protected override Task Execute(CancellationToken token)
{
log.LogInformation("Downloading message...");
var downloadMessages = File.ReadLines(config.Ids).Select(long.Parse).ToArray();
log.LogInformation("Total messages to download: {0}", downloadMessages.Length);
var cred = auth.Authenticate();
var extractor = new MessageCleanup();
var monitor = new PerformanceMonitor(downloadMessages.Length);
using (var streamWriter = new StreamWriter(config.Out, false, new UTF8Encoding(false)))
using (var csvDataTarget = new CsvWriter(streamWriter))
{
csvDataTarget.WriteField("Id");
csvDataTarget.WriteField("Date");
csvDataTarget.WriteField("Author");
csvDataTarget.WriteField("Message");
csvDataTarget.NextRecord();
Auth.ExecuteOperationWithCredentials(
cred,
() =>
{
using (Observable.Interval(TimeSpan.FromSeconds(30)).Subscribe(item => log.LogInformation(monitor.ToString())))
{
downloader.Download(downloadMessages)
.ToObservable()
.Select(
item =>
{
try
{
csvDataTarget.WriteField(item.Id);
csvDataTarget.WriteField(item.CreatedAt);
csvDataTarget.WriteField(item.CreatedBy.Id);
var text = item.Text;
if (config.Clean)
{
text = extractor.Cleanup(text);
}
csvDataTarget.WriteField(text);
csvDataTarget.NextRecord();
monitor.Increment();
}
catch (Exception e)
{
log.LogError(e, "Error");
}
return(item);
})
.LastOrDefaultAsync()
.Wait();
}
});
}
return(Task.CompletedTask);
}
private async Task Process(string[] files)
{
var monitor = new PerformanceMonitor(files.Length);
using (Observable.Interval(TimeSpan.FromSeconds(30)).Subscribe(item => log.LogInformation(monitor.ToString())))
{
var inputBlock = new BufferBlock <ProcessingChunk <string> >(new DataflowBlockOptions {
BoundedCapacity = 1000000
});
var deserializeBlock = new TransformBlock <ProcessingChunk <string>, ProcessingChunk <TweetDTO> >(
json => new ProcessingChunk <TweetDTO>(json.FileName, json.ChunkId, json.TotalChunks, jsonConvert.DeserializeObject <TweetDTO>(json.Data)),
new ExecutionDataflowBlockOptions
{
BoundedCapacity = 2,
MaxDegreeOfParallelism = Environment.ProcessorCount
});
var outputBlock = new ActionBlock <ProcessingChunk <TweetDTO> >(
Deserialized,
new ExecutionDataflowBlockOptions {
MaxDegreeOfParallelism = Environment.ProcessorCount
});
inputBlock.LinkTo(deserializeBlock, new DataflowLinkOptions {
PropagateCompletion = true
});
deserializeBlock.LinkTo(outputBlock, new DataflowLinkOptions {
PropagateCompletion = true
});
foreach (var file in files)
{
try
{
var data = fileLoader.Load(file);
for (var i = 0; i < data.Length; i++)
{
await inputBlock.SendAsync(new ProcessingChunk <string>(file, i, data.Length, data[i])).ConfigureAwait(false);
}
monitor.Increment();
}
catch (Exception ex)
{
log.LogError(ex, "Failed");
}
}
inputBlock.Complete();
await Task.WhenAll(inputBlock.Completion, outputBlock.Completion).ConfigureAwait(false);
}
}
protected override Task Execute(CancellationToken token)
{
log.LogInformation("Starting twitter monitoring...");
SetupWords();
RateLimit.RateLimitTrackerMode = RateLimitTrackerMode.TrackAndAwait;
var cleanup = new MessageCleanup();
var monitor = new PerformanceMonitor(100000);
var cred = auth.Authenticate();
using (Observable.Interval(TimeSpan.FromSeconds(30)).Subscribe(item => log.LogInformation(monitor.ToString())))
using (var streamWriter = new StreamWriter(config.Out, true, new UTF8Encoding(false)))
using (var csvDataTarget = new CsvWriter(streamWriter))
{
Auth.ExecuteOperationWithCredentials(
cred,
() =>
{
var enrichments = Enrichment().ToArray();
foreach (var enrichment in enrichments)
{
enrichment.Discovery.BatchSize = 5;
enrichment.Discovery.AddProcessed(enrichments.SelectMany(p => p.Discovery.Processed).ToArray());
enrichment.Discovery.Process()
.ToObservable()
.ObserveOn(TaskPoolScheduler.Default)
