIEWSTATE"
},
new string[] {
opValue,
leveVlaue, "",
pageValiCode,
"", "",
i.ToString(),
"ctl00$ContentPlaceHolder1$AspNetPager1", "",
viewState
}
);
html = ToolWeb.GetHtmlByUrl(SiteUrl, nvc, Encoding.UTF8, ref cookiestr);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "tab_ent")));
if (nodeList != null && nodeList.Count > 0)
{
#region 循环列表
TableTag table = nodeList[0] as TableTag;
for (int j = 1; j < table.RowCount; j++)
{
string CorpName = string.Empty, CorpCode = string.Empty, CorpAddress = string.Empty,
RegDate = string.Empty, RegFund = string.Empty, BusinessCode = string.Empty,
BusinessType = string.Empty, LinkMan = string.Empty, LinkPhone = string.Empty,
Fax = string.Empty, Email = string.Empty, CorpSite = string.Empty, cUrl = string.Empty,
ISOQualNum = string.Empty, ISOEnvironNum = string.Empty, corpType = string.Empty,
qualCode = string.Empty, corpMgr = string.Empty, businessMgr = string.Empty, tecMgr = string.Empty;
string htlCtx = string.Empty, QualType = string.Empty, CorpLevey = string.Empty;
TableRow tr = table.Rows[j];
string qualStr = tr.Columns[2].ToHtml();
CorpName = tr.Columns[1].ToNodePlainString();
QualType = tr.Columns[2].ToPlainTextString();
CorpLevey = tr.Columns[3].ToNodePlainString();
qualCode = tr.Columns[4].ToNodePlainString();
if (QualType == "--")
{
QualType = "";
}
cUrl = "http://113.108.219.40/PlatForm/SearchCenter/" + tr.Columns[1].GetATagHref();
List <string> quaList = new List <string>();
parser = new Parser(new Lexer(tr.Columns[4].ToHtml()));
NodeList quaNodeList = parser.ExtractAllNodesThatMatch(new TagNameFilter("a"));
if (quaNodeList != null && quaNodeList.Count > 0)
{
for (int q = 0; q < quaNodeList.Count; q++)
{
quaList.Add("http://113.108.219.40/PlatForm/SearchCenter/" + quaNodeList[q].GetATagHref());
}
}
string quaUrl = "http://113.108.219.40/PlatForm/SearchCenter/" + tr.Columns[4].GetATagHref();
string htldtl = string.Empty;
try
{
htldtl = ToolWeb.GetHtmlByUrl(cUrl, Encoding.UTF8);
}
catch { continue; }
parser = new Parser(new Lexer(htldtl));
NodeList dtlList = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (dtlList != null && dtlList.Count > 0)
{
TableTag tab = dtlList[0] as TableTag;
string ctx = string.Empty;
for (int k = 0; k < tab.RowCount; k++)
{
for (int d = 0; d < tab.Rows[k].ColumnCount; d++)
{
if ((d + 1) % 2 == 0)
{
ctx += tab.Rows[k].Columns[d].ToNodePlainString() + "\r\n";
}
else
{
ctx += tab.Rows[k].Columns[d].ToNodePlainString().Replace("：", "").Replace(":", "") + "：";
}
}
}
corpType = ctx.GetRegex(new string[] { "企业类型", "类型" });
CorpAddress = ctx.GetRegex(new string[] { "企业注册地址", "地址" });
BusinessCode = ctx.GetRegex(new string[] { "营业执照注册号", "注册号" });
RegDate = ctx.GetRegex(new string[] { "成立时间", "成立日期", "时间", "日期" }).GetDateRegex();
LinkMan = ctx.GetRegex(new string[] { "企业法定代表人", "法定代表人" });
RegFund = ctx.GetRegex(new string[] { "注册资金", "资金" });
if (!RegFund.Contains("万"))
{
RegFund += "万";
}
corpMgr = ctx.GetRegex(new string[] { "企业经理" });
if (corpMgr.Contains("暂无"))
{
corpMgr = string.Empty;
}
businessMgr = ctx.GetRegex(new string[] { "经营负责人" });
if (businessMgr.Contains("暂无"))
{
businessMgr = string.Empty;
}
tecMgr = ctx.GetRegex(new string[] { "技术负责人" });
