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; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "pagination f_right")), true), new TagNameFilter("a"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode[pageNode.Count - 1].GetATagValue("onclick").GetRegexBegEnd("Info", ",").GetReplace("("); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl + "&pageSize=15&pageNum=" + i); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "newsList")), true), new TagNameFilter("li"))); if (listNode != null && listNode.Count > 0) { for (int j = 0; j < listNode.Count; j++) { INode node = listNode[j]; ATag aTag = node.GetATag(); if (aTag == null) { continue; } 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, area = string.Empty; area = node.ToNodePlainString().GetReplace("[", "【").GetReplace("]", "】").GetRegexBegEnd("【", "】"); prjName = aTag.GetAttribute("title"); beginDate = node.ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.ahtba.org.cn" + aTag.Link.GetReplace("amp;"); string id = aTag.Link.Substring(aTag.Link.IndexOf("id="), aTag.Link.Length - aTag.Link.IndexOf("id=")).GetReplace("id="); string htmldtl = string.Empty; try { NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "id" }, new string[] { id }); htmldtl = this.ToolWebSite.GetHtmlByUrl("http://www.ahtba.org.cn/Notice/NoticeContent", nvc).GetJsString(); } catch { } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "new_detail"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.ToLower().GetReplace("</p>,<br/>,<br>", "\r\n").ToCtxString(); buildUnit = bidCtx.GetBuildRegex(); code = bidCtx.GetCodeRegex().GetCodeDel(); bidUnit = bidCtx.GetBidRegex(); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = bidCtx.GetRegex("中标候选人名称,中签单位,第一成交候选人,成交候选人"); } bidMoney = bidCtx.GetMoneyRegex(); if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = bidCtx.GetMoneyRegex(null, true); } if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = bidCtx.GetRegex("总额").GetMoney(); } prjMgr = bidCtx.GetMgrRegex(); if (string.IsNullOrEmpty(bidUnit)) { parser = new Parser(new Lexer(HtmlTxt)); NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (tableNode != null && tableNode.Count > 0) { for (int t = 0; t < tableNode.Count; t++) { TableTag tag = tableNode[t] as TableTag; string classStr = tag.GetAttribute("class"); if (!string.IsNullOrEmpty(classStr) && classStr.ToLower().Contains("table_detail")) { continue; } string ctx = string.Empty; for (int r = 0; r < tag.RowCount; r++) { for (int c = 0; c < tag.Rows[r].ColumnCount; c++) { string temp = tag.Rows[r].Columns[c].ToNodePlainString().GetReplace(":,:"); if ((c + 1) % 2 == 0) { ctx += temp + "\r\n"; } else { ctx += temp + ":"; } } } bidUnit = ctx.GetBidRegex(); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = ctx.GetRegex("成交候选人,中标单位名称,第一中标候选人,第一成交候选人"); } if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = ctx.GetMoneyRegex(); } if (string.IsNullOrEmpty(prjMgr)) { prjMgr = ctx.GetMgrRegex(); } if (string.IsNullOrEmpty(prjMgr)) { prjMgr = ctx.GetRegex("拟任总监,拟任项目经理"); } if (!bidUnit.Contains("公司")) { ctx = string.Empty; try { for (int r = 1; r < tag.Rows[4].ColumnCount; r++) { string temp = tag.Rows[4].Columns[r].ToNodePlainString().GetReplace(":,:"); ctx += temp + ":"; ctx += tag.Rows[5].Columns[r].ToNodePlainString().GetReplace(":,:") + "\r\n"; } bidUnit = ctx.GetBidRegex(null, true, 200); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = ctx.GetRegex("成交候选人,中标单位名称,第一中标候选人,第一成交候选人"); } if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = ctx.GetMoneyRegex(); } if (string.IsNullOrEmpty(prjMgr)) { prjMgr = ctx.GetMgrRegex(); } if (string.IsNullOrEmpty(prjMgr)) { prjMgr = ctx.GetRegex("拟任总监,拟任项目经理"); } } catch { } } } } } try { if (Convert.ToDecimal(bidMoney) > 100000) { bidMoney = (decimal.Parse(bidMoney) / 10000).ToString(); } } catch { } if (prjMgr.Contains("联系")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("联系")); } if (prjMgr.Contains("电话")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("电话")); } if (prjMgr.Contains("2")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("2")); } if (prjMgr.Contains("(")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("(")); } if (prjMgr.Contains("(")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("(")); } if (prjMgr.Contains("二")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("二")); } if (prjMgr.Contains("注册")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("注册")); } if (prjMgr.Contains("业绩")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("业绩")); } if (prjMgr.Contains("I")) { prjMgr = prjMgr.Remove(prjMgr.IndexOf("I")); } if (prjMgr.Contains("投标") || prjMgr.IsNumber()) { prjMgr = ""; } if (bidUnit.Contains("公司")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("公司")) + "公司"; } bidUnit = bidUnit.GetReplace("名称,1,、I标段"); prjMgr = prjMgr.GetReplace("1,、,一,第一中标人,第一中标,第中标人,第名,I标段,第中标候选人,标段").GetCodeDel(); specType = bidType = "建设工程"; msgType = "安徽省发展和改革委员会"; BidInfo info = ToolDb.GenBidInfo("安徽省", "安徽省及地市", area, 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; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.Default, ref cookiestr); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "lblPageCount"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode[0].ToNodePlainString(); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(html); eventValidation = this.ToolWebSite.GetAspNetEventValidation(html); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "__EVENTTARGET", "__EVENTARGUMENT", "__LASTFOCUS", "__VIEWSTATE", "__VIEWSTATEGENERATOR", "__EVENTVALIDATION", "gcbh_Text_Box", "gcmc_TextBox", "num_TextBox", "ImageButton3.x", "ImageButton3.y" }, new string[] { "", "", "", viewState, "B0108473", eventValidation, "", "", "", "5", "12" }); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc, Encoding.Default, ref cookiestr); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "DataGrid1"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 0; j < table.RowCount; j++) { TableRow tr = table.Rows[j]; ATag aTag = tr.Columns[1].GetATag(); if (aTag == null) { continue; } 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 = aTag.LinkText.GetReplace(" "); beginDate = tr.Columns[2].ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.bcactc.com/home/gcxx/" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "hei_text"))); if (dtlNode != null && dtlNode.Count > 0) { TableTag dtlTable = dtlNode[0] as TableTag; HtmlTxt = dtlTable.ToHtml(); bidCtx = ""; for (int r = 0; r < dtlTable.RowCount; r++) { for (int c = 0; c < dtlTable.Rows[r].ColumnCount; c++) { string temp = dtlTable.Rows[r].Columns[c].ToHtml().GetReplace("<br>,<br/>", "\r\n").ToCtxString(); if (!temp.Contains("\r\n")) { temp = dtlTable.Rows[r].Columns[c].ToNodePlainString(); } if (!IsTable(dtlTable.Rows[r].ToHtml())) { if ((c + 1) % 2 == 0) { bidCtx += temp + "\r\n"; } else { bidCtx += temp.GetReplace(":,:") + ":"; } } else { bidCtx += GetTableBid(dtlTable.Rows[r].ToHtml()); } } } bidCtx = bidCtx.GetReplace(":\r\n", ":"); code = bidCtx.GetCodeRegex(); buildUnit = bidCtx.GetBuildRegex(); if (string.IsNullOrEmpty(buildUnit)) { buildUnit = bidCtx.GetRegex("建设单位名称"); } bidUnit = bidCtx.GetBidRegex(); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = bidCtx.GetRegex("中标侯选人"); } bidMoney = bidCtx.GetMoneyRegex(); if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = bidCtx.GetRegex("中标合同额").GetMoney(); } prjMgr = bidCtx.GetMgrRegex(); msgType = "北京市建设工程发包承包交易中心"; specType = "建设工程"; bidType = "勘察设计"; BidInfo info = ToolDb.GenBidInfo("北京市", "北京市区", "", string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType, bidMoney, InfoUrl, prjMgr, HtmlTxt); list.Add(info); 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; } else { link = "http://www.bcactc.com/" + 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); }
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; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.Default); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "pagebox"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode.AsString().GetRegexBegEnd("共", "页"); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl("http://www.ynbidding.net/classlist.aspx?no-cache=0.7657945008653804&id=032430863326&id=://www.ynbidding.net/list&page=" + i + "&_=", Encoding.Default); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("width", "100%"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.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, area = string.Empty; TableRow tr = table.Rows[j]; ATag aTag = tr.GetATag(); if (aTag == null) { continue; } prjName = aTag.LinkText; beginDate = tr.Columns[0].ToNodePlainString().GetDateRegex("yyyy/MM/dd"); InfoUrl = "http://www.ynbidding.net" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "Content"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.ToCtxString(); buildUnit = bidCtx.GetBuildRegex(); prjAddress = bidCtx.GetAddressRegex(); code = bidCtx.GetCodeRegex(); bidMoney = bidCtx.GetMoneyRegex(); bidUnit = bidCtx.GetBidRegex(); if (string.IsNullOrWhiteSpace(bidUnit)) { bidUnit = bidCtx.GetRegex("第一候选供应商"); } if (bidUnit.Contains("第一")) { bidUnit = string.Empty; } if (string.IsNullOrWhiteSpace(bidUnit)) { parser = new Parser(new Lexer(HtmlTxt)); NodeList bidNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (bidNode != null && bidNode.Count > 1) { string ctx = string.Empty; TableTag tag = bidNode[1] as TableTag; if (tag.RowCount > 1) { for (int c = 0; c < tag.Rows[0].ColumnCount; c++) { try { ctx += tag.Rows[0].Columns[c].ToNodePlainString().GetReplace("sourcefromwww.ynbidding.net") + ":"; ctx += tag.Rows[1].Columns[c].ToNodePlainString().GetReplace("sourcefromwww.ynbidding.net") + "\r\n"; } catch { } } } bidUnit = ctx.GetBidRegex(); if (string.IsNullOrWhiteSpace(bidUnit)) { bidUnit = ctx.GetRegex("投标单位,单位名称,投标人名称"); } if (bidMoney == "0" || string.IsNullOrWhiteSpace(bidMoney)) { bidMoney = ctx.GetMoneyRegex(); } if (bidMoney == "0" || string.IsNullOrWhiteSpace(bidMoney)) { bidMoney = ctx.GetMoneyString().GetMoney(); } } } if (bidUnit.Contains("公司")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("公司")) + "公司"; } if (bidUnit.Contains("第一")) { bidUnit = string.Empty; } bidType = prjName.GetInviteBidType(); specType = "建设工程"; msgType = "云南省发展和改革委员会"; BidInfo info = ToolDb.GenBidInfo("云南省", "云南省及地市", area, 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; } else { link = "http://www.ynbidding.net/" + 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); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <InviteInfo>(); foreach (string area in this.DicSiteUrl.Keys) { int pageInt = 1, count = 0; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.DicSiteUrl[area], 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", "page"))); if (sNode != null && sNode.Count > 0) { try { string page = sNode.AsString().ToNodeString().Replace("createPageHTML(", ""); string temp = page.Remove(page.IndexOf(",")); pageInt = Convert.ToInt32(temp); } catch (Exception) { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(this.DicSiteUrl[area] + "index_" + (i - 1) + ".html".ToString(), Encoding.UTF8); } catch { continue; } } parser = new Parser(new Lexer(html)); sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "secondrightlistbox"))), new TagNameFilter("ul"))), new TagNameFilter("li"))); if (sNode != null && sNode.Count > 0) { for (int t = 0; t < sNode.Count; t++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; beginDate = sNode[t].ToNodePlainString().GetDateRegex(); prjName = sNode[t].GetATagValue("title"); InfoUrl = this.DicSiteUrl[area] + sNode[t].GetATagHref().Replace("./", ""); string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Trim(); Parser dtlparserHTML = new Parser(new Lexer(htmldetail)); NodeList dtnodeHTML = dtlparserHTML.