protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new ArrayList();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.Default);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList noList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "yema")));
if (noList != null && noList.Count > 0)
{
string temp = noList.AsString();
try
{
Regex reg = new Regex(@"/[^页]+页");
string result = reg.Match(temp).Value.Replace("页", "").Replace("/", "");
pageInt = Convert.ToInt32(result);
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl("http://www.gy-center.net/announce/list.jhtml?visi_id=&cid=97&chid=&gid=&thistype=&searchcid=&keyword=&action=yes&interval=&page=" + i.ToString(), Encoding.Default);
}
catch
{
continue;
}
}
parser = new Parser(new Lexer(html));
NodeList dtlList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "tab01"))), new TagNameFilter("ul"))), new TagNameFilter("li")));
if (dtlList != null && dtlList.Count > 0)
{
for (int j = 0; j < dtlList.Count - 1; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
string temp = dtlList[j].ToPlainTextString();
string tempHtl = dtlList[j].ToHtml();
prjName = ToolHtml.GetHtmlAtagValue("title", tempHtl);
beginDate = ToolHtml.GetRegexDateTime(temp);
InfoUrl = "http://www.gy-center.net/announce/" + ToolHtml.GetHtmlAtagValue("href", tempHtl);
string htlDtl = string.Empty;
try
{
htlDtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default);
htlDtl = System.Text.RegularExpressions.Regex.Replace(htlDtl, "(<script)[\\s\\S]*?(</script>)", "");
}
catch
{
continue;
}
parser = new Parser(new Lexer(htlDtl));
NodeList htlList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "r_content_right_main")));
if (htlList != null && htlList.Count > 0)
{
HtmlTxt = htlList.ToHtml();
bidCtx = Regex.Replace(HtmlTxt, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\t\t", "").Replace("\r\r", "\r").Replace("\n\n", "\n");
bidType = ToolHtml.GetInviteTypes(prjName);
string bidStr = string.Empty;
parser = new Parser(new Lexer(HtmlTxt));
NodeList bidList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "MsoNormalTable")));
if (bidList != null && bidList.Count > 0)
{
try
{
TableTag tab = bidList[0] as TableTag;
if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 6)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[2].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[2].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[3].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[3].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[4].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[4].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[5].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[5].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[6].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[6].ToPlainTextString().ToNodeString() + "\r\n";
}
else if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 5)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[2].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[2].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[3].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[3].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[4].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[4].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[5].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[5].ToPlainTextString().ToNodeString() + "\r\n";
}
else if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 4)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[2].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[2].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[3].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[3].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[4].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[4].ToPlainTextString().ToNodeString() + "\r\n";
}
else if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 3)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[2].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[2].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[3].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[3].ToPlainTextString().ToNodeString() + "\r\n";
}
