private void Init()
{
_redisDesHash = CacheProductDesciptioHash.Instance();
_jobClientLogChangePrice = new MqLogChangePrice();
_jobClientLogChangeProduct = MQLogChangeProduct.Instance();
_dicDuplicate = new Dictionary <long, long>();
_dicCacheProduct = new Dictionary <long, ProductCache>();
_linksQueue = new Queue <NSCrawler.Job>();
_productAdapter = new ProductAdapter(new SqlDb(Server.ConnectionString));
_redisWaitCrawler = RedisCompanyWaitCrawler.Instance();
_redisProduct = RedisCacheProductInfo.Instance();
_redisProductLastUpdate = RedisLastUpdateProduct.Instance();
_publiserDesciption = new PublisherDesciption();
}
private static void SyncCompanyCrawler()
{
int count = 0;
RedisCacheCompanyCrawler redisCompany = RedisCacheCompanyCrawler.Instance();
RedisCompanyWaitCrawler redisCompanyWaitCrawler = RedisCompanyWaitCrawler.Instance();
RedisLastUpdateProduct redisLstProduct = RedisLastUpdateProduct.Instance();
CacheProductInfo cacheProductInfo = new CacheProductInfo(new SqlDb(Server.ConnectionString));
ProductAdapter productAdapter = new ProductAdapter(new SqlDb(Server.ConnectionString));
List <long> lstCrawler = productAdapter.GetAllCompanyIdCrawlerReload();
for (int i = 0; i < lstCrawler.Count; i++)
{
redisCompanyWaitCrawler.SetNexReload(lstCrawler[i], -10);
//long companyID = lstCrawler[i];
//if (!redisCompanyWaitCrawler.CheckHaveItemReload(companyID))
//{
// count++;
// redisCompanyWaitCrawler.SetNexReload(companyID, 1);
// Company cmp = new Company(companyID);
// redisCompany.SetCompanyInfo(companyID, cmp.Domain, 1, 1);
// cacheProductInfo.RefeshCacheProductInfo(companyID);
// redisLstProduct.ResetLastUpdateForCompany(companyID, productAdapter.GetListProductIdOfCompany(companyID));
//}
log.Info(string.Format("sync company {0} {1}/{2}", count, i, lstCrawler.Count));
}
List <long> lstCrawlerFindNew = productAdapter.GetAllCompanyIdCrawlerFindNew();
for (int i = 0; i < lstCrawler.Count; i++)
{
redisCompanyWaitCrawler.SetNexFindNew(lstCrawler[i], -10);
long companyID = lstCrawler[i];
//if (!redisCompanyWaitCrawler.CheckHaveItemFindNew(companyID))
//{
// count++;
// redisCompanyWaitCrawler.SetNexFindNew(companyID, 1);
// Company cmp = new Company(companyID);
// redisCompany.SetCompanyInfo(companyID, cmp.Domain, 1, 1);
// cacheProductInfo.RefeshCacheProductInfo(companyID);
// redisLstProduct.ResetLastUpdateForCompany(companyID, productAdapter.GetListProductIdOfCompany(companyID));
//}
log.Info(string.Format("sync company {0} {1}/{2}", count, i, lstCrawler.Count));
}
Console.WriteLine("Success sync company crawl!");
}
private void InitData()
{
_productAdapter = new ProductAdapter(new SqlDb(Server.ConnectionString));
_linkQueue = new Queue <string>();
_crcProductOldGroup = new Dictionary <long, bool>();
_visitedCrc = new Dictionary <long, bool>();
_productsReloaded = new List <long>();
_redisQueueFindNew = RedisQueueFindNew.Instance();
_redisCacheCompanyCrawler = RedisCacheCompanyCrawler.Instance();
_redisLastUpdateProduct = RedisLastUpdateProduct.Instance();
_redisWaitCrawler = RedisCompanyWaitCrawler.Instance();
_redisCrcVisited = RedisCrcVisitedFindNew.Instance();
_mqLogQueueVisit = MQLogQueueVisit.Instance();
_mqLogWarning = MQLogWarningFindNew.Instance();
_mqLogChangePrice = new MqLogChangePrice();
}
public bool Init()
{
try
{
var rabbitMqCrawler = RabbitMQManager.GetRabbitMQServer(ConfigCrawler.KeyRabbitMqCrawler);
_producerReportSessionRunning = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeSessionRunning, ConfigCrawler.RoutingkeySessionRunning);
_producerReportError = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeErorrCrawler, ConfigCrawler.RoutingKeyErrorCrawler);
_producerProductChange = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeChangeProduct, ConfigCrawler.RoutingkeyChangeProduct);
_producerDuplicateProduct = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeDuplicateProductToCache, ConfigCrawler.ExchangeDuplicateProductToCache);