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "contentrightlistbox2"))); HtmlTxt = dtnodeHTML.AsHtml(); htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Replace("</br>", "\r\n").Replace("<br>", "\r\n").Replace("<br/>", "\r\n"); } catch (Exception ex) { continue; } Parser dtlparser = new Parser(new Lexer(htmldetail)); NodeList dtnode = dtlparser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "contentrightlistbox2"))); Regex regexCtx = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>"); inviteCtx = regexCtx.Replace(dtnode.AsString(), "").Replace(" ", ""); Regex regPrjAdd = new Regex(@"(工程地点|工程地址|项目地址)[:|:][^\r\n]+[\r\n]{1}"); prjAddress = regPrjAdd.Match(inviteCtx).Value.Replace("工程地点", "").Replace("工程地址", "").Replace("项目地址", "").Replace(":", "").Replace(":", "").Replace(")", "").Trim(); Regex regbuildUnit = new Regex(@"(招标单位|招标人|采购人):[^\r\n]+[\r\n]{1}"); buildUnit = regbuildUnit.Match(inviteCtx).Value.Replace("招标单位:", "").Replace("招标人:", "").Replace("采购人:", "").Trim(); if (buildUnit.Contains("招标代理机构")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("招标代理机构")); } if (buildUnit.Contains("联系人")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("联系人")); } if (buildUnit.Contains(";")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf(";")); } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")); } msgType = "佛山市建设工程交易中心"; specType = "政府采购"; inviteType = ToolHtml.GetInviteTypes(prjName); string are = area != "市直" ? area : ""; InviteInfo info = ToolDb.GenInviteInfo("广东省", "佛山市区", are, string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, HtmlTxt); count++; 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; } else { link = "http://www.fsggzy.cn/" + a.Link.GetReplace("../,./"); } if (Encoding.Default.GetByteCount(link) > 500) { continue; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } if (!crawlAll && count >= this.MaxCount) { goto Funcs; } } } } Funcs :; } 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 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(this.ToolWebSite.UrlEncode(SiteUrl), Encoding.Default, ref cookiestr); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(htl)); NodeList tableNodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "fanyie"))); 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 j = 1; j < page; j++) { //if (j > 1) //{ // try // { // htl = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(GetStartUrl() + "&ipage=" + j.ToString()), Encoding.Default); // } // catch (Exception ex) { continue; } //} parser = new Parser(new Lexer(htl)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("ul"), new HasAttributeFilter("class", "list_headnews")), true), new TagNameFilter("li"))); for (int i = 0; i < nodeList.Count; i++) { ATag aTag = nodeList.SearchFor(typeof(ATag), true)[i] as ATag; string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; prjName = nodeList[i].ToPlainTextString().Replace(" ", ""); InfoUrl = "http://gcjs.yunfu.gov.cn" + aTag.Link; Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}"); beginDate = regDate.Match(prjName).Value.Trim(); prjName = prjName.Replace(beginDate, "").Trim(); if (prjName.Contains("招标公告") || prjName.Contains("补充公告")) { inviteType = ToolHtml.GetInviteTypes(prjName); string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(InfoUrl), Encoding.Default).Replace(" ", ""); } catch (Exception) { Logger.Error("InviteYunFu"); continue; } Parser parserdetail = new Parser(new Lexer(htmldetail)); NodeList dtnode = parserdetail.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("class", "STYLE10"))); if (dtnode.Count <= 0) { parserdetail = new Parser(new Lexer(htmldetail)); dtnode = parserdetail.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("TABLE"), new HasAttributeFilter("id", "Table1"))); } if (dtnode.Count > 0) { inviteCtx = dtnode.AsString().Replace("\n", "\r\n"); HtmlTxt = dtnode.AsHtml(); Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>"); inviteCtx = regexHtml.Replace(inviteCtx, ""); Regex regBuidUnit = new Regex(@"(招 标 人|招标人|招 标人)(:|:)[^\r\n]+\r\n"); buildUnit = regBuidUnit.Match(inviteCtx).Value.Replace("招 标 人:", "").Replace("招 标人:", "").Replace("招标人:", "").Trim(); if (buildUnit.Contains("招标代理")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("招标代理")).ToString().Trim(); } msgType = "云浮市工程建设交易中心"; specType = "建设工程"; if (buildUnit == "") { buildUnit = ""; } prjAddress = "见招标信息"; InviteInfo info = ToolDb.GenInviteInfo("广东省", "云浮市区", "", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, HtmlTxt); list.Add(info); if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { string url = "http://www.hljztb.com/ajaxtools.ashx"; Dictionary <string, string> ggType = new Dictionary <string, string>(); ggType.Add("勘察设计", "18101"); ggType.Add("施工", "18102"); ggType.Add("监理", "18103"); ggType.Add("设备", "18104"); IList list = new List <BidInfo>(); foreach (string key in ggType.Keys) { int pageInt = 1; int count = 0; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "dopost", "pagesize", "CategoryID", "sort", "keyword", "pageno" }, new string[] { "product_list", "5", "6", ggType[key], "", "1" }); try { html = this.ToolWebSite.GetHtmlByUrl(url, nvc); } catch { return(list); } JavaScriptSerializer serializer = new JavaScriptSerializer(); Dictionary <string, object> smsTypeJson = (Dictionary <string, object>)(((Dictionary <string, object>)serializer.DeserializeObject(html))["listpage"]); pageInt = Convert.ToInt32(smsTypeJson["pagecount"]); for (int i = 1; i <= pageInt; i++) { if (i > 1) { nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "dopost", "pagesize", "CategoryID", "sort", "keyword", "pageno" }, new string[] { "product_list", "5", "6", ggType[key], "", i.ToString() }); try { html = this.ToolWebSite.GetHtmlByUrl(url, nvc); } catch { continue; } serializer = new JavaScriptSerializer(); smsTypeJson = (Dictionary <string, object>)(((Dictionary <string, object>)serializer.DeserializeObject(html))["listpage"]); } object[] listDatas = (object[])smsTypeJson["listdata"]; foreach (object obj in listDatas) { Dictionary <string, object> dic = (Dictionary <string, object>)obj; 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, area = string.Empty; string tempName = Convert.ToString(dic["Name"]); area = tempName.GetRegexBegEnd("【", "】"); prjName = tempName.GetReplace("【" + area + "】"); beginDate = Convert.ToString(dic["FTime"]); InfoUrl = "http://www.hljztb.com/" + dic["SUrl"]; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } Parser parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "lblFZBContent"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.GetReplace("<br/>,<br>,</p>", "\r\n").ToCtxString(); //TableTag table = dtlNode[0] as TableTag; //for (int r = 0; r < table.RowCount; r++) //{ // for (int c = 0; c < table.Rows[r].ColumnCount; c++) // { // string temp = table.Rows[r].Columns[c].ToNodePlainString(); // if ((c + 1) % 2 == 0) // bidCtx += temp.GetReplace(":,:") + "\r\n"; // else // bidCtx += temp.GetReplace(":,:") + ":"; // } //} prjAddress = bidCtx.GetAddressRegex(); buildUnit = bidCtx.GetBuildRegex(); bidUnit = bidCtx.GetBidRegex(); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = bidCtx.GetRegex("第一名,预中标单位"); } bidMoney = bidCtx.GetMoneyRegex(); if (bidMoney == "0" || string.IsNullOrEmpty(bidMoney)) { bidMoney = bidCtx.GetRegex("中标造价,造价,预 中 标 价,预中标价格").GetMoney(); } prjMgr = bidCtx.GetMgrRegex(); code = bidCtx.GetCodeRegex().GetCodeDel(); if (buildUnit.Contains("公司")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("公司")) + "公司"; } if (buildUnit.Contains("联系")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("联系")); } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")); } if (string.IsNullOrEmpty(code)) { code = bidCtx.GetRegex("编码"); } if (bidUnit.Contains("公司")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("公司")) + "公司"; } if (bidUnit.Contains("研究院")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("研究院")) + "研究院"; } if (bidUnit.Contains("研究所")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("研究所")) + "研究所"; } bidType = key; specType = "建设工程"; msgType = "黑龙江住房和城乡建设厅"; BidInfo info = ToolDb.GenBidInfo("黑龙江省", "黑龙江省及地市", area, string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType, bidMoney, InfoUrl, prjMgr, HtmlTxt); list.Add(info); count++; 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; } else { link = "http://www.hljztb.com/" + a.Link.GetReplace("../,./"); } if (Encoding.Default.GetByteCount(link) > 500) { continue; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } if (!crawlAll && count >= this.MaxCount) { goto Found; } } } } Found :; } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8); } catch { try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8); } catch { return(list); } } int startIndex = html.LastIndexOf("<xml"); int endIndex = html.LastIndexOf("</xml>"); string xmlstr = html.Substring(startIndex, endIndex - startIndex).ToLower().GetReplace("infourl", "span").GetReplace("info", "div").GetReplace("publishedtime", "p"); Parser parser = new Parser(new Lexer(xmlstr)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("div")); if (pageNode != null && pageNode.Count > 0) { for (int i = 0; i < pageNode.Count; i++) { parser = new Parser(new Lexer(pageNode[i].ToHtml())); NodeList dateNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("p")); string beginDate = dateNode[0].ToPlainTextString().GetDateRegex(); parser.Reset(); NodeList urlNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("span")); string InfoUrl = "http://www.zhongtang.gov.cn/business/htmlfiles/" + urlNode[0].ToPlainTextString(); string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } } parser = new Parser(new Lexer(htmldtl)); NodeList titleNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "zoomtitl"))); string prjName = string.Empty; if (titleNode != null && titleNode.Count > 0) { prjName = titleNode[0].ToNodePlainString().GetReplace(" "); } parser.Reset(); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("width", "778"))); if (dtlNode != null && dtlNode.Count > 0) { if (prjName.Contains("中标")) { string buildUnit = string.Empty, bidUnit = string.Empty, bidMoney = string.Empty, code = string.Empty, bidDate = string.Empty, endDate = string.Empty, bidType = string.Empty, specType = string.Empty, msgType = string.Empty, bidCtx = string.Empty, prjAddress = string.Empty, remark = string.Empty, prjMgr = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; HtmlTxt = dtlNode[0].ToHtml(); bidCtx = HtmlTxt.ToLower().GetReplace("</p>,</br>,<br>", "\r\n").ToCtxString(); buildUnit = bidCtx.GetBuildRegex(); code = bidCtx.GetCodeRegex().GetCodeDel(); bidUnit = bidCtx.GetBidRegex(); bidMoney = bidCtx.GetRegex("中标值").GetMoney(); if (string.IsNullOrEmpty(bidMoney) || bidMoney == "0") { bidMoney = bidCtx.GetMoneyRegex(); } try { if (decimal.Parse(bidMoney) < 1) { bidMoney = "0"; } } catch { } prjMgr = bidCtx.GetMgrRegex(); specType = "政府采购"; bidType = prjName.GetInviteBidType(); msgType = "东莞市中堂镇政府"; BidInfo info = ToolDb.GenBidInfo("广东省", "东莞市区", "中堂镇", 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].GetATag(); if (a.IsAtagAttach()) { string link = string.Empty; if (a.Link.ToLower().Contains("http")) { link = a.Link; } else { link = "http://www.zhongtang.gov.cn/" + a.Link; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } else { string code = string.Empty, buildUnit = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty, area = string.Empty; HtmlTxt = dtlNode[0].ToHtml(); inviteCtx = HtmlTxt.ToLower().GetReplace("</p>,</br>,<br>", "\r\n").ToCtxString(); buildUnit = inviteCtx.GetBuildRegex(); prjAddress = inviteCtx.GetAddressRegex(); code = inviteCtx.GetCodeRegex().GetCodeDel(); if (buildUnit.Contains("招标代理")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("招标代理")); } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")); } specType = "政府采购"; inviteType = prjName.GetInviteBidType(); msgType = "东莞市中堂镇政府"; InviteInfo info = ToolDb.GenInviteInfo("广东省", "东莞市区", "中堂镇", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, 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].GetATag(); if (a.IsAtagAttach()) { string link = string.Empty; if (a.Link.ToLower().Contains("http")) { link = a.Link; } else { link = "http://www.zhongtang.gov.cn/" + 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); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <InviteInfo>(); int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl); } catch { } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("class", "countPage"))); if (pageNode != null && pageNode.Count > 0) { try { pageInt = int.Parse(pageNode[0].ToNodePlainString().GetRegexBegEnd("共", "页")); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "method", "page", "id", "step", "view", "st", "ldateQGE", "ldateQLE" }, new string[] { "view", i.ToString(), "1665", "1", "Infor", "1", "", "" }); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "reflshPage")), true), new TagNameFilter("li"))); if (listNode != null && listNode.Count > 0) { for (int j = 0; j < listNode.Count; j++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; ATag aTag = listNode[j].GetATag(); if (aTag == null) { continue; } prjName = aTag.GetAttribute("title"); string tempCode = prjName.GetReplace(" (项目编号:", "kdxx").GetReplace(")", ")").GetRegexBegEnd("kdxx", ")"); code = tempCode.GetReplace("目编号:,目编号:"); beginDate = listNode[j].ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.tjgp.gov.cn" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("body")); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); inviteCtx = System.Web.HttpUtility.HtmlDecode(HtmlTxt.ToCtxString()); buildUnit = inviteCtx.GetBuildRegex(); prjAddress = inviteCtx.GetAddressRegex(); inviteType = prjName.GetInviteBidType(); specType = "政府采购"; msgType = "天津政府采购办公室"; InviteInfo info = ToolDb.GenInviteInfo("天津市", "天津市区", "", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, 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 a = 0; a < aNode.Count; a++) { ATag aFile = aNode[a] as ATag; if (aFile.IsAtagAttach()) { string link = string.Empty; if (aFile.Link.Contains("http")) { link = aFile.Link; } else { link = "http://www.tjgp.gov.cn/" + aFile.Link; } string text = System.Web.HttpUtility.HtmlDecode(aFile.LinkText); base.AttachList.Add(ToolDb.GenBaseAttach(text, info.Id, link)); } } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <NotifyInfo>(); int pageInt = 1, sqlCount = 0; string html = string.Empty; string viewState = string.Empty; string cookiestr = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "pager")), true), new TagNameFilter("a"))); if (pageNode != null && pageNode.Count > 0) { try { Regex reg = new Regex(@"[0-9]+"); string temp = reg.Match(pageNode[pageNode.Count - 1].GetATagHref().Replace("'", "")).Value; pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(html); eventValidation = this.ToolWebSite.GetAspNetEventValidation(html); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection( new string[] { "__VIEWSTATE", "__EVENTTARGET", "__EVENTARGUMENT", "__VIEWSTATEENCRYPTED", "__EVENTVALIDATION", "hsa1$DD_LX", "hsa1$wd", "pager_input" }, new string[] { viewState, "pager", i.ToString(), "", eventValidation, "综合搜索", "", (i - 1).ToString() } ); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc, Encoding.UTF8, ref cookiestr); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "GV1"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.RowCount - 1; j++) { string headName = string.Empty, releaseTime = string.Empty, infoScorce = string.Empty, msgType = string.Empty, infoUrl = string.Empty, ctxHtml = string.Empty, infoCtx = string.Empty, infoType = string.Empty; TableRow tr = table.Rows[j]; headName = tr.Columns[0].ToNodePlainString(); releaseTime = tr.Columns[1].ToPlainTextString(); infoUrl = "http://www.ycsggzy.cn/" + tr.Columns[0].GetATagHref().Replace("&", "&"); string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(infoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "Lb_nr"))); if (dtlNode != null && dtlNode.Count > 0) { ctxHtml = dtlNode.AsHtml(); infoCtx = ctxHtml.ToCtxString(); msgType = "银川市公共资源交易中心"; NotifyInfo info = ToolDb.GenNotifyInfo(headName, releaseTime, infoScorce, msgType, infoUrl, ctxHtml, "宁夏回族自治区", "宁夏回族自治区及地市", "银川市", infoCtx, "通知公告"); sqlCount++; if (!crawlAll && sqlCount >= this.MaxCount) { return(null); } if (ToolDb.SaveEntity(info, this.ExistCompareFields, this.ExistsUpdate)) { parser = new Parser(new Lexer(ctxHtml)); NodeList aNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a")); if (aNode != null && aNode.Count > 0) { for (int k = 0; k < aNode.Count; k++) { ATag fileATag = aNode[k].GetATag(); if (fileATag.IsAtagAttach()) { BaseAttach obj = null; try { if (fileATag.Link.ToLower().Contains("http")) { obj = ToolHtml.GetBaseAttach(fileATag.Link, headName, info.Id); } else { obj = ToolHtml.GetBaseAttach("http://www.ycsggzy.cn" + fileATag.Link, headName, info.Id); } } catch { } if (obj != null) { ToolDb.SaveEntity(obj, string.Empty); } } } } else { parser.Reset(); NodeList imgNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("img")); if (imgNode != null && imgNode.Count > 0) { for (int k = 0; k < imgNode.Count; k++) { ImageTag img = imgNode[0] as ImageTag; BaseAttach obj = null; try { if (img.ImageURL.ToLower().Contains("http")) { obj = ToolHtml.GetBaseAttach(img.ImageURL, headName, info.Id); } else { obj = ToolHtml.GetBaseAttach("http://www.ycsggzy.cn" + img.ImageURL, headName, info.Id); } } catch { } if (obj != null) { ToolDb.SaveEntity(obj, string.Empty); } } } } } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); //取得页码 int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(SiteUrl), Encoding.UTF8); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("name", "__ec_pages"))); if (pageList != null && pageList.Count > 0) { try { SelectTag selectTag = pageList[0] as SelectTag; pageInt = selectTag.OptionTags.Length; } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "ec_i", "topicChrList_20070702_crd", "topicChrList_20070702_f_a", "topicChrList_20070702_p", "topicChrList_20070702_s_name", "id", "method", "__ec_pages", "topicChrList_20070702_rd", "topicChrList_20070702_f_name", "topicChrList_20070702_f_ldate" }, new string[] { "topicChrList_20070702", "20", "", i.ToString(), "", "709", "view", i.ToString(), "20", "", "" }); try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, nvc, Encoding.UTF8); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList dtList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasAttributeFilter("id", "topicChrList_20070702_table"), new TagNameFilter("table"))); if (dtList != null && dtList.Count > 0) { TableTag table = dtList[0] as TableTag; for (int j = 3; j < table.RowCount; j++) { TableRow tr = table.Rows[j]; string headName = string.Empty, releaseTime = string.Empty, infoScorce = string.Empty, msgType = string.Empty, infoUrl = string.Empty, ctxHtml = string.Empty, infoCtx = string.Empty, infoType = string.Empty; headName = tr.Columns[1].ToPlainTextString().Trim(); releaseTime = tr.Columns[2].ToPlainTextString().Trim(); infoType = "政策法规"; ATag aTag = tr.Columns[1].SearchFor(typeof(ATag), true)[0] as ATag; infoUrl = "http://www.szzfcg.cn/portal/documentView.do?method=view&id=" + aTag.Link.Replace("/viewer.do?id=", ""); string htmldeil = string.Empty; try { htmldeil = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(infoUrl), Encoding.UTF8); } catch { continue; } Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>|<style[^<]*</style>|<xml[^<]*</xml>"); htmldeil = regexHtml.Replace(htmldeil, ""); parser = new Parser(new Lexer(htmldeil)); NodeFilter filter = new TagNameFilter("body"); NodeList noList = parser.ExtractAllNodesThatMatch(filter); if (noList != null && noList.Count > 0) { ctxHtml = noList.AsHtml().Replace("<br/>", "\r\n").Replace("<BR/>", ""); infoCtx = noList.AsString().Replace(" ", "").Replace(" ", "").Replace("\t\t", "\t").Replace("\t\t", "\t"); infoCtx = Regex.Replace(infoCtx, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase).Replace(" ", "").Replace("\t", ""); msgType = "深圳政府采购"; infoScorce = infoScorce.Replace(" ", ""); NotifyInfo info = ToolDb.GenNotifyInfo(headName, releaseTime, infoScorce, msgType, infoUrl, ctxHtml, "广东省", "深圳市工程", string.Empty, infoCtx, infoType); list.Add(info); if (crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); string htl = string.Empty; string cookiestr = string.Empty; string viewState = string.Empty; int pageInt = 1; string eventValidation = string.Empty; try { htl = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr); } catch { return(null); } Parser parser = new Parser(new Lexer(htl)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "ctl00_cph_context_ZsjyjgsList2_GridViewPaging1_PagingDescTd"))); if (nodeList != null && nodeList.Count > 0) { try { string pagestr = nodeList[0].ToPlainTextString().Trim(); string[] page = pagestr.Split(','); pageInt = int.Parse(page[page.Length - 1].Replace("共", "").Replace("页", "")); } catch { pageInt = 1; } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(htl); eventValidation = this.ToolWebSite.GetAspNetEventValidation(htl); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "ctl00$ScriptManager1", "__EVENTTARGET", "__EVENTARGUMENT", "__VIEWSTATE", "ctl00$cph_context$ZsjyjgsList2$ddlSearch", "ctl00$cph_context$ZsjyjgsList2$txtTitle", "ctl00$cph_context$ZsjyjgsList2$txtStartTime", "ctl00$cph_context$ZsjyjgsList2$txtEndTime", "ctl00$cph_context$ZsjyjgsList2$GridViewPaging1$txtGridViewPagingForwardTo", "__VIEWSTATEENCRYPTED", "__EVENTVALIDATION", "ctl00$cph_context$ZsjyjgsList2$GridViewPaging1$btnForwardToPage" }, new string[] { "ctl00$cph_context$ZsjyjgsList2$UpdatePanel2|ctl00$cph_context$ZsjyjgsList2$GridViewPaging1$btnForwardToPage", "", "", viewState, "xxbt", "", "", "", i.ToString(), "", eventValidation, "GO" }); try { htl = this.ToolWebSite.GetHtmlByUrl(SiteUrl, nvc, Encoding.UTF8, ref cookiestr); } catch { } } parser = new Parser(new Lexer(htl)); NodeList dtList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "ctl00_cph_context_ZsjyjgsList2_GridView1"))); if (dtList != null && dtList.Count > 0) { TableTag table = dtList[0] as TableTag; for (int j = 1; j < table.RowCount; j++) { string bProv = string.Empty, bCity = string.Empty, bArea = string.Empty, bPrjno = string.Empty, bPrjname = string.Empty, bBidresultendtime = string.Empty, bBaseprice = string.Empty, bBiddate = string.Empty, bBuildunit = string.Empty, bBidmethod = string.Empty, bRemark = string.Empty, bInfourl = string.Empty; TableRow tr = table.Rows[j]; bPrjname = tr.Columns[2].ToPlainTextString().Trim(); bBuildunit = tr.Columns[3].ToPlainTextString().Trim(); bBiddate = tr.Columns[4].ToPlainTextString().Trim(); ATag aTag = tr.Columns[2].SearchFor(typeof(ATag), true)[0] as ATag; bInfourl = "http://jyzx.cb.gov.cn/LGjyzxWeb/SiteManage/" + aTag.Link; BidProject info = ToolDb.GenResultProject("广东省", "深圳市", "龙岗区", bPrjno, bPrjname, bBidresultendtime, bBaseprice, bBiddate, bBuildunit, bBidmethod, bRemark, bInfourl); string sql = string.Format("select Id from BidProject where 1=1 and PrjNo='{0}' and PrjName='{1}'", info.PrjNo, info.PrjName); string result = Convert.ToString(ToolDb.ExecuteScalar(sql)); if (!string.IsNullOrEmpty(result)) { if (ToolDb.SaveEntity(info, this.ExistCompareFields)) { string htltxt = string.Empty; try { htltxt = this.ToolWebSite.GetHtmlByUrl(bInfourl, Encoding.UTF8); } catch { } Parser par = new Parser(new Lexer(htltxt)); NodeList fileList = par.