else if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 2)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[2].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[2].ToPlainTextString().ToNodeString() + "\r\n";
}
else if (tab.RowCount > 1 && tab.Rows[0].ColumnCount > 1)
{
bidStr = tab.Rows[0].Columns[0].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[0].ToPlainTextString().ToNodeString() + "\r\n";
bidStr += tab.Rows[0].Columns[1].ToPlainTextString().ToNodeString() + "：" + tab.Rows[1].Columns[1].ToPlainTextString().ToNodeString() + "\r\n";
}
}
catch { }
}
buildUnit = ToolHtml.GetRegexString(bidCtx, ToolHtml.BuildRegex);
prjAddress = ToolHtml.GetRegexString(bidCtx, ToolHtml.AddressRegex);
code = ToolHtml.GetRegexString(bidCtx, ToolHtml.CodeRegex);
bidUnit = ToolHtml.GetRegexString(bidCtx, ToolHtml.BidRegex);
if (string.IsNullOrEmpty(bidUnit))
{
bidUnit = ToolHtml.GetRegexString(bidStr.Replace(" ", ""), ToolHtml.BidRegex, false);
}
if (string.IsNullOrEmpty(bidUnit))
{
bidUnit = bidCtx.GetRegexBegEnd("确认", "为");
}
bidMoney = ToolHtml.GetRegexString(bidCtx, ToolHtml.MoneyRegex);
bidMoney = ToolHtml.GetRegexMoney(bidMoney);
if (string.IsNullOrEmpty(bidMoney) || bidMoney == "0")
{
bidMoney = bidCtx.GetRegexBegEnd("￥", "元").GetMoney();
}
buildUnit = ToolHtml.GetSubString(buildUnit, 150);
prjAddress = ToolHtml.GetSubString(prjAddress, 150);
code = ToolHtml.GetSubString(code, 50);
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
bidUnit = ToolHtml.GetStringTemp(bidUnit);
buildUnit = ToolHtml.GetStringTemp(buildUnit);
if (string.IsNullOrEmpty(code))
{
code = "见中标信息";
}
if (string.IsNullOrEmpty(prjAddress))
{
prjAddress = "见中标信息";
}
specType = "其他";
msgType = "工网在线";
BidInfo info = ToolDb.GenBidInfo("广东省", "电网专项工程", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
parser = new Parser(new Lexer(HtmlTxt));
NodeList nodeAtag = parser.ExtractAllNodesThatMatch(new TagNameFilter("a"));
if (nodeAtag != null && nodeAtag.Count > 0)
{
for (int c = 0; c < nodeAtag.Count; c++)
{
ATag a = nodeAtag[c] as ATag;
if (a.Link.IsAtagAttach())
{
string alink = "http://www.bidding.csg.cn/" + a.Link;
BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText.Replace(" ", ""), info.Id, alink);
base.AttachList.Add(attach);
}
}
}
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new List <BidInfo>();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "venycms-page")), true), new TagNameFilter("script")));
if (sNode != null && sNode.Count > 0)
{
try
{
string page = sNode.ToString().Replace("createPageHTML(", "").Replace(",", "kd").Replace("****", "").Replace("\n", "");
page = page.GetRegexBegEnd("Code", "kd");
pageInt = int.Parse(page);
}
catch { }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl("http://sgjd.baoan.gov.cn/zbcg/zhbgg_139208/index_" + (i - 1) + ".html", Encoding.UTF8);
}
catch
{
continue;
}
}
parser = new Parser(new Lexer(html));
NodeList viewList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "content clearfix")), true), new TagNameFilter("li")));
if (viewList != null && viewList.Count > 0)
{
for (int j = 0; j < viewList.Count; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
ATag aTag = viewList[j].GetATag();
prjName = aTag.GetAttribute("title");
beginDate = viewList[j].ToPlainTextString().GetDateRegex();
InfoUrl = aTag.Link;
InfoUrl = InfoUrl.GetRegexBegEnd("./", ".html");
InfoUrl = "http://sgjd.baoan.gov.cn/zbcg/zhbgg_139208/" + InfoUrl + ".html";
string htlDtl = string.Empty;
try
{
htlDtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).GetJsString();
}
catch { continue; }
parser = new Parser(new Lexer(htlDtl));
NodeList dtlList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "con")));
if (dtlList != null && dtlList.Count > 0)
{
HtmlTxt = dtlList.AsHtml();
bidCtx = HtmlTxt.ToCtxString();
bidType = prjName.GetInviteBidType();
Regex regPrjCode = new Regex(@"(工程编号|项目编号|招标编号|中标编号|编号)(:|：)[^\r\n]+\r\n");
code = regPrjCode.Match(bidCtx.Replace(" ", "")).Value.Replace("工程编号", "").Replace("项目编号", "").Replace("招标编号", "").Replace("中标编号", "").Replace("编号", "").Replace(":", "").Replace("：", "").Trim();
Regex regBuidUnit = new Regex(@"(建设单位|招标人|承包人|招标单位|招标方|招标代理机构)(：|:)[^\r\n]+\r\n");
