protected override IList ExecuteCrawl(bool crawlAll)
{
IList list = new ArrayList();
string htl = string.Empty;
string cookiestr = string.Empty;
string viewState = string.Empty;
int page = 1;
string eventValidation = string.Empty;
try
{
htl = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr);
}
catch (Exception ex)
{
return(list);
}
Parser parser = new Parser(new Lexer(htl));
NodeList tableNodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("align", "right")));
if (tableNodeList != null && tableNodeList.Count > 0)
{
Regex regexPage = new Regex(@"共\d+页");
page = int.Parse(regexPage.Match(tableNodeList.AsString()).Value.Trim(new char[] { '共', '页' }));
}
for (int i = 1; i <= page; i++)
{
if (i > 1)
{
try
{
htl = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode("http://www.yantian.gov.cn/cn/zwgk/zfcg/zb/index_" + (i - 1).ToString() + ".shtml"), Encoding.UTF8);
}
catch (Exception ex) { continue; }
}
parser = new Parser(new Lexer(htl));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("width", "565")), true), new TagNameFilter("table")));
if (nodeList != null && nodeList.Count > 0)
{
string url = string.Empty;
for (int j = 0; j < nodeList.Count; j++)
{
string beg = nodeList[j].ToPlainTextString().GetDateRegex();
if (string.IsNullOrEmpty(beg))
{
continue;
}
else if (j > 0 && nodeList[j].GetATagHref() == url)
{
continue;
}
url = nodeList[j].GetATagHref().GetReplace("=", "=").GetReplace("&", "&");
TableTag table = nodeList[j] as TableTag;
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;
prjName = table.GetATagValue("title");
beginDate = beg;
string htmldetail = string.Empty;
if (!url.ToLower().Contains("http"))
{
InfoUrl = "http://www.yantian.gov.cn" + table.GetATagHref();
}
else
{
InfoUrl = url;
}
string htmltext = string.Empty;
string htmldttext = string.Empty;
try
{
htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Replace("<br />", "\r\n").Trim();
htmltext = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Trim();
}
catch
{
continue;
}
Parser dtlparser = new Parser(new Lexer(htmldetail));
Parser texthtml = new Parser(new Lexer(htmltext));
NodeList listtext = texthtml.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "content")));
NodeList dtnode = dtlparser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "content")));
if (dtnode != null && dtnode.Count > 0)
{
HtmlTxt = dtnode.AsHtml();
htmldttext = listtext.AsHtml();
bidCtx = dtnode.AsString().Replace(" ", "").Trim();
Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>");
bidCtx = regexHtml.Replace(bidCtx, "");
dtlparser = new Parser(new Lexer(HtmlTxt));
NodeList spanNode = dtlparser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "holder")), true), new TagNameFilter("table")));
if (spanNode != null && spanNode.Count > 1)
{
TableTag spanTable = spanNode[1] as TableTag;
string ctx = string.Empty;
for (int t = 1; t < spanTable.RowCount; t++)
{
for (int c = 0; c < spanTable.Rows[t].ColumnCount; c++)
{
ctx += spanTable.Rows[0].Columns[c].ToNodePlainString() + "：" + spanTable.Rows[t].Columns[c].ToNodePlainString() + "\r\n";
}
}
bidUnit = ctx.GetBidRegex();
if (!string.IsNullOrEmpty(bidUnit))
{
bool isBreak = false;
spanTable = spanNode[0] as TableTag;
for (int t = 1; t < spanTable.RowCount; t++)
{
for (int c = 0; c < spanTable.Rows[t].ColumnCount; c++)
{
string unit = spanTable.Rows[t].Columns[c].ToNodePlainString();
if (unit == bidUnit)
{
try
{
bidMoney = spanTable.Rows[t].Columns[c + 1].ToNodePlainString().Replace(",", "").GetMoney();
}
catch { }
isBreak = true;
break;
}
}
if (isBreak)
{
break;
}
}
}
}
else if (spanNode != null && spanNode.Count > 0)
{
TableTag spanTable = spanNode[0] as TableTag;
string ctx = string.Empty;
for (int t = 1; t < spanTable.RowCount; t++)
{
for (int c = 0; c < spanTable.Rows[t].ColumnCount; c++)
{
ctx += spanTable.Rows[0].Columns[c].ToNodePlainString() + "：" + spanTable.Rows[t].Columns[c].ToNodePlainString() + "\r\n";
}
}
bidUnit = ctx.GetBidRegex();
bidMoney = ctx.GetMoneyRegex();
}
else
{
bidUnit = bidCtx.GetBidRegex();
bidMoney = bidCtx.GetMoneyRegex();
}
if (bidCtx.Contains("项目编号："))
{
Regex regCode = new Regex(@"\w{14}");
code = regCode.Match(bidCtx.Substring(bidCtx.IndexOf("项目编号："))).Value.Trim();
}
buildUnit = bidCtx.GetBuildRegex();
bidCtx = bidCtx.Replace("<ahref=", "").Replace("/service/", "").Replace("</a>", "").Replace("您是第", "").Replace("位访问者粤ICP备06000803号", "").Replace(">", "").Trim();
bidCtx = bidCtx.Replace("<chsdatest="on"year="2012"month="01"day="16"islunardate="False"isrocdate="False">", "").Replace("</chsdate>", "").Trim();
prjName = prjName.Replace("“", "").Replace("”", "").Trim();
msgType = "深圳市盐田区政府采购中心";
specType = "建设工程";
bidType = ToolHtml.GetInviteTypes(prjName);
prjName = ToolDb.GetPrjName(prjName);
if (Encoding.Default.GetByteCount(code) > 50)
{
code = "";
}
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, htmldttext);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
else
{
BidInfo info = GetBidInfo(prjName, InfoUrl, beginDate, htmldetail);
list.Add(info);
if (!crawlAll && list.Count >= this.MaxCount)
{
return(list);
}
}
}
}
}
return(list);
}