public void SaveAllSiteClass()
{
HasBindClasslist = new SiteClassInfoDB().getAllSiteCatInfo(Baseinfo.SiteId);
string url = "http://www.j1.com/sitemap.html";
string page = HtmlAnalysis.Gethtmlcode(url);
string content = RegGroupsX<string>(page, "<div class=\"sitemap_sortwrap qbfl\">(?<x>.*?)</div>");
var list = RegGroupCollection(content, "<a target='_blank' href='(?<y>.*?)'>(?<x>.*?)</a>");
for (int i = 0; i < list.Count; i++)
{
string catUrl = list[i].Groups["y"].Value;
string catid = RegGroupsX<string>(catUrl, "http://www.j1.com/p-(?<x>\\d+)");
string catName = list[i].Groups["x"].Value;
if (!HasBindClasslist.Exists(p => p.ClassId == catid))
{
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo = catUrl,
ClassId = catid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = catName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
}
private void GetYhdClassInfo(string url)
{
if (!url.Contains("b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/"))
url += "b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/";
string pageInfo = Gethtmlcode(url);
string classInfo = RegGroupsX<string>(pageInfo, "<ul class=\"listCon clearfix\">(?<x>.*?)<a class=\"c_btn c_next iconSearch\"");
var list = RegGroupCollection(classInfo, "href=\"(?<x>.*?)\".*?title=\"(?<y>.*?)\"");
if (list == null)
return;
string pcatUrl = "";
string pcatName = "";
string pcatId = "";
string classCrumble = "";
int total = RegGroupsX<int>(pageInfo, "共(?<x>\\d+?)条");
if (list.Count == 1)
{
string categoryname = RegGroupsX<string>(pageInfo, "var categoryName = '(?<x>.*?)'");
string extid = RegGroupsX<string>(pageInfo, "var expectCategoryId = \"(?<x>\\d+)\"");
string tempcurcatid = "c" + extid + "-" + categoryname;
if(HasBindClasslist.Exists(p => p.ClassId == tempcurcatid))
return;
string current = RegGroupsX<string>(pageInfo, "<title>(?<x>.*?)品种齐全|<div class=\"guide_title\"><span title=\"(?<x>.*?)\">");
SiteClassInfo catInfo = new SiteClassInfo
{
Urlinfo = $"http://list.yhd.com/{tempcurcatid}/b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k",
ClassId = tempcurcatid,
ClassName = current,
BindClassId = 0,
BindClassName = "",
CreateDate = DateTime.Now,
UpdateTime = DateTime.Now,
SiteId = Baseinfo.SiteId,
IsBind = false,
IsDel = false,
IsHide = false
};
catInfo.HasChild = HasBindClasslist.Exists(p => p.ParentClass == catInfo.ClassId);
catInfo.ParentClass = pcatId;
catInfo.ParentName = pcatName;
catInfo.ParentUrl = pcatUrl;
if (regIsMatch(tempcurcatid, "^(?<x>c\\d+-\\d+(-\\d+)?)$"))
{
if (!HasBindClasslist.Exists(p => p.ClassId == catInfo.ClassId))
{
new SiteClassInfoDB().AddSiteClass(catInfo);
LogServer.WriteLog("线程id：" + Thread.CurrentThread.ManagedThreadId + "\t" + catInfo.ClassId + "\t" + catInfo.ClassName + "111111111111111", "addpro");
HasBindClasslist.Add(catInfo);
}
}
return;
}
//SiteClassInfoDB db = new SiteClassInfoDB();
for (int i = 0; i < list.Count; i++)
{
SiteClassInfo catInfo = new SiteClassInfo();
catInfo.Urlinfo = list[i].Groups["x"].Value;
catInfo.ClassId = RegGroupsX<string>(catInfo.Urlinfo, "http://list.yhd.com/(?<x>.*?)/");
catInfo.ClassName = list[i].Groups["y"].Value;
if (catInfo.Urlinfo == null || catInfo.ClassId == null || catInfo.ClassName == null)
{
LogServer.WriteLog(Baseinfo.SiteName + "分类抓取错误1url\t" + url, "AddClassError");
continue;
}
if (!HasBindClasslist.Exists(p => p.ClassId == catInfo.ClassId) )
{
catInfo.ParentClass = pcatId;
catInfo.ParentName = pcatName;
catInfo.ParentUrl = pcatUrl;
if (i != 0 && pcatId!="")
classCrumble += pcatId + ",";
catInfo.ClassCrumble = classCrumble.TrimEnd(',');
catInfo.BindClassId = 0;
catInfo.BindClassName = "";
catInfo.CreateDate = DateTime.Now;
catInfo.UpdateTime = DateTime.Now;
catInfo.SiteId = Baseinfo.SiteId;
catInfo.IsBind = false;
catInfo.IsDel = false;
catInfo.IsHide = false;
catInfo.HasChild = true;
if( list.Count - 1 == i)
{
catInfo.HasChild = false;
catInfo.TotalProduct = total;
}
HasBindClasslist.Add(catInfo);
LogServer.WriteLog("线程id："+Thread.CurrentThread.ManagedThreadId+ "\t" + catInfo.ClassId +"\t"+catInfo.ClassName+"111111111111111","addpro");
new SiteClassInfoDB().AddSiteClass(catInfo);
}
pcatUrl = catInfo.Urlinfo;
pcatName = catInfo.ClassName;
pcatId = catInfo.ClassId;
}
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string pageinfo = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string crumb = RegGroupsX<string>(pageinfo, "<div class=detailnav>(?<x>.*?)</div>");
if (crumb == null)
return;
var deep = RegGroupCollection(crumb, "href=\"(?<y>.*?)\"( target=\"_blank\")?>(?<x>.*?)</a>");
string parentUrl = "";
string parentName = "";
string parentId = "";
for (int i = 0; i < deep.Count; i++)
{
if (deep[i].Groups["x"].Value.Contains("首页") || deep[i].Groups["x"].Value == siteClassInfo.ClassName)
continue;
parentUrl = deep[i].Groups["y"].Value;
parentName = deep[i].Groups["x"].Value;
parentId = RegGroupsX<string>(parentUrl, "http://www.j1.com/p-(?<x>\\d+)");
}
string children = RegGroupsX<string>(pageinfo, "<div class=\"listpageChooseBox\">(?<x>.*?)</div>");
var catlist = RegGroupCollection(children, "<a href=\"(?<y>.*?)\">(?<x>.*?)<span>");
siteClassInfo.HasChild = !children.Contains(siteClassInfo.ClassName);
if (catlist != null)
{
for (int i = 0; i < catlist.Count; i++)
{
string url = catlist[i].Groups["y"].Value;
string catid = RegGroupsX<string>(url, "http://www.j1.com/p-(?<x>\\d+)");
if (!HasBindClasslist.Exists(p => p.ClassId == catid))
{
string catName = catlist[i].Groups["x"].Value;
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo = url,
ClassId = catid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = catName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
siteClassInfo.ParentUrl = parentUrl;
siteClassInfo.ParentClass = parentId;
siteClassInfo.ParentName = parentName;
siteClassInfo.TotalProduct = RegGroupsX<int>(pageinfo, "共(?<x>\\d+)个商品");
siteClassInfo.UpdateTime = DateTime.Now;
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string page = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string cromb = RegGroupsX<string>(page, "您现在的位置：</span>(?<x>.*?)</div>");
if (cromb == null)
return;
var plist = RegGroupCollection(cromb, "<a class=\"\" href=\"(?<x>.*?)\">(?<y>.*?)</a>");
if (plist == null)
return;
string parentUrl = "";
string parentName = "";
string parentId = "";
foreach (Match item in plist)
{
parentUrl = item.Groups["x"].Value;
parentName = item.Groups["y"].Value;
if(parentName=="首页")
{
continue;
}
if (!string.IsNullOrEmpty(parentName))
{
parentName = parentName.Trim();
}
if (parentName == "")
{ parentUrl = ""; continue; }
parentId = RegGroupsX<string>(parentUrl, "category/(?<x>\\d+)-");
if (!ValidCatId(parentId))
{
parentId = RegGroupsX<string>(parentUrl, "/(?<x>.*?).html");
if (string.IsNullOrEmpty(parentId))
continue;
}
parentUrl = string.Format(domain+"{0}", parentUrl);
if (!HasBindClasslist.Exists(c => c.ClassId == parentId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = parentName,
ClassId = parentId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = parentUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