buildUnit = regBuidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("招标代理机构", "").Replace("建设单位", "").Replace("招标人", "").Replace("承包人", "").Replace("招标单位", "").Replace("招标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regMoney = new Regex(@"(中标价|投标价|总投资|发包价|投标报价|价格|金额|总价)(：|:|)[^\r\n]+\r\n");
bidMoney = regMoney.Match(bidCtx.Replace(" ", "")).Value.Replace("中标价", "").Replace("总投资", "").Replace("发包价", "").Replace("总价", "").Replace("投标报价", "").Replace("投标价", "").Replace("价格", "").Replace("金额", "").Replace("：", "").Replace(":", "").Replace("，", "").Replace(",", "").Trim();
Regex regBidUnit = new Regex(@"(成交供应商|中标供应商|第一候选人|中标候选人|中标单位|中标人|中标方)(：|:)[^\r\n]+\r\n");
bidUnit = regBidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("成交供应商", "").Replace("中标供应商", "").Replace("中标候选人", "").Replace("第一候选人", "").Replace("中标单位", "").Replace("中标人", "").Replace("中标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regprjMgr = new Regex(@"(项目经理姓名|项目经理（或建造师）|项目经理|项目负责人|项目总监|建造师|总工程师|监理师)(：|:)[^\r\n]+\r\n");
prjMgr = regprjMgr.Match(bidCtx.Replace(" ", "")).Value.Replace("项目经理（或建造师）", "").Replace("项目经理姓名", "").Replace("总工程师", "").Replace("项目经理", "").Replace("项目总监", "").Replace("建造师", "").Replace("监理师", "").Replace("项目负责人", "").Replace("：", "").Replace(":", "").Trim();
Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+");
if (prjMgr.Contains("资格"))
{
prjMgr = prjMgr.Remove(prjMgr.IndexOf("资格"));
}
bidUnit = ToolHtml.GetStringTemp(bidUnit);
buildUnit = ToolHtml.GetSubString(buildUnit, 150);
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
code = ToolHtml.GetSubString(code, 50);
prjMgr = ToolHtml.GetSubString(prjMgr, 50);
if (string.IsNullOrWhiteSpace(bidUnit))
{
parser = new Parser(new Lexer(HtmlTxt));
NodeList tableNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "MsoNormalTable")));
if (tableNode != null && tableNode.Count > 0)
{
for (int t = 0; t < tableNode.Count; t++)
{
TableTag table = tableNode[t] as TableTag;
string ctx = string.Empty;
for (int r = 0; r < table.Rows[0].ColumnCount; r++)
{
try
{
ctx += table.Rows[0].Columns[r].ToNodePlainString() + "：";
ctx += table.Rows[1].Columns[r].ToNodePlainString() + "\r\n";
}
catch { }
}
bidUnit = ctx.GetBidRegex();
if (string.IsNullOrEmpty(bidUnit))
{
bidUnit = ctx.GetRegex("中标供应商");
}
if (string.IsNullOrWhiteSpace(code))
{
code = ctx.GetCodeRegex();
}
// break;
bidMoney = ctx.GetMoneyRegex();
}
}
}
try
{
if (Convert.ToDecimal(bidMoney) > 100000)
{
bidMoney = (decimal.Parse(bidMoney) / 10000).ToString();
}
}
catch { }
if (bidMoney.Contains("万"))
{
bidMoney = bidMoney.Remove(bidMoney.IndexOf("万")).Trim();
bidMoney = regBidMoney.Match(bidMoney).Value;
}
if (string.IsNullOrEmpty(buildUnit))
{
buildUnit = "深圳市宝安区松岗街道办事处";
}
msgType = "深圳市宝安区松岗街道办事处";
specType = "建设工程";
bidType = "小型工程";
BidInfo info = ToolDb.GenBidInfo("广东省", "深圳区及街道工程", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new ArrayList();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.Default);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("width", "50%")));
if (sNode != null && sNode.Count > 0)
{
try
{
string temp = sNode.AsString().Replace(" ", "");
Regex reg = new Regex(@"条,[^页]+页");
pageInt = Convert.ToInt32(reg.Match(temp).Value.Replace("条,", "").Replace("页", ""));
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl + "&p=" + i.ToString(), Encoding.Default);
}
catch
{ continue; }
}
parser = new Parser(new Lexer(html));
NodeList viewList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "Datagrid")));
if (viewList != null && viewList.Count > 0)
{
TableTag tab = viewList[0] as TableTag;
for (int j = 0; j < tab.RowCount; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
TableRow tr = tab.Rows[j];
prjName = tr.Columns[1].ToPlainTextString().Replace("\r", "").Replace("\t", "").Replace("\n", "");
Regex regDate = new Regex(@"\d{4}/\d{1,2}/\d{1,2}");
beginDate = regDate.Match(tr.Columns[2].ToPlainTextString()).Value;
ATag aTag = tr.Columns[1].SearchFor(typeof(ATag), true)[0] as ATag;
InfoUrl = "http://www.xixiang.gov.cn/" + aTag.Link;
string htmDtl = string.Empty;
try
{
htmDtl = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(InfoUrl), Encoding.Default);
Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>|<style[^<]*</style>|<xml[^<]*</xml>");
htmDtl = regexHtml.Replace(htmDtl, "");
}
catch { continue; }