_producerEndCrawler = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeEndSession, ConfigCrawler.RoutingEndSession);
_producerVisitedLinkFindNew = new ProducerBasic(rabbitMqCrawler, ConfigCrawler.ExchangeVisitedLinkFindNew, ConfigCrawler.RoutingKeyVisitedLinkFindNew);
_company = new Company(_companyId);
_config = new Configuration(_companyId);
if (_config.LimitProductValid == 0)
{
this._limitProductValid = 1000000;
}
_rootUri = new Uri(_company.Website);
_cacheCrcVisited = RedisCrcVisitedFindNew.Instance();
_cacheWaitCrawler = RedisCompanyWaitCrawler.Instance();
_cacheLastUpdateProduct = RedisLastUpdateProduct.Instance();
_cacheProductHash = CacheProductHash.Instance();
_cacheCacheCompanyCrawler = RedisCacheCompanyCrawler.Instance();
_cacheDuplicateProduct = CacheDuplicateProduct.Instance();
_company = new Company(_companyId);
_config = new Configuration(_companyId);
_visitedCrc = new HashSet <long>();
_linkQueue = new Queue <JobFindNew>();
_crcProductOldGroup = new HashSet <long>();
_dicDuplicate = new Dictionary <long, long>();
_countVisited = 0;
_countNewProduct = 0;
_tokenCrawler.ThrowIfCancellationRequested();
_visitRegexs = _config.VisitUrlsRegex;
_detailLinkRegexs = _config.ProductUrlsRegex;
_noCrawlerRegexs = _config.NoVisitUrlRegex ?? new List <string>();
_noCrawlerRegexs.AddRange(UtilCrawlerProduct.NoCrawlerRegexDefault);
_timeStart = DateTime.Now;
_rootUri = Common.GetUriFromUrl(_company.Website);
_hsDuplicateProduct = _cacheDuplicateProduct.GetHashDuplicate(_companyId);
ClearOldCache();
LoadCrcOldProduct();
LoadOldQueue();
return(true);
}
catch (Exception ex)
{
_log.Error(string.Format("Company:{0} {1} {2}", _companyId, ex.Message, ex.StackTrace));
string mss =
Newtonsoft.Json.JsonConvert.SerializeObject(new ErrorCrawler()
{
CompanyId = _companyId, ProductId = 0, TimeError = DateTime.Now, Message = "Init" + ex.Message + ex.StackTrace
});
_producerReportError.PublishString(mss, true);
if (_producerEndCrawler != null)
{
_producerEndCrawler.PublishString(new CrawlerSessionLog()
{
CompanyId = _companyId,
CountChange = 0,
CountProduct = 0,
CountVisited = 0,
Domain = "",
EndAt = DateTime.Now,
Ip = Dns.GetHostName(),
NumberDuplicateProduct = 0,
Session = this._session,
StartAt = this._timeStart,
TotalProduct = 0,
TypeCrawler = 0,
TypeEnd = "Error Init",
TypeRun = "Auto"
}.ToJson());
}
return(false);
}
}
public bool Init()
{
try
{
_cacheWaitCrawler = RedisCompanyWaitCrawler.Instance();
_redisLastCrl = RedisLastUpdateProduct.Instance();
_config = new Configuration(_companyId, true);
_company = new Company(_companyId);
_cacheDesHash = CacheProductDesciptioHash.Instance();
var rabbitMQCrawler = RabbitMQManager.GetRabbitMQServer(ConfigCrawler.KeyRabbitMqCrawler);
_producerReportError = new ProducerBasic(rabbitMQCrawler, ConfigCrawler.ExchangeErorrCrawler, ConfigCrawler.RoutingKeyErrorCrawler);
_producerProductChange = new ProducerBasic(rabbitMQCrawler, ConfigCrawler.ExchangeChangeProduct, ConfigCrawler.RoutingkeyChangeProduct);
_producerDuplicateProduct = new ProducerBasic(rabbitMQCrawler, ConfigCrawler.ExchangeDuplicateProductToCache, ConfigCrawler.ExchangeDuplicateProductToCache);
_producerPushCompanyReload = new ProducerBasic(rabbitMQCrawler, ConfigCrawler.ExchangeCompanyReload, ConfigCrawler.RoutingkeyCompanyReload);
_producerEndCrawler = new ProducerBasic(rabbitMQCrawler, ConfigCrawler.ExchangeEndSession, ConfigCrawler.RoutingEndSession);
_cacheCheckDelete = CacheTrackDeleteProduct.Instance();
_cacheProductHash = CacheProductHash.Instance();
_dicTrackDie = _cacheCheckDelete.GetDicTrackOfCompany(_companyId);
_dicDuplicate = new Dictionary <long, long>();
_dicCacheProduct = new Dictionary <long, ProductHash>();
_dicHashDesc = new Dictionary <long, long>();
_dicCacheProduct = new Dictionary <long, ProductHash>();
_linksQueue = new Queue <Job>();