if (tecMgr.Contains("暂无"))
{
tecMgr = string.Empty;
}
CorpInfo info = ToolDb.GenCorpInfo(CorpName, CorpCode, CorpAddress, RegDate, RegFund, BusinessCode, BusinessType, LinkMan, LinkPhone, Fax, Email, CorpSite, corpType, "广东省", "广东地区", "广东省住房和城乡建设厅", cUrl, ISOQualNum, ISOEnvironNum, string.Empty);
string strSql = string.Format("select Id from CorpInfo where CorpName='{0}' and Url='{1}'", info.CorpName, info.Url);
object obj = ToolDb.ExecuteScalar(strSql);
if (obj != null && obj.ToString() != "")
{
StringBuilder delCorpQual = new System.Text.StringBuilder();
StringBuilder delCorpLeader = new System.Text.StringBuilder();
delCorpQual.AppendFormat("delete from CorpQual where CorpId='{0}'", obj);
delCorpLeader.AppendFormat("delete from CorpLeader where CorpId='{0}'", obj);
ToolDb.ExecuteSql(delCorpQual.ToString());
ToolDb.ExecuteSql(delCorpLeader.ToString());
string corpSql = string.Format("delete from CorpInfo where Id='{0}'", obj);
ToolCoreDb.ExecuteSql(corpSql);
}
if (ToolDb.SaveEntity(info, string.Empty))
{
if (!string.IsNullOrEmpty(LinkMan))
{
CorpLeader leader = ToolDb.GenCorpLeader(info.Id, LinkMan, "", "企业法定代表人", cUrl);
ToolDb.SaveEntity(leader, string.Empty);
}
if (!string.IsNullOrEmpty(corpMgr))
{
CorpLeader leader = ToolDb.GenCorpLeader(info.Id, corpMgr, "", "企业经理", cUrl);
ToolDb.SaveEntity(leader, string.Empty);
}
if (!string.IsNullOrEmpty(businessMgr))
{
CorpLeader leader = ToolDb.GenCorpLeader(info.Id, businessMgr, "", "经营负责人", cUrl);
ToolDb.SaveEntity(leader, string.Empty);
}
if (!string.IsNullOrEmpty(tecMgr))
{
CorpLeader leader = ToolDb.GenCorpLeader(info.Id, tecMgr, "", "技术负责人", cUrl);
ToolDb.SaveEntity(leader, string.Empty);
}
if (!string.IsNullOrEmpty(qualStr))
{
List <CorpQual> corpQuals = new List <CorpQual>();
string quaCtx = string.Empty;
for (int c = 0; c < quaList.Count; c++)
{
string quaHtl = string.Empty;
try
{
quaHtl = ToolWeb.GetHtmlByUrl(quaList[c], Encoding.UTF8);
}
catch { }
parser = new Parser(new Lexer(quaHtl));
NodeList quaNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (quaNode != null && quaNode.Count > 0)
{
TableTag quaTable = quaNode[0] as TableTag;
for (int k = 0; k < quaTable.RowCount; k++)
{
for (int d = 0; d < quaTable.Rows[k].ColumnCount; d++)
{
string temp = quaTable.Rows[k].Columns[d].ToNodePlainString();
//string quatemp = quaTable.Rows[k].ToNodePlainString();
if ((d + 1) % 2 == 0)
{
quaCtx += temp + "\r\n";
}
else
{
quaCtx += temp.Replace("：", "").Replace(":", "") + "：";
}
}
}
}
string qualctx = string.Empty;
parser.Reset();
NodeList spanNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "lblQuaInfo")));
if (spanNode != null && spanNode.Count > 0)
{
qualctx = spanNode.ToHtml().GetReplace("<br/>,<br />,<br>,</br>", "∈").ToCtxString();
}
string str = qualctx.ToLower().Replace("<br/>", "∈").Replace("</br>", "∈").Replace("<br>", "∈");
str = Regex.Replace(str, "<[^>]*>", "");
string[] qual = str.Split('∈');
for (int q = 0; q < qual.Length; q++)
{
if (string.IsNullOrEmpty(qual[q]) || qual[q] == "--")
{
continue;
}
string CorpId = string.Empty, QualName = string.Empty, quaCode = string.Empty, QualSeq = string.Empty, qualNum = string.Empty, QualLevel = string.Empty, ValidDate = string.Empty, LicDate = string.Empty, LicUnit = string.Empty, quaType = string.Empty;