parser = new Parser(new Lexer(htmDtl));
NodeList dtl = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "Lblcontent")));
if (dtl != null && dtl.Count > 0)
{
HtmlTxt = dtl.AsHtml();
parser = new Parser(new Lexer(HtmlTxt));
NodeList span = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "Lblcontent")));
if (span != null && span.Count > 0)
{
bidCtx = Regex.Replace(span.AsHtml().ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n"), "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n").Replace("\r\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
parser = new Parser(new Lexer(span.AsHtml().ToLower().Replace("th", "td")));
NodeList dtlTab = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (dtlTab != null && dtlTab.Count > 0)
{
bidCtx = "";
TableTag table = dtlTab[0] as TableTag;
for (int k = 0; k < table.RowCount; k++)
{
for (int c = 0; c < table.Rows[k].ColumnCount; c++)
{
if (table.RowCount > 1 && k == 0)
{
string strCtx = table.Rows[k].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "").Replace("\r\n", "").Replace("\n", "");
bidCtx += strCtx + "：" + table.Rows[k + 1].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "").Replace("\r\n", "").Replace("\n", "") + "\r\n";
}
}
break;
}
bidCtx = bidCtx.Replace("\n", "").Replace("\r\n\r\n", "\r\n").Replace("\r", "\r\n") + "\r\n";
}
else
{
string ctx = HtmlTxt.ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n");
bidCtx = Regex.Replace(ctx, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
}
}
else
{
parser = new Parser(new Lexer(HtmlTxt.ToLower().Replace("th", "td")));
NodeList dtlTab = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (dtlTab != null && dtlTab.Count > 0)
{
TableTag table = dtlTab[0] as TableTag;
for (int k = 0; k < table.RowCount; k++)
{
for (int c = 0; c < table.Rows[k].ColumnCount; c++)
{
if (table.RowCount > 1 && k == 0)
{
string strCtx = table.Rows[k].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "").Replace("\r\n", "").Replace("\n", "");
bidCtx += strCtx + "：" + table.Rows[k + 1].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "").Replace("\r\n", "").Replace("\n", "") + "\r\n";
}
}
break;
}
bidCtx = bidCtx.Replace("\n", "").Replace("\r\n\r\n", "\r\n").Replace("\r", "\r\n") + "\r\n";
}
else
{
string ctx = HtmlTxt.ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n");
bidCtx = Regex.Replace(ctx, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
}
}
bidCtx = bidCtx.Replace(" ", "");
bidType = "工程";
if (prjName.Contains("施工"))
{
bidType = "施工";
}
if (prjName.Contains("监理"))
{
bidType = "监理";
}
if (prjName.Contains("设计"))
{
bidType = "设计";
}
if (prjName.Contains("勘察"))
{
bidType = "勘察";
}
if (prjName.Contains("服务"))
{
bidType = "服务";
}
if (prjName.Contains("劳务分包"))
{
bidType = "劳务分包";
}
if (prjName.Contains("专业分包"))
{
bidType = "专业分包";
}
if (prjName.Contains("小型施工"))
{
bidType = "小型工程";
}
if (prjName.Contains("设备材料"))
{
bidType = "设备材料";
}
Regex regPrjCode = new Regex(@"(工程编号|项目编号|招标编号|中标编号|编号)(:|：)[^\r\n]+\r\n");
code = regPrjCode.Match(bidCtx.Replace(" ", "")).Value.Replace("工程编号", "").Replace("项目编号", "").Replace("招标编号", "").Replace("中标编号", "").Replace("编号", "").Replace(":", "").Replace("：", "").Trim();
Regex regBuidUnit = new Regex(@"(建设单位|招标人|承包人|招标单位|招标方|招标代理机构)(：|:)[^\r\n]+\r\n");
buildUnit = regBuidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("招标代理机构", "").Replace("建设单位", "").Replace("招标人", "").Replace("承包人", "").Replace("招标单位", "").Replace("招标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regMoney = new Regex(@"(中标价|投标价|总投资|发包价|投标报价|价格|金额|报价)(：|:|)[^\r\n]+\r\n");
bidMoney = regMoney.Match(bidCtx.Replace(" ", "")).Value.Replace("报价", "").Replace("中标价", "").Replace("总投资", "").Replace("发包价", "").Replace("投标报价", "").Replace("投标价", "").Replace("价格", "").Replace("金额", "").Replace("：", "").Replace(":", "").Replace("￥", "").Replace(",", "").Trim();
Regex regBidUnit = new Regex(@"(第一候选人|投标供应商名称|中标候选人|中标供应商|中标单位|中标人|中标方)(：|:)[^\r\n]+\r\n");
bidUnit = regBidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("投标供应商名称", "").Replace("中标供应商", "").Replace("中标候选人", "").Replace("第一候选人", "").Replace("中标单位", "").Replace("中标人", "").Replace("中标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regprjMgr = new Regex(@"(项目经理姓名|项目经理|项目负责人|项目总监|建造师|总工程师|监理师)(：|:)[^\r\n]+\r\n");