var templist = RegGroupCollection(page, "getJSON\\(\"(?<x>.*?)\"");
for (int i = 0; i < templist.Count; i++)
{
var caturl = templist[i].Groups["x"].Value;
string temppage = HtmlAnalysis.Gethtmlcode(domain+ caturl);
var catlist = RegGroupCollection(temppage, "n_(?<x>.*?)\"EntityState");
if (catlist == null)
continue;
foreach (Match item in catlist)
{
string cat = item.Groups["x"].Value;
string catid = RegGroupsX<string>(cat, "id\":(?<x>\\d+),");
string catName = RegGroupsX<string>(cat, "\"n_name\":\"(?<x>.*?)\"");
string catpid = RegGroupsX<string>(cat, "\"parentid\":(?<x>.*?),");
string tempurl = string.Format("http://www.360kxr.com/category/{0}-0-2-1-15-1.html", catid);
if (!HasBindClasslist.Exists(c => c.ClassId == catid))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = catpid,
ParentName = "",
ClassName = catName,
ClassId = catid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = tempurl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
siteClassInfo.HasChild = HasBindClasslist.Exists(c => c.ParentClass == siteClassInfo.ClassId);
siteClassInfo.ParentClass = parentId;
siteClassInfo.ParentName = parentName;
siteClassInfo.ParentUrl = parentUrl;
siteClassInfo.UpdateTime = DateTime.Now;
siteClassInfo.TotalProduct = RegGroupsX<int>(page, "<div class=\"goods-total\">共<b>(?<x>\\d+)</b>个商品</div>");
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
public bool updateSpiderOnly(SiteClassInfo catinfo)
{
if (catinfo == null) throw new ArgumentNullException("catinfo");
using (var db = _dbFactory.OpenDbConnection())
{
try
{
//db.UpdateNonDefaults(new SiteClassInfo { Id = catinfo.Id, ClassName = catinfo.ClassName, ClassId = catinfo.ClassId, ParentClass = catinfo.ParentClass, ParentName = catinfo.ParentName, ParentUrl = catinfo.ParentUrl, UpdateTime = DateTime.Now, HasChild = catinfo.HasChild, Urlinfo = catinfo.Urlinfo, ClassCrumble = catinfo.ClassCrumble }, p => p.Id == catinfo.Id);
int res = db.UpdateOnly(catinfo,
u => new { u.ClassId, u.ClassName, u.Urlinfo, u.UpdateTime, u.ParentClass, u.ParentUrl, u.ParentName, u.ClassCrumble, u.TotalProduct, u.HasChild }, u => u.Id == catinfo.Id);
if (res > 0)
return true;
return false;
}
catch (Exception ex)
{
LogServer.WriteLog(ex, "DBError");
return false;
}
//int res = db.Update(catinfo);
//db.Update(catinfo, p => p.ClassId == catinfo.ClassId);
}
}
public void SetIsDel(SiteClassInfo catinfo)
{
if (catinfo == null) throw new ArgumentNullException("catinfo");
using (var db = _dbFactory.OpenDbConnection())
{
try
{
db.UpdateOnly(new SiteClassInfo{IsDel=true}, p => p.IsDel,p=>p.Id==catinfo.Id);
}
catch (Exception ex)
{
LogServer.WriteLog(ex, "DBError");
// throw;
}
}
}
private void GetCatInfo(string directoryHtml)
{
string catArea = RegGroupsX<string>(directoryHtml,
"<div class=\"all-category-box\">(?<x>.*?)<div class=\"category-contact\">");
if (catArea == null)
return;
var list = RegGroupCollection(catArea, "href=\"(?<x>.*?)\".*?>(?<y>.*?)</a>");
foreach (Match item in list)
{
string tempUrl = item.Groups["x"].Value;
string tempName = item.Groups["y"].Value;
if (tempName == "商品分类")
continue;
string tempid = RegGroupsX<string>(tempUrl, "http://www.ehaier.com/l/(?<x>\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+-\\d+).html");
if (!ValidCatId(tempid))
continue;
if (!HasBindClasslist.Exists(c => c.ClassId == tempid))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild =tempid.Contains("-"),
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = tempUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void GetCatInfo(string directoryHtml)
{
string catArea = RegGroupsX<string>(directoryHtml, "<div class=\"category-box\" id=\"category_box\">(?<x>.*?)</ul>");
if (catArea == null)
return;
var list = RegGroupCollection(catArea, "<a(?<x>.*?)</a>");
foreach (Match item in list)
{
string tempUrl = RegGroupsX<string>(item.ToString(), "href=\"(?<x>.*?)\"");
if (string.IsNullOrEmpty(tempUrl))
continue;
tempUrl = string.Format("http://www.hangowa.com{0}", tempUrl);
string tempName = RegGroupsX<string>(item.ToString(), ">(?<x>.*?)</a>");
string tempid = RegGroupsX<string>(tempUrl, "gallery-(?<x>\\d+?).html");
if (!HasBindClasslist.Exists(c => c.ClassId == tempid))
{
int page = RegGroupsX<int>(directoryHtml, "共<b class=\"op-search-result\">(?<x>\\d+?)</b>件");
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = !item.ToString().Contains("class=\"level3\""),
ClassCrumble = "",
TotalProduct = page,
SiteId = Baseinfo.SiteId,
Urlinfo = tempUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
/// <summary>
/// 批量添加分类
/// </summary>
/// <param name="catid"></param>
private void bathAddCat(string catid)
{
SiteClassBll bll = new SiteClassBll();
int siteClassId;
int parentId;
SiteClassInfo siteCat = new SiteClassInfo();
if (int.TryParse(catid, out siteClassId))
{
siteCat = bll.GetCatById(siteClassId);
SiteCatName = siteCat.ClassName.Replace(" ","");
}
if (!int.TryParse(Request.Form["parCatName"], out parentId))
return;
#region 添加分类
ClassInfo cat = new ClassInfo();
cat.CatName = SiteCatName;
cat.SpellWord = WordCenter.GetShortPinyin(cat.CatName);
cat.SEOWords = "";
cat.Sort = 0;
cat.CreateDate = DateTime.Now;
cat.UpdateTime = DateTime.Now;
cat.HasChild = false;
ClassInfoBll catbll = new ClassInfoBll();
if (parentId != 0)
{
ClassInfo parCat = catbll.getCat(parentId);
cat.ParentId = parCat.Id;
cat.Level = parCat.Level + 1;
cat.ParentName = parCat.CatName;
if (!string.IsNullOrEmpty(parCat.CatCrumbleIds))
{
cat.CatCrumbleIds = parCat.CatCrumbleIds + "," + parCat.Id;
cat.CatCrumbleNames = parCat.CatCrumbleNames + "," + parCat.CatName;
}
else
{
cat.CatCrumbleIds = parCat.Id.ToString(CultureInfo.InvariantCulture);
cat.CatCrumbleNames = parCat.CatName;
}
if (!parCat.HasChild)
{
parCat.HasChild = true;
catbll.UpdateCat(parCat);
}
}
else
{
cat.ParentId = 0;
cat.Level = 1;
cat.ParentName = "";
cat.CatCrumbleIds = "";
cat.CatCrumbleNames = "";
}
cat.Id = catbll.AddCat(cat);
#endregion
#region 添加子分类
AllSiteCat = bll.GetClassInfo(siteCat.SiteId);
addChildCat(siteCat.ClassId, cat);
#endregion
}
private void UpdateTmallNode(SiteClassInfo item)
{
if(item.ParentClass!="")
return;
string url = string.Format("http://list.tmall.com/search_product.htm?cat={0}" , item.ClassId);
HtmlAnalysis reqest = new HtmlAnalysis();
reqest.Headers.Add("Cookie", "_med=dw:1440&dh:900&pw:1440&ph:900&ist:0; pnm_cku822=126UW5TcyMNYQwiAiwQRHhBfEF8QXtHcklnMWc%3D%7CUm5Ockt%2BQXVPdUp%2BQH9Dfyk%3D%7CU2xMHDJxPk82UjVOI1h2VnhCbExiPl85VTJMNhhOGA%3D%3D%7CVGhXd1llXGlWYlhiXWlXaFRoX2JAekN3TndMdUB1S3RAeUx0TmA2%7CVWldfS0SMgwzCCgULg4gWz0ReEB2Aix6LA%3D%3D%7CVmhIGCcYJAQ%2FAyMXLRc3DTQNORklHCUYOAwxDCwQKRAtDTgDPmg%2B%7CV25Tbk5zU2xMcEl1VWtTaUlwJg%3D%3D; cq=ccp%3D1; tt=login.taobao.com; res=scroll%3A990*776-client%3A977*290-offset%3A977*290-screen%3A1440*900; hng=; uss=BqRyb7nd5KLIbC5D91VCamaiwt66iy8KP0cAS24EJNQWFeWsxGZv%2FwEo%2BAs%3D; cna=cFJaEEwJdRsCATyy24A1yMNe; l=AkZGKGJIZ/WDVSsY65u6dVSj1jLItYph; isg=Alpa8TTm1nmgf1rVi7OVW5M1rADEst5lFaLZTWTTFO241_oRTBsudSCt8xs0; OZ_1U_2061=vid=v801c15a894bb1.0&ctime=1478143053<ime=1476512356; otherx=e%3D1%26p%3D*%26s%3D0%26c%3D0%26f%3D0%26g%3D0%26t%3D0; x=__ll%3D-1%26_ato%3D0; _tb_token_=eeed7bb353eb5; ck1=; uc1=cookie14=UoW%2FX9QwsnjAzg%3D%3D&lng=zh_CN&cookie16=W5iHLLyFPlMGbLDwA%2BdvAGZqLg%3D%3D&existShop=false&cookie21=V32FPkk%2FhSg%2F&tag=0&cookie15=UIHiLt3xD8xYTw%3D%3D&pas=0; uc3=sg2=AQI4ctClVx2ycnFp5kyAa%2F3VFKDYjzhZBJFC8KK2LVw%3D&nk2=D9ZNP7htc6w%3D&id2=UU8Lx7%2BmPirPbw%3D%3D&vt3=F8dARHfHI%2BnGtn3VuNA%3D&lg2=UtASsssmOIJ0bQ%3D%3D; lgc=lunce188; tracknick=lunce188; cookie2=10682dca3e46d779e26f299924785699; cookie1=AV0h8l61cg4iTp3AqqPZRlYP3nQGpHHQCAg%2FB5Sm3VI%3D; unb=2731635449; t=65336f3349d3648c68445898ef92bec2; skt=2c4d55251dbb75a9; _nk_=lunce188; _l_g_=Ug%3D%3D; cookie17=UU8Lx7%2BmPirPbw%3D%3D; login=true");