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "ctl00_cph_context_AccessoriesControl1_GridView1"))); if (fileList != null && fileList.Count > 0) { string sqlDelete = string.Format("delete from BaseAttach where SourceId='{0}'", result); ToolDb.ExecuteSql(sqlDelete); TableTag tab = fileList[0] as TableTag; for (int k = 1; k < tab.RowCount; k++) { TableRow dr = tab.Rows[k]; ATag aLink = dr.Columns[1].SearchFor(typeof(ATag), true)[0] as ATag; string data = DateTime.Now.Year.ToString() + DateTime.Now.Month.ToString() + "\\"; string annexName = ToolDb.NewGuid; FilesClass file = new FilesClass(); file.strUrl = "http://jyzx.cb.gov.cn/LGjyzxWeb/" + aLink.Link.Replace("../", ""); int index = aLink.LinkText.IndexOf("."); string fileName = annexName + aLink.LinkText.Substring(index, aLink.LinkText.Length - index); file.strFileName = fileName; file.strFile = data; long size = file.DownLoadFile(); if (size > 1024) { BaseAttach baseInfo = ToolDb.GenBaseAttach(annexName, aLink.LinkText, info.Id, data + fileName, size.ToString(), ""); ToolDb.SaveEntity(baseInfo, ""); } } } } } else { //开始下载附件 if (ToolDb.SaveEntity(info, this.ExistCompareFields)) { string htltxt = string.Empty; try { htltxt = this.ToolWebSite.GetHtmlByUrl(bInfourl, Encoding.UTF8); } catch { } Parser par = new Parser(new Lexer(htltxt)); NodeList fileList = par.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "ctl00_cph_context_AccessoriesControl1_GridView1"))); if (fileList != null && fileList.Count > 0) { TableTag tab = fileList[0] as TableTag; for (int k = 1; k < tab.RowCount; k++) { TableRow dr = tab.Rows[k]; ATag aLink = dr.Columns[1].SearchFor(typeof(ATag), true)[0] as ATag; string data = DateTime.Now.Year.ToString() + DateTime.Now.Month.ToString() + "\\"; string annexName = ToolDb.NewGuid; FilesClass file = new FilesClass(); file.strUrl = "http://jyzx.cb.gov.cn/LGjyzxWeb/" + aLink.Link.Replace("../", ""); int index = aLink.LinkText.IndexOf("."); string fileName = annexName + aLink.LinkText.Substring(index, aLink.LinkText.Length - index); file.strFileName = fileName; file.strFile = data; long size = file.DownLoadFile(); if (size > 1024) { BaseAttach baseInfo = ToolDb.GenBaseAttach(annexName, aLink.LinkText, info.Id, data + fileName, size.ToString(), ""); ToolDb.SaveEntity(baseInfo, ""); } } } } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookieStr = string.Empty; int pageInt = 1; try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "ctl00_cph_context_GridViewPaingTwo1_PagingDescTd"))); if (pageList != null && pageList.Count > 0) { try { string temp = pageList.AsString().GetRegexBegEnd("共", "页"); pageInt = int.Parse(temp); } catch { pageInt = 1; } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { viewState = this.ToolWebSite.GetAspNetViewState(html); eventValidation = this.ToolWebSite.GetAspNetEventValidation(html); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection( new string[] { "__VIEWSTATE", "__EVENTVALIDATION", "ctl00$cph_context$drp_selSeach", "ctl00$cph_context$txt_strWhere", "ctl00$cph_context$drp_Rq", "ctl00$cph_context$GridViewPaingTwo1$txtGridViewPagingForwardTo", "ctl00$cph_context$GridViewPaingTwo1$btnForwardToPage" }, new string[] { viewState, eventValidation, "1", "", "3", i.ToString(), "GO" }); html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, nvc, Encoding.UTF8); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "ctl00_cph_context_GridView1"))); if (nodeList != null && nodeList.Count > 0) { TableTag table = nodeList[0] as TableTag; for (int j = 1; j < table.RowCount; j++) { string InfoTitle = string.Empty, InfoType = string.Empty, PublistTime = string.Empty, InfoCtx = string.Empty, InfoUrl = string.Empty, prjCode = string.Empty, buildUnit = string.Empty, htmlTxt = string.Empty; TableRow tr = table.Rows[j]; InfoType = "补充通知"; InfoTitle = tr.Columns[2].ToNodePlainString(); prjCode = tr.Columns[1].ToNodePlainString(); PublistTime = tr.Columns[4].ToPlainTextString().GetDateRegex(); InfoUrl = SiteUrl; InfoCtx = "招标编号:" + prjCode + "\r\n工程名称:" + InfoTitle + "\r\n附件类型:" + tr.Columns[3].ToNodePlainString() + "\r\n上传时间:" + PublistTime; htmlTxt = InfoCtx; NoticeInfo info = ToolDb.GenNoticeInfo("广东省", "东莞市区", string.Empty, string.Empty, InfoTitle, InfoType, InfoCtx, PublistTime, string.Empty, MsgTypeCosnt.DongGuanMsgType, InfoUrl, prjCode, buildUnit, string.Empty, string.Empty, string.Empty, string.Empty, htmlTxt); list.Add(info); ATag fileUrl = tr.Columns[2].GetATag(); if (fileUrl != null) { string alink = "http://www.dgzb.com.cn/DGJYWEB/SiteManage/" + fileUrl.Link; BaseAttach attach = ToolDb.GenBaseAttach(fileUrl.LinkText.Replace(" ", "").Replace(";", "").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 <InviteInfo>(); int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string cookiestr = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8, ref cookiestr); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "pager")), true), new TagNameFilter("a"))); if (pageNode != null && pageNode.Count > 0) { try { Regex reg = new Regex(@"[0-9]+"); string temp = reg.Match(pageNode[pageNode.Count - 1].GetATagHref().Replace("'", "")).Value; pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(html); eventValidation = this.ToolWebSite.GetAspNetEventValidation(html); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection( new string[] { "__VIEWSTATE", "__EVENTTARGET", "__EVENTARGUMENT", "__VIEWSTATEENCRYPTED", "__EVENTVALIDATION", "hsa1$DD_LX", "hsa1$wd", "pager_input" }, new string[] { viewState, "pager", i.ToString(), "", eventValidation, "综合搜索", "", (i - 1).ToString() } ); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc, Encoding.UTF8, ref cookiestr); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "GV1"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.RowCount - 1; j++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; TableRow tr = table.Rows[j]; prjName = tr.Columns[0].ToNodePlainString(); beginDate = tr.Columns[1].ToPlainTextString(); InfoUrl = "http://www.ycsggzy.cn/" + tr.Columns[0].GetATagHref().Replace("&", "&"); string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "Lb_nr"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.ToHtml(); inviteCtx = HtmlTxt.ToCtxString(); code = inviteCtx.GetCodeRegex(); buildUnit = inviteCtx.GetBuildRegex(); if (buildUnit.Contains(" ")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf(" ")); } prjAddress = inviteCtx.GetAddressRegex().Replace(" ", ""); inviteType = prjName.GetInviteBidType(); specType = "政府采购"; msgType = "银川市公共资源交易中心"; InviteInfo info = ToolDb.GenInviteInfo("宁夏回族自治区", "宁夏回族自治区及地市", "银川市", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, HtmlTxt); list.Add(info); if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <InviteInfo>(); int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "pages")), true), new TagNameFilter("a"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode[pageNode.Count - 1].GetATagHref(); temp = temp.Substring(temp.ToLower().IndexOf("p="), temp.Length - temp.ToLower().IndexOf("p=")).GetReplace("P,p,="); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl("http://ggzy.tylr.gov.cn/JiaoYiInfo.aspx?pn=22-26-1.xtj&p=" + i); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "newlist pList")), true), new TagNameFilter("a"))); if (listNode != null && listNode.Count > 0) { for (int j = 0; j < listNode.Count; j++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty, area = string.Empty; INode node = listNode[j]; ATag aTag = node.GetATag(); prjName = aTag.LinkText.GetReplace(" "); beginDate = node.ToPlainTextString().GetDateRegex(); if (!string.IsNullOrEmpty(beginDate)) { prjName = prjName.GetReplace(beginDate); } InfoUrl = "http://ggzy.tylr.gov.cn/" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "Details"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); inviteCtx = HtmlTxt.ToCtxString(); prjAddress = inviteCtx.GetAddressRegex(); buildUnit = inviteCtx.GetBuildRegex().GetReplace("&mdash,&emsp"); if (buildUnit.Contains("公司")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("公司")) + "公司"; } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")) + "地址"; } code = inviteCtx.GetCodeRegex().GetCodeDel(); msgType = "太原市公共资源交易中心"; specType = "政府采购"; inviteType = "建设工程"; InviteInfo info = ToolDb.GenInviteInfo("山西省", "山西省及地市", "太原市", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, 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; } else { link = "http://ggzy.tylr.gov.cn/" + a.Link.GetReplace("../,./"); } if (Encoding.Default.GetByteCount(link) > 500) { continue; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <InviteInfo>(); string html = string.Empty; int pageInt = 1; string nextPage = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("class", "Normal"))); if (pageNode != null && pageNode.Count > 0) { string temp = pageNode.AsString().GetRegexBegEnd("总记录数:", ",每页显示"); string sum = pageNode.AsString().GetRegexBegEnd("每页显示", "条记录"); try { pageInt = int.Parse(temp) / int.Parse(sum) + 1; } catch { } parser = new Parser(new Lexer(pageNode.AsHtml())); NodeList aNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a")); if (aNode != null && aNode.Count > 0) { for (int a = 0; a < aNode.Count; a++) { ATag aTag = aNode[a].GetATag(); if (aTag.LinkText.Contains("下一页")) { nextPage = "http://www.szns.gov.cn" + aTag.Link; break; } } } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { string inurl = "http://www.szns.gov.cn/jyj/xxgk6/qt74/zbgg79/14b4b8b0-" + i + ".html"; html = this.ToolWebSite.GetHtmlByUrl(inurl); } catch { continue; } parser = new Parser(new Lexer(html)); NodeList aNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "partPage")), true), new TagNameFilter("a"))); if (aNode != null && aNode.Count > 0) { for (int a = 0; a < aNode.Count; a++) { ATag aTag = aNode[a].GetATag(); if (aTag.LinkText.Contains("下一页")) { nextPage = "http://www.szns.gov.cn" + aTag.Link; break; } } } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("style", "width:100%;border-collapse:collapse;"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.RowCount; j++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty, area = string.Empty; TableRow tr = table.Rows[j]; ATag aTag = tr.Columns[0].GetATag(); prjName = aTag.GetAttribute("title"); string m = tr.ChildrenHTML.ToString(); beginDate = m.GetRegexBegEnd("<span>", "</span>").GetDateRegex(); InfoUrl = aTag.Link.GetReplace("&", "&"); InfoUrl = "http://www.szns.gov.cn" + InfoUrl; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "page_con"))); if (dtlNode == null || dtlNode.Count < 1) { parser.Reset(); dtlNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("body")); } if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); inviteCtx = HtmlTxt.GetReplace("<br/>,</p>,<br>,<br />", "\r\n").ToCtxString(); buildUnit = inviteCtx.GetBuildRegex(); prjAddress = inviteCtx.GetAddressRegex(); code = inviteCtx.GetCodeRegex().GetCodeDel(); specType = "政府采购"; inviteType = prjName.GetInviteBidType(); msgType = "深圳市南山区教育局"; InviteInfo info = ToolDb.GenInviteInfo("广东省", "深圳社会招标", area, string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, 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 a = 0; a < aNode.Count; a++) { ATag file = aNode[a].GetATag(); if (file.IsAtagAttach()) { string link = file.Link; if (!link.ToLower().Contains("http")) { link = "http://exoa.nsjy.com" + file.Link; } base.AttachList.Add(ToolDb.GenBaseAttach(file.LinkText, info.Id, link)); } } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
