void Snapshot(IW2S_SG_BaiduCommend tsk)
{
SogouQuery wc = new SogouQuery();
wc.Query(tsk);
// SaveKeyRecord(searchTask);
}
string get_urls(IW2S_SG_BaiduCommend tsk)
{
string searchKeywords = tsk.Keyword.RemoveSpace().GetLower();
if (!string.IsNullOrEmpty(searchKeywords))
{
string baiduUrlFormat = "https://www.sogou.com/web?query={0}&cid=&page=1&ie=utf8&dr=1";// "https://www.sogou.com/web?query={0}&ie=utf8";
return(baiduUrlFormat.FormatStr(searchKeywords.GetUrlEncodedString("utf-8")));
}
return(string.Empty);
}
public void Query(IW2S_SG_BaiduCommend tsk)
{
List <ResultLiks> result = new List <ResultLiks>();
var link = get_urls(tsk);
try
{
GetLinks(link, tsk);
//if (list != null && list.Count > 0)
// result.AddRange(list);
}
catch (Exception ex)
{
log(ex.Message + ex.StackTrace);
}
}
public List <SG_links> Query(IW2S_SG_BaiduCommend searchTsk, List <IW2S_SG_BaiduCommend> taskList)
{
List <SG_links> linkvaluelist = new List <SG_links>();
SG_BotTaskService SG_bt = new SG_BotTaskService();
List <IW2S_SG_level1link> linklist = SG_bt.GetLinkTitleList(searchTsk.ProjectId);
int index = 0;
for (int i = 0; i < taskList.Count; i++)
{
if (searchTsk.CommendKeyword == taskList[i].CommendKeyword)
{
index = i;
}
}
for (int i = 0; i < taskList.Count; i++)
{
SG_links lk = new SG_links();
//if (searchTsk.CommendKeyword == taskList[i].CommendKeyword)
//{
//}
//else
//{
lk.source = index;
lk.target = i;
lk.KeywordId = searchTsk._id;
lk.ProjectId = searchTsk.ProjectId;
int linkNum = 0;
foreach (var item in linklist)
{
if (!string.IsNullOrEmpty(item.Title))
{
if (item.Title.Contains(searchTsk.CommendKeyword) && item.Title.Contains(taskList[i].CommendKeyword))
{
linkNum = linkNum + 1;
}
}
}
lk.value = linkNum;
lk.Gid = IDHelper.GetGuid("{0}/{1}/{2}/{3}".FormatStr(lk.source, lk.target, lk.KeywordId, lk.ProjectId));
linkvaluelist.Add(lk);
// }
}
SG_SaveResult(linkvaluelist);
return(linkvaluelist);
}
public void Run()
{
while (true)
{
BotTaskService bt = new BotTaskService();
Random r = new Random();
IW2S_SG_BaiduCommend keyTask = bt.GetBotTask(); //get_task();
if (keyTask == null || keyTask.CommendKeyword == "" || keyTask.CommendKeyword == null)
{
SetReady();
log("没有搜索任务 !!!");
Thread.Sleep(3000);
continue;
}
SetBusy();
var update = new UpdateDocument {
{ "$set", new QueryDocument {
{ "WXStatus", 1 }, { "BotStatus", 1 }
} }
};
var result = MongoDBHelper.Instance.Get_IW2S_SG_BaiduCommend().UpdateOne(new QueryDocument {
{ "_id", keyTask._id }
}, update);
Snapshot(keyTask);
try
{
update = new UpdateDocument {
{ "$set", new QueryDocument {
{ "WXStatus", 2 }, { "BotStatus", 2 }
} }
};
result = MongoDBHelper.Instance.Get_IW2S_SG_BaiduCommend().UpdateOne(new QueryDocument {
{ "_id", keyTask._id }
}, update);
}
catch (Exception ex)
{
log(DateTime.Now + "ERROR ." + ex.Message);
Thread.Sleep(2000);
}
}
}
void SG_Snapshot(IW2S_SG_BaiduCommend searchTask, List <IW2S_SG_BaiduCommend> listKey)
{
List <IW2S_SG_Data> xListings = new List <IW2S_SG_Data>();
SG_WeChatQuery wc = new SG_WeChatQuery(searchTask.Keyword + searchTask.CommendKeyword);
var links360 = wc.Query(searchTask, listKey);
}
void GetLinks(string link, IW2S_SG_BaiduCommend tsk)
{
List <IW2S_SG_level1link> result = new List <IW2S_SG_level1link>();
int nohist_pages = 0;
int quried_pages = 0;
while (!string.IsNullOrEmpty(link) && quried_pages <= 2)
{
log(link);
CookieContainer cc = new CookieContainer();
Encoding enc = null;
CookieCollection cookiesColl = new CookieCollection();
CookieCollection cookieCollection = new CookieCollection();
string Rurl = "https://www.sogou.com/";
string cookie = "";
string hhhtml = GetContentByIndex(Rurl, 8000, cc, ref enc, out Rurl, ref cookiesColl, out cookieCollection);
cookiesColl = cookieCollection;