LicDate = quaCtx.GetRegex("发证日期,发证时间").GetDateRegex();
LicUnit = quaCtx.GetRegex("发证机关,发证机构");
ValidDate = quaCtx.GetRegex("证书有效期").GetDateRegex();
quaType = quaCtx.GetRegex("证书类型");
string value = qual[q];
int len = value.IndexOf("/");
if (len != -1)
{
QualLevel = value.Substring(len, value.Length - len).Replace("/", "");
value = value.Remove(len);
}
else
{
QualLevel = CorpLevey;
}
string[] dtl = value.Split(' ');
CorpId = info.Id;
QualName = dtl[0].Trim();
if (string.IsNullOrEmpty(QualName))
{
QualName = dtl[dtl.Length - 1];
}
quaCode = quaCtx.GetRegex("证书编号");//qualCode;
for (int ty = 1; ty < dtl.Length; ty++)
{
quaType += dtl[ty].Trim() + ",";
}
if (!string.IsNullOrEmpty(quaType) && quaType.Contains(","))
{
quaType = quaType.Substring(0, quaType.Length - 1);
if (quaType[0] == ',' || quaType[0] == '，')
{
quaType = quaType.Substring(1, quaType.Length - 1);
}
}
qualNum = QualLevel.GetLevel();
CorpQual corpQual = null;
corpQual = ToolDb.GenCorpQual(info.Id, QualName, quaCode, QualSeq, quaType, QualLevel, ValidDate, LicDate, LicUnit, quaUrl, qualNum, "广东省", "广东地区");
ToolDb.SaveEntity(corpQual, string.Empty);
}
}
}
}
}
}
#endregion
}
}
}
}
}
ToolCoreDb.ExecuteProcedure();
return(null);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
string html = string.Empty;
string cookiestr = string.Empty;
string viewState = string.Empty;
int pageInt = 1;
string eventValidation = string.Empty;
string pageHtl = string.Empty;
try
{
html = ToolWeb.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr);
}
catch { return(null); }
Parser parser = new Parser(new Lexer(html));
NodeList enttypeNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("name", "selected2")), true), new TagNameFilter("option")));
parser.Reset();
NodeList typeNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("name", "selected")), true), new TagNameFilter("option")));
if (enttypeNode != null && enttypeNode.Count > 0 && typeNode != null && typeNode.Count > 0)
{
for (int t = 0; t < enttypeNode.Count; t++)
{
string entTag = (enttypeNode[t] as OptionTag).GetAttribute("value");
string entText = enttypeNode[t].ToNodePlainString();
for (int d = 0; d < typeNode.Count; d++)
{
string typeTag = (typeNode[d] as OptionTag).GetAttribute("value");
string corpType = typeNode[d].ToNodePlainString();
if (t == 1 && d == 0)
{
typeTag = "16"; corpType = "房地产开发企业";
}
if (t == 1 && d == 1)
{
typeTag = "17"; corpType = "预拌商品混凝土企业";
}
if (t == 1 && d == 2)
{
typeTag = "19"; corpType = "建筑业施工企业";
}
try
{
NameValueCollection nvc = ToolWeb.GetNameValueCollection(
new string[] {
"pageMethod",
"method",
"selected2",
"selected",
"_state",
"keyword",
"currentPage",
"currentPage_temp"
},
new string[] {
"",
"searchHandBook",
entTag,
typeTag,
"1",
"", "1", "1"
});
html = ToolWeb.GetHtmlByUrl("http://www.dgjs.gov.cn/dgweb/search.do", nvc, Encoding.UTF8, ref cookiestr);
}
catch { continue; }
parser = new Parser(new Lexer(html));
NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "mainNextPage")));
if (pageNode != null && pageNode.Count > 0)
{
try
{
string temp = pageNode.AsString().GetRegexBegEnd("/", "页").Replace("\r", "").Replace("\t", "").Replace("\n", "");
