public bool CrawlData(string Id, string createdBy, ref string msg)
{
NSLog.Logger.Info("CrawlData: " + Id);
var result = true;
try
{
using (var _db = new CMS_Context())
{
/* get key by ID */
var keyWord = _db.CMS_KeyWord.Where(o => o.ID == Id && o.Status == (byte)Commons.EStatus.Active).FirstOrDefault();
if (keyWord != null)
{
/* check time span crawl */
var timeSpanCrawl = DateTime.Now - keyWord.UpdatedDate;
if (timeSpanCrawl.Value.TotalMinutes > 5 || keyWord.UpdatedDate == keyWord.CreatedDate) /* 5min to crawl data again */
{
/* update crawer date */
var bkTime = keyWord.UpdatedDate;
keyWord.UpdatedDate = DateTime.Now;
keyWord.UpdatedBy = createdBy;
keyWord.KeyWord = keyWord.KeyWord.Trim();
_db.SaveChanges();
/* cookies
* User: [email protected]
* Pass: pitool.org79
*/
if (!string.IsNullOrEmpty(keyWord.CrawlAccountID))
{
var _Cookie = keyWord.CMS_Account.Cookies;
if (!string.IsNullOrEmpty(_Cookie))
{
CrawlerBoardHelper._Cookies = _Cookie;
}
else
{
CrawlerHelper._Cookies = "_b = \"AS+B1gn0GdpGgLQl83JubKX1bG19kiuUUvX8lnvITKDHNq2tJcgqXNIQ0cLN+kjq4KM=\"; _pinterest_pfob = enabled; _ga = GA1.2.229901352.1528170174; pnodepath = \"/pin4\"; fba = True; G_ENABLED_IDPS = google; bei = false; logged_out = True; sessionFunnelEventLogged = 1; cm_sub = denied; _auth = 1; csrftoken = fkrSitmDb4vW2kT1G3GfOkcC8mPvl0kV; _pinterest_sess = \"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\"";
}
}
else
{
CrawlerHelper._Cookies = "_b = \"AS+B1gn0GdpGgLQl83JubKX1bG19kiuUUvX8lnvITKDHNq2tJcgqXNIQ0cLN+kjq4KM=\"; _pinterest_pfob = enabled; _ga = GA1.2.229901352.1528170174; pnodepath = \"/pin4\"; fba = True; G_ENABLED_IDPS = google; bei = false; logged_out = True; sessionFunnelEventLogged = 1; cm_sub = denied; _auth = 1; csrftoken = fkrSitmDb4vW2kT1G3GfOkcC8mPvl0kV; _pinterest_sess = \"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\"";
}
var searchStr = HttpUtility.UrlEncode(keyWord.KeyWord);
/* get first class result */
var model = new CMS_CrawlerModels();
CMSPinFactory _fac = new CMSPinFactory();
CrawlerHelper.Get_Tagged_Pins(ref model, searchStr, Commons.PinDefault);
if (model != null && model.Pins != null && model.Pins.Any())
{
/* get second class result */
var listPinID = model.Pins.Select(o => o.ID).ToList();
Parallel.ForEach(listPinID, pinID =>
{
CrawlerHelper.Get_Tagged_OrtherPins(ref model, searchStr, Commons.PinOrtherDefault, "", 1, pinID);
});
}
/* create or update pin */
var res = _fac.CreateOrUpdate(model.Pins, keyWord.ID, createdBy, ref msg);
if (res == false)
{
/* back to last crawl data */
//keyWord.UpdatedDate = bkTime;
//_db.SaveChanges();
result = false;
}
else
{
keyWord.UpdatedDate = DateTime.Now;
_db.SaveChanges();
}
}
}
}
NSLog.Logger.Info("ResponseCrawlData: " + Id, result);
}
catch (Exception ex)
{
msg = "Crawl data is unsuccessfully.";
result = false;
LogHelper.WriteLogs("ErrorCrawlData: " + Id, JsonConvert.SerializeObject(ex));
NSLog.Logger.Error("ErrorCrawlData: " + Id, ex);
}
return(result);
}
public bool CrawlData(string Id, string createdBy, ref string msg)
{
NSLog.Logger.Info("CrawlData: " + Id);
var model = new CMS_CrawlerModels();