_timeStart = DateTime.Now;
_countChange = 0;
_countVisited = 0;
return(true);
}
catch (Exception ex)
{
_log.Error(ex);
if (_producerEndCrawler != null)
{
_producerEndCrawler.PublishString(new CrawlerSessionLog()
{
CompanyId = _companyId,
CountChange = 0,
CountProduct = 0,
CountVisited = 0,
Domain = "",
EndAt = DateTime.Now,
Ip = Dns.GetHostName(),
NumberDuplicateProduct = 0,
Session = this._session,
StartAt = this._timeStart,
TotalProduct = 0,
TypeCrawler = 0,
TypeEnd = "Error Init",
TypeRun = "Auto"
}.ToJson());
}
string mss =
Newtonsoft.Json.JsonConvert.SerializeObject(new ErrorCrawler()
{
CompanyId = _companyId, ProductId = 0, TimeError = DateTime.Now, Message = "Init" + ex.Message + ex.StackTrace
});
_producerReportError.PublishString(mss, true, 20);
return(false);
}
}
public void Run(System.Threading.CancellationToken token)
{
ProductAdapter productAdapter = new ProductAdapter(new SqlDb("Data Source=42.112.28.93;Initial Catalog=QT_2;Persist Security Info=True;User ID=wss_price;Password=HzlRt4$$axzG-*UlpuL2gYDu;connection timeout=200"));
log.InfoFormat("Start run at {0}", DateTime.Now.ToString(CultureInfo.InvariantCulture));
CacheProductHash cashProductHash = CacheProductHash.Instance();
RedisLastUpdateProduct cacheLastUpdateProduct = RedisLastUpdateProduct.Instance();
int countProduct = 0;
try
{
var lstFn = productAdapter.GetAllCompanyIdCrawlerFindNew();
var lstRl = productAdapter.GetAllCompanyIdCrawlerReload();
RedisCompanyWaitCrawler redisCache = RedisCompanyWaitCrawler.Instance();
redisCache.SyncCompanyFindNew(lstFn);
redisCache.SyncCompanyReload(lstRl);
}
catch (Exception ex)
{
log.Error(ex);
}
var lst = new List <QT.Entities.CrawlerProduct.Cache.ProductHash>();
var lstLastUpdate = new List <long>();
var lstCompany = productAdapter.GetAllCompanyIdCrawler();
foreach (var companyID in lstCompany)
{
Company cmp = new Company(companyID);
productAdapter.DeleteProductUnvalidOfCOmpany(companyID);
DataTable tbl = productAdapter.GetProductResetColumnDuplicateAndChange(companyID);
foreach (DataRow rowProduct in tbl.Rows)
{
long productId = QT.Entities.Common.Obj2Int64(rowProduct["ID"]);
long originPrice = QT.Entities.Common.Obj2Int64(rowProduct["OriginPrice"]);
string name = rowProduct["Name"].ToString();
long price = QT.Entities.Common.Obj2Int64(rowProduct["Price"]);
string imageUrl = Convert.ToString(rowProduct["ImageUrls"]);
string detailUrl = Convert.ToString(rowProduct["DetailUrl"]);
int inStock = QT.Entities.Common.Obj2Int(rowProduct["InStock"]);
bool valid = QT.Entities.Common.Obj2Bool(rowProduct["Valid"]);
string shortDescription = QT.Entities.Common.CellToString(rowProduct["ShortDescription"], "");
long categoryId = rowProduct["ClassificationID"] == DBNull.Value ? 0 : QT.Entities.Common.Obj2Int64(rowProduct["ClassificationID"]);
long hashChange = ProductEntity.GetHashChangeInfo(inStock, valid, price, name, imageUrl, categoryId, shortDescription, originPrice);
long hashDuplicate = Product.GetHashDuplicate(cmp.Domain, price, name, imageUrl);
long hashImage = Product.GetHashImageInfo(imageUrl);
lst.Add(new QT.Entities.CrawlerProduct.Cache.ProductHash()
{
HashChange = hashChange,
HashDuplicate = hashDuplicate,
HashImage = hashImage,
Id = productId,
Price = price,
url = detailUrl
});
lstLastUpdate.Add(productId);
}
cashProductHash.SetCacheProductHash(companyID, lst, 100); cacheLastUpdateProduct.RemoveAllLstProduct(companyID);
cacheLastUpdateProduct.UpdateBathLastUpdateProduct(companyID, lstLastUpdate, DateTime.Now.AddDays(-1));
productAdapter.UpdateCountProductForCompany(companyID, lstLastUpdate.Count, lstLastUpdate.Count);
lst.Clear();
lstLastUpdate.Clear();
log.Info(string.Format("Complete Company: {0} {1}/{2}", companyID, countProduct++, lstCompany.Count));
}
log.Info("Complete all company");
NextRun = DateTime.Now.AddHours(MAX_HOUR_LOOP);
log.InfoFormat("End at {0}", DateTime.Now.ToString());
}