prjMgr = regprjMgr.Match(bidCtx.Replace(" ", "")).Value.Replace("项目经理姓名", "").Replace("总工程师", "").Replace("项目经理", "").Replace("项目总监", "").Replace("建造师", "").Replace("监理师", "").Replace("项目负责人", "").Replace("：", "").Replace(":", "").Trim();
Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+");
if (bidMoney.Contains("万"))
{
//bidMoney = bidMoney.Remove(bidMoney.IndexOf("万")).Trim();
bidMoney = regBidMoney.Match(bidMoney).Value;
}
else
{
try
{
bidMoney = (decimal.Parse(regBidMoney.Match(bidMoney).Value) / 10000).ToString();
if (decimal.Parse(bidMoney) < decimal.Parse("0.1"))
{
bidMoney = "0";
}
}
catch (Exception)
{
bidMoney = "0";
}
}
if (prjMgr.Contains("资格"))
{
prjMgr = prjMgr.Remove(prjMgr.IndexOf("资格"));
}
string ctxs = HtmlTxt.ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n");
bidCtx = Regex.Replace(ctxs, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
bidCtx = bidCtx.Replace(" ", "");
bidUnit = ToolHtml.GetStringTemp(bidUnit);
buildUnit = ToolHtml.GetSubString(buildUnit, 150);
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
code = ToolHtml.GetSubString(code, 50);
prjMgr = ToolHtml.GetSubString(prjMgr, 50);
if (string.IsNullOrEmpty(buildUnit))
{
buildUnit = "深圳市宝安区西乡街道办事处";
}
msgType = "深圳市宝安区西乡街道办事处";
specType = "建设工程";
bidType = "小型工程";
prjName = ToolDb.GetPrjName(prjName);
BidInfo info = ToolDb.GenBidInfo("广东省", "深圳区及街道工程", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new ArrayList();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "NewsPage")));
if (sNode != null && sNode.Count > 0)
{
try
{
string temp = sNode.AsString().Replace("createPageHTML(", "").Replace("index", "").Replace("html", "").Replace(", 0,", "").Replace(");", "").Replace(",", "").Replace(";", "").Replace(")", "").Replace("\"", "").Replace(" ", "");
pageInt = Convert.ToInt32(temp);
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl("http://fyjdb.baoan.gov.cn/xxgk_11867/ywxx/zbcg/zbxxgg/index_" + (i - 1).ToString() + ".html", Encoding.UTF8);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList viewList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "NewsLiks01Text"))), new TagNameFilter("ul"))), new TagNameFilter("li")));
if (viewList != null && viewList.Count > 0)
{
for (int j = 0; j < viewList.Count; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}");
beginDate = regDate.Match(viewList[j].ToPlainTextString().Trim()).Value;
string temp = viewList[j].ToPlainTextString().Trim().Replace(beginDate, "");
try
{
int beg = temp.IndexOf("else"), end = temp.Length;
temp = temp.Substring(beg, end - beg);
beg = temp.IndexOf("<a");
end = temp.IndexOf("/a>");
temp = temp.Substring(beg, (end - beg) + 3);
beg = temp.IndexOf(">");
end = temp.IndexOf("</");
prjName = temp.Substring(beg + 1, end - beg - 1);
Parser p = new Parser(new Lexer(temp));
NodeList l = p.ExtractAllNodesThatMatch(new TagNameFilter("a"));
ATag aTag = l.SearchFor(typeof(ATag), true)[0] as ATag;
InfoUrl = "http://fyjdb.baoan.gov.cn/xxgk_11867/ywxx/zbcg/zbxxgg/" + aTag.Link.Replace("../", "").Replace("./", "");
}
catch { continue; }
string htlDtl = string.Empty;
try
{
htlDtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8);
Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>|<style[^<]*</style>|<xml[^<]*</xml>");
htlDtl = regexHtml.Replace(htlDtl, "");
}
catch { continue; }
parser = new Parser(new Lexer(htlDtl));
NodeList dtl = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "DivContent")));
if (dtl != null && dtl.Count > 0)
{
HtmlTxt = dtl.AsHtml();
parser = new Parser(new Lexer(HtmlTxt.ToLower().Replace("th", "td")));
NodeList dtlTab = parser.ExtractAllNodesThatMatch(new TagNameFilter("table"));
if (dtlTab != null && dtlTab.Count > 0)
{
bidCtx = "";
TableTag table = dtlTab[0] as TableTag;
for (int k = 0; k < table.RowCount; k++)
{
for (int c = 0; c < table.Rows[k].ColumnCount; c++)
{
if (table.RowCount > 1 && k == 0)
{
string strCtx = table.Rows[k].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "");
bidCtx += strCtx + "：" + table.Rows[k + 1].Columns[c].ToPlainTextString().Replace(" ", "").Replace(" ", "") + "\r\n";
}
}
break;
}
bidCtx = bidCtx.Replace("\n", "").Replace("\r\n\r\n", "\r\n").Replace("\r", "\r\n").Replace(" ", "").Replace(" ", "").Trim() + "\r\n";
}
else
{