pageInt = int.Parse(temp);
}
catch { }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
NameValueCollection nvc = ToolWeb.GetNameValueCollection(
new string[] {
"pageMethod",
"method",
"selected2",
"selected",
"_state",
"keyword",
"currentPage",
"currentPage_temp",
},
new string[] {
"next",
"searchHandBook",
entTag,
typeTag,
"1",
"", (i - 1).ToString(), i.ToString()
});
html = ToolWeb.GetHtmlByUrl("http://www.dgjs.gov.cn/dgweb/search.do", nvc, Encoding.UTF8, ref cookiestr);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList tableNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("align", "center")), true), new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "jsxmtb"))));
if (tableNode != null && tableNode.Count > 0)
{
TableTag table = tableNode[0] as TableTag;
for (int j = 2; t == 1 ? j <= table.RowCount : j < table.RowCount; j++)
{
string CorpName = string.Empty, CorpCode = string.Empty, CorpAddress = string.Empty,
RegDate = string.Empty, RegFund = string.Empty, BusinessCode = string.Empty,
BusinessType = string.Empty, LinkMan = string.Empty, LinkPhone = string.Empty,
Fax = string.Empty, Email = string.Empty, CorpSite = string.Empty, cUrl = string.Empty,
ISOQualNum = string.Empty, ISOEnvironNum = string.Empty, CorpLevey = string.Empty;
TableRow tr = null;
#region 信用手册
if (entText.Contains("手册"))
{
tr = table.Rows[j];
CorpName = tr.Columns[1].ToNodePlainString();
LinkMan = tr.Columns[3].ToNodePlainString();
CorpAddress = tr.Columns[5].ToNodePlainString();
CorpLevey = tr.Columns[2].ToNodePlainString();
if (corpType.Contains("担保企业"))
{
cUrl = "http://www.dgjs.gov.cn/dgweb/" + tr.Columns[10].GetATagHref();
}
else
{
cUrl = "http://www.dgjs.gov.cn/dgweb/" + tr.Columns[9].GetATagHref();
}
string htlDtl = string.Empty;
try
{
htlDtl = ToolWeb.GetHtmlByUrl(cUrl, Encoding.UTF8).GetJsString();
}
catch { continue; }
parser = new Parser(new Lexer(htlDtl.Replace("th", "td").Replace("TH", "TD")));
NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "dgjsj")), true), new TagNameFilter("table")));
if (dtlNode != null && dtlNode.Count > 1)
{
string ctx = string.Empty;
TableTag dtlTable = dtlNode[0] as TableTag;
for (int c = 1; c < dtlTable.RowCount; c++)
{
for (int v = 0; v < dtlTable.Rows[c].ColumnCount; v++)
{
if (string.IsNullOrEmpty(dtlTable.Rows[c].Columns[v].ToNodePlainString()))
{
continue;
}
if ((v + 1) % 2 == 0)
{
ctx += dtlTable.Rows[c].Columns[v].ToNodePlainString() + "\r\n";
}
else
{
ctx += dtlTable.Rows[c].Columns[v].ToNodePlainString() + "：";
}
}
}
RegDate = ctx.GetRegex("设立时间,设立日期");
LinkPhone = ctx.GetRegex("联系电话");
Fax = ctx.GetRegex("传真");
Email = ctx.GetRegex("电子邮箱");
BusinessType = ctx.GetRegex("经济性质");
BusinessCode = ctx.GetRegex("营业执照注册号");
}
CorpInfo info = ToolDb.GenCorpInfo(CorpName, CorpCode, CorpAddress, RegDate, RegFund, BusinessCode, BusinessType, LinkMan, LinkPhone, Fax, Email, CorpSite, corpType, "广东省", "东莞市", "东莞市住房和城乡建设局", cUrl, ISOQualNum, ISOEnvironNum, string.Empty);
if (!string.IsNullOrEmpty(CorpName.GetNotChina()))
{
string strSql = string.Format("select Id from CorpInfo where CorpName='{0}' and InfoSource='{1}' and CorpType='{2}'", info.CorpName, info.InfoSource, info.CorpType);
