private void DelayedExampleCallback(VowpalWabbitJsonSerializer serializer)
{
try
{
this.perfCounters.Feature_Requests_Pending.IncrementBy(-1);
var data = (PipelineData)serializer.UserContext;
data.Example = serializer.CreateExamples();
// fire and forget
// must not block to avoid dead lock
this.trainProcessorFactory.LearnBlock
.SendAsync(data)
.ContinueWith(async ret =>
{
if (!await ret)
{
this.telemetry.TrackTrace("Unable to enqueue delayed examples", SeverityLevel.Error);
// since we couldn't enqueue, need to dispose here
data.Example.Dispose();
}
});
}
catch (Exception e)
{
this.telemetry.TrackException(e);
}
finally
{
serializer.Dispose();
}
}
public void TestJsonDictReverse()
{
var vec = new float[] { 1, 2, 3 };
var jsonResolver = new RefResolve();
var settings = new JsonSerializerSettings {
ReferenceResolverProvider = () => jsonResolver
};
var ctx1 = new Context(vec, 1, settings);
var ctx2 = new Context(vec, 2, settings);
VowpalWabbitJsonSerializer delayedSerializer = null;
using (var validator = new VowpalWabbitExampleJsonValidator())
using (var vw = new VowpalWabbit(new VowpalWabbitSettings {
EnableStringExampleGeneration = true
}))
using (var resolver = new VowpalWabbitJsonReferenceResolver(serializer => delayedSerializer = serializer))
{
var serializer2 = new VowpalWabbitJsonSerializer(vw, resolver);
var example2 = serializer2.ParseAndCreate(ctx2.JSON);
// incomplete data
Assert.IsNull(example2);
// triggers example2 completion
using (var serializer1 = new VowpalWabbitJsonSerializer(vw, resolver))
using (var example1 = serializer1.ParseAndCreate(ctx1.JSON))
{
validator.Validate("| Id:1 :1 :2 :3", example1);
}
Assert.IsNotNull(delayedSerializer);
using (var delayedExample2 = delayedSerializer.CreateExamples())
{
validator.Validate("| Id:2 :1 :2 :3", delayedExample2);
}
delayedSerializer.Dispose();
}
}
private IEnumerable <PipelineData> Stage1_Deserialize(PipelineData data)
{
try
{
using (var jsonReader = new JsonTextReader(new StringReader(data.JSON)))
{
//jsonReader.FloatParser = Util.ReadDoubleString;
// jsonReader.ArrayPool = pool;
VowpalWabbitJsonSerializer vwJsonSerializer = null;
try
{
vwJsonSerializer = new VowpalWabbitJsonSerializer(this.trainer.VowpalWabbit, this.trainer.ReferenceResolver);
vwJsonSerializer.RegisterExtension((state, property) =>
{
if (TryExtractProperty(state, property, "_eventid", JsonToken.String, reader => data.EventId = (string)reader.Value))
{
return(true);
}
else if (TryExtractProperty(state, property, "_timestamp", JsonToken.Date, reader => data.Timestamp = (DateTime)reader.Value))
{
return(true);
}
else if (TryExtractProperty(state, property, "_ProbabilityOfDrop", JsonToken.Float, reader => data.ProbabilityOfDrop = (float)(reader.Value ?? 0f)))
{
return(true);
}
else if (TryExtractArrayProperty <float>(state, property, "_p", arr => data.Probabilities = arr))
{
return(true);
}
else if (TryExtractArrayProperty <int>(state, property, "_a", arr => data.Actions = arr))
{
return(true);
}
return(false);
});
data.Example = vwJsonSerializer.ParseAndCreate(jsonReader);
if (data.Probabilities == null)
{
throw new ArgumentNullException("Missing probabilities (_p)");
}
if (data.Actions == null)
{
throw new ArgumentNullException("Missing actions (_a)");
}
if (data.Example == null)
{
// unable to create example due to missing data
// will be trigger later
vwJsonSerializer.UserContext = data.Example;
// make sure the serialize is not deallocated
vwJsonSerializer = null;
}
}
finally
{
if (vwJsonSerializer != null)
{
vwJsonSerializer.Dispose();
}
}
performanceCounters.Stage1_JSON_DeserializePerSec.Increment();
// delayed
if (data.Example == null)
{
this.performanceCounters.Feature_Requests_Pending.Increment();
yield break;
}
}
}
catch (Exception ex)
{