string ctx = HtmlTxt.ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n");
bidCtx = Regex.Replace(ctx, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
}
bidType = "工程";
if (prjName.Contains("施工"))
{
bidType = "施工";
}
if (prjName.Contains("监理"))
{
bidType = "监理";
}
if (prjName.Contains("设计"))
{
bidType = "设计";
}
if (prjName.Contains("勘察"))
{
bidType = "勘察";
}
if (prjName.Contains("服务"))
{
bidType = "服务";
}
if (prjName.Contains("劳务分包"))
{
bidType = "劳务分包";
}
if (prjName.Contains("专业分包"))
{
bidType = "专业分包";
}
if (prjName.Contains("小型施工"))
{
bidType = "小型工程";
}
if (prjName.Contains("设备材料"))
{
bidType = "设备材料";
}
Regex regPrjCode = new Regex(@"(工程编号|项目编号|招标编号|中标编号|编号)(:|：)[^\r\n]+\r\n");
code = regPrjCode.Match(bidCtx.Replace(" ", "")).Value.Replace("工程编号", "").Replace("项目编号", "").Replace("招标编号", "").Replace("中标编号", "").Replace("编号", "").Replace(":", "").Replace("：", "").Trim();
Regex regBuidUnit = new Regex(@"(建设单位|招标人|承包人|招标单位|招标方|招标代理机构)(：|:)[^\r\n]+\r\n");
buildUnit = regBuidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("招标代理机构", "").Replace("建设单位", "").Replace("招标人", "").Replace("承包人", "").Replace("招标单位", "").Replace("招标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regMoney = new Regex(@"(中标价|投标价|总投资|发包价|投标报价|价格|金额)(：|:|)[^\r\n]+\r\n");
bidMoney = regMoney.Match(bidCtx.Replace(" ", "")).Value.Replace("中标价", "").Replace("总投资", "").Replace("发包价", "").Replace("投标报价", "").Replace("投标价", "").Replace("价格", "").Replace("金额", "").Replace("：", "").Replace(":", "").Trim();
Regex regBidUnit = new Regex(@"(第一候选人|中标候选人|中标单位|中标人|中标方|中标供应商)(：|:)[^\r\n]+\r\n");
bidUnit = regBidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("中标供应商", "").Replace("中标候选人", "").Replace("第一候选人", "").Replace("中标单位", "").Replace("中标人", "").Replace("中标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regprjMgr = new Regex(@"(项目经理姓名|项目经理|项目负责人|项目总监|建造师|总工程师|监理师)(：|:)[^\r\n]+\r\n");
prjMgr = regprjMgr.Match(bidCtx.Replace(" ", "")).Value.Replace("项目经理姓名", "").Replace("总工程师", "").Replace("项目经理", "").Replace("项目总监", "").Replace("建造师", "").Replace("监理师", "").Replace("项目负责人", "").Replace("：", "").Replace(":", "").Trim();
Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+");
if (bidMoney.Contains("万"))
{
bidMoney = bidMoney.Remove(bidMoney.IndexOf("万")).Trim();
bidMoney = regBidMoney.Match(bidMoney).Value;
}
else
{
try
{
bidMoney = (decimal.Parse(regBidMoney.Match(bidMoney).Value) / 10000).ToString();
if (decimal.Parse(bidMoney) < decimal.Parse("0.1"))
{
bidMoney = "0";
}
}
catch (Exception)
{
bidMoney = "0";
}
}
if (prjMgr.Contains("资格"))
{
prjMgr = prjMgr.Remove(prjMgr.IndexOf("资格"));
}
string ctxs = HtmlTxt.ToLower().Replace("<br/>", "\r\n").Replace("<br>", "\r\n");
bidCtx = Regex.Replace(ctxs, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n") + "\r\n";
bidUnit = ToolHtml.GetStringTemp(bidUnit);
buildUnit = ToolHtml.GetSubString(buildUnit, 150);
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
code = ToolHtml.GetSubString(code, 50);
prjMgr = ToolHtml.GetSubString(prjMgr, 50);
if (string.IsNullOrEmpty(buildUnit))
{
buildUnit = "深圳市宝安区福永街道办事处";
}
msgType = "深圳市宝安区福永街道办事处";
specType = "建设工程";
bidType = "小型工程";
prjName = ToolDb.GetPrjName(prjName);
BidInfo info = ToolDb.GenBidInfo("广东省", "深圳区及街道工程", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new ArrayList();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "pagination")));
if (sNode != null && sNode.Count > 0)
{
try
{
string temp = sNode.AsString().Replace(" ", "");
Regex reg = new Regex(@"/[^页]+页");
pageInt = Convert.ToInt32(reg.Match(temp).Value.Replace("/", "").Replace("页", ""));
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl("http://zyjy.huizhou.gov.cn/pages/cms/hzggzyjyzx/html/artList.html?cataId=a000dc84e53b4dc88e1e05d15d7c90f7&pageNo=" + i.ToString(), Encoding.UTF8);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList viewList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "div_list"))), new TagNameFilter("ul")));
if (viewList != null && viewList.Count > 0)
{
for (int j = 0; j < viewList.Count; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}");
beginDate = regDate.Match(viewList[j].ToPlainTextString()).Value;
//prjName = viewList[j].ToPlainTextString().Replace("\r", "").Replace("\n", "").Replace(beginDate, "");