reqest.RequestUserAgent = "Mozilla/5.0 (SymbianOS/9.3; U; Series60/3.2 NokiaE75-1 /110.48.125 Profile/MIDP-2.1 Configuration/CLDC-1.1 ) AppleWebKit/413 (KHTML, like Gecko) Safari/413";
string catPage= reqest.HttpRequest(url);
Thread.Sleep(new Random().Next(6, 30)*1000);
//string catPage = HtmlAnalysis.Gethtmlcode("http://list.tmall.com/search_product.htm?cat=" + item.ClassId);
if (catPage.Contains("ResponseUri:http://www.tmall.com/") && item.UpdateTime.AddDays(15)<DateTime.Now)
{
new SiteClassBll().delClass(item);
return;
}
var crumbsList = RegGroupCollection(catPage,
"<li data-tag=\"cat\">(?<x>.*?)</li>");
if (crumbsList == null)
return;
SiteClassInfo catinfo = new SiteClassInfo();
string paraInfo = "";
string paraUrl = "";
string paraCatId = "";
string paraName = "";
for (int i = 0; i < crumbsList.Count; i++)
{
catinfo.ParentName = paraName;
catinfo.ParentClass = paraCatId;
catinfo.ParentUrl = paraCatId == "" ? "" : "http://list.tmall.com/search_product.htm?cat=" + paraCatId;
//添加父类
if (paraCatId != "" && paraCatId != "" &&
!HasBindClasslist.Exists(c => c.ClassId == paraCatId))
{
string tempparaInfo = "";
string tempparaUrl = "";
string tempparaCatId = "";
string tempparaName = "";
if (i > 1)
{
Match pnode = crumbsList[i - 2];
tempparaInfo = pnode.Groups["x"].Value;
// tempparaUrl = RegGroupsX<string>(tempparaInfo, "href=\"(?<x>.*?)\"");
tempparaCatId = RegGroupsX<string>(paraUrl, "cat=(?<x>\\d+)");
tempparaUrl = "http://list.tmall.com/search_product.htm?cat=" + tempparaCatId;
tempparaName = RegGroupsX<string>(tempparaInfo, "title=\"(?<x>.*?)\"");
}
SiteClassInfo catPareInfo = new SiteClassInfo
{
ClassName = paraName,
ClassId = paraCatId,
SiteId = Baseinfo.SiteId,
CreateDate = DateTime.Now,
UpdateTime = DateTime.Now,
IsHide = false,
ParentUrl = tempparaUrl,
ParentName = tempparaName,
ClassCrumble = tempparaCatId + ",",
ParentClass = tempparaCatId,
Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + paraCatId,
IsDel=false,
HasChild = true,
IsBind = false
};
catPareInfo.ClassCrumble = catPareInfo.ClassCrumble.TrimEnd(',');
HasBindClasslist.Add(catPareInfo);
shopClasslist.Add(catPareInfo);
//父类的同级分类
GetAllBrotherCats(catPareInfo);
//子类
GetChildCats(catPareInfo, catPage);
}
Match node = crumbsList[i];
paraInfo = node.Groups["x"].Value;
paraUrl = RegGroupsX<string>(paraInfo, "href=\"(?<x>.*?)\"");
paraCatId = RegGroupsX<string>(paraUrl, "cat=(?<x>\\d+)");
paraName = RegGroupsX<string>(paraInfo, "title=\"(?<x>.*?)\"");
catinfo.ClassName = paraName;
catinfo.ClassId = paraCatId;
catinfo.SiteId = Baseinfo.SiteId;
catinfo.CreateDate = DateTime.Now;
catinfo.UpdateTime = DateTime.Now;
catinfo.Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + paraCatId;
catinfo.TotalProduct = RegGroupsX<int>(catPage, "共<span> (?<x>\\d+)</span>件相关商品");
catinfo.IsHide = false;
catinfo.IsBind = false;
catinfo.IsDel = false;
catinfo.ClassCrumble += paraCatId + ",";
GetAllBrotherCats(catinfo);
}
if (string.IsNullOrEmpty(catinfo.ClassId))
return;
if(regIsMatch(catPage, "<div class=\"cateAttrs\" data-spm=\".*?\">(?<x>.*?)<div class=\"propAttrs\""))
{
GetChildCats(catinfo, catPage);
catinfo.HasChild = true;
}
else
catinfo.HasChild = false;
catinfo.ClassCrumble = catinfo.ClassCrumble.TrimEnd(',');
var oldCatInfo = HasBindClasslist.Find(c => c.ClassId == catinfo.ClassId);
if (oldCatInfo==null)
{
catinfo.HasChild = true;
HasBindClasslist.Add(catinfo);
shopClasslist.Add(catinfo);
}
else
{
oldCatInfo.Urlinfo = catinfo.Urlinfo;
oldCatInfo.ClassId = catinfo.ClassId;
oldCatInfo.ClassName = catinfo.ClassName;
oldCatInfo.TotalProduct = catinfo.TotalProduct;
oldCatInfo.ParentUrl = catinfo.ParentUrl;
oldCatInfo.ParentClass = catinfo.ParentClass;
oldCatInfo.ParentUrl = catinfo.ParentUrl;
oldCatInfo.UpdateTime = DateTime.Now;
new SiteClassBll().UpdateSiteCat(oldCatInfo);
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void SaveCat(string catId)
{
if (!ValidCatId(catId)) return;
Thread.Sleep(new Random().Next(10, 60) * 1000);
string catPage = HtmlAnalysis.Gethtmlcode("http://list.tmall.com/search_product.htm?cat=" + catId);
var crumbsList = RegGroupCollection(catPage,
"<li data-tag=\"cat\">(?<x>.*?)</li>");
if (crumbsList == null)
return;
SiteClassInfo catinfo = new SiteClassInfo();
string paraInfo = "";
string paraUrl = "";
string paraCatId = "";
string paraName = "";
for (int i = 0; i < crumbsList.Count; i++)
{
catinfo.ParentName = paraName;
catinfo.ParentClass = paraCatId;
catinfo.ParentUrl = paraCatId=="" ?"": "http://list.tmall.com/search_product.htm?cat=" + paraCatId;
//添加父类
if (ValidCatId(paraCatId) &&!HasBindClasslist.Exists(c => c.ClassId == paraCatId))
{
string tempparaInfo = "";
string tempparaUrl = "";
string tempparaCatId = "";
string tempparaName = "";
if (i > 1)
{
Match pnode = crumbsList[i - 2];
tempparaInfo = pnode.Groups["x"].Value;
tempparaUrl = RegGroupsX<string>(tempparaInfo, "href=\"(?<x>.*?)\"");
tempparaCatId = RegGroupsX<string>(paraUrl, "cat=(?<x>\\d+)");
tempparaUrl = "http://list.tmall.com/search_product.htm?cat=" + tempparaCatId;
tempparaName = RegGroupsX<string>(tempparaInfo, "title=\"(?<x>.*?)\"");
}
SiteClassInfo catPareInfo = new SiteClassInfo
{
ClassName = paraName,
ClassId = paraCatId,
SiteId = Baseinfo.SiteId,
CreateDate = DateTime.Now,
UpdateTime = DateTime.Now,
IsHide = false,
ParentUrl = tempparaUrl,
ParentName = tempparaName,
ClassCrumble = tempparaCatId + ",",
IsDel=false,
ParentClass = tempparaCatId,
Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + paraCatId,
HasChild = true,
IsBind = false
};
catPareInfo.ClassCrumble = catPareInfo.ClassCrumble.TrimEnd(',');
HasBindClasslist.Add(catPareInfo);
shopClasslist.Add(catPareInfo);
//父类的同级分类
GetAllBrotherCats(catPareInfo);
//子类
GetChildCats(catPareInfo, "");
}
Match node = crumbsList[i];
paraInfo = node.Groups["x"].Value;
paraUrl = RegGroupsX<string>(paraInfo, "href=\"(?<x>.*?)\"");
paraCatId = RegGroupsX<string>(paraUrl, "cat=(?<x>\\d+)");
paraName = RegGroupsX<string>(paraInfo, "title=\"(?<x>.*?)\"");
catinfo.ClassName = paraName;
catinfo.ClassId = paraCatId;
catinfo.SiteId = Baseinfo.SiteId;
catinfo.CreateDate = DateTime.Now;
catinfo.UpdateTime = DateTime.Now;
catinfo.Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + paraCatId;
catinfo.TotalProduct = RegGroupsX<int>(catPage, "共<span> (?<x>\\d+)</span>件相关商品");
catinfo.IsHide = false;
catinfo.IsBind = false;
catinfo.ClassCrumble += paraCatId + ",";
catinfo.IsDel = false;
