public static void InitializeRssCrawler(ILogger logger, RandomTimeSpan channelFetchingDelay, TimeSpan channelErrorDelay, TimeSpan errorDelay, TimeSpan loopDelay)
{
System.Threading.Tasks.Task.Run(() =>
{
//var channelHubCtx = GlobalHost.ConnectionManager.GetHubContext<ChannelHub>();
ObjectCache cache = MemoryCache.Default;
while (true)
{
if (_currentDate != DateTime.Now.Day)
{
_feedCache = new HashSet <string>();
_currentDate = DateTime.Now.Day;
}
var feedUrl = string.Empty;
try
{
List <RssChannelsRow> channels = channels = SimpleFeedlyDatabaseAccess.GetActiveChannels().OrderBy(x => x.Id).ToList();
var count = 0;
foreach (var channel in channels)
{
feedUrl = channel.Link;
count++;
logger.Info($"- [{count}/{channels.Count}] Working on channel: {channel.Id} | {feedUrl}");
//channelHubCtx.Clients.All.updateChannelProgress(new { Message = $"<strong>Fetching</strong> <a href='{channel.Link}' target='_blank'>{channel.Link}</a>", IsSleeping = false });
if (string.IsNullOrWhiteSpace(feedUrl))
{
logger.Warn($"=> Channel has empty link: {channel.Id}");
continue;
}
var channelSleepingCacheKey = "channel_is_sleeping|" + channel.Id;
var isSleeping = cache[channelSleepingCacheKey] as bool?;
if (isSleeping == null || isSleeping == false)
{
try
{
RssCrawlerEngine usedEngine = RssCrawlerEngine.CodeHollowFeedReader;
var feed = RssCrawler.GetFeedsFromChannel(feedUrl, channel.RssCrawlerEngine, false, out usedEngine, out Exception fetchFeedError);
// update default engine for channel
SimpleFeedlyDatabaseAccess.UpdateChannelDefaultEngine((long)channel.Id, feed == null ? (RssCrawlerEngine?)null : usedEngine);
if (feed != null)
{
logger.Info($" + Number of items: {feed.Items.Count}");
var hasNew = false;
foreach (var fItem in feed.Items)
{
if (!StringUtils.IsUrl(fItem.Link))
{
continue;
}
var feedItemKey = GenerateFeedItemKey(fItem);
var feedCacheKey = GenerateFeedCacheKey((long)channel.Id, feedItemKey);
if (string.IsNullOrWhiteSpace(feedItemKey) || string.IsNullOrWhiteSpace(fItem.Link))
{
logger.Info($" + Skipped item: {JsonConvert.SerializeObject(fItem)}");
continue;
}
if (!_feedCache.Contains(feedCacheKey))
{
if (!SimpleFeedlyDatabaseAccess.CheckExistFeedItem((long)channel.Id, feedItemKey))
{
var feedItem = new RssFeedItemsRow
{
ChannelId = channel.Id,
FeedItemKey = feedItemKey,
Title = string.IsNullOrWhiteSpace(fItem.Title) ? fItem.Link : fItem.Title,
Link = fItem.Link,
Description = fItem.Description,
PublishingDate = fItem.PublishingDate,
Author = fItem.Author,
Content = fItem.Content
};
SimpleFeedlyDatabaseAccess.InsertFeedItem(feedItem);
hasNew = true;
}
_feedCache.Add(feedCacheKey);
}
}
SimpleFeedlyDatabaseAccess.UpdateChannelErrorStatus((long)channel.Id, false, null);
if (!hasNew)
{
var randomExpiryTime =
cache.Add(channelSleepingCacheKey, true, DateTime.Now.Add(channelFetchingDelay.GenerateRamdomValue()));
}
}
else
{
SimpleFeedlyDatabaseAccess.UpdateChannelErrorStatus((long)channel.Id, true, fetchFeedError == null ? null : JsonConvert.SerializeObject(fetchFeedError));
if (fetchFeedError != null)
{
ErrorHandle(fetchFeedError, feedUrl);
}
}
}
catch (Exception err)
{
SimpleFeedlyDatabaseAccess.UpdateChannelErrorStatus((long)channel.Id, true, JsonConvert.SerializeObject(err));
cache.Add(channelSleepingCacheKey, true, DateTime.Now.Add(channelErrorDelay));
logger.Error(err, $"An error occurred on channel: {channel.Id} | {feedUrl}");
ErrorHandle(err, feedUrl);
}
}
else
{
logger.Info($" + sleeping...");
}
}
}
catch (Exception ex)
{
logger.Error(ex, "An error occurred");
ErrorHandle(ex, feedUrl);
System.Threading.Thread.Sleep(errorDelay);
}
//channelHubCtx.Clients.All.updateChannelProgress(new { Message = "<span class='link-muted'>Crawler's sleeping...</span>", IsSleeping = true });
_currentDate = DateTime.Now.Day;
// we should delay a little bit, some seconds maybe
System.Threading.Thread.Sleep(loopDelay);
}
});
}
private static async Task <int> FakeAsync(int minSecond, int maxSecond)
{
await Task.Delay(RandomTimeSpan.Between(minSecond, maxSecond));
return(0);
}