object obj = ToolDb.ExecuteScalar(strSql);
if (obj != null && obj.ToString() != "")
{
StringBuilder delCorpQual = new System.Text.StringBuilder();
StringBuilder delCorpLeader = new System.Text.StringBuilder();
StringBuilder delCorpTecStaff = new System.Text.StringBuilder();
delCorpQual.AppendFormat("delete from CorpQual where CorpId='{0}'", obj);
delCorpLeader.AppendFormat("delete from CorpLeader where CorpId='{0}'", obj);
delCorpTecStaff.AppendFormat("delete from CorpTecStaff where CorpId='{0}'", obj);
ToolDb.ExecuteSql(delCorpQual.ToString());
ToolDb.ExecuteSql(delCorpLeader.ToString());
ToolDb.ExecuteSql(delCorpTecStaff.ToString());
string corpSql = string.Format("delete from CorpInfo where Id='{0}'", obj);
ToolCoreDb.ExecuteSql(corpSql);
}
if (ToolDb.SaveEntity(info, string.Empty))
{
object corpId = ToolDb.ExecuteScalar("select Id from CorpInfo where Url='" + info.Url + "' and InfoSource='东莞市住房和城乡建设局' ");
ToolDb.ExecuteSql("delete from CorpQual where CorpId='" + corpId + "'");
#region 企业资质
TableTag quaTable = dtlNode[1] as TableTag;
for (int q = 2; q < quaTable.RowCount; q++)
{
TableRow quaTr = quaTable.Rows[q];
string CorpId = string.Empty, QualName = string.Empty, QualCode = string.Empty, QualSeq = string.Empty, qualNum = string.Empty, QualType = string.Empty, QualLevel = string.Empty, ValidDate = string.Empty, LicDate = string.Empty, LicUnit = string.Empty;
CorpId = info.Id;
QualName = quaTr.Columns[0].ToNodePlainString();
QualLevel = quaTr.Columns[1].ToNodePlainString();
QualCode = quaTr.Columns[5].ToNodePlainString();
LicUnit = quaTr.Columns[6].ToNodePlainString();
QualType = quaTr.Columns[0].ToNodePlainString();
ValidDate = quaTr.Columns[3].ToPlainTextString().GetDateRegex();
qualNum = QualLevel.GetLevel();
CorpQual qual = ToolDb.GenCorpQual(CorpId, QualName, QualCode, QualSeq, QualType, QualLevel, ValidDate, LicDate, LicUnit, cUrl, qualNum, "广东省", "东莞市");
ToolDb.SaveEntity(qual, "");
}
#endregion
#region 企业负责人
parser = new Parser(new Lexer(htlDtl));
NodeList leaderNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "head2")));
if (leaderNode != null && leaderNode.Count > 0)
{
ToolDb.ExecuteSql("delete from CorpLeader where CorpId='" + corpId + "'");
ATag leaderTag = leaderNode.GetATag(1);
if (!leaderTag.LinkText.Contains("负责人"))
{
leaderTag = leaderNode.GetATag(2);
}
if (!leaderTag.LinkText.Contains("负责人"))
{
leaderTag = leaderNode.GetATag(3);
}
if (!leaderTag.LinkText.Contains("负责人"))
{
leaderTag = leaderNode.GetATag(4);
}
if (leaderTag.LinkText.Contains("负责人"))
{
string leaderUrl = "http://www.dgjs.gov.cn/dgweb/" + leaderTag.Link;
string leaderDtl = string.Empty;
try
{
leaderDtl = ToolWeb.GetHtmlByUrl(leaderUrl, Encoding.UTF8).GetJsString();
}
catch { }
parser = new Parser(new Lexer(leaderDtl));
NodeList leaderDtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "dgjsj")), true), new TagNameFilter("table")));
if (leaderDtlNode != null && leaderDtlNode.Count > 0)
{
TableTag leaderTable = leaderDtlNode[0] as TableTag;
for (int l = 3; l < leaderTable.RowCount; l++)
{
TableRow leaderTr = leaderTable.Rows[l];
if (leaderTr.ToHtml().ToLower().Contains("none"))
{
continue;
}