GetAllBrotherCats(catinfo);
}
GetChildCats(catinfo, catPage);
catinfo.ClassCrumble = catinfo.ClassCrumble.TrimEnd(',');
if (catinfo.ClassId != "" && !HasBindClasslist.Exists(c => c.ClassId == catinfo.ClassId))
{
catinfo.HasChild = true;
HasBindClasslist.Add(catinfo);
shopClasslist.Add(catinfo);
}
else if (catinfo.ClassName != "" &&
!HasBindClasslist.Exists(c => c.ClassName == catinfo.ClassName))
{
catinfo.HasChild = true;
HasBindClasslist.Add(catinfo);
shopClasslist.Add(catinfo);
}
if (shopClasslist.Count > 100)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void SaveBand(SiteClassInfo siteClassInfo)
{
string brandUrl = string.Format("http://list.tmall.com/ajax/allBrandShowForGaiBan.htm?cat={0}", siteClassInfo.ClassId);
string page = HtmlAnalysis.Gethtmlcode(brandUrl);
if (string.IsNullOrEmpty(page))
return;
page = page.Replace("\r", "").Replace("\n", "").Trim();
if (page == "")
return;
var list = RegGroupCollection(page, "\\{(?<x>.*?)\\}");
if (list == null)
return;
List<SiteClassBand> catBands = new List<SiteClassBand>();
List<SiteBandInfo> siteBand = new List<SiteBandInfo>();
for(int i=0;i<list.Count;i++)
{
string templist = list[i].ToString();
string url = RegGroupsX<string>(templist, "\"href\":\"(?<x>.*?)\"");
if (string.IsNullOrEmpty(url))
continue;
url = url.Replace("&", "&");
if (!url.Contains("http://"))
{
url = "http://list.tmall.com/search_product.htm" + url;
}
string bandId = RegGroupsX<string>(templist, "brand=(?<x>\\d+)");
string disName = RegGroupsX<string>(templist, "\"title\":\"(?<x>.*?)\"");
string img = RegGroupsX<string>(templist, "\"img\":\"(?<x>.*?)\"");
string cnName = "";
string enName = "";
string key = Baseinfo.SiteId + "_" + siteClassInfo.ClassId + "_" + bandId;
string key1 = Baseinfo.SiteId + "_" + bandId;
string[] names = disName.Split('/');
foreach (string obj in names)
{
if (regIsMatch(obj, @"[\u4e00-\u9fa5]"))
cnName = obj;
else
enName = obj;
}
SiteClassBand tempBand = new SiteClassBand
{
ImgUrl=img,
UniqueKey = key,
DisplayName = disName,
CnName=cnName,
EnName = enName,
CommentCount=0,
ProductCount=0,
Urlinfo=url,
SiteBandId = bandId,
SiteClassId= siteClassInfo.ClassId,
SiteId=Baseinfo.SiteId,
IsHid=false,
UpdateDate=DateTime.Now,
CreateDate =DateTime.Now
};
if (!HasBindBandlist.Exists(p => p.UniqueKey == key))
{
HasBindBandlist.Add(tempBand);
catBands.Add(tempBand);
}
if (!HasSiteBandlist.Exists(p => p.UniqueKey == key1))
{
SiteBandInfo tempsband = new SiteBandInfo
{
CatArea="",
EnName=tempBand.EnName,
ImgUrl=tempBand.ImgUrl,
Introduction="",
IsHid=false,
Remark="",
TotalComments=0,
TotalProduts=0,
UniqueKey = key1,
SiteId=tempBand.SiteId,
SiteBandId=tempBand.SiteBandId,
DisplayName=tempBand.DisplayName,
CnName=tempBand.CnName,
CreateDate=DateTime.Now,
UpdateDate=DateTime.Now
};
HasSiteBandlist.Add(tempsband);
siteBand.Add(tempsband);
}
}
try
{
new SiteClassBandDb().Save(catBands);
new SiteBandDb().Save(siteBand);
}
catch (Exception ex)
{
LogServer.WriteLog(ex);
}
}
/// <summary>
/// 获取子分类
/// </summary>
/// <param name="catinfo"></param>
private void GetChildCats(SiteClassInfo catinfo,string pageinfo)
{
if (pageinfo == "")
{
Thread.Sleep(new Random().Next(10, 30)*1000);
pageinfo = HtmlAnalysis.Gethtmlcode(catinfo.Urlinfo);
}
string catInfo = RegGroupsX<string>(pageinfo, "<div class=\"cateAttrs\" data-spm=\".*?\">(?<x>.*?)<div class=\"propAttrs\"");
if (catInfo == null)
return;
var catList = RegGroupCollection(catInfo, "<a title=\"(?<x>.*?)\">\r\n <b>(?<y>.*?)</b><span>\\((?<z>\\d+)\\)</span>\r\n </a>");
if (catList == null)
return;
foreach (Match item in catList)
{
string catUrl = item.Groups["x"].Value;
string catId = RegGroupsX<string>(catUrl, "cat=(?<x>\\d+)");
if (string.IsNullOrEmpty(catId))
continue;
int total = 0;
int.TryParse(item.Groups["z"].Value, out total);
if (!HasBindClasslist.Exists(c => c.ClassId == catId))
{
SiteClassInfo cat = new SiteClassInfo {
ClassId =catId,
ClassCrumble = catinfo.ClassCrumble+","+catinfo.ClassId,
ParentClass = catinfo.ClassId,
ParentName = catinfo.ClassName,
ClassName=item.Groups["y"].Value,
IsHide=false,
ParentUrl=catinfo.Urlinfo,
UpdateTime=DateTime.Now,
IsBind=false,
IsDel=false,
SiteId=Baseinfo.SiteId,
Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + catId,
TotalProduct=total,
CreateDate=DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
}
}
/// <summary>
/// 获取同父级的分类
/// </summary>
/// <param name="catinfo"></param>
private void GetAllBrotherCats(SiteClassInfo catinfo)
{
// Thread.Sleep(new Random().Next(5, 30) * 1000);
string catPage = HtmlAnalysis.Gethtmlcode("http://list.tmall.com/ajax/getAllBrotherCats.htm?cat=" + catinfo.ClassId);
var catList = RegGroupCollection(catPage, "\"href\":\"(?<x>.*?)\",\r\n\"title\":\"(?<y>.*?)\",\r\n\"atp\"");
if (catList == null)
return;
foreach (Match item in catList)
{
string catUrl = item.Groups["x"].Value;
string catId = RegGroupsX<string>(catUrl, "cat=(?<x>\\d+)");
if (string.IsNullOrEmpty(catId))
continue;
if (!HasBindClasslist.Exists(c => c.ClassId == catId))
{
SiteClassInfo cat = new SiteClassInfo {
ClassId =catId,
ClassCrumble = catinfo.ClassCrumble,
ParentClass = catinfo.ParentClass,
ParentName = catinfo.ParentName,
ClassName=item.Groups["y"].Value,
IsHide=false,
ParentUrl=catinfo.ParentUrl,
UpdateTime=DateTime.Now,
IsBind=false,
IsDel=false,
SiteId=Baseinfo.SiteId,
Urlinfo = "http://list.tmall.com/search_product.htm?cat=" + catId,
CreateDate=DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
}
}
private void UpdateNode(SiteClassInfo siteClassInfo)
{
//if (!siteClassInfo.Urlinfo.Contains("b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k"))
// siteClassInfo.Urlinfo += "b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/";
siteClassInfo.Urlinfo = $"http://list.yhd.com/{siteClassInfo.ClassId}/b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k";
string pageInfo = Gethtmlcode(siteClassInfo.Urlinfo);
string categoryname = RegGroupsX<string>(pageInfo, "var categoryName = '(?<x>.*?)'");
string extid = RegGroupsX<string>(pageInfo, "var expectCategoryId = \"(?<x>\\d+)\"");
string tempcurcatid = "c" + extid + "-" + categoryname;
if (!regIsMatch(tempcurcatid, "^(?<x>c\\d+-\\d+(-\\d+)?)$"))
{
var aa = "ddd";
return;
}
if (categoryname == "0-0")
tempcurcatid = "c" + extid;
if (extid=="0")
return;
//string classInfo = RegGroupsX<string>(pageInfo, "<div id=\"searchColSub\"(?<x>.*?)<div id=\"bodyRight\"|<div class=\"crumbClip\">(?<x>.*?)<li class=\"crumb_search search_empty\">");
//string classInfo = RegGroupsX<string>(pageInfo, "<div class=\"classWrap\">(?<x>.*?)<div class=\"brandWrap\">");
string classInfo = RegGroupsX<string>(pageInfo, "<ul class=\"listCon clearfix\">(?<x>.*?)<a class=\"c_btn c_next iconSearch\"");
string classinfo2 = RegGroupsX<string>(pageInfo, "<ul class=\"guide_con clearfix\">(?<x>.*?)</ul>");
if (!string.IsNullOrEmpty(classinfo2))
{
classInfo += classinfo2;
}
string tempcatid = RegGroupsX<string>(siteClassInfo.Urlinfo, "c\\d+-0-(?<x>\\d+)/");
if (!string.IsNullOrEmpty(tempcatid))
{
string classurl =
string.Format(
"http://list.yhd.com/lazyLoadBrotherCategory/nc{0}-a-f0d-mid0-k/?urlFilterSuffix=/b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/",