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(this.ToolWebSite.UrlEncode(SiteUrl), Encoding.UTF8, ref cookiestr); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(htl)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxma03"))); Regex regexPage = new Regex(@"共\d+页"); try { page = int.Parse(regexPage.Match(nodeList.AsString()).Value.Trim(new char[] { '共', '页' })); } catch (Exception) { } for (int i = 1; i <= page; i++) { if (i > 1) { try { htl = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode("http://www.szns.gov.cn/publish/main/1/19/26/zbtbxx/5465/index_" + i.ToString() + ".html"), Encoding.UTF8); } catch (Exception ex) { continue; } } parser = new Parser(new Lexer(htl)); NodeList tableNodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxdianbeijing"))); if (tableNodeList.Count > 0) { for (int j = 0; j < tableNodeList.Count; j++) { ATag aTag = tableNodeList.SearchFor(typeof(ATag), true)[j] as ATag; string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, bidType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; prjName = aTag.LinkText; Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}"); beginDate = regDate.Match(tableNodeList[j].ToPlainTextString()).Value.Trim(); InfoUrl = "http://www.szns.gov.cn" + aTag.Link.Replace("amp;", "").Trim(); string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(InfoUrl), Encoding.UTF8).Replace(" ", ""); } catch (Exception) { continue; } Parser parserdetail = new Parser(new Lexer(htmldetail)); NodeList dtnode = parserdetail.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxzf2"))); if (dtnode.Count > 0) { HtmlTxt = dtnode.AsHtml(); Regex regeximg = new Regex(@"<img[^>]*>");//去掉图片 HtmlTxt = regeximg.Replace(HtmlTxt, ""); inviteCtx = dtnode.AsString().Replace("\n", "\r\n"); Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>"); inviteCtx = regexHtml.Replace(inviteCtx, ""); Regex regBuidUnit = new Regex(@"(招标人|建设单位)(:|:)[^\r\n]+\r\n"); buildUnit = regBuidUnit.Match(inviteCtx).Value.Replace("招标人:", "").Replace("建设单位:", "").Trim(); Regex regPrjAddr = new Regex(@"(工程地点|工程地址)(:|:)[^\r\n]+\r\n"); prjAddress = regPrjAddr.Match(inviteCtx).Value.Replace("工程地点:", "").Replace("工程地址", "").Replace(":", "").Trim(); Regex regCode = new Regex(@"工程编号(:|:)[^\r\n]+\r\n"); code = regCode.Match(inviteCtx).Value.Replace("工程编号:", "").Replace("工程编号:", "").Trim(); if (Encoding.Default.GetByteCount(code) > 50) { code = ""; } msgType = "深圳市南山区政府采购及招标中心"; specType = "建设工程"; if (buildUnit == "") { buildUnit = ""; } if (prjAddress == "") { prjAddress = "见招标信息"; } bidType = ToolHtml.GetInviteTypes(prjName); InviteInfo info = ToolDb.GenInviteInfo("广东省", "深圳区及街道工程", "", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, bidType, specType, otherType, InfoUrl, 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>(); string html = string.Empty; string cookiestr = string.Empty; string viewState = string.Empty; int pageInt = 1; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.Default); } catch { return(null); } try { string temp = html.GetRegexBegEnd("<strong>", "</strong>").GetReplace("<fontcolor=red>1</font>/"); pageInt = int.Parse(temp); } catch { } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl + "&SpecialID=0&page=" + i, Encoding.Default); } catch { continue; } } Parser parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("height", "200"))); if (listNode != null && listNode.Count > 0) { parser = new Parser(new Lexer(listNode.AsHtml())); NodeList fontNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a")); if (fontNode != null && fontNode.Count > 0) { for (int j = 0; j < fontNode.Count; j++) { ATag aTag = fontNode[j] as ATag; if (aTag == null) { continue; } 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, area = string.Empty; string temp = aTag.GetAttribute("title"); prjName = temp.GetRegex("文章标题"); code = temp.GetRegex("招标代码"); beginDate = temp.GetRegex("更新时间").GetDateRegex("yyyy/MM/dd"); InfoUrl = "http://www.lntb.gov.cn/" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("height", "200"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.ToLower().GetReplace("</p>,<br/>,<br>", "\r\n").ToCtxString(); buildUnit = bidCtx.GetBuildRegex(); bidUnit = bidCtx.GetBidRegex().GetReplace("名称"); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = bidCtx.GetRegex("中标人名称").GetReplace("名称"); } if (string.IsNullOrEmpty(bidUnit)) { bidUnit = bidCtx.GetReplace("中标人名称:,:中标人名称,中标人:,中标人:", "\r\n").GetBidRegex().GetReplace("名称"); } if (string.IsNullOrEmpty(bidUnit)) { parser = new Parser(new Lexer(HtmlTxt)); NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (tableNode != null && tableNode.Count > 0) { TableTag tag = tableNode[tableNode.Count - 1] as TableTag; string ctx = string.Empty; for (int r = 0; r < tag.RowCount; r++) { for (int c = 0; c < tag.Rows[r].ColumnCount; c++) { string tempStr = tag.Rows[r].Columns[c].ToNodePlainString(); if ((c + 1) % 2 == 0) { ctx += tempStr.GetReplace(":,:") + "\r\n"; } else { ctx += tempStr.GetReplace(":,:") + ":"; } } } bidUnit = ctx.GetBidRegex().GetReplace("名称"); if (string.IsNullOrEmpty(bidUnit)) { bidUnit = ctx.GetRegex("中标人名称").GetReplace("名称"); } } } bidMoney = bidCtx.GetMoneyRegex(); prjMgr = bidCtx.GetMgrRegex(); if (buildUnit.Contains("公司")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("公司")) + "公司"; } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")); } msgType = "辽宁省招标投标协调管理办公室"; specType = "建设工程"; bidType = "建设工程"; BidInfo info = ToolDb.GenBidInfo("辽宁省", "辽宁省及地市", area, 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) { int sqlCount = 0; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl + this.MaxCount); } catch { return(null); } int startIndex = html.IndexOf("{"); int endIndex = html.LastIndexOf("}"); html = html.Substring(startIndex, (endIndex + 1) - startIndex); JavaScriptSerializer serializer = new JavaScriptSerializer(); Dictionary <string, object> smsTypeJson = (Dictionary <string, object>)serializer.DeserializeObject(html); foreach (KeyValuePair <string, object> obj in smsTypeJson) { if (obj.Key != "rows") { continue; } object[] array = (object[])obj.Value; foreach (object arrValue in array) { string headName = string.Empty, releaseTime = string.Empty, infoScorce = string.Empty, msgType = string.Empty, infoUrl = string.Empty, ctxHtml = string.Empty, infoCtx = string.Empty, infoType = string.Empty; Dictionary <string, object> dic = (Dictionary <string, object>)arrValue; headName = Convert.ToString(dic["title"]); releaseTime = Convert.ToString(dic["faBuStartDate"]); infoUrl = "http://jyzx.cb.gov.cn/tongzhigonggao/tzgg_view.html?guid=" + Convert.ToString(dic["tongZhiGuid"]); string attachId = Convert.ToString(dic["attachFileGroupGuid"]); string attachUrl = "http://jyzx.cb.gov.cn/common/nofilter/queryFuJianList.do?groupGuid=" + attachId; ctxHtml = Convert.ToString(dic["content"]); infoCtx = ctxHtml.ToCtxString(); infoType = "通知公告"; msgType = "深圳市建设工程交易中心龙岗分中心"; NotifyInfo info = ToolDb.GenNotifyInfo(headName, releaseTime, infoScorce, msgType, infoUrl, ctxHtml, "广东省", "深圳龙岗区工程", "龙岗区", infoCtx, infoType); sqlCount++; if (!crawlAll && sqlCount > this.MaxCount) { return(null); } if (ToolDb.SaveEntity(info, this.ExistCompareFields)) { if (!string.IsNullOrWhiteSpace(attachId)) { string attchJson = this.ToolWebSite.GetHtmlByUrl(attachUrl); JavaScriptSerializer attachSerializer = new JavaScriptSerializer(); object[] dicTypeJson = (object[])attachSerializer.DeserializeObject(attchJson); foreach (object objJson in dicTypeJson) { Dictionary <string, object> dicFiles = objJson as Dictionary <string, object>; string attachName = Convert.ToString(dicFiles["attachName"]); string attachDelId = Convert.ToString(dicFiles["attachGuid"]); string attachDelUrl = "http://jyzx.cb.gov.cn/longgang-jyw-file/downloadFile?fileId=" + attachDelId; BaseAttach attach = null; try { attach = ToolHtml.GetBaseAttach(attachDelUrl, attachName, info.Id); } catch { } if (attach != null) { ToolDb.SaveEntity(attach, "AttachServerPath,SourceID"); } } } } } } return(null); }
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; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.Default); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("nowrap", "true"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode.AsString().GetRegexBegEnd("总页数", "当前页").Replace(":", ""); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(html); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "__VIEWSTATE", "__VIEWSTATEGENERATOR", "__EVENTTARGET", "__EVENTARGUMENT" }, new string[] { viewState, "69C4529D", "MoreInfoList1$Pager", i.ToString() }); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc, Encoding.Default); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "MoreInfoList1_DataGrid1"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 0; j < table.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, area = string.Empty; TableRow tr = table.Rows[j]; ATag aTag = tr.Columns[1].GetATag(); prjName = aTag.GetAttribute("title"); beginDate = tr.Columns[2].ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.xysztb.gov.cn" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "TDContent"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.GetReplace("</p>,<br />,<br/>", "\r\n").ToCtxString().GetReplace("\t", "\r\n"); prjAddress = bidCtx.GetAddressRegex(); buildUnit = bidCtx.GetBuildRegex(); bidUnit = bidCtx.GetBidRegex(); bidMoney = bidCtx.GetMoneyRegex(); if (string.IsNullOrWhiteSpace(bidUnit)) { parser = new Parser(new Lexer(HtmlTxt)); NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (tableNode != null && tableNode.Count > 0) { TableTag tableTag = tableNode[0] as TableTag; string ctx = string.Empty; for (int r = 0; r < tableTag.RowCount; r++) { for (int c = 0; c < tableTag.Rows[r].ColumnCount; c++) { string temp = tableTag.Rows[r].Columns[c].ToNodePlainString(); try { ctx += temp.GetReplace(":,:") + ":" + tableTag.Rows[r + 1].Columns[c].ToNodePlainString() + "\r\n"; ctx += tableTag.Rows[r].Columns[c + 1].ToNodePlainString().GetReplace(":,:") + ":" + tableTag.Rows[r + 1].Columns[c + 1].ToNodePlainString() + "\r\n"; } catch { } r++; break; } } bidUnit = ctx.GetBidRegex(); if (bidUnit.Contains("中标")) { bidUnit = ""; } bidMoney = ctx.GetMoneyRegex(); if (string.IsNullOrWhiteSpace(bidUnit)) { ctx = ""; for (int r = 0; r < tableTag.RowCount; r++) { for (int c = 0; c < tableTag.Rows[r].ColumnCount; c++) { string temp = tableTag.Rows[r].Columns[c].ToNodePlainString().GetReplace(":,:"); if (c % 2 == 0) { ctx += temp + ":"; } else { ctx += temp + "\r\n"; } } } bidUnit = ctx.GetBidRegex(); if (bidUnit.Contains("中标")) { bidUnit = ""; } bidMoney = ctx.GetMoneyRegex(); } if (string.IsNullOrWhiteSpace(bidUnit)) { tableTag = tableNode[1] as TableTag; if (tableTag != null) { ctx = string.Empty; for (int r = 0; r < tableTag.RowCount; r++) { for (int c = 0; c < tableTag.Rows[r].ColumnCount; c++) { string temp = tableTag.Rows[r].Columns[c].ToNodePlainString(); try { ctx += temp.GetReplace(":,:") + ":" + tableTag.Rows[r + 1].Columns[c].ToNodePlainString() + "\r\n"; ctx += tableTag.Rows[r].Columns[c + 1].ToNodePlainString().GetReplace(":,:") + ":" + tableTag.Rows[r + 1].Columns[c + 1].ToNodePlainString() + "\r\n"; } catch { } r++; break; } } bidUnit = ctx.GetBidRegex(); if (bidUnit.Contains("中标")) { bidUnit = ""; } bidMoney = ctx.GetMoneyRegex(); if (string.IsNullOrWhiteSpace(bidUnit)) { ctx = ""; for (int r = 0; r < tableTag.RowCount; r++) { for (int c = 0; c < tableTag.Rows[r].ColumnCount; c++) { string temp = tableTag.Rows[r].Columns[c].ToNodePlainString().GetReplace(":,:"); if (c % 2 == 0) { ctx += temp + ":"; } else { ctx += temp + "\r\n"; } } } bidUnit = ctx.GetBidRegex(); if (bidUnit.Contains("中标")) { bidUnit = ""; } bidMoney = ctx.GetMoneyRegex(); } } } } } if (buildUnit.Contains("公司")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("公司")) + "公司"; } if (buildUnit.Contains("地址")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("地址")); } if (buildUnit.Contains("联系")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("联系")); } if (buildUnit.Contains("指挥部")) { buildUnit = buildUnit.Remove(buildUnit.IndexOf("指挥部")); } code = bidCtx.GetCodeRegex().GetCodeDel().GetReplace("."); if (bidUnit.Contains("日历天") || bidUnit.Contains("预期中标") || bidUnit.Contains("投标人") || bidUnit.Contains("中标价")) { bidUnit = string.Empty; } try { if (decimal.Parse(bidMoney) < 1) { bidMoney = "0"; } } catch { } msgType = "襄阳市公共资源交易中心"; specType = "建设工程"; bidType = prjName.GetInviteBidType(); buildUnit = buildUnit.Replace(" ", ""); BidInfo info = ToolDb.GenBidInfo("湖北省", "湖北省及地市", "襄阳市", 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; } else { link = "http://www.xysztb.gov.cn/" + 