var sequence = 0;
var key = "";
var result = true;
try
{
using (var _db = new CMS_Context())
{
/* get key by ID */
var keyWord = _db.CMS_KeyWord.Where(o => o.ID == Id).FirstOrDefault();
if (keyWord != null)
{
sequence = keyWord.Sequence;
key = keyWord.KeyWord;
/* check time span crawl */
var timeSpanCrawl = DateTime.Now - keyWord.UpdatedDate;
if (timeSpanCrawl.Value.TotalMinutes > 5 || keyWord.UpdatedDate == keyWord.CreatedDate) /* 5min to crawl data again */
{
/* update crawer date */
var bkTime = keyWord.UpdatedDate;
keyWord.UpdatedDate = DateTime.Now;
keyWord.UpdatedBy = createdBy;
_db.SaveChanges();
/* call drawler api to crawl data */
CMSPinFactory _fac = new CMSPinFactory();
var listAcc = _db.CMS_Account.Where(o => o.Status == (byte)Commons.EStatus.Active && o.IsActive).ToList();
var listCookie = listAcc.Select(x => x.Cookies).ToList();
var _cookie = CommonHelper.RamdomCookie(listCookie);
CrawlerFbHelpers_v2.CrawlerAllFb(keyWord.KeyWord, _cookie, ref model);
var res = false;
if (model.Pins.Count > 0)
{
res = _fac.CreateOrUpdate(model.Pins, keyWord.ID, createdBy, ref msg);
}
if (res == false)
{
/* back to last crawl data */
//keyWord.UpdatedDate = bkTime;
//_db.SaveChanges();
result = false;
}
else
{
keyWord.UpdatedDate = DateTime.Now;
_db.SaveChanges();
}
}
}
}
LogHelper.WriteLogs(sequence.ToString() + " " + key, "Num post: " + model.Pins.Count().ToString());
NSLog.Logger.Info("ResponseCrawlData", result.ToString());
}
catch (Exception ex)
{
msg = "Crawl data is unsuccessfully.";
result = false;
LogHelper.WriteLogs("ErrorCrawlData: " + Id, JsonConvert.SerializeObject(ex));
NSLog.Logger.Error("ErrorCrawlData: " + Id, ex);
}
return(result);
}
public bool CrawlData(string Id, string createdBy, ref string msg)
{
NSLog.Logger.Info("CrawlData: " + Id);
var model = new CMS_CrawlerModels();
var sequence = 0;
var key = "";
var _cookie = "";
DateTime lastdate = DateTime.Now.AddDays(-7);
DateTime datenow = DateTime.Now;
var result = true;
try
{
using (var _db = new CMS_Context())
{
/* get key by ID */
var keyWord = _db.CMS_KeyWord.Where(o => o.ID == Id).FirstOrDefault();
if (keyWord != null)
{
sequence = keyWord.Sequence;
key = keyWord.KeyWord;
/* check time span crawl */
var timeSpanCrawl = DateTime.Now - keyWord.UpdatedDate;
if (timeSpanCrawl.Value.TotalMinutes > 5 || keyWord.UpdatedDate == keyWord.CreatedDate) /* 5min to crawl data again */
{
/* update crawer date */
var bkTime = keyWord.UpdatedDate;
keyWord.UpdatedDate = DateTime.Now;
keyWord.UpdatedBy = createdBy;
_db.SaveChanges();
/* call drawler api to crawl data */
CMSPinFactory _fac = new CMSPinFactory();
var listAcc = _db.CMS_Account.Where(o => o.Status == (byte)Commons.EStatus.Active && o.IsActive && !string.IsNullOrEmpty(o.Cookies)).ToList();
var listCookie = listAcc.Select(x => x.Cookies).ToList();
_cookie = CommonHelper.RamdomCookie(listCookie);
/* crawler tab post */
var PageSize = Convert.ToInt32(Commons.PageSize);
var modelPost = new CMS_CrawlerModels();
string q = "keywords_search(" + keyWord.KeyWord.Replace(" ", "+") + ")";
string ref_path = "/search/str/" + keyWord.KeyWord + "/stories-keyword/stories-public";