tempcatid);
string catpage = Gethtmlcode(classurl);
if (!string.IsNullOrEmpty(catpage))
classInfo += catpage.Replace("\\\"","\"");
}
var catList = RegGroupCollection(classInfo, "href=\"(?<x>.*?)\".*?title=\"(?<y>.*?)\"");
if (catList == null)
{
if (pageInfo.Contains("没有找到符合条件的商品，建议您更改下搜索条件")||pageInfo.Contains("很抱歉！没有找到与<span class=\"color_red\">\"\"</span>相关的商品，要不你换个关键词我帮你再找找吧"))
new SiteClassBll().delClass(siteClassInfo);
return;
}
for (int i = 0; i < catList.Count; i++)
{
string url = catList[i].Groups["x"].Value;
string catname = catList[i].Groups["y"].Value;
if (string.IsNullOrEmpty(url))
continue;
string catId = RegGroupsX<string>(url, "http://list.yhd.com/(?<x>.*?)/");
if (!HasBindClasslist.Exists(p => p.ClassId == catId) && !HasBindClasslist.Exists(p => p.ClassName == catname))
{
GetYhdClassInfo(url);
}
}
//string crumb = RegGroupsX<string>(pageInfo,
// "<div class=\"mod_search_crumb clearfix\"(?<x>.*?)<div id=\"searchColSub\"");
//string crumb = RegGroupsX<string>(pageInfo,"<div class=\"crumbClip\">(?<x>.*?)<li class=\"crumb_search search_empty\">|<div class=\"mod_search_crumb clearfix\"(?<x>.*?)<div id=\"searchColSub\"");
//string crumb = RegGroupsX<string>(pageInfo, "<div class=\"crumbClip\">(?<x>.*?)<li class=\"crumb_search search_empty\">|<div class=\"mod_search_crumb clearfix\"(?<x>.*?)<div id=\"searchColSub\"");
//if (crumb == null)
// return;
//var list = RegGroupCollection(crumb, "<div class=\"crumb_list\">(?<x>.*?)</div>");
var list = RegGroupCollection(classInfo, "<li class=\"crumb_list\">(?<x>.*?)</li>");
if (list == null)
return;
string current = RegGroupsX<string>(pageInfo, "<title>(?<x>.*?)品种齐全|<div class=\"guide_title\"><span title=\"(?<x>.*?)\">");
string pcatUrl = "";
string pcatName = "";
string pcatId = "";
string classCrumble = "";
int total = RegGroupsX<int>(pageInfo, "共(?<x>\\d+?)条");
SiteClassInfo catInfo = new SiteClassInfo();
for (int i = 0; i < list.Count; i++)
{
string div = list[i].Groups["x"].Value;
catInfo.Urlinfo = RegGroupsX<string>(div, "href=\"(?<x>.*?)\"");
catInfo.ClassId = RegGroupsX<string>(catInfo.Urlinfo, "http://list.yhd.com/(?<x>.*?)/");
if(string.IsNullOrEmpty(catInfo.Urlinfo))
continue;
if (!catInfo.Urlinfo.Contains("b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/"))
catInfo.Urlinfo += "b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k/";
catInfo.ClassName = RegGroupsX<string>(div, "\" >(?<x>.*?)</a>|title=\"(?<x>.*?)\"");
if (catInfo.ClassName== "全部结果")
continue;
if (catInfo.Urlinfo == null || catInfo.ClassId == null || catInfo.ClassName == null)
{
LogServer.WriteLog(Baseinfo.SiteName + "分类抓取错误1", "AddClassError");
continue;
}
if (!HasBindClasslist.Exists(p => p.ClassId == catInfo.ClassId) && !HasBindClasslist.Exists(p => p.ClassName == catInfo.ClassName))
{
GetYhdClassInfo(catInfo.Urlinfo);
}
catInfo.ParentClass = pcatId;
catInfo.ParentName = pcatName;
catInfo.ParentUrl = pcatUrl;
if (i != 0 && pcatId != "")
classCrumble += pcatId + ",";
catInfo.ClassCrumble = classCrumble.TrimEnd(',');
if (list.Count - 1 == i)
{
catInfo.HasChild = false;
catInfo.TotalProduct = total;
}
if (catInfo.ClassName == current)
{
break;
}
pcatUrl = catInfo.Urlinfo;
pcatName = catInfo.ClassName;
pcatId = catInfo.ClassId;
}
if (!string.IsNullOrEmpty(pcatId))
{
if (HasBindClasslist.Any(p => p.ClassId == pcatId))
{
GetYhdClassInfo(pcatUrl);
}
}
string childCat = RegGroupsX<string>(pageInfo,"<div class=\"classWrap\">(?<x>.*?)</ul>");
if (childCat != null)
{
var childList = RegGroupCollection(childCat, "href=\"(?<x>.*?)\".*?<span title=\"(?<y>.*?)\">");
foreach (Match item in childList)
{
string tempUrl = item.Groups["x"].Value;
string tempName = item.Groups["y"].Value;
string tempid = RegGroupsX<string>(tempUrl, "http://list.yhd.com/(?<x>.*?)/");
if (!HasBindClasslist.Exists(c => c.ClassId == tempid) && !HasBindClasslist.Exists(p => p.ClassName == tempName))
{
GetYhdClassInfo(tempUrl);
}
}
siteClassInfo.HasChild = true;
}
else
{ siteClassInfo.HasChild = false; }
if (siteClassInfo.ClassId != catInfo.ClassId)
{
LogServer.WriteLog("分类id 更改 old id:" + siteClassInfo.Id + siteClassInfo.Id + "oldclass:" + siteClassInfo.ClassId + "newclass:" + catInfo.ClassId);
if (HasBindClasslist.Exists(c => c.ClassId == catInfo.ClassId))
{
new SiteClassBll().delClass(siteClassInfo);
return;
}
siteClassInfo.Urlinfo = catInfo.Urlinfo;
siteClassInfo.ClassId = catInfo.ClassId;
}
if (tempcurcatid != siteClassInfo.ClassId && regIsMatch(tempcurcatid, "^(?<x>c\\d+-\\d+(-\\d+)?)$"))
{
siteClassInfo.ClassId = tempcurcatid;
siteClassInfo.Urlinfo =$"http://list.yhd.com/{tempcurcatid}/b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k";
}
siteClassInfo.ClassName = catInfo.ClassName;
siteClassInfo.TotalProduct = catInfo.TotalProduct;
siteClassInfo.ParentUrl = catInfo.ParentUrl;
siteClassInfo.ParentClass = catInfo.ParentClass;
if (regIsMatch(siteClassInfo.ParentClass, "c\\d+-0"))
{
string parentpage = Gethtmlcode(catInfo.ParentUrl);
string pcategoryname = RegGroupsX<string>(parentpage, "var categoryName = '(?<x>.*?)'");
string pextid = RegGroupsX<string>(parentpage, "var expectCategoryId = \"(?<x>\\d+)\"");
string ptempcurcatid = "c" + pextid + "-" + pcategoryname;
if (regIsMatch(ptempcurcatid, "^(?<x>c\\d+-\\d+(-\\d+)?)$"))
{
siteClassInfo.ParentClass = ptempcurcatid;
siteClassInfo.ParentUrl =
$"http://list.yhd.com/{ptempcurcatid}/b/a-s1-v4-p1-price-d0-f0d-m1-rt0-pid-mid0-k";
}
}
siteClassInfo.HasChild = HasBindClasslist.Exists(p => p.ParentClass == siteClassInfo.ClassId);
siteClassInfo.ParentName = catInfo.ParentName;
siteClassInfo.UpdateTime = DateTime.Now;
new SiteClassBll().UpdateSiteCat(siteClassInfo);
}
public void SaveAllSiteClass()
{
HasBindClasslist = new SiteClassInfoDB().getAllSiteCatInfo(Baseinfo.SiteId);
string page = HtmlAnalysis.Gethtmlcode(domain);
string content = RegGroupsX<string>(page, "<h2>所有商品分类</h2><div class=\"submenu\">(?<x>.*?)<div class=\"nav\">");
var catList = RegGroupCollection(content, "<a href=\"(?<x>.*?)\">(?<y>.*?)</a>");
for (int i = 0; i < catList.Count; i++)
{
string tempurl = catList[i].Groups["x"].Value;
if (string.IsNullOrEmpty(tempurl))
continue;
tempurl = domain + tempurl;
string tempName = catList[i].Groups["y"].Value;
string catid = RegGroupsX<string>(tempurl, "/list-(?<x>\\d+)");
if (!ValidCatId(catid))
continue;
if (!HasBindClasslist.Exists(p => p.ClassId == catid))
{
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo = tempurl,
ClassId = catid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = tempName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string page = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string cromb = RegGroupsX<string>(page, "<span class=\"pos-front\">(?<x>.*?)</div>");
if (cromb == null)
return;
var plist = RegGroupCollection(cromb, "<a href=\"(?<x>.*?)\" alt=\"\" title=\"\">(?<y>.*?)</a></span>");
if (plist == null)
return;
string parentUrl = "";
string parentName="";
string parentId = "";
foreach (Match item in plist)
{
if (item.ToString().Contains("首页"))
continue;
parentUrl = item.Groups["x"].Value;
