protected override IList ExecuteCrawl(bool crawlAll)
{
//取得页码
int pageInt = 1, sqlCount = 0;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.Default);
}
catch (Exception ex)
{
return(null);
}
Parser parser = new Parser(new Lexer(html));
NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("select"), new HasAttributeFilter("id", "dnn_ctr467_ArticleList_cboPages")), true), new TagNameFilter("option")));
if (pageList != null && pageList.Count > 0)
{
try
{
pageInt = pageList.Count;
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
viewState = this.ToolWebSite.GetAspNetViewState(html);
NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(
new string[] { "__EVENTARGUMENT", "dnn:ctr467:ArticleList:cboPages",
"ScrollTop", "__dnnVariable", "__VIEWSTATE" },
new string[] { "", (i - 1).ToString(), "", "", viewState }
);
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, nvc, Encoding.Default);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "dnn_ctr467_ArticleList_PanelA")), true), new TagNameFilter("table")));
if (nodeList != null && nodeList.Count > 0)
{
TableTag table = nodeList[0] as TableTag;
for (int j = 0; j < table.RowCount; 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];
infoType = "通知公告";
releaseTime = "20" + tr.Columns[2].ToPlainTextString().GetDateRegex("yy-MM-dd");
headName = tr.Columns[1].ToNodePlainString();
infoUrl = "http://www.szmea.net" + tr.Columns[1].GetATagHref();
string htldtl = string.Empty;
try
{
htldtl = ToolHtml.GetHtmlByUrl(SiteUrl, infoUrl, Encoding.Default).GetJsString(); //ToolHtml.GetHtmlByUrl(infoUrl, Encoding.Default).GetJsString();
}
catch { }
parser = new Parser(new Lexer(htldtl));
NodeList noList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("id", "dnn_ctr391_ArticleShow_lblContent")));
if (noList != null && noList.Count > 0)
{
ctxHtml = noList.AsHtml().Replace("<br/>", "\r\n").Replace("<BR/>", "");
infoCtx = ctxHtml.ToCtxString().Replace(" ", "").Replace(" ", "").Replace("\t\t", "\t").Replace("\t\t", "\t");
infoCtx = Regex.Replace(infoCtx, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase).Replace(" ", "").Replace("\t", "").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n");
msgType = MsgTypeCosnt.ShenZhenJLGCMsgType;
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);
sqlCount++;
if (!crawlAll && sqlCount >= this.MaxCount)
{
return(null);
}
if (ToolDb.SaveEntity(info, this.ExistCompareFields))
{
parser = new Parser(new Lexer(ctxHtml));
NodeList imgList = parser.ExtractAllNodesThatMatch(new TagNameFilter("img"));
if (imgList != null && imgList.Count > 0)
{
for (int m = 0; m < imgList.Count; m++)
{
try
{
ImageTag img = imgList[m] as ImageTag;
string src = img.GetAttribute("src");
if (src.ToLower().Contains(".gif"))
{
continue;
}
BaseAttach obj = null;
if (src.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(src, headName, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.szmea.net" + src.Replace("../", "/").Replace("./", "/"), headName, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { }
}
}
parser = new Parser(new Lexer(ctxHtml));
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] as ATag;
if (aTag.IsAtagAttach())
{
try
{
BaseAttach obj = null;
string href = aTag.GetATagHref();
if (href.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(href, aTag.LinkText, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.szmea.net" + href.Replace("../", "/").Replace("./", "/"), aTag.LinkText, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { continue; }
}
}
}
}
}
}
}
}
return(null);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
//取得页码
int pageInt = 1, sqlCount = 0;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.UTF8);
}
catch (Exception ex)
{
return(null);
}
Parser parser = new Parser(new Lexer(html));
NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("a"), new HasAttributeFilter("id", "PageDataList__ctl7_LinkButton1")));
if (pageList != null && pageList.Count > 0)
{
try
{
string temp = pageList.AsString();
pageInt = Convert.ToInt32(temp.GetRegexBegEnd("共", "页"));
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
viewState = this.ToolWebSite.GetAspNetViewState(html);
NameValueCollection nvc = this.ToolWebSite.GetNameValueCollection(
new string[] {
"__EVENTTARGET", "__EVENTARGUMENT", "__VIEWSTATE", "head1:username",
"head1:Password", "head1:rbLoginType", "Tb_keyword", "ddlNewsType", "ddlistaddnewsdate"
},
new string[] {
"PageDataList$_ctl" + (i + 1).ToString() + "$LinkButton1", "", viewState, "", "", "unit", "", "20", ""
}
);
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("class", " tb_list")));
if (nodeList != null && nodeList.Count > 0)
{
TableTag table = nodeList[0] as TableTag;
for (int j = 0; j < table.RowCount; 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];
infoType = "通知公告";
releaseTime = tr.Columns[2].ToPlainTextString().GetDateRegex();
headName = tr.Columns[1].ToNodePlainString();
infoUrl = "http://www.szpark.com.cn" + tr.Columns[1].GetATagHref();
string htldtl = string.Empty;
try
{
htldtl = ToolHtml.GetHtmlByUrl(infoUrl, Encoding.UTF8).GetJsString();
}
catch { }
parser = new Parser(new Lexer(htldtl));
NodeList noList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("div"), new HasAttributeFilter("id", "newsinfo")));
if (noList != null && noList.Count > 0)
{
