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);
}
}