if (string.IsNullOrEmpty(parentUrl))
continue;
parentUrl = string.Format("http://www.hangowa.com{0}", parentUrl);
parentName = item.Groups["y"].Value;
parentId = RegGroupsX<string>(parentUrl, "gallery-(?<x>\\d+?).html");
if (!HasBindClasslist.Exists(c => c.ClassId == parentId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = parentName,
ClassId = parentId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = parentUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
string chlidCat = RegGroupsX<string>(page, "<dt class=\"filter-entries-label\">分类：</dt>\n <dd class=\"filter-entries-values\">(?<x>.*?)</dd>");
if (chlidCat != null)
{
var blist = RegGroupCollection(chlidCat, "<a href=\"(?<x>.*?)\" class=\"handle action-cat-filter\">(?<y>.*?)</a>");
if (blist != null)
{
foreach (Match item in blist)
{
string burl = item.Groups["x"].Value;
if (string.IsNullOrEmpty(burl))
continue;
burl = "http://www.hangowa.com" + burl.TrimEnd('?');
string bName = item.Groups["y"].Value;
string bId = RegGroupsX<string>(burl, "gallery-(?<x>\\d+?).html");
if (!HasBindClasslist.Exists(c => c.ClassId == bId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass =siteClassInfo.ClassId,
ParentName =siteClassInfo.ClassName,
ClassName = bName,
ClassId = bId,
ParentUrl = siteClassInfo.Urlinfo,
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = burl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
//if (chlidCat != null && chlidCat.Contains(siteClassInfo.ClassId))
// siteClassInfo.HasChild = false;
//else
// siteClassInfo.HasChild = true;
siteClassInfo.ParentClass = parentId;
siteClassInfo.ParentName = parentName;
siteClassInfo.ParentUrl =parentUrl;
siteClassInfo.UpdateTime = DateTime.Now;
siteClassInfo.TotalProduct = RegGroupsX<int>(page, "共<b class=\"op-search-result\">(?<x>\\d+)</b>件商品");
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string pageinfo = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string crumb = RegGroupsX<string>(pageinfo, "<div class=\"curr_position\">(?<x>.*?)</div>");
if (crumb == null)
{
siteClassInfo.IsDel = true;
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
LogServer.WriteLog(Baseinfo.SiteName + "分类抓取错误\turl:" + siteClassInfo.Urlinfo, "AddClassError");
return;
}
var list = RegGroupCollection(crumb, "<a href=\"(?<x>.*?)\">(?<y>.*?)</a>");
if (list == null)
return;
string parentid = "";
string parentName = "";
string parentUrl = "";
for (int i = 0; i < list.Count; i++)
{
string tempName = list[i].Groups["y"].Value;
if (tempName.Contains("首页"))
continue;
string tempUrl = list[i].Groups["x"].Value;
if (!tempUrl.Contains(domain))
tempUrl = domain + tempUrl;
string tempid = RegGroupsX<string>(tempUrl, "(?<x>\\d+)");
if (string.IsNullOrEmpty(tempid))
{
continue;
}
if (tempid == siteClassInfo.ClassId)
break;
parentid = tempid;
parentName = tempName;
parentUrl = string.Format("http://www.jxdyf.com/category/{0}.html", tempid);
}
if (ValidCatId(parentid))
{
siteClassInfo.ParentName = parentName;
siteClassInfo.ParentUrl = parentUrl;
siteClassInfo.ParentClass = parentid;
}
siteClassInfo.TotalProduct = RegGroupsX<int>(pageinfo, "共有(?<x>\\d+)个商品");
siteClassInfo.UpdateTime = DateTime.Now;
siteClassInfo.HasChild = HasBindClasslist.Exists(c => c.ParentClass == siteClassInfo.ClassId);
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
string areaCat = RegGroupsX<string>(pageinfo, "<div class=\"menu outline_01\">(?<x>.*?)<div class=\"web_surfer outline_01\">");
if (areaCat == null)
return;
var catList = RegGroupCollection(areaCat, "<a href=\"/category/(?<x>\\d+).html\" >(?<y>.*?)");
for (int i = 0; i < catList.Count; i++)
{
string tempid = catList[i].Groups["x"].Value;
if (!ValidCatId(tempid) || HasBindClasslist.Exists(c => c.ClassId == tempid))
continue;
string tempName = catList[i].Groups["y"].Value;
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo = string.Format("http://www.jxdyf.com/category/{0}.html", tempid),
ClassId = tempid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = tempName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string page = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string cromb = RegGroupsX<string>(page, "您现在的位置： <a href=\"http://www.ehaier.com\">海尔商城</a>(?<x>.*?)</div>");
if (cromb == null)
return;
var plist = RegGroupCollection(cromb, "<a href=\"(?<x>.*?)\">(?<y>.*?)</a>");
if (plist == null)
return;
string parentUrl = "";
string parentName="";
string parentId = "";
foreach (Match item in plist)
{
parentUrl = item.Groups["x"].Value;
parentName = item.Groups["y"].Value;
parentId = RegGroupsX<string>(parentUrl, "http://www.ehaier.com/l/(?<x>\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+-\\d+).html");
if (!ValidCatId(parentId))
continue;
if (!HasBindClasslist.Exists(c => c.ClassId == parentId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = parentName,
ClassId = parentId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = parentUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
string brotherCat = RegGroupsX<string>(page, "<div class=\"list-subnav\">(?<x>.*?)<dl class=\"dl-subnav dl-subnav-now\">");
if (brotherCat != null)
{
var blist = RegGroupCollection(brotherCat, "href=\"(?<x>.*?)\">(?<y>.*?)</a>");
if (blist != null)
{
foreach (Match item in blist)
{
string burl = item.Groups["x"].Value;
string bName = item.Groups["y"].Value;
string bId = RegGroupsX<string>(burl, "http://www.ehaier.com/l/(?<x>\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+).html|http://www.ehaier.com/l/(?<x>\\d+-\\d+-\\d+).html");
if (!HasBindClasslist.Exists(c => c.ClassId == bId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = bName,
ClassId = bId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = burl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
if(HasBindClasslist.Exists(c=>c.ParentClass==siteClassInfo.ClassId))
siteClassInfo.HasChild = true;
else
siteClassInfo.HasChild = false;
siteClassInfo.ParentClass = parentId;
siteClassInfo.ParentName = parentName;
siteClassInfo.ParentUrl = parentUrl;
siteClassInfo.UpdateTime = DateTime.Now;
siteClassInfo.TotalProduct = RegGroupsX<int>(page, "共<strong class=\"haierred\">(?<x>\\d+)</strong> 件");
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
public void SaveAllSiteClass()
{
string url = "http://www.jxdyf.com/category";
List<SiteClassInfo> shopClasslist = new List<SiteClassInfo>();
HasBindClasslist = new SiteClassInfoDB().getAllSiteCatInfo(Baseinfo.SiteId);
string popHtml = HtmlAnalysis.Gethtmlcode(url);
string content = RegGroupsX<string>(popHtml, "<div class=\"fl\">(?<x>.*?)<div id=\"footer\"");
var catlist = RegGroupCollection(content, "<a href='(?<y>.*?)'( class='.*?')?>(?<x>.*?)</a>");
for (int i = 0; i < catlist.Count; i++)
{
string tempurl = catlist[i].Groups["y"].Value;
string catid = RegGroupsX<string>(tempurl, "(?<x>\\d+)");
if (!ValidCatId(catid) || HasBindClasslist.Exists(c => c.ClassId == catid))
{
continue;
}
string catName = catlist[i].Groups["x"].Value;
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo =domain+ tempurl,