ATag aTag = viewList.SearchFor(typeof(ATag), true)[j] as ATag;
prjName = aTag.GetAttribute("title");
InfoUrl = "http://zyjy.huizhou.gov.cn" + aTag.Link;
string htmDtl = string.Empty;
try
{
htmDtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8);
Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>|<style[^<]*</style>|<xml[^<]*</xml>");
htmDtl = regexHtml.Replace(htmDtl, "");
}
catch { continue; }
parser = new Parser(new Lexer(htmDtl));
NodeList dtl = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "divZoom")));
if (dtl != null && dtl.Count > 0)
{
HtmlTxt = dtl.AsHtml();
bidCtx = HtmlTxt.ToCtxString();
NodeList ifrm = new Parser(new Lexer(htmDtl)).ExtractAllNodesThatMatch(new TagNameFilter("iframe"));
if (ifrm != null && ifrm.Count > 0)
{
IFrameTag frame = ifrm[0] as IFrameTag;
string url = frame.GetAttribute("src");
try
{
string htm = this.ToolWebSite.GetHtmlByUrl(url, Encoding.Default);
NodeList tabNode = new Parser(new Lexer(htm)).ExtractAllNodesThatMatch(new TagNameFilter("table"));
string ctx = tabNode.AsHtml().ToCtxString().Replace("\r\n\t\r\n\t", "\r\n\t").Replace("\r\n\t\r\n\t", "\r\n\t").Replace("\r\n\t\r\n\t", "\r\n\t");
bidCtx = ctx + bidCtx;
}
catch { }
}
//bidCtx = System.Text.RegularExpressions.Regex.Replace(HtmlTxt, "(<script)[\\s\\S]*?(</script>)", "");
//bidCtx = System.Text.RegularExpressions.Regex.Replace(bidCtx, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\r\n", "\r\n").Replace("\r\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\t", "").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n");
Regex regPrjCode = new Regex(@"(工程编号|项目编号|招标编号|中标编号|编号)(:|：)[^\r\n]+\r\n");
code = regPrjCode.Match(bidCtx.Replace(" ", "")).Value.Replace("工程编号", "").Replace("项目编号", "").Replace("招标编号", "").Replace("中标编号", "").Replace("编号", "").Replace(":", "").Replace("：", "").Trim();
Regex regBuidUnit = new Regex(@"(建设单位|招标人|承包人|招标单位|招标方|招标代理机构)(：|:)[^\r\n]+\r\n");
buildUnit = regBuidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("招标代理机构", "").Replace("建设单位", "").Replace("招标人", "").Replace("承包人", "").Replace("招标单位", "").Replace("招标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regMoney = new Regex(@"(中标价|投标价|总投资|发包价|投标报价|价格|金额|总价)(：|:|)[^\r\n]+\r\n");
bidMoney = regMoney.Match(bidCtx.Replace(" ", "")).Value.Replace("中标价", "").Replace("总投资", "").Replace("发包价", "").Replace("总价", "").Replace("投标报价", "").Replace("投标价", "").Replace("价格", "").Replace("金额", "").Replace("：", "").Replace(":", "").Replace("，", "").Replace(",", "").Trim();
Regex regBidUnit = new Regex(@"(成交供应商|中标供应商|第一候选人|中标候选人|中标单位|中标人|中标方)(：|:)[^\r\n]+\r\n");
bidUnit = regBidUnit.Match(bidCtx.Replace(" ", "")).Value.Replace("成交供应商", "").Replace("中标供应商", "").Replace("中标候选人", "").Replace("第一候选人", "").Replace("中标单位", "").Replace("中标人", "").Replace("中标方", "").Replace("：", "").Replace(":", "").Trim();
Regex regprjMgr = new Regex(@"(项目经理姓名|项目经理（或建造师）|项目经理|项目负责人|项目总监|建造师|总工程师|监理师)(：|:)[^\r\n]+\r\n");
prjMgr = regprjMgr.Match(bidCtx.Replace(" ", "")).Value.Replace("项目经理（或建造师）", "").Replace("项目经理姓名", "").Replace("总工程师", "").Replace("项目经理", "").Replace("项目总监", "").Replace("建造师", "").Replace("监理师", "").Replace("项目负责人", "").Replace("：", "").Replace(":", "").Trim();
Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+");
if (bidMoney.Contains("万"))
{
bidMoney = bidMoney.Remove(bidMoney.IndexOf("万")).Trim();
bidMoney = regBidMoney.Match(bidMoney).Value;
}
else
{
try
{
bidMoney = (decimal.Parse(regBidMoney.Match(bidMoney).Value) / 10000).ToString();
if (decimal.Parse(bidMoney) < decimal.Parse("0.1"))
{
bidMoney = "0";
}
}
catch (Exception)
{
bidMoney = "0";
}
}
if (prjMgr.Contains("资格"))
{
prjMgr = prjMgr.Remove(prjMgr.IndexOf("资格"));
}
bidUnit = ToolHtml.GetStringTemp(bidUnit).Replace("；", "");
buildUnit = ToolHtml.GetSubString(buildUnit, 150);
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
code = ToolHtml.GetSubString(code, 50);
prjMgr = ToolHtml.GetSubString(prjMgr, 50);
msgType = "惠州市公共资源交易中心";
specType = "建设工程";
if (string.IsNullOrEmpty(buildUnit))
{
buildUnit = "惠州市公共资源交易中心";
}
bidType = ToolHtml.GetInviteTypes(prjName);
BidInfo info = ToolDb.GenBidInfo("广东省", "惠州市区", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new List <BidInfo>();