//CrawlerFBToolHelpers.CrawlerNow(q, ref_path, "list", (byte)Commons.EType.Post, _cookie, PageSize, ref modelPost);
//string q = "stories-public(stories-keyword(" + keyWord.KeyWord + "))";
//string ref_path = "/search/str/" + keyWord.KeyWord + "/stories-keyword/stories-public";
NSLog.Logger.Info("done crawler tab post : ", modelPost.Pins.Count);
if (modelPost.Pins != null && modelPost.Pins.Any())
{
model.Pins.AddRange(modelPost.Pins);
}
/* crawler tab people */
var modelPeople = new CMS_CrawlerModels();
q = "stories-opinion(stories-keyword(" + keyWord.KeyWord + "))";
ref_path = "/search/str/" + keyWord.KeyWord + "/stories-keyword/stories-opinion";
//CrawlerFBToolHelpers.CrawlerNow(q, ref_path, "list", (byte)Commons.EType.People, _cookie, PageSize, ref modelPeople);
NSLog.Logger.Info("done crawler tab people : ", modelPeople.Pins.Count);
if (modelPeople.Pins != null && modelPeople.Pins.Any())
{
model.Pins.AddRange(modelPeople.Pins);
}
/* crawler tab photo */
var modelPhoto = new CMS_CrawlerModels();
q = "photos-keyword(" + keyWord.KeyWord.Replace(" ", "+") + ")";
ref_path = "/search/str/" + keyWord.KeyWord.Replace(" ", "+") + "/photos-keyword";
CrawlerFBToolHelpers.CrawlerNow(q, ref_path, "grid", (byte)Commons.EType.Photo, _cookie, 70, ref modelPhoto);
/*crawler detail tab photo */
PinsModels refmodelPhoto = new PinsModels();
var options = new ParallelOptions {
MaxDegreeOfParallelism = 10
};
//for (int i = 0; i < modelPhoto.Pins.Count; i++)
//{
// CrawlerFBToolHelpers.CrawlerDetail(modelPhoto.Pins[i].PhotoID, _cookie, (byte)Commons.EType.Photo, ref refmodelPhoto);
//}
Parallel.ForEach(modelPhoto.Pins, options, pin =>
{
CrawlerFBToolHelpers.CrawlerDetail(pin.PhotoID, _cookie, (byte)Commons.EType.Photo, ref pin);
});
NSLog.Logger.Info("done crawler tab photo : ", modelPhoto.Pins.Count);
if (modelPhoto.Pins != null && modelPhoto.Pins.Any())
{
model.Pins.AddRange(modelPhoto.Pins);
}
var res = false;
if (model.Pins.Count > 0)
{
NSLog.Logger.Info("done crawler before 7 days ago : ", model.Pins.Count);
/* check 7 days ago */
model.Pins = model.Pins.Where(o => o.Created_At >= lastdate && o.Created_At <= datenow).ToList();
NSLog.Logger.Info("done crawler after 7 days ago : ", model.Pins.Count);
Parallel.ForEach(model.Pins, options, pin =>
{
if (pin.Type != (byte)Commons.EType.Photo)
{
CrawlerFBToolHelpers.CrawlerDetail(pin.PhotoID, _cookie, (byte)Commons.EType.Post, ref pin);
}
});
res = _fac.CreateOrUpdate(model.Pins, keyWord.ID, createdBy, keyWord.KeyWord, ref msg);
}
if (res == false)
{
/* back to last crawl data */
//keyWord.UpdatedDate = bkTime;
//_db.SaveChanges();
result = false;
}
else
{
keyWord.UpdatedDate = DateTime.Now;
_db.SaveChanges();
}
}
}
}
LogHelper.WriteLogs(sequence.ToString() + " " + key, "Num post: " + model.Pins.Count().ToString());
NSLog.Logger.Info("ResponseCrawlData", result.ToString());
}
catch (Exception ex)
{
msg = "Crawl data is unsuccessfully.";
result = false;
LogHelper.WriteLogs("ErrorCrawlData: " + Id, JsonConvert.SerializeObject(ex));
NSLog.Logger.Error("ErrorCrawlData: " + Id, ex);
}
return(result);
}