ClassId = catid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = catName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
;
}
public bool BingCatInfo(SiteClassInfo cat)
{
using (var db = _dbFactory.OpenDbConnection())
{
int res = db.UpdateOnly(cat, u => new {u.IsBind, u.BindClassId,u.BindClassName}, u => u.Id == cat.Id);
return res > 0;
}
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string page = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string cromb = RegGroupsX<string>(page, "<div class=\"breadcrumb-area fcn\">(?<x>.*?)</div>");
if (cromb == null)
return;
var plist = RegGroupCollection(cromb, "<a href=\"(?<x>.*?)\" title=\"(?<y>.*?)\">");
if (plist == null)
return;
string parentUrl = "";
string parentName="";
string parentId = "";
foreach (Match item in plist)
{
if (item.ToString().Contains("首页"))
continue;
parentUrl = item.Groups["x"].Value;
parentName = item.Groups["y"].Value;
parentId = RegGroupsX<string>(parentUrl, "list-(?<x>\\d+)");
if (!ValidCatId(parentId))
{
parentId = "";
continue;
}
if (!HasBindClasslist.Exists(c => c.ClassId == parentId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = parentName,
ClassId = parentId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = string.Format("http://www.vmall.com/{0}", parentUrl),
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
string brotherCat = RegGroupsX<string>(page, "<div class=\"p-title\">分类：</div>(?<x>.*?)<div class=\"pro-cate-sort clearfix\">");
if (brotherCat != null)
{
var blist = RegGroupCollection(brotherCat, " <li ><a href=\"(?<x>.*?)\">(?<y>.*?)</a></li>");
if (blist != null)
{
foreach (Match item in blist)
{
string burl = item.Groups["x"].Value;
string bName = item.Groups["y"].Value;
string bId = RegGroupsX<string>(burl, "list-(?<x>\\d+)$");
if (!ValidCatId(bId))
{
continue;
}
if (!HasBindClasslist.Exists(c => c.ClassId == bId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = bName,
ClassId = bId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = string.Format("http://www.vmall.com/{0}", burl),
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
if (brotherCat != null)
siteClassInfo.HasChild = false;
else
siteClassInfo.HasChild = true;
siteClassInfo.ParentClass = parentId;
siteClassInfo.ParentName = parentName;
if (!string.IsNullOrEmpty(parentUrl))
siteClassInfo.ParentUrl = string.Format("http://www.vmall.com/{0}", parentUrl);
siteClassInfo.UpdateTime = DateTime.Now;
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
public void UpdateSiteClass(SiteClassInfo catinfo)
{
_dbFactory = new OrmLiteConnectionFactory(MmbpriceDbConnectionString, SqlServerDialect.Provider);
if (catinfo == null) throw new ArgumentNullException("catinfo");
using (var db = _dbFactory.OpenDbConnection())
{
try
{
db.Update(catinfo);
}
catch (Exception ex)
{
LogServer.WriteLog(ex, "DBError");
// throw;
}
}
}
private void GetCatInfo(string directoryHtml)
{
string catArea = RegGroupsX<string>(directoryHtml,
"<ol class=\"category-list\">(?<x>.*?)<a href=\"http://app.vmall.com\" target=\"_blank\"><span>应用市场");
if (catArea == null)
return;
catArea = catArea.Replace("\r", "").Replace("\n", "").Replace("\t", "");
var list = RegGroupCollection(catArea, "<a href=\"(?<x>.*?)\" (target=\"_blank\")?><span>(?<y>.*?)</span>");
foreach (Match item in list)
{
string tempUrl = item.Groups["x"].Value;
string tempName = item.Groups["y"].Value;
string tempid = RegGroupsX<string>(tempUrl, "list-(?<x>\\d+)$");
if (!HasBindClasslist.Exists(c => c.ClassId == tempid))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = false,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = string.Format("http://www.vmall.com{0}", tempUrl),
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
public void AddSiteClass(SiteClassInfo siteclass)
{
if (siteclass == null) throw new ArgumentNullException("siteclass");
using (var db = _dbFactory.OpenDbConnection())
{
try
{
if (db.Exists<SiteClassInfo>(p => p.SiteId == siteclass.SiteId && p.ClassId == siteclass.ClassId))
return;
db.Insert(siteclass);
}
catch (Exception ex)
{
LogServer.WriteLog(ex, "DBError");
throw;
}
}
}
private void AddNode(string url)
{
string classid = RegGroupsX<string>(url, "^http://searchex.yixun.com/(?<x>.*?)\\-");
if (string.IsNullOrEmpty(classid))
return;
if (classid.Contains("t"))
{
var tempids = classid.Split('t');
string catid = tempids[tempids.Length - 1];
if (HasBindClasslist.Exists(p => p.ClassId == catid))
return;
}
else
{
if (HasBindClasslist.Exists(p => p.ClassId == classid))
return;
}
string pageinfo = HtmlAnalysis.Gethtmlcode(url);
if (!pageinfo.Contains("<div class=\"goods\""))
return;
string cromp = RegGroupsX<string>(pageinfo, "<div id=\"crumbBox\" class=\"crumb \">(?<x>.*?)<div class=\"crumb_search \">");
if (string.IsNullOrEmpty(cromp)) return;
var caplist = RegGroupCollection(cromp, "<a class=\"crumb_lk\" href=\"(?<x>.*?)\" rg=\"[0-9_]+\" ytag=\"\\d+\">(?<y>.*?)</a>");
if (caplist == null || caplist.Count < 1)
return;
string proName = caplist[caplist.Count - 1].Groups["y"].Value;
string parentUrl = "";
string parentName = "";
string parentid = "";
string classCrumble = "";
List<string> lessCat = new List<string>();
for (int i = 0; i < caplist.Count - 1; i++)
{
if (i == 0)
continue;
parentUrl = caplist[i].Groups["x"].Value;
parentName = caplist[i].Groups["y"].Value;
parentid = RegGroupsX<string>(parentUrl, "path=(?<x>[A-Za-z0-9]+)|^http://searchex.yixun.com/(?<x>.*?)\\-");
if (!string.IsNullOrEmpty(parentid))
{
if (parentid.Contains(','))
parentid = parentid.Substring(parentid.LastIndexOf(',') + 1);
classCrumble += parentid + ",";
if (!HasBindClasslist.Exists(p => p.ClassId == parentid))
{
lessCat.Add(string.Format("http://searchex.yixun.com/{0}-1-/", parentid));
}
}
}
if (classid.Contains("t"))
{
var tempids = classid.Split('t');
parentid = tempids[0];
classid = tempids[1];
if (classCrumble == "")
classCrumble = parentid;
}
classCrumble = classCrumble.TrimEnd(',');
if (HasBindClasslist.Exists(p => p.ClassId == classid))
return;
int total = RegGroupsX<int>(pageinfo, "共<b>(?<x>\\d+)</b>件商品");
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = parentUrl,
ParentClass = parentid,
ParentName = parentName,
TotalProduct = total,
Urlinfo = url,
ClassId = classid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = false,
IsBind = false,
IsHide = false,
ClassName = proName,
SiteId = Baseinfo.SiteId,
ClassCrumble = classCrumble,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
string catList = RegGroupsX<string>(pageinfo, "<div class=\"cate_bd\">(?<x>.*?)<div id=\"viewedGoods\"");
var temCats = RegGroupCollection(catList,"href=\"(?<x>.*?)\"");
if (temCats == null)
return;
for (int i = 0; i < temCats.Count; i++)
{
string tempCatUrl = temCats[i].Groups["x"].Value;
string tempcatid = RegGroupsX<string>(tempCatUrl, "path=(?<x>[A-Za-z0-9]+)|^http://searchex.yixun.com/(?<x>.*?)\\-");
if (string.IsNullOrEmpty(tempcatid))
continue;
if (!HasBindClasslist.Exists(p => p.ClassId == parentid))
{
lessCat.Add(string.Format("http://searchex.yixun.com/{0}-1-/", tempcatid));
}
}
for (int i = 0; i < lessCat.Count; i++)
{
AddNode(lessCat[i]);
}
}
private void GetCatInfo(string directoryHtml)