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); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); //取得页码 int pageInt = 1; int crawlMax = 0; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(SiteUrl + "&page=0"), Encoding.Default).Replace(" ", ""); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("id", "page_PageList"))); if (sNode != null && sNode.Count > 0) { SelectTag select = sNode[0] as SelectTag; pageInt = select.OptionTags.Length; } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl + "&page=" + (i - 1).ToString(), Encoding.Default); } catch (Exception ex) { continue; } } parser = new Parser(new Lexer(html)); sNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("tr"), new HasAttributeFilter("onmouseover", "this.style.backgroundColor=\"#EFFCD0\";"))); if (sNode != null && sNode.Count > 0) { for (int n = 0; n < sNode.Count; n++) { 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 = sNode[n] as TableRow; prjName = tr.Columns[0].ToPlainTextString().Trim(); bidUnit = tr.Columns[1].ToPlainTextString().Trim(); ATag aTag = tr.Columns[0].SearchFor(typeof(ATag), true)[0] as ATag; Regex regexLink = new Regex(@"id=[^-]+"); InfoUrl = "http://www.sdcin.com.cn/viewzbgg.php?" + regexLink.Match(aTag.Link).Value; string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.Default).Replace(" ", "").GetJsString(); } catch (Exception ex) { continue; } Parser dtlparser = new Parser(new Lexer(htmldetail)); NodeList dtnode = dtlparser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("width", "98%"))); if (dtnode != null && dtnode.Count > 0) { HtmlTxt = dtnode.AsHtml(); string ctx = string.Empty; TableTag table = dtnode[0] as TableTag; for (int k = 0; k < table.RowCount; k++) { for (int d = 0; d < table.Rows[k].ColumnCount; d++) { if (d == 0) { ctx += table.Rows[k].Columns[d].ToNodePlainString().Replace(":", "").Replace(":", "") + ":"; } else { ctx += table.Rows[k].Columns[d].ToNodePlainString() + "\r\n"; } } } bidCtx = ctx; buildUnit = bidCtx.GetBuildRegex(); if (string.IsNullOrEmpty(buildUnit)) { buildUnit = bidCtx.GetRegex("招标代理"); } bidMoney = bidCtx.GetMoneyRegex(); prjAddress = bidCtx.GetAddressRegex(); if (prjAddress.Contains("邮政编码")) { prjAddress = prjAddress.Remove(prjAddress.IndexOf("邮政编码")); } if (string.IsNullOrEmpty(beginDate)) { beginDate = DateTime.Now.ToString("yyyy-MM-dd"); } prjMgr = bidCtx.GetMgrRegex(); code = bidCtx.GetCodeRegex(); msgType = "佛山市顺德区建设工程交易中心"; specType = "建设工程"; prjName = ToolDb.GetPrjName(prjName); bidType = ToolHtml.GetInviteTypes(prjName); BidInfo info = ToolDb.GenBidInfo("广东省", "佛山市区", "顺德区", 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 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); } catch (Exception ex) { return(list); } if (htl.Contains("RowCount")) { try { int index = htl.IndexOf("RowCount"); string pageStr = htl.Substring(index, htl.Length - index).Replace("RowCount", "").Replace("}", "").Replace(":", "").Replace("\"", ""); decimal b = decimal.Parse(pageStr) / 20; if (b.ToString().Contains(".")) { page = Convert.ToInt32(b) + 1; } else { page = Convert.ToInt32(b); } } catch { } } for (int i = 1; i <= page; i++) { if (i > 1) { try { htl = this.ToolWebSite.GetHtmlByUrl("http://www.szjs.gov.cn/build/build.ashx?_=1352593922281&menu=%E9%A1%B9%E7%9B%AE%E4%BF%A1%E6%81%AF&type=%E6%96%BD%E5%B7%A5%E8%AE%B8%E5%8F%AF&pageSize=20&pageIndex=" + i.ToString(), Encoding.UTF8); } catch { continue; } } JavaScriptSerializer serializer = new JavaScriptSerializer(); Dictionary <string, object> smsTypeJson = (Dictionary <string, object>)serializer.DeserializeObject(htl); foreach (KeyValuePair <string, object> obj in smsTypeJson) { if (obj.Key != "DataList") { continue; } object[] array = (object[])obj.Value; foreach (object obj2 in array) { Dictionary <string, object> dicSmsType = (Dictionary <string, object>)obj2; string pPrjName = string.Empty, pBuildUnit = string.Empty, pBuildAddress = string.Empty, pBuildManager = string.Empty, pBuildScale = string.Empty, pPrjPrice = string.Empty, pPrjStartDate = string.Empty, PrjEndDate = string.Empty, pConstUnit = string.Empty, pConstUnitManager = string.Empty, pSuperUnit = string.Empty, pSuperUnitManager = string.Empty, pProspUnit = string.Empty, pProspUnitManager = string.Empty, pDesignUnit = string.Empty, pDesignUnitManager = string.Empty, pPrjManager = string.Empty, pSpecialPerson = string.Empty, pLicUnit = string.Empty, pPrjLicCode = string.Empty, PrjLicDate = string.Empty, pPrjDesc = string.Empty, pProvince = string.Empty, pCity = string.Empty, pInfoSource = string.Empty, pUrl = string.Empty, pCreatetime = string.Empty, pPrjCode = string.Empty; try { pPrjCode = Convert.ToString(dicSmsType["AnnSerial"]); pPrjName = Convert.ToString(dicSmsType["PrjName"]); pBuildUnit = Convert.ToString(dicSmsType["ConstOrg"]); PrjLicDate = Convert.ToString(dicSmsType["IssueDate"]); pUrl = "http://www.szjs.gov.cn/build/sgxk_detail.aspx?id=" + pPrjCode; string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(pUrl, Encoding.UTF8, ref cookiestr).Trim(); } catch (Exception) { continue; } Parser parser = new Parser(new Lexer(htmldetail)); NodeList dtList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "js-table mar-l-4"))); if (dtList != null && dtList.Count > 0) { TableTag table = dtList[0] as TableTag; for (int j = 0; j < table.RowCount; j++) { TableRow dr = table.Rows[j]; string ctx = string.Empty; for (int k = 0; k < dr.ColumnCount; k++) { ctx += dr.Columns[k].ToPlainTextString().Trim().Replace("\r", "").Replace("\n", ""); } pInfoSource += ctx + "\r\n"; } Regex regpDesignUnit = new Regex(@"设计单位(:|:)[^\r\n]+\r\n"); pDesignUnit = regpDesignUnit.Match(pInfoSource).Value.Replace("设计单位", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regAddree = new Regex(@"(工程地址|工程地点)(:|:)[^\r\n]+\r\n"); pBuildAddress = regAddree.Match(pInfoSource).Value.Replace("工程地址", "").Replace("工程地点", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regSca = new Regex(@"(建筑面积|建设规模)(:|:)[^\r\n]+\r\n"); pBuildScale = regSca.Match(pInfoSource).Value.Replace("建设规模", "").Replace("建筑面积", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regpSuperUnit = new Regex(@"监理单位(:|:)[^\r\n]+\r\n"); pSuperUnit = regpSuperUnit.Match(pInfoSource).Value.Replace("监理单位", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regpConstUnit = new Regex(@"施工单位(:|:)[^\r\n]+\r\n"); pConstUnit = regpConstUnit.Match(pInfoSource).Value.Replace("施工单位", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regpLicUnit = new Regex(@"发证机关(:|:)[^\r\n]+\r\n"); pLicUnit = regpLicUnit.Match(pInfoSource).Value.Replace("发证机关", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regpPosoUnit = new Regex(@"勘察单位(:|:)[^\r\n]+\r\n"); pProspUnit = regpPosoUnit.Match(pInfoSource).Value.Replace("勘察单位", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regMan = new Regex(@"(项目经理|项目负责人)(:|:)[^\r\n]+\r\n"); pPrjManager = regMan.Match(pInfoSource).Value.Replace("项目负责人", "").Replace("项目经理", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regBeg = new Regex(@"计划开工日期(:|:)[^\r\n]+\r\n"); pPrjStartDate = regBeg.Match(pInfoSource).Value.Replace("计划开工日期", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regEnd = new Regex(@"计划竣工日期(:|:)[^\r\n]+\r\n"); PrjEndDate = regEnd.Match(pInfoSource).Value.Replace("计划竣工日期", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regpPrice = new Regex(@"工程造价(:|:)[^\r\n]+\r\n"); pPrjPrice = regpPrice.Match(pInfoSource).Value.Replace("工程造价", "").Replace("/", "").Replace(":", "").Replace(":", "").Trim(); Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+"); if (pPrjPrice.Contains("万")) { pPrjPrice = pPrjPrice.Remove(pPrjPrice.IndexOf("万")).Trim(); pPrjPrice = regBidMoney.Match(pPrjPrice).Value; } else { try { pPrjPrice = (decimal.Parse(regpPrice.Match(pPrjPrice).Value) / 10000).ToString(); if (decimal.Parse(pPrjPrice) < decimal.Parse("0.1")) { pPrjPrice = "0"; } } catch (Exception) { pPrjPrice = "0"; } } if (string.IsNullOrEmpty(pLicUnit)) { pLicUnit = "深圳市住房和建设局"; } ProjectLic info = ToolDb.GenProjectLic(pPrjName, pBuildUnit, pBuildAddress, pBuildManager, pBuildScale, pPrjPrice, pPrjStartDate, PrjEndDate, pConstUnit, pConstUnitManager, pSuperUnit, pSuperUnitManager, pProspUnit, pProspUnitManager, pDesignUnit, pDesignUnitManager, pPrjManager, pSpecialPerson, pLicUnit, pPrjLicCode, PrjLicDate, pPrjDesc, "广东省", "深圳市区", pInfoSource, pUrl, pCreatetime, pPrjCode, "深圳市住房和建设局"); list.Add(info); if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } catch { continue; } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <NoticeInfo>(); int pageInt = 37; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; string url = "http://www.gsggzyjy.cn/ajax/Controls_InfoList,App_Web_rzplwhmc.ashx?_method=getCurrentData&_session=rw"; try { this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8, ref cookiestr); string post = "currentPage=1\r\nQuery="; html = ToolHtml.GetHtmlByUrlPost(url, post, Encoding.UTF8, ref cookiestr); } catch { return(null); } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { string post = "currentPage=" + i + "\r\nQuery="; html = ToolHtml.GetHtmlByUrlPost(url, post, Encoding.UTF8, ref cookiestr); } catch { continue; } } Parser parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("li")); if (listNode != null && listNode.Count > 0) { for (int j = 0; j < listNode.Count; j++) { string InfoTitle = string.Empty, InfoType = string.Empty, PublistTime = string.Empty, InfoCtx = string.Empty, InfoUrl = string.Empty, prjCode = string.Empty, buildUnit = string.Empty, htmlTxt = string.Empty, area = string.Empty; INode node = listNode[j]; ATag aTag = node.GetATag(); InfoTitle = aTag.GetAttribute("title"); InfoType = "控制价公示"; PublistTime = node.GetSpan().StringText; InfoUrl = "http://www.gsggzyjy.cn" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "ContentPlaceHolder1_InfoHtml"))); if (dtlNode != null && dtlNode.Count > 0) { htmlTxt = dtlNode.AsHtml(); InfoCtx = htmlTxt.ToCtxString(); parser = new Parser(new Lexer(htmlTxt)); NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (tableNode != null && tableNode.Count > 0) { string ctx = string.Empty; TableTag table = tableNode[0] as TableTag; for (int r = 0; r < table.RowCount; r++) { for (int c = 0; c < table.Rows[r].ColumnCount; c++) { string temp = table.Rows[r].Columns[c].ToNodePlainString(); if ((c + 1) % 2 == 0) { ctx += temp.GetReplace(":,:") + "\r\n"; } else { ctx += temp.GetReplace(":,:") + ":"; } } } buildUnit = ctx.GetBuildRegex(); } NoticeInfo info = ToolDb.GenNoticeInfo("甘肃省", "甘肃省及地市", area, string.Empty, InfoTitle, InfoType, InfoCtx, PublistTime, string.Empty, "甘肃省公共资源交易中心", InfoUrl, prjCode, buildUnit, string.Empty, string.Empty, "交通运输工程", string.Empty, htmlTxt); 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].GetATag(); if (a.IsAtagAttach()) { string link = string.Empty; if (a.Link.ToLower().Contains("http")) { link = a.Link; } else { link = "http://www.gsggzyjy.cn/" + a.Link; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } list.Add(info); if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <BidInfo>(); Dictionary <string, string> citys = this.GetCitys(); foreach (string area in citys.Keys) { int count = 0; int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(citys[area], Encoding.UTF8, ref cookiestr); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("nowrap", "true"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode.AsString().GetRegexBegEnd("总页数", "当前页").Replace(":", ""); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { viewState = this.ToolWebSite.GetAspNetViewState(html); eventValidation = this.ToolWebSite.GetAspNetEventValidation(html); string viewSTATEGENERATOR = ToolHtml.GetHtmlInputValue(html, "__VIEWSTATEGENERATOR"); NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "__VIEWSTATE", "__VIEWSTATEGENERATOR", "__EVENTTARGET", "__EVENTARGUMENT", "__EVENTVALIDATION", "MoreInfoList1$txtTitle" }, new string[] { viewState, viewSTATEGENERATOR, "MoreInfoList1$Pager", i.ToString(), eventValidation, "" }); try { html = this.ToolWebSite.GetHtmlByUrl(citys[area], nvc, Encoding.UTF8, ref cookiestr); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "MoreInfoList1_DataGrid1"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.