int pageInt = 1;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(html));
NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("class", "Normal")));
if (sNode != null && sNode.Count > 0)
{
try
{
string temp = sNode.AsString().Replace("createPageHTML(", "").Replace("index", "").Replace("html", "").Replace(", 0,", "").Replace(");", "").Replace(",", "").Replace(";", "").Replace(")", "").Replace("\"", "").Replace(" ", "").GetRegexBegEnd("/", "跳");
pageInt = Convert.ToInt32(temp);
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
string url = "http://lhbsc.szlhxq.gov.cn/lhbsc/bsdt43/qyfw78/zbcg2/zbxxgg/065b33d5-" + i.ToString() + ".html";
html = this.ToolWebSite.GetHtmlByUrl(url, Encoding.UTF8);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList viewList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("tr"), new HasAttributeFilter("class", "")));
if (viewList != null && viewList.Count > 0)
{
for (int j = 0; j < viewList.Count; j++)
{
string prjName = string.Empty,
buildUnit = string.Empty, bidUnit = string.Empty,
bidMoney = string.Empty, code = string.Empty,
bidDate = string.Empty,
beginDate = string.Empty,
endDate = string.Empty, bidType = string.Empty,
specType = string.Empty, InfoUrl = string.Empty,
msgType = string.Empty, bidCtx = string.Empty,
prjAddress = string.Empty, remark = string.Empty,
prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty;
Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}");
INode node = viewList[j];
ATag aTag = node.GetATag();
beginDate = regDate.Match(viewList[j].ToPlainTextString().Trim()).Value;
prjName = aTag.GetAttribute("title");
InfoUrl = "http://lhbsc.szlhxq.gov.cn" + aTag.Link.Replace("../", "").Replace("./", "");
string htlDtl = string.Empty;
try
{
htlDtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8);
Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>|<style[^<]*</style>|<xml[^<]*</xml>");
htlDtl = regexHtml.Replace(htlDtl, "");
}
catch { continue; }
parser = new Parser(new Lexer(htlDtl));
NodeList dtl = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "contentbox")));
if (dtl != null && dtl.Count > 0)
{
Regex.Replace(dtl.AsHtml(), "(<script)[\\s\\S]*?(</script>)", "");
Regex.Replace(HtmlTxt, "(<script)[\\s\\S]*?(</script>)", "");
Regex.Replace(bidCtx, "<[^>]*>", "").Replace(" ", "").Replace(" ", "").Replace("\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("¥", "");
bidType = ToolHtml.GetInviteTypes(prjName);
buildUnit = ToolHtml.GetRegexString(bidCtx, "按（建设单位）", "(提供)");
bidMoney = ToolHtml.GetRegexString(bidCtx, "(中标金额)", "(元)|(万元)|(；)").GetReplace("：", "").GetMoney("万元");
bidUnit = bidCtx.GetBidRegex();
if (string.IsNullOrWhiteSpace(bidUnit))
{
bidUnit = bidCtx.GetRegex("中标供应商名称");
}
if (bidUnit.Contains("公司"))
{
bidUnit = bidUnit.Remove(bidUnit.IndexOf("公司")) + "公司";
}
if (prjMgr.Contains("资格"))
{
prjMgr = prjMgr.Remove(prjMgr.IndexOf("资格"));
}
if (string.IsNullOrWhiteSpace(bidMoney))
{
bidMoney = bidCtx.GetRegex("中标金额").GetReplace("：", "");
}
bidUnit = ToolHtml.GetStringTemp(bidUnit);
if (string.IsNullOrWhiteSpace(buildUnit))
{
buildUnit = bidCtx.GetRegex("采购人名称");
}
bidUnit = ToolHtml.GetSubString(bidUnit, 150);
code = bidCtx.GetCodeRegex().GetReplace("）", "");
if (string.IsNullOrWhiteSpace(code))
{
code = bidCtx.GetRegexBegEnd("招标编号：", "）");
}
prjMgr = bidCtx.GetMgrRegex();
try
{
if (Convert.ToDecimal(bidMoney) > 100000)
{
bidMoney = (decimal.Parse(bidMoney) / 10000).ToString();
}
}
catch { }
if (string.IsNullOrEmpty(buildUnit))
{
buildUnit = "深圳市龙华新区龙华街道办事处";
}
msgType = "深圳市龙华新区龙华街道办事处";
specType = "建设工程";
bidType = "小型工程";
prjName = ToolDb.GetPrjName(prjName);
BidInfo info = ToolDb.GenBidInfo("广东省", "深圳区及街道工程", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType,
bidMoney, InfoUrl, prjMgr, HtmlTxt);
list.Add(info);
parser = new Parser(new Lexer(HtmlTxt));
NodeList aNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a"));
if (aNode != null && aNode.Count > 0)
{
for (int k = 0; k < aNode.Count; k++)
{
ATag a = aNode[k] as ATag;
if (a.IsAtagAttach())
{
string link = string.Empty;
if (a.Link.ToLower().Contains("http"))
{
link = a.Link;
}
BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link);
base.AttachList.Add(attach);
}
}
}
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}