string realUrl = "";
var html = GetContent(link, 8000, ref enc, out Rurl, cookie, ref cookiesColl, out cookieCollection); //(link);
cookiesColl = cookieCollection;
if (html == null)
{
break;
}
//处理 vrwrap
var tags = html.SubAfter("<body").SubAfter("results").SubBefore("hint_container").SplitWith("<h3");
if (tags == null || tags.Length == 0)
{
log("已被sogou屏蔽，请调试！ " + tsk.Keyword);
break;
}
bool nohit = true;
foreach (var tag in tags)
{
try
{
if (!tag.Contains("<a"))
{
continue;
}
string title = RemoveInivalidChar(tag.SubAfter("<a").SubBefore("</a>").GetTxtFromHtml2().RemoveSpace());
string href = tag.SubAfter("<a").SubBefore("</a>").GetFirstHref2();
string Jianjie = "";
if (tag.Contains("简介："))
{
Jianjie = tag.SubAfter("简介：").SubBefore("</").GetTxtFromHtml2().RemoveSpace();
}
if (tag.Contains("cacheresult_summary"))
{
Jianjie = tag.SubAfter("cacheresult_summary").SubBefore("</div>").GetTxtFromHtml2().RemoveSpace();
}
if (string.IsNullOrEmpty(Jianjie))
{
Jianjie = tag.SubAfter("summary_beg").SubBefore("summary_end").GetTxtFromHtml2().RemoveSpace();
}
int n = new Random().Next(8000, 15000);
Thread.Sleep(n);
var tuplehtml = GetContent(href, 8000, ref enc, out Rurl, cookie, ref cookiesColl, out cookieCollection); // get_htmlUrl(href);
if (!string.IsNullOrEmpty(tuplehtml))
{
if (tuplehtml.Contains("window.location.replace"))
{
realUrl = tuplehtml.SubAfter("window.location.replace").SubBefore("</script>").Replace('"', ' ').Replace("(", "").Replace(")", "").RemoveSpace();
}
else
{
realUrl = Rurl;
}
}
string domain = "";
if (!string.IsNullOrEmpty(realUrl))
{
domain = GetDomain(realUrl);
}
else
{
realUrl = href;
domain = GetDomain(href);
}
string topDomain = GetLevel1Domain(domain);
bool IsContains = false;
int States = 0;
int blackid = 0;
realUrl = realUrl.Replace("amp;", "");
int nn = new Random().Next(6000, 15000);
Thread.Sleep(nn);
var htmldetail = GetContent(realUrl, 8000, ref enc, out Rurl, cookie, ref cookiesColl, out cookieCollection);// get_Detailehtml(href, 8000, cc, ref enc, out Rurl, cookie, ref cookiesColl, out cookieCollection);// GetContent(href, 8000, cc, ref enc, out Rurl);
Regex reg = new Regex("(20\\d{2}[-/]\\d{1,2}[-/]\\d{1,2})|(20\\d{2}年\\d{1,2}月\\d{1,2}日)");
Match m = reg.Match(htmldetail);
//MatchCollection cols = reg.Matches(item.Html);
string time = "";
if (m.Groups.Count > 0)
{
time = m.Groups[0].Value;
}
//foreach (var item in BLtb)
//{
// if (item.Domain.Trim().ToLower().Equals(topDomain.Trim().ToLower()))
// {
// States = 2;
// blackid = item.Id;
// }
//}
//foreach (var item in excludedKeywords)
//{
// if (item.AuthorizedUrl1.Contains(topDomain))
// {
// IsContains = true;
// States = 1;
// }
//}
//if (IsContains == true)
// continue;
IW2S_SG_level1link l1 = new IW2S_SG_level1link
{
UsrId = tsk.UsrId,
Domain = domain,
TopDomain = topDomain,
Keywords = string.Format("{0}", tsk.Keyword),
LinkUrl = realUrl,
Html = htmldetail,
BizId = IDHelper.GetGuid("{0}/{1}".FormatStr(realUrl, tsk._id.ToString())),
SearchkeywordId = tsk._id.ToString(),
CreatedAt = DateTime.UtcNow.AddHours(8),
Description = Jianjie,
Title = title,
ProjectId = tsk.ProjectId,
PublishTime = time,
AlternateFields = "0",
DataCleanStatus = 0
};
result.Add(l1);
nohit = false;
nohist_pages = 0;
}
catch (Exception ex)
{
log("有错误信息！" + ex.Message);
}
}
if (nohit)
{
nohist_pages++;
}
//如果连续3页都没有结果，就跳出
if (nohist_pages > 3)
{
break;
}
quried_pages++;
pages++;
link = html.SubAfter("sogou_next").SubBefore("下一页").GetLastHref2();
if (!string.IsNullOrEmpty(link) && !link.IsStartWith("http"))
{
if (link.IsStartWith("/"))
{
link = link.SubAfter("/");
}
link = "https://www.sogou.com/web".GetContact(link);
}
SaveResult(result);
result.Clear();
int nn1 = new Random().Next(6000, 15000);
Thread.Sleep(nn1);
}
}