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 = table.Rows[j]; ATag aTag = tr.Columns[1].GetATag(); prjName = aTag.GetAttribute("title").GetReplace("【正在报名】,【报名结束】"); beginDate = tr.Columns[2].ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.gxzbtb.cn" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("id", "TDContent"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); bidCtx = HtmlTxt.GetReplace(new string[] { "<br/>", "<br />", "<br>" }, "\r\n").ToCtxString(); prjAddress = bidCtx.GetAddressRegex(); buildUnit = bidCtx.GetBuildRegex(); bidUnit = bidCtx.GetBidRegex(); bidMoney = bidCtx.GetMoneyRegex(); prjMgr = bidCtx.GetMgrRegex(); code = bidCtx.GetCodeRegex().GetCodeDel(); if (string.IsNullOrEmpty(bidUnit)) { parser = new Parser(new Lexer(HtmlTxt)); NodeList bidNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (bidNode != null && bidNode.Count > 0) { string ctx = string.Empty; TableTag bidTable = bidNode[0] as TableTag; for (int r = 0; r < bidTable.RowCount; r++) { for (int c = 0; c < bidTable.Rows[r].ColumnCount; c++) { if ((c + 1) % 2 == 0) { ctx += bidTable.Rows[r].Columns[c].ToNodePlainString() + "\r\n"; } else { ctx += bidTable.Rows[r].Columns[c].ToNodePlainString() + ":"; } } } bidUnit = ctx.GetBidRegex(); if (string.IsNullOrEmpty(bidMoney) || bidMoney == "0") { bidMoney = ctx.GetMoneyString().GetMoney("万元"); } if (string.IsNullOrEmpty(prjAddress)) { prjAddress = ctx.GetAddressRegex(); } if (string.IsNullOrEmpty(buildUnit)) { buildUnit = ctx.GetBuildRegex(); } if (string.IsNullOrEmpty(code)) { code = ctx.GetCodeRegex().GetCodeDel(); } if (bidUnit.Contains("推荐") || bidUnit.Contains("中标") || bidUnit.Contains("地址")) { bidUnit = string.Empty; } if (string.IsNullOrEmpty(bidUnit)) { if (bidTable.RowCount > 1) { ctx = string.Empty; for (int d = 0; d < bidTable.Rows[0].ColumnCount; d++) { ctx += bidTable.Rows[0].Columns[d].ToNodePlainString() + ":"; try { ctx += bidTable.Rows[1].Columns[d].ToNodePlainString() + "\r\n"; } catch { } } bidUnit = ctx.GetBidRegex(); if (string.IsNullOrEmpty(bidMoney) || bidMoney == "0") { bidMoney = ctx.GetMoneyString().GetMoney(); } if (string.IsNullOrEmpty(prjAddress)) { prjAddress = ctx.GetAddressRegex(); } if (string.IsNullOrEmpty(buildUnit)) { buildUnit = ctx.GetBuildRegex(); } if (string.IsNullOrEmpty(code)) { code = ctx.GetCodeRegex().GetCodeDel(); } } } } } try { if (decimal.Parse(bidMoney) > 10000) { bidMoney = (decimal.Parse(bidMoney) / 10000).ToString(); } } catch { } bidUnit = bidUnit.Replace("名称", "").Replace("单位", "").Replace("№", "").Replace("1", "").Replace("2", "").Replace("联合体", "").Replace("(", ""); if (bidUnit.Contains("公司")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("公司")) + "公司"; } if (bidUnit.Contains("研究院")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("研究院")) + "研究院"; } if (bidUnit.Contains("研究所")) { bidUnit = bidUnit.Remove(bidUnit.IndexOf("研究所")) + "研究所"; } bidType = "房建市政"; specType = "建设工程"; msgType = "广西壮族自治区公共资源交易中心"; BidInfo info = ToolDb.GenBidInfo("广西壮族自治区", "广西壮族自治区及地市", area, string.Empty, code, prjName, buildUnit, beginDate, bidUnit, beginDate, endDate, bidCtx, string.Empty, msgType, bidType, specType, otherType, bidMoney, InfoUrl, prjMgr, HtmlTxt); list.Add(info); count++; 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; } else { link = "http://www.gxzbtb.cn/" + a.Link.GetReplace("../,./"); } if (Encoding.Default.GetByteCount(link) > 500) { continue; } BaseAttach attach = ToolDb.GenBaseAttach(a.LinkText, info.Id, link); base.AttachList.Add(attach); } } } if (!crawlAll && count >= this.MaxCount) { goto Funcs; } } } } } Funcs :; } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new ArrayList(); //取得页码 int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode("http://www.szsszx.com/tender/pager?key=zhongbiao&pagenumber=20&pageindex=1"), Encoding.UTF8); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList tdNodes = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "pagelist"))); if (tdNodes != null && tdNodes.Count > 0) { string pageTemp = tdNodes.AsString().Replace(" ", ""); Regex regpage = new Regex(@"1/\d+"); try { pageInt = int.Parse(regpage.Match(pageTemp).Value.Replace("1/", "")); } catch (Exception ex) { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { html = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode("http://www.szsszx.com/tender/pager?key=zhongbiao&pagenumber=20&pageindex=" + i.ToString()), Encoding.UTF8); } parser = new Parser(new Lexer(html)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new TagNameFilter("li")); if (nodeList != null && nodeList.Count > 0) { for (int j = 0; j < nodeList.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; CompositeTag obj = nodeList[j] as CompositeTag; ATag aTag = obj.SearchFor(typeof(ATag), true)[0] as ATag; Span dateSpan = obj.SearchFor(typeof(Span), true)[0] as Span; prjName = aTag.GetAttribute("title"); beginDate = dateSpan.ToPlainTextString().Trim(new char[] { '[', ']' }); InfoUrl = "http://www.szsszx.com" + aTag.Link; string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Trim(); Parser dtlparserHTML = new Parser(new Lexer(htmldetail)); NodeList dtnodeHTML = dtlparserHTML.ExtractAllNodesThatMatch(new AndFilter(new HasAttributeFilter("class", "news-content"), new TagNameFilter("div"))); HtmlTxt = dtnodeHTML.AsHtml(); htmldetail = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8).Replace(" ", "").Replace("</br>", "\r\n").Replace("<br>", "\r\n"); Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>"); htmldetail = regexHtml.Replace(htmldetail, ""); } catch (Exception ex) { continue; } Parser dtlparser = new Parser(new Lexer(htmldetail)); NodeList dtnode = dtlparser.ExtractAllNodesThatMatch(new AndFilter(new HasAttributeFilter("class", "news-content"), new TagNameFilter("div"))); bidCtx = dtnode.ToHtml().ToCtxString(); bidCtx = Regex.Replace(bidCtx, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase).Replace("<", "").Replace(">", "").Replace("\n\n\n\t", "\r\n").Replace("\n\n", "\r\n"); Regex regCode = new Regex(@"招标编号(:|:)[^\r\n]+\r\n"); code = regCode.Match(bidCtx).Value.Replace("招标编号", "").Replace(":", "").Replace(":", "").Trim(); if (Encoding.Default.GetByteCount(code) > 50) { code = ""; } Regex regbuildUnit = new Regex(@"(采购人|采购单位|采购代理机构)(:|:)[^\r\n]+\r\n"); buildUnit = regbuildUnit.Match(bidCtx).Value.Replace("采购人", "").Replace("采购单位", "").Replace("采购代理机构", "").Replace(":", "").Replace(":", "").Trim(); prjAddress = bidCtx.GetAddressRegex(); bidUnit = bidCtx.GetBidRegex();// regBidUnit.Match(bidCtx).Value.Replace("中标单位", "").Replace(":", "").Replace(":", "").Trim(); Regex regBidMoneystr = new Regex(@"(中标价|价格|金额)(:|:)[^\r\n]+\r\n"); string monerystr = regBidMoneystr.Match(bidCtx).Value.Replace("中标价", "").Replace("价格", "").Replace("金额", "").Replace("万元整", "").Replace(":", "").Replace(":", "").Replace(",", "").Replace(",", "").Trim(); Regex regBidMoney = new Regex(@"[0-9]+[.]{0,1}[0-9]+"); if (!string.IsNullOrEmpty(regBidMoney.Match(monerystr).Value)) { if ((monerystr.Contains("万元") || monerystr.Contains("万美元")) && !monerystr.Contains("万元整")) { bidMoney = regBidMoney.Match(monerystr).Value; } else { try { bidMoney = (decimal.Parse(regBidMoney.Match(monerystr).Value) / 10000).ToString(); if (decimal.Parse(bidMoney) < decimal.Parse("0.1")) { bidMoney = "0"; } } catch (Exception) { bidMoney = "0"; } } } specType = "其他"; msgType = "深圳市深水水务咨询有限公司"; prjName = ToolDb.GetPrjName(prjName); bidType = ToolHtml.GetInviteTypes(prjName); BidInfo info = ToolDb.GenBidInfo("广东省", "深圳社会招标", "", 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 <NoticeInfo>(); int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(SiteUrl), Encoding.UTF8); } catch { return(list); } Parser parser = new Parser(new Lexer(html)); NodeList tdNodes = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("class", "input-group-addon"))); if (tdNodes != null && tdNodes.Count > 0) { try { string reTemp = tdNodes.AsString().GetRegexBegEnd("共", "项"); string pageTemp = tdNodes.AsString().GetRegexBegEnd("项", "页").GetReplace("共,项,页," + reTemp + ",,"); pageInt = int.Parse(pageTemp); } catch (Exception) { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(SiteUrl + "?pi=" + (i - 1), Encoding.UTF8); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "inside_table"))); if (nodeList != null && nodeList.Count > 0) { TableTag table = nodeList[0] as TableTag; for (int j = 1; j < table.RowCount; j++) { string InfoTitle = string.Empty, InfoType = string.Empty, PublistTime = string.Empty, InfoCtx = string.Empty, InfoUrl = string.Empty, htmlTxt = string.Empty, prjCode = string.Empty, buildUnit = string.Empty, prjType = string.Empty; TableRow tr = table.Rows[j]; prjCode = tr.Columns[1].ToNodePlainString().Replace(" ", ""); InfoTitle = tr.Columns[2].ToPlainTextString().Trim(); buildUnit = tr.Columns[3].ToNodePlainString(); prjType = tr.Columns[4].ToNodePlainString(); InfoType = "标底公示"; PublistTime = tr.Columns[5].ToPlainTextString().Trim(); InfoUrl = "http://www.bajsjy.com/" + tr.Columns[2].GetATagHref(); string ctxhtml = string.Empty; try { ctxhtml = this.ToolWebSite.GetHtmlByUrl(InfoUrl, Encoding.UTF8); } catch { continue; } parser = new Parser(new Lexer(ctxhtml)); NodeList dtnode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasAttributeFilter("class", "showContent"), new TagNameFilter("div"))); if (dtnode != null && dtnode.Count > 0) { htmlTxt = dtnode.AsHtml(); InfoCtx = htmlTxt.ToLower().Replace("</p>", "\r\n").Replace("<br>", "\r\n").Replace("<br/>", "\r\n").ToCtxString(); NoticeInfo info = ToolDb.GenNoticeInfo("广东省", "深圳宝安区工程", string.Empty, string.Empty, InfoTitle, InfoType, InfoCtx, PublistTime, string.Empty, "深圳市建设工程交易中心宝安分中心", InfoUrl, prjCode, buildUnit, string.Empty, string.Empty, prjType, string.Empty, htmlTxt); list.Add(info); parser = new Parser(new Lexer(htmlTxt)); NodeList fileNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "inside_table"))); if (fileNode != null && fileNode.Count > 0 && fileNode[0] is TableTag) { TableTag fileTable = fileNode[0] as TableTag; for (int f = 1; f < fileTable.Rows.Length; f++) { BaseAttach attach = ToolDb.GenBaseAttach(fileTable.Rows[f].Columns[1].ToPlainTextString().Trim(), info.Id, "http://www.bajsjy.com/" + (fileTable.Rows[f].Columns[1].SearchFor(typeof(ATag), true)[0] as ATag).Link.Replace("../", "/")); base.AttachList.Add(attach); } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }
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(this.ToolWebSite.UrlEncode(SiteUrl), Encoding.UTF8, ref cookiestr); } catch (Exception ex) { return(list); } Parser parser = new Parser(new Lexer(htl)); NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxma03"))); Regex regexPage = new Regex(@"共\d+页"); try { page = int.Parse(regexPage.Match(nodeList.AsString()).Value.Trim(new char[] { '共', '页' })); } catch (Exception) { } for (int i = 1; i <= page; i++) { if (i > 1) { try { htl = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode("http://www.szns.gov.cn/publish/main/1/19/26/zbtbxx/5466/index_" + i.ToString() + ".html"), Encoding.UTF8); } catch (Exception ex) { } } parser = new Parser(new Lexer(htl)); NodeList tableNodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxdianbeijing"))); if (tableNodeList.Count > 0) { for (int j = 0; j < tableNodeList.Count; j++) { ATag aTag = tableNodeList.SearchFor(typeof(ATag), true)[j] as ATag; 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 = aTag.LinkText; Regex regDate = new Regex(@"\d{4}-\d{1,2}-\d{1,2}"); beginDate = regDate.Match(tableNodeList[j].ToPlainTextString()).Value.Trim(); InfoUrl = "http://www.szns.gov.cn" + aTag.Link.Replace("amp;", "").Trim(); string htmldetail = string.Empty; try { htmldetail = this.ToolWebSite.GetHtmlByUrl(this.ToolWebSite.UrlEncode(InfoUrl), Encoding.UTF8).Replace(" ", ""); } catch (Exception) { continue; } Parser parserdetail = new Parser(new Lexer(htmldetail)); NodeList dtnode = parserdetail.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "hyxzf2"))); if (dtnode.Count > 0) { HtmlTxt = dtnode.AsHtml(); Regex regeximg = new Regex(@"<img[^>]*>");//去掉图片 HtmlTxt = regeximg.Replace(HtmlTxt, ""); bidCtx = dtnode.AsString().Replace("\n", "\r\n").Replace(" ", "").Trim(); Regex regexHtml = new Regex(@"<script[^<]*</script>|<\?xml[^/]*/>"); bidCtx = regexHtml.Replace(bidCtx, ""); Regex regBuidUnit = new Regex(@"(招标人|建设单位)(:|:)[^\r\n]+\r\n"); buildUnit = regBuidUnit.Match(bidCtx).Value.Replace("招标人:", "").Replace("建设单位:", "").Trim(); Regex regCode = new Regex(@"工程编号(:|:)[^\r\n]+\r\n"); code = regCode.Match(bidCtx).Value.Replace("工程编号:", "").Trim(); Regex regBidUnit = new Regex(@"中标人(:|:)[^\r\n]+\r\n"); bidUnit = regBidUnit.Match(bidCtx).Value.Replace("中标人:", "").Trim(); Regex regMoney = new Regex(@"(中标价|中标价格)(:|:)[^\r\n]+\r\n"); bidMoney = regMoney.Match(bidCtx).Value.Replace("中标价:", "").Replace("中标价格:", "").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 (buildUnit == "") { buildUnit = ""; } if (bidUnit == "") { bidUnit = ""; } Regex regprjMgr = new Regex(@"(总监|建造师|建造师(总监))(:|:)[^\r\n]+\r\n"); prjMgr = regprjMgr.Match(bidCtx).Value.Replace("建造师:", "").Replace("总监:", "").Replace("建造师(总监):", "").Trim(); msgType = "深圳市南山区政府采购及招标中心"; specType = "建设工程"; bidType = ToolHtml.GetInviteTypes(prjName); 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(null); }
protected override IList ExecuteCrawl(bool crawlAll) { int pageInt = 1, sqlCount = 0; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8); } catch { return(null); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "pages"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode.AsString().GetRegexBegEnd("共", "页"); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl + "index_" + (i - 1) + ".htm"); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("ul"), new HasAttributeFilter("class", "page_list_ul")), true), new TagNameFilter("li"))); if (listNode != null && listNode.Count > 0) { for (int j = 1; j < listNode.Count; j++) { INode node = listNode[j]; string headName = string.Empty, releaseTime = string.Empty, infoScorce = string.Empty, msgType = string.Empty, infoUrl = string.Empty, ctxHtml = string.Empty, infoCtx = string.Empty, infoType = string.Empty; ATag aTag = node.GetATag(); headName = aTag.GetAttribute("title"); releaseTime = node.ToPlainTextString().GetDateRegex(); infoType = "通知公告"; msgType = MsgTypeCosnt.ShenZhenMsgType; infoUrl = "http://www.szjs.gov.cn/jsjy/zxsz/tzgg/" + aTag.Link.GetReplace("../,./"); string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(infoUrl).GetJsString(); } catch { continue; } if (aTag.IsAtagAttach()) { sqlCount++; NotifyInfo info = ToolDb.GenNotifyInfo(headName, releaseTime, infoScorce, msgType, infoUrl, ctxHtml, "广东省", "深圳市工程", "", infoCtx, infoType); if (ToolDb.SaveEntity(info, this.ExistCompareFields, this.ExistsUpdate)) { try { BaseAttach obj = ToolHtml.GetBaseAttach(infoUrl, headName, info.Id); if (obj != null) { ToolDb.SaveEntity(obj, string.Empty); } } catch { } } continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "detail_contect"))); if (dtlNode != null && dtlNode.Count > 0) { ctxHtml = dtlNode.AsHtml(); infoCtx = ctxHtml.ToCtxString(); NotifyInfo info = ToolDb.GenNotifyInfo(headName, releaseTime, infoScorce, msgType, infoUrl, ctxHtml, "广东省", "深圳市工程", "", infoCtx, infoType); if (!crawlAll && sqlCount >= this.MaxCount) { return(null); } sqlCount++; if (ToolDb.SaveEntity(info, this.ExistCompareFields, this.ExistsUpdate)) { parser = new Parser(new Lexer(ctxHtml)); NodeList aNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a")); if (aNode != null && aNode.Count > 0) { for (int k = 0; k < aNode.Count; k++) { ATag tag = aNode[k].GetATag(); if (tag.IsAtagAttach()) { string temp = aTag.Link.GetReplace("../,./"); string alink = "http://www.szjs.gov.cn/jsjy/zxsz/tzgg/" + temp.Substring(0, temp.LastIndexOf("/")) + tag.Link.GetReplace("./", "/"); try { BaseAttach obj = ToolHtml.GetBaseAttach(alink, tag.LinkText, info.Id); if (obj != null) { ToolDb.SaveEntity(obj, string.Empty); } } catch { } } } } } } } } } return(null); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <ItemPlan>(); Dictionary <string, string> dic = GetCityList(); if (dic == null || dic.Count < 1) { return(list); } foreach (string key in dic.Keys) { string html = string.Empty; string cookiestr = string.Empty; string viewState = string.Empty; int pageInt = 1, sqlCount = 0; string eventValidation = string.Empty; try { this.ToolWebSite.GetHtmlByUrl(dic[key], Encoding.UTF8, ref cookiestr); html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8, ref cookiestr); } catch { } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "badoo")), true), new TagNameFilter("a"))); if (pageNode != null && pageNode.Count > 0) { try { string temp = pageNode[pageNode.Count - 1].GetATag().Link.Replace("javascript", "").Replace("jumpPage(", "").Replace(")", ""); pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(new string[] { "param.name", "param.proofCode", "page.pageNo", "page.orderBy", "page.order" }, new string[] { "", "", i.ToString(), "", "" }); try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, nvc, Encoding.UTF8, ref cookiestr); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("id", "hytab"))); if (listNode != null && listNode.Count > 0) { TableTag table = listNode[0] as TableTag; for (int j = 1; j < table.RowCount; j++) { string ItemCode = string.Empty, ItemName = string.Empty, ItemAddress = string.Empty, BuildUnit = string.Empty, BuildNature = string.Empty, TotalInvest = string.Empty, PlanInvest = string.Empty, IssuedPlan = string.Empty, InvestSource = string.Empty, ApprovalUnit = string.Empty, ApprovalDate = string.Empty, ApprovalCode = string.Empty, MsgUnit = string.Empty, PlanDate = string.Empty, PlanType = string.Empty, PlanBeginDate = string.Empty, PlanEndDate = string.Empty, CtxHtml = string.Empty, ItemCtx = string.Empty, ItemContent = string.Empty, InfoUrl = string.Empty, MsgType = string.Empty; TableRow tr = table.Rows[j]; ItemName = tr.Columns[0].ToNodePlainString(); PlanDate = tr.Columns[2].ToPlainTextString().GetDateRegex(); InfoUrl = "http://www.gdtz.gov.cn" + tr.Columns[0].GetATagHref(); string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "xmgknr"))); if (dtlNode != null && dtlNode.Count > 0) { CtxHtml = dtlNode.AsHtml(); parser = new Parser(new Lexer(CtxHtml)); NodeList tableNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("table")); if (tableNode != null && tableNode.Count > 0) { TableTag tab = tableNode[0] as TableTag; for (int k = 1; k < tab.RowCount; k++) { TableRow dr = tab.Rows[k]; if (dr.ColumnCount < 2) { break; } try { ItemCtx += dr.Columns[0].ToNodePlainString() + ":"; ItemCtx += dr.Columns[1].ToNodePlainString() + "\r\n"; } catch (Exception ex) { Logger.Error(InfoUrl + ItemName + key + i); Logger.Error(ex); } } } else { ItemCtx = CtxHtml.ToCtxString(); } ApprovalCode = ItemCtx.GetRegex("备案项目编号"); ItemAddress = ItemCtx.GetRegex("项目所在地"); TotalInvest = ItemCtx.GetRegex("项目总投资").Replace("万元", "").Replace("万", ""); ItemContent = ItemCtx.GetRegex("项目规模及内容"); ApprovalUnit = ItemCtx.GetRegex("备案机关"); ApprovalDate = ItemCtx.GetRegex("复核通过日期"); string temp = ItemCtx.GetRegex("项目起止年限"); string[] tempPlan = temp.Split('-'); if (tempPlan.Length == 2) { PlanBeginDate = tempPlan[0]; PlanEndDate = tempPlan[1]; } PlanType = "项目公开"; MsgType = "广东省发展和改革委员会"; string city = key; if (key.Contains("顺德")) { city = "佛山市区"; } ItemPlan info = ToolDb.GenItemPlan("广东省", city, "", ItemCode, ItemName, ItemAddress, BuildUnit, BuildNature, TotalInvest, PlanInvest, IssuedPlan, InvestSource, ApprovalUnit, ApprovalDate, ApprovalCode, MsgUnit, PlanDate, PlanType, PlanBeginDate, PlanEndDate, CtxHtml, ItemCtx, ItemContent, MsgType, InfoUrl); list.Add(info); sqlCount++; if (!crawlAll && sqlCount >= this.MaxCount) { goto type; } } } } } type : continue; } return(list); }
protected override IList ExecuteCrawl(bool crawlAll) { IList list = new List <InviteInfo>(); int pageInt = 1; string html = string.Empty; string viewState = string.Empty; string eventValidation = string.Empty; string cookiestr = string.Empty; try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl, Encoding.UTF8); } catch { return(null); } Parser parser = new Parser(new Lexer(html)); NodeList pageNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "navigation"))); if (pageNode != null && pageNode.Count > 0) { string temp = pageNode[0].ToNodePlainString().GetRegexBegEnd("总共", "页").GetReplace("【,】,[,]"); try { pageInt = int.Parse(temp); } catch { } } for (int i = 1; i <= pageInt; i++) { if (i > 1) { try { html = this.ToolWebSite.GetHtmlByUrl(this.SiteUrl + "&page=" + i, Encoding.UTF8); } catch { continue; } } parser = new Parser(new Lexer(html)); NodeList listNode = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("class", "slidingList")), true), new TagNameFilter("li"))); if (listNode != null && listNode.Count > 0) { for (int j = 0; j < listNode.Count; j++) { string code = string.Empty, buildUnit = string.Empty, prjName = string.Empty, prjAddress = string.Empty, inviteCtx = string.Empty, inviteType = string.Empty, specType = string.Empty, beginDate = string.Empty, endDate = string.Empty, remark = string.Empty, inviteCon = string.Empty, InfoUrl = string.Empty, CreateTime = string.Empty, msgType = string.Empty, otherType = string.Empty, HtmlTxt = string.Empty; INode node = listNode[j]; ATag aTag = node.GetATag(); prjName = aTag.GetAttribute("title"); beginDate = node.GetSpan().StringText; if (!string.IsNullOrEmpty(beginDate)) { beginDate = beginDate.Substring(0, 4) + "-" + beginDate.Substring(4, 2) + "-" + beginDate.Substring(6, 2); } InfoUrl = "http://www.gsggzyjy.cn" + aTag.Link; string htmldtl = string.Empty; try { htmldtl = this.ToolWebSite.GetHtmlByUrl(InfoUrl).GetJsString(); } catch { continue; } parser = new Parser(new Lexer(htmldtl)); NodeList dtlNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "ContentPlaceHolder1_AnnoGoodsHtml"))); if (dtlNode != null && dtlNode.Count > 0) { HtmlTxt = dtlNode.AsHtml(); inviteCtx = HtmlTxt.ToCtxString(); msgType = "甘肃省公共资源交易中心"; specType = "政府采购"; inviteType = "水利及其他工程"; InviteInfo info = ToolDb.GenInviteInfo("甘肃省", "甘肃省及地市", "", string.Empty, code, prjName, prjAddress, buildUnit, beginDate, endDate, inviteCtx, remark, msgType, inviteType, specType, otherType, InfoUrl, HtmlTxt); list.Add(info); parser = new Parser(new Lexer(htmldtl)); NodeList aNode = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("Iframe"), new HasAttributeFilter("id", "Iframe"))); if (aNode != null && aNode.Count > 0) { for (int k = 0; k < aNode.Count; k++) { IFrameTag itag = aNode[k] as IFrameTag; string link = itag.GetAttribute("src"); if (!string.IsNullOrEmpty(link)) { BaseAttach attach = ToolDb.GenBaseAttach(prjName + ".pdf", info.Id, link); base.AttachList.Add(attach); } } } parser = new Parser(new Lexer(HtmlTxt)); NodeList atagNode = parser.ExtractAllNodesThatMatch(new TagNameFilter("a")); if (atagNode != null && atagNode.Count > 0) { for (int a = 0; a < atagNode.Count; a++) { ATag fileTag = atagNode[a] as ATag; if (fileTag.IsAtagAttach()) { string link = string.Empty; if (fileTag.Link.Contains("http")) { link = fileTag.Link; } else { link = "http://www.gsggzyjy.cn/" + fileTag.Link; } BaseAttach attach = ToolDb.GenBaseAttach(fileTag.LinkText, info.Id, link); if (!base.AttachList.Exists(x => x.AttachServerPath == link)) { base.AttachList.Add(attach); } } } } if (!crawlAll && list.Count >= this.MaxCount) { return(list); } } } } } return(list); }