.Select(
x =>
{
var text = cleanup.Cleanup(x.Message.Text).Replace("\r\n", " ");
if (!CanInclude(text, enrichment.Type))
{
return(x);
}
text = Regex.Replace(text, @"[^\u0000-\u007F]+", string.Empty);
csvDataTarget.WriteField(x.Message.Id);
csvDataTarget.WriteField(x.Topic);
csvDataTarget.WriteField(enrichment.Type);
csvDataTarget.WriteField(text);
csvDataTarget.NextRecord();
streamWriter.Flush();
monitor.Increment();
return(x);
})
.Wait();
}
});
}
return(Task.CompletedTask);
}
public void Construct()
{
Assert.AreEqual("Processed: 0/9 Operations per second: 0", instance.ToString());
for (int i = 0; i < 5; i++)
{
instance.ManualyCount();
instance.Increment();
}
instance.Increment();
Assert.AreEqual("Processed: 6/9 Opera", instance.ToString().Substring(0, 20));
for (int i = 0; i < 10; i++)
{
instance.ManualyCount();
}
for (int i = 0; i < 10; i++)
{
instance.Increment();
}
Assert.AreEqual("Processed: 16/16 Ope", instance.ToString().Substring(0, 20));
}
public async Task Process(IConnectionContext target, SentimentMessage message, CancellationToken token)
{
if (message == null)
{
throw new ArgumentNullException(nameof(message));
}
var request = message.Request;
if (request?.Documents == null)
{
throw new Exception("Nothing to process");
}
if (request.Documents.Length > 500)
{
throw new Exception("Too many documents. Maximum is 500");
}
var completed = new CompletedMessage();
try
{
var monitor = new PerformanceMonitor(request.Documents.Length);
using (Observable.Interval(TimeSpan.FromSeconds(10))
.Subscribe(item => logger.LogInformation(monitor.ToString())))
{
ISentimentDataHolder lexicon = default;
if (request.Dictionary != null &&
request.Dictionary.Count > 0)
{
logger.LogInformation("Creating custom dictionary with {0} words", request.Dictionary.Count);
lexicon = SentimentDataHolder.Load(request.Dictionary.Select(item =>
new WordSentimentValueData(
item.Key,
new SentimentValueData(item.Value))));
}
if ((lexicon == null || request.AdjustDomain) &&
!string.IsNullOrEmpty(request.Domain))
{
logger.LogInformation("Using Domain dictionary [{0}]", request.Domain);
var previous = lexicon;
lexicon = lexiconLoader.GetLexicon(request.Domain);
if (previous != null)
{
lexicon.Merge(previous);
}
}
string modelLocation = null;
if (!string.IsNullOrEmpty(request.Model))
{
logger.LogInformation("Using model path: {0}", request.Model);
modelLocation = storage.GetLocation(target.Connection.User, request.Model, ServiceConstants.Model);
if (!Directory.Exists(modelLocation))
{
throw new ApplicationException($"Can't find model {request.Model}");
}
}
using (var scope = provider.CreateScope())
{
var container = scope.ServiceProvider.GetService <ISessionContainer>();
container.Context.NGram = 3;
container.Context.ExtractAttributes = request.Emotions;
var client = container.GetTesting(modelLocation);
var converter = scope.ServiceProvider.GetService <IDocumentConverter>();
client.Init();
client.Pipeline.ResetMonitor();
if (lexicon != null)
{
client.Lexicon = lexicon;
}
await client.Process(request.Documents.Select(item => converter.Convert(item, request.CleanText))
.ToObservable())
.Select(item =>
{
monitor.Increment();
return(item);
})
.Buffer(TimeSpan.FromSeconds(5), 10, scheduler)
.Select(async item =>
{
var result = new ResultMessage <Document> {
Data = item.Select(x => x.Processed).ToArray()
};
await target.Write(result, token).ConfigureAwait(false);
return(Unit.Default);
})
.Merge();
}
logger.LogInformation("Completed with final performance: {0}", monitor);
completed.Message = "Testing Completed";
await target.Write(completed, token).ConfigureAwait(false);
}
}
catch (Exception e)
{
completed.Message = e.Message;
await target.Write(completed, token).ConfigureAwait(false);
completed.IsError = true;
throw;
}
}
