cloutier--bird.makeup/src/BirdsiteLive.Pipeline/Processors/RetrieveTweetsProcessor.cs

118 lines
4.8 KiB
C#
Raw Normal View History

2021-01-22 21:23:27 -05:00
using System;
2023-01-10 21:00:21 -05:00
using System.Collections.Concurrent;
2021-01-22 21:23:27 -05:00
using System.Collections.Generic;
2020-07-18 23:35:19 -04:00
using System.Linq;
using System.Threading;
2020-07-16 01:19:41 -04:00
using System.Threading.Tasks;
2020-07-18 23:35:19 -04:00
using BirdsiteLive.DAL.Contracts;
2020-07-16 01:19:41 -04:00
using BirdsiteLive.DAL.Models;
using BirdsiteLive.Pipeline.Contracts;
using BirdsiteLive.Pipeline.Models;
2020-07-18 23:35:19 -04:00
using BirdsiteLive.Twitter;
2020-07-22 20:19:40 -04:00
using BirdsiteLive.Twitter.Models;
using BirdsiteLive.Common.Settings;
using Microsoft.Extensions.Logging;
2020-07-16 01:19:41 -04:00
2023-02-22 11:30:02 -05:00
namespace BirdsiteLive.Pipeline.Processors.SubTasks
2020-07-16 01:19:41 -04:00
{
public class RetrieveTweetsProcessor : IRetrieveTweetsProcessor
{
private readonly ITwitterTweetsService _twitterTweetsService;
2021-02-02 22:49:37 -05:00
private readonly ICachedTwitterUserService _twitterUserService;
2020-07-18 23:35:19 -04:00
private readonly ITwitterUserDal _twitterUserDal;
private readonly ILogger<RetrieveTweetsProcessor> _logger;
private readonly InstanceSettings _settings;
2020-07-18 23:35:19 -04:00
#region Ctor
public RetrieveTweetsProcessor(ITwitterTweetsService twitterTweetsService, ITwitterUserDal twitterUserDal, ICachedTwitterUserService twitterUserService, InstanceSettings settings, ILogger<RetrieveTweetsProcessor> logger)
2020-07-18 23:35:19 -04:00
{
_twitterTweetsService = twitterTweetsService;
2020-07-18 23:35:19 -04:00
_twitterUserDal = twitterUserDal;
_twitterUserService = twitterUserService;
_logger = logger;
_settings = settings;
2020-07-18 23:35:19 -04:00
}
#endregion
public async Task<UserWithDataToSync[]> ProcessAsync(UserWithDataToSync[] syncTwitterUsers, CancellationToken ct)
2020-07-18 23:35:19 -04:00
{
if (_settings.ParallelTwitterRequests == 0)
{
while(true)
await Task.Delay(1000);
}
var usersWtTweets = new ConcurrentBag<UserWithDataToSync>();
2023-01-10 21:00:21 -05:00
List<Task> todo = new List<Task>();
2022-12-26 15:05:46 -05:00
int index = 0;
foreach (var userWtData in syncTwitterUsers)
2020-07-18 23:35:19 -04:00
{
2022-12-26 15:05:46 -05:00
index++;
2023-01-10 21:00:21 -05:00
var t = Task.Run(async () => {
2023-03-17 19:45:51 -04:00
var user = userWtData.User;
var now = DateTime.UtcNow;
2023-01-20 14:17:32 -05:00
try
2023-01-10 21:00:21 -05:00
{
2023-01-20 14:17:32 -05:00
var tweets = await RetrieveNewTweets(user);
_logger.LogInformation(index + "/" + syncTwitterUsers.Count() + " Got " + tweets.Length + " tweets from user " + user.Acct + " " );
if (tweets.Length > 0 && user.LastTweetPostedId == -1)
{
// skip the first time to avoid sending backlog of tweet
var tweetId = tweets.Last().Id;
await _twitterUserDal.UpdateTwitterUserAsync(user.Id, tweetId, tweetId, user.FetchingErrorCount, now);
}
else if (tweets.Length > 0 && user.LastTweetPostedId != -1)
2023-01-20 14:17:32 -05:00
{
userWtData.Tweets = tweets;
usersWtTweets.Add(userWtData);
var tweetId = tweets.Last().Id;
await _twitterUserDal.UpdateTwitterUserAsync(user.Id, tweetId, tweetId, user.FetchingErrorCount, now);
}
else
{
await _twitterUserDal.UpdateTwitterUserAsync(user.Id, user.LastTweetPostedId, user.LastTweetSynchronizedForAllFollowersId, user.FetchingErrorCount, now);
}
}
catch(Exception e)
2023-01-10 21:00:21 -05:00
{
2023-01-20 14:17:32 -05:00
_logger.LogError(e.Message);
2023-03-17 19:45:51 -04:00
await _twitterUserDal.UpdateTwitterUserAsync(user.Id, user.LastTweetPostedId, user.LastTweetSynchronizedForAllFollowersId, user.FetchingErrorCount, now);
2023-01-10 21:00:21 -05:00
}
});
todo.Add(t);
if (todo.Count > _settings.ParallelTwitterRequests)
2021-01-23 18:20:13 -05:00
{
2023-01-10 21:00:21 -05:00
await Task.WhenAll(todo);
todo.Clear();
2021-01-23 18:20:13 -05:00
}
2022-12-18 16:12:16 -05:00
2020-07-18 23:35:19 -04:00
}
2023-01-10 21:00:21 -05:00
await Task.WhenAll(todo);
2020-07-18 23:35:19 -04:00
return usersWtTweets.ToArray();
}
2022-12-26 11:13:00 -05:00
private async Task<ExtractedTweet[]> RetrieveNewTweets(SyncTwitterUser user)
2020-07-16 01:19:41 -04:00
{
var tweets = new ExtractedTweet[0];
2021-09-05 14:00:18 -04:00
try
{
if (user.LastTweetPostedId == -1)
2023-01-20 09:28:32 -05:00
tweets = await _twitterTweetsService.GetTimelineAsync(user.Acct);
else
2023-01-20 09:28:32 -05:00
tweets = await _twitterTweetsService.GetTimelineAsync(user.Acct, user.LastTweetSynchronizedForAllFollowersId);
}
catch (Exception e)
{
2021-02-02 22:49:37 -05:00
_logger.LogError(e, "Error retrieving TL of {Username} from {LastTweetPostedId}, purging user from cache", user.Acct, user.LastTweetPostedId);
_twitterUserService.PurgeUser(user.Acct);
}
2020-07-18 23:35:19 -04:00
return tweets;
2020-07-16 01:19:41 -04:00
}
}
}