string LeaderName = string.Empty, LeaderDuty = string.Empty, LeaderType = string.Empty, htlCtx = string.Empty;
try
{
LeaderName = leaderTr.Columns[0].ToNodePlainString();
LeaderDuty = leaderTr.Columns[4].ToNodePlainString();
LeaderType = leaderTr.Columns[1].ToNodePlainString();
}
catch
{ }
if (!string.IsNullOrEmpty(LeaderName))
{
CorpLeader corpLeader = ToolDb.GenCorpLeader(info.Id, LeaderName, LeaderDuty, LeaderType, leaderUrl);
ToolDb.SaveEntity(corpLeader, string.Empty);
}
}
}
}
}
#endregion
#region 企业技术力量
parser = new Parser(new Lexer(htlDtl));
NodeList tecNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "head2")));
if (tecNode != null && tecNode.Count > 0)
{
ToolDb.ExecuteSql("delete from CorpTecStaff where CorpId='" + corpId + "'");
ATag leaderTag = tecNode.GetATag(1);
if (!leaderTag.LinkText.Contains("技术"))
{
leaderTag = tecNode.GetATag(2);
}
if (!leaderTag.LinkText.Contains("技术"))
{
leaderTag = tecNode.GetATag(3);
}
if (!leaderTag.LinkText.Contains("技术"))
{
leaderTag = tecNode.GetATag(4);
}
if (!leaderTag.LinkText.Contains("技术"))
{
leaderTag = tecNode.GetATag(5);
}
if (leaderTag.LinkText.Contains("技术"))
{
string leaderUrl = "http://www.dgjs.gov.cn/dgweb/" + leaderTag.Link;
string leaderDtl = string.Empty;
try
{
leaderDtl = ToolWeb.GetHtmlByUrl(leaderUrl, Encoding.UTF8).GetJsString();
}
catch { }
parser = new Parser(new Lexer(leaderDtl));
NodeList leaderDtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "maintable")));
if (leaderDtlNode != null && leaderDtlNode.Count > 0)
{
TableTag leaderTable = leaderDtlNode[0] as TableTag;
for (int l = 2; l < leaderTable.RowCount - 1; l++)
{
TableRow leaderTr = leaderTable.Rows[l];
string StaffName = string.Empty, IdCard = string.Empty, CertLevel = string.Empty, CertNo = string.Empty, stffType = string.Empty;
try
{
StaffName = leaderTr.Columns[1].ToNodePlainString();
stffType = leaderTr.Columns[6].ToNodePlainString();
if (stffType == "/")
{
stffType = null;
}
CertNo = leaderTr.Columns[8].ToNodePlainString();
}
catch { }
if (!string.IsNullOrEmpty(StaffName))
{
CorpTecStaff staff = ToolDb.GenCorpTecStaff(info.Id, StaffName, IdCard, CertLevel, CertNo, leaderUrl, stffType);
ToolDb.SaveEntity(staff, string.Empty);
}
}
}
}
}
#endregion
}
}
}
#endregion
#region 资质证书企业
else
{
tr = table.Rows[j - 1];
try
{
CorpName = tr.Columns[0].ToNodePlainString();
CorpAddress = tr.Columns[1].ToNodePlainString();
LinkMan = tr.Columns[2].ToNodePlainString();
CorpInfo info1 = ToolDb.GenCorpInfo(CorpName, CorpCode, CorpAddress, RegDate, RegFund, BusinessCode, BusinessType, LinkMan, LinkPhone, Fax, Email, CorpSite, corpType, "广东省", "东莞市", "东莞市住房和城乡建设局", cUrl, ISOQualNum, ISOEnvironNum, string.Empty);
if (!string.IsNullOrEmpty(CorpName.GetNotChina()))
{
string strSql = string.Format("select Id from CorpInfo where CorpName='{0}' and InfoSource='{1}' and CorpType='{2}'", info1.CorpName, info1.InfoSource, info1.CorpType);
object obj = ToolDb.ExecuteScalar(strSql);
if (obj != null && obj.ToString() != "")
{
string corpSql = string.Format("delete from CorpInfo where Id='{0}'", obj);
ToolCoreDb.ExecuteSql(corpSql);
}
ToolDb.SaveEntity(info1, string.Empty);
}
}
catch (Exception ex) { }
}
#endregion
}
}
}
}
}
}
return(null);
}