ctxHtml = noList.AsHtml().Replace("<br/>", "\r\n").Replace("<BR/>", "");
infoCtx = ctxHtml.ToCtxString().Replace(" ", "").Replace(" ", "").Replace("\t\t", "\t").Replace("\t\t", "\t");
infoCtx = Regex.Replace(infoCtx, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase).Replace(" ", "").Replace("\t", "").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n");
msgType = MsgTypeCosnt.ShenZhenFJYLMsgType;
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);
sqlCount++;
if (!crawlAll && sqlCount >= this.MaxCount)
{
return(null);
}
if (ToolDb.SaveEntity(info, this.ExistCompareFields))
{
parser = new Parser(new Lexer(ctxHtml));
NodeList imgList = parser.ExtractAllNodesThatMatch(new TagNameFilter("img"));
if (imgList != null && imgList.Count > 0)
{
for (int m = 0; m < imgList.Count; m++)
{
try
{
ImageTag img = imgList[m] as ImageTag;
string src = img.GetAttribute("src");
if (src.ToLower().Contains(".gif"))
{
continue;
}
BaseAttach obj = null;
if (src.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(src, headName, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.szpark.com.cn" + src.Replace("../", "/").Replace("./", "/"), headName, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { }
}
}
parser = new Parser(new Lexer(ctxHtml));
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] as ATag;
if (aTag.IsAtagAttach())
{
try
{
BaseAttach obj = null;
string href = aTag.GetATagHref();
if (href.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(href, aTag.LinkText, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.szpark.com.cn" + href.Replace("../", "/").Replace("./", "/"), aTag.LinkText, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { }
}
}
}
}
}
}
}
}
return(null);
}
protected override IList ExecuteCrawl(bool crawlAll)
{
//取得页码
int pageInt = 1, sqlCount = 0;
string html = string.Empty;
string viewState = string.Empty;
string eventValidation = string.Empty;
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl, Encoding.Default);
}
catch (Exception ex)
{
return(null);
}
Parser parser = new Parser(new Lexer(html));
NodeList pageList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("span"), new HasAttributeFilter("class", "f12")));
if (pageList != null && pageList.Count > 0)
{
try
{
string temp = pageList.AsString();
pageInt = Convert.ToInt32(temp.GetRegexBegEnd("/", "页"));
}
catch { pageInt = 1; }
}
for (int i = 1; i <= pageInt; i++)
{
if (i > 1)
{
try
{
html = this.ToolWebSite.GetHtmlByUrl(SiteUrl + "&page=" + i.ToString(), Encoding.Default);
}
catch { continue; }
}
parser = new Parser(new Lexer(html));
NodeList nodeList = parser.ExtractAllNodesThatMatch(new AndFilter(new HasParentFilter(new AndFilter(new TagNameFilter("td"), new HasAttributeFilter("height", "32")), true), new TagNameFilter("table")));
if (nodeList != null && nodeList.Count > 0)
{
TableTag table = nodeList[0] as TableTag;
for (int j = 0; j < table.RowCount - 1; j++)
{
if ((j + 1) % 2 == 0)
{
continue;
}
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];
infoType = "通知公告";
releaseTime = tr.Columns[2].ToPlainTextString().GetDateRegex();
headName = tr.Columns[1].ToNodePlainString();
infoUrl = "http://www.jianzhuxh.com/news/" + tr.Columns[1].GetATagValue("onclick").GetRegexBegEnd("'", "'");
string htldtl = string.Empty;
try
{
htldtl = ToolHtml.GetHtmlByUrl(infoUrl, Encoding.Default).GetJsString();
}
catch { continue; }
parser = new Parser(new Lexer(htldtl));
NodeList noList = parser.ExtractAllNodesThatMatch(new AndFilter(new TagNameFilter("table"), new HasAttributeFilter("class", "text18")));
if (noList != null && noList.Count > 0)
{
ctxHtml = noList[0].ToHtml().Replace("<br/>", "\r\n").Replace("<BR/>", "");
infoCtx = ctxHtml.ToCtxString().Replace(" ", "").Replace(" ", "").Replace("\t\t", "\t").Replace("\t\t", "\t");
infoCtx = Regex.Replace(infoCtx, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase).Replace(" ", "").Replace("\t", "").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n").Replace("\r\n\r\n", "\r\n");
msgType = MsgTypeCosnt.ShenZhenJZYMsgType;
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);
sqlCount++;
if (!crawlAll && sqlCount >= this.MaxCount)
{
return(null);
}
if (ToolDb.SaveEntity(info, this.ExistCompareFields))
{
parser = new Parser(new Lexer(ctxHtml));
NodeList imgList = parser.ExtractAllNodesThatMatch(new TagNameFilter("img"));
if (imgList != null && imgList.Count > 0)
{
for (int m = 0; m < imgList.Count; m++)
{
try
{
ImageTag img = imgList[m] as ImageTag;
string src = img.GetAttribute("src");
if (src.ToLower().Contains(".gif"))
{
continue;
}
BaseAttach obj = null;
if (src.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(src, headName, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.jianzhuxh.com" + src.Replace("../", "/").Replace("./", "/"), headName, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { }
}
}
parser = new Parser(new Lexer(ctxHtml));
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] as ATag;
if (aTag.IsAtagAttach())
{
try
{
BaseAttach obj = null;
string href = aTag.GetATagHref();
if (href.Contains("http"))
{
obj = ToolHtml.GetBaseAttach(href, aTag.LinkText, info.Id);
}
else
{
obj = ToolHtml.GetBaseAttach("http://www.jianzhuxh.com" + href.Replace("../", "/").Replace("./", "/"), aTag.LinkText, info.Id);
}
if (obj != null)
{
ToolDb.SaveEntity(obj, string.Empty);
}
}
catch { }
}
}
}
}
}
}
}
}
return(null);
}