{
string catArea = RegGroupsX<string>(directoryHtml,
"全部商品分类</h2>(?<x>.*?)<div class=\"nav-right\">");
if (catArea == null)
return;
//catArea = catArea.Replace("\t", "").Replace("\r", "").Replace("\n", "");
var list = RegGroupCollection(catArea, "href=('|\")(?<x>.*?)('|\")\\s*>(?<y>.*?)</a>");
foreach (Match item in list)
{
string tempUrl = item.Groups["x"].Value;
string tempName = item.Groups["y"].Value;
if (!string.IsNullOrEmpty(tempName))
{
tempName = tempName.Trim();
}
string tempid = RegGroupsX<string>(tempUrl, "category/(?<x>\\d+)-");
if (ValidCatId(tempid) &&!HasBindClasslist.Exists(c => c.ClassId == tempid))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = false,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = tempUrl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string pageinfo = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
string crumble = RegGroupsX<string>(pageinfo,
"<div class=\"crumb_wrap\">(?<x>.*?)<div class=\"crumb_search \">");
if (crumble == null)
{
LogServer.WriteLog(Baseinfo.SiteName + "分类抓取错误1\turl:" + siteClassInfo.Urlinfo, "AddClassError");
return;
}
var crumblelist = RegGroupCollection(crumble,
"<a class=\"crumb_lk\" href=\"(?<x>.*?)\" rg=\"\\d+_?\\d+\" ytag=\"\\d+\">(?<y>.*?)</a>");
siteClassInfo.TotalProduct = RegGroupsX<int>(pageinfo,
"<div class=\"sort_page_txt\">共<b>(?<x>\\d+)</b>件商品</div>");
if (siteClassInfo.TotalProduct == 0)
{
siteClassInfo.IsDel = true;
new SiteClassBll().delClass(siteClassInfo);
}
if (crumblelist == null || crumblelist.Count == 0)
{
LogServer.WriteLog(Baseinfo.SiteName + "分类抓取错误2\turl:" + siteClassInfo.Urlinfo, "AddClassError");
return;
}
string pcatUrl = "";
string pcatName = "";
string pcatId = "";
string classCrumble = "";
foreach (Match item in crumblelist)
{
if (item.ToString().Contains("首页"))
continue;
if (item.ToString().Contains(siteClassInfo.ClassName))
{
string itemurl = item.Groups["x"].Value;
string tempid = RegGroupsX<string>(itemurl, "http://searchex.yixun.com/(?<x>.*?)-1-/");
if (!ValidCatId(tempid))
continue;
if (tempid.Contains('t'))
{
var tempids = tempid.Split('t');
string catid = tempids[tempids.Length - 1];
siteClassInfo.ClassId = catid;
if (tempids.Length > 1)
siteClassInfo.ParentClass = tempids[tempids.Length - 2];
}
else
{
siteClassInfo.ClassId = tempid;
}
siteClassInfo.ClassName = item.Groups["y"].Value;
siteClassInfo.Urlinfo = itemurl;
}
else
{
pcatUrl = item.Groups["x"].Value;
pcatName = item.Groups["y"].Value;
pcatId = RegGroupsX<string>(pcatUrl, "http://searchex.yixun.com/(?<x>.*?)-1-/");
if (!string.IsNullOrEmpty(pcatId))
classCrumble += pcatId + ",";
}
}
if ( siteClassInfo.ClassId.Contains('t'))
{
siteClassInfo.ParentClass = siteClassInfo.ClassId.Substring(0, siteClassInfo.ClassId.IndexOf('t'));
}
string catArea = RegGroupsX<string>(pageinfo, "<div class=\"cate cate_2\" id=\"cateList\">(?<x>.*?)<div id=\"zdmArticle\" class=\"article_relative hide\">");
var tempcatlist = RegGroupCollection(catArea, "<a class=\"cate_lk2 \" href=\"(?<x>.*?)\" title=(?<y>.*?) navvalue");
foreach (Match item in tempcatlist)
{
string tempurl = item.Groups["x"].Value;
string tempId = RegGroupsX<string>(tempurl, "http://searchex.yixun.com/(?<x>.*?)-1-/");
if (tempId.Contains("t"))
{
tempId = tempId.Substring(tempId.IndexOf('t') + 1);
}
string tempName = item.Groups["y"].Value;
if (!HasBindClasslist.Exists(c => c.ClassId == tempId))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempId,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = tempurl,
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
//siteClassInfo.ParentUrl = pcatUrl;
if (pcatName != "")
siteClassInfo.ParentName = pcatName;
//if (pcatId!="")
//siteClassInfo.ParentClass = pcatId;
siteClassInfo.ClassCrumble = classCrumble;
siteClassInfo.UpdateTime = DateTime.Now;
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
private void UpdateCat(SiteClassInfo siteClassInfo)
{
string pageinfo = HtmlAnalysis.Gethtmlcode(siteClassInfo.Urlinfo);
//if (pageinfo.Contains("很抱歉，没有找到相关的商品。"))
// return;
string crumb = RegGroupsX<string>(pageinfo,"<div class=\"w1200 breadNav\">(?<x>.*?)<div class=\"w1200\">");
var catlist = RegGroupCollection(crumb, "<a href=\"(?<x>.*?)\">(?<y>.*?)</a>");
if (catlist == null)
return;
int deep = catlist.Count;
if (deep > 1)
{
siteClassInfo.ParentUrl =domain+ catlist[deep - 2].Groups["x"].Value;
siteClassInfo.ParentName = catlist[deep - 2].Groups["y"].Value;
siteClassInfo.ParentClass = RegGroupsX<string>(siteClassInfo.ParentUrl, "/list-(?<x>\\d+)");
}
siteClassInfo.HasChild = deep <= 1;
siteClassInfo.TotalProduct = RegGroupsX<int>(pageinfo, "搜索到 <span class=\"red\">(?<x>\\d+)</span> 件相关商品");
string catArea = RegGroupsX<string>(pageinfo, "<div class=\"sortlist mb10\">(?<x>.*?)<!--左侧产品分类列表 E--");
if (!string.IsNullOrEmpty(catArea))
{
var list = RegGroupCollection(catArea, "<a href=\"(?<x>.*?)\"\\s*>(?<y>.*?)</a>");
if (list != null)
{
foreach (Match match in list)
{
string tempurl =domain+ match.Groups["x"].Value;
string tempid = RegGroupsX<string>(tempurl, "/list-(?<x>\\d+)");
if (!ValidCatId(tempid))
{
continue;
}
if (!HasBindClasslist.Exists(p => p.ClassId == tempid))
{
string tempName = match.Groups["y"].Value;
if(string.IsNullOrEmpty(tempName))
{ continue; }
SiteClassInfo cat = new SiteClassInfo
{
ParentUrl = "",
ParentClass = "",
ParentName = "",
TotalProduct = 0,
Urlinfo = tempurl,
ClassId = tempid,
UpdateTime = DateTime.Now,
IsDel = false,
BindClassId = 0,
BindClassName = "",
HasChild = true,
IsBind = false,
IsHide = false,
ClassName = tempName,
SiteId = Baseinfo.SiteId,
ClassCrumble = "",
CreateDate = DateTime.Now
};
HasBindClasslist.Add(cat);
shopClasslist.Add(cat);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}
}
siteClassInfo.UpdateTime = DateTime.Now;
new mmbSiteClassInfoDB().UpdateSiteClass(siteClassInfo);
}
private void GetCatInfo(string directoryHtml)
{
string catArea = RegGroupsX<string>(directoryHtml,
"<div class=\"brandContent\">(?<x>.*?)<div id=\"div2\"");
if (catArea == null)
return;
catArea = catArea.Replace("\t", "").Replace("\r", "").Replace("\n", "");
var list = RegGroupCollection(catArea, "<a href=\"(?<x>.*?)\" target=\"_blank\" title=\"(?<y>.*?)\">");
foreach (Match item in list)
{
string tempUrl = item.Groups["x"].Value;
string tempName = item.Groups["y"].Value;
if (!string.IsNullOrEmpty(tempName))
{
tempName = tempName.Trim();
}
string tempid = RegGroupsX<string>(tempUrl, "/Category/(?<x>\\d+)-");
if (!HasBindClasslist.Exists(c => c.ClassId == tempid))
{
SiteClassInfo iteminfo = new SiteClassInfo
{
ParentClass = "",
ParentName = "",
ClassName = tempName,
ClassId = tempid,
ParentUrl = "",
IsDel = false,
IsBind = false,
IsHide = false,
BindClassId = 0,
BindClassName = "",
HasChild = false,
ClassCrumble = "",
TotalProduct = 0,
SiteId = Baseinfo.SiteId,
Urlinfo = string.Format("http://www.lbxcn.com/{0}", tempUrl),
UpdateTime = DateTime.Now,
CreateDate = DateTime.Now
};
HasBindClasslist.Add(iteminfo);
shopClasslist.Add(iteminfo);
}
}
if (shopClasslist.Count > 0)
{
new SiteClassInfoDB().AddSiteClass(shopClasslist);
shopClasslist.Clear();
}
}