this.telemetry.TrackException(ex, new Dictionary <string, string> {
{ "JSON", data.JSON }
});
this.performanceCounters.Stage2_Faulty_Examples_Total.Increment();
this.performanceCounters.Stage2_Faulty_ExamplesPerSec.Increment();
yield break;
}
yield return(data);
}
public void TestJsonDictThreading()
{
var jsonResolver = new RefResolve();
var settings = new JsonSerializerSettings {
ReferenceResolverProvider = () => jsonResolver
};
var rnd = new Random(123);
var examples = new List <Context>();
var id = 0;
// different reference objects
for (int i = 0; i < 10; i++)
{
var data = Enumerable.Range(1, 5).Select(_ => (float)rnd.Next(10)).ToArray();
// referencing the same data
for (int j = 0; j < 5; j++)
{
examples.Add(new Context(data, id++, settings));
}
}
for (int i = 0; i < 4; i++)
{
Permute(examples, rnd);
for (int maxDegreeOfParallelism = 1; maxDegreeOfParallelism < 4; maxDegreeOfParallelism++)
{
var examplesFound = 0;
using (var vw = new VowpalWabbit(new VowpalWabbitSettings {
EnableStringExampleGeneration = true, EnableThreadSafeExamplePooling = true
}))
using (var resolver = new VowpalWabbitJsonReferenceResolver(serializer =>
{
using (var example = serializer.CreateExamples())
{
ValidateExample(example, (Context)serializer.UserContext);
}
serializer.Dispose();
Interlocked.Increment(ref examplesFound);
}))
{
Parallel.ForEach(
Partitioner.Create(0, examples.Count),
new ParallelOptions {
MaxDegreeOfParallelism = maxDegreeOfParallelism
},
range =>
{
for (int j = range.Item1; j < range.Item2; j++)
{
var ctx = examples[j];
var serializer = new VowpalWabbitJsonSerializer(vw, resolver)
{
UserContext = ctx
};
var example = serializer.ParseAndCreate(ctx.JSON);
// example not ready yet
if (example == null)
{
continue;
}
ValidateExample(example, ctx);
example.Dispose();
serializer.Dispose();
Interlocked.Increment(ref examplesFound);
}
});
}
Assert.AreEqual(examples.Count, examplesFound);
}
}
}
private IEnumerable <PipelineData> Stage1_Deserialize(PipelineData data)
{
try
{
using (var jsonReader = new JsonTextReader(new StringReader(data.JSON)))
{
//jsonReader.FloatParser = Util.ReadDoubleString;
// jsonReader.ArrayPool = pool;
VowpalWabbitJsonSerializer vwJsonSerializer = null;
try
{
vwJsonSerializer = new VowpalWabbitJsonSerializer(this.trainer.VowpalWabbit, this.trainer.ReferenceResolver);
vwJsonSerializer.RegisterExtension((state, property) =>
{
if (property.Equals("_eventid", StringComparison.OrdinalIgnoreCase))
{
if (!state.Reader.Read() && state.Reader.TokenType != JsonToken.String)
{
throw new VowpalWabbitJsonException(state.Reader, "Expected string");
}
data.EventId = (string)state.Reader.Value;
return(true);
}
else if (property.Equals("_timestamp", StringComparison.OrdinalIgnoreCase))
{
if (!state.Reader.Read() && state.Reader.TokenType != JsonToken.Date)
{
throw new VowpalWabbitJsonException(state.Reader, "Expected date");
}
data.Timestamp = (DateTime)state.Reader.Value;
}
return(false);
});
data.Example = vwJsonSerializer.ParseAndCreate(jsonReader);
if (data.Example == null)
{
// unable to create example due to missing data
// will be trigger later
vwJsonSerializer.UserContext = data.Example;
// make sure the serialize is not deallocated
vwJsonSerializer = null;
}
}
finally
{
if (vwJsonSerializer != null)
{
vwJsonSerializer.Dispose();
}
}
performanceCounters.Stage1_JSON_DeserializePerSec.Increment();
// delayed
if (data.Example == null)
{
this.performanceCounters.Feature_Requests_Pending.Increment();
yield break;
}
}
}
catch (Exception ex)
{
this.telemetry.TrackException(ex, new Dictionary <string, string> {
{ "JSON", data.JSON }
});
this.performanceCounters.Stage2_Faulty_Examples_Total.Increment();
this.performanceCounters.Stage2_Faulty_ExamplesPerSec.Increment();
yield break;
}
yield return(data);
}
