public void Validate(string[] lines, string json, IVowpalWabbitLabelComparator labelComparator = null, ILabel label = null, int? index = null, VowpalWabbitJsonExtension extension = null)
{
VowpalWabbitExample[] strExamples = new VowpalWabbitExample[lines.Count()];
try
{
for (int i = 0; i < lines.Length; i++)
strExamples[i] = this.vw.ParseLine(lines[i]);
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw))
{
if (extension != null)
jsonSerializer.RegisterExtension(extension);
using (var jsonExample = (VowpalWabbitMultiLineExampleCollection)jsonSerializer.ParseAndCreate(json, label, index))
{
var jsonExamples = new List<VowpalWabbitExample>();
if (jsonExample.SharedExample != null)
jsonExamples.Add(jsonExample.SharedExample);
jsonExamples.AddRange(jsonExample.Examples);
Assert.AreEqual(strExamples.Length, jsonExamples.Count);
for (int i = 0; i < strExamples.Length; i++)
{
using (var strJsonExample = this.vw.ParseLine(jsonExamples[i].VowpalWabbitString))
{
var diff = strExamples[i].Diff(this.vw, jsonExamples[i], labelComparator);
Assert.IsNull(diff, diff + " generated string: '" + jsonExamples[i].VowpalWabbitString + "'");
diff = strExamples[i].Diff(this.vw, strJsonExample, labelComparator);
Assert.IsNull(diff, diff);
}
}
}
}
}
finally
{
foreach (var ex in strExamples)
if (ex != null)
ex.Dispose();
}
}
public void TestJsonLabelExtraction()
{
using (var vw = new VowpalWabbit("--cb_adf --rank_all"))
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
{
string eventId = null;
jsonSerializer.RegisterExtension((state, property) =>
{
Assert.AreEqual(property, "_eventid");
Assert.IsTrue(state.Reader.Read());
eventId = (string)state.Reader.Value;
return true;
});
jsonSerializer.Parse("{\"_eventid\":\"abc123\",\"a\":1,\"_label_cost\":-1,\"_label_probability\":0.3}");
Assert.AreEqual("abc123", eventId);
using (var examples = jsonSerializer.CreateExamples())
{
var single = examples as VowpalWabbitSingleLineExampleCollection;
Assert.IsNotNull(single);
var label = single.Example.Label as ContextualBanditLabel;
Assert.IsNotNull(label);
Assert.AreEqual(-1, label.Cost);
Assert.AreEqual(0.3, label.Probability, 0.0001);
}
}
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
{
jsonSerializer.Parse("{\"_multi\":[{\"_text\":\"w1 w2\", \"a\":{\"x\":1}}, {\"_text\":\"w2 w3\"}], \"_labelindex\":1, \"_label_cost\":-1, \"_label_probability\":0.3}");
using (var examples = jsonSerializer.CreateExamples())
{
var multi = examples as VowpalWabbitMultiLineExampleCollection;
Assert.IsNotNull(multi);
Assert.AreEqual(2, multi.Examples.Length);
var label = multi.Examples[0].Label as ContextualBanditLabel;
Assert.AreEqual(0, label.Cost);
Assert.AreEqual(0, label.Probability);
label = multi.Examples[1].Label as ContextualBanditLabel;
Assert.IsNotNull(label);
Assert.AreEqual(-1, label.Cost);
Assert.AreEqual(0.3, label.Probability, 0.0001);
}
}
}
}
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;
}
