public void Activate(string processId, ClassifierSettingsElastic settings, CancellationToken token)
{
try
{
GC.Collect();
machineResourceService.UpdateResourcesManually();
var freeMemInBytes = machineResourceService.Status.FreeMemory * 1024 * 1024;
var dictionaryPaths = new List <string>();
foreach (var nGram in settings.ActivatedNGramList)
{
var directoryPath = string.Format("{0}/{1}/{2}", _dictionaryRootPath, settings.ServiceId, nGram);
var fileList = IOHelper.GetFilesInFolder(directoryPath, DictionaryProtoBuf.GetExtension())
.Where(file => settings.ActivatedTagIdList.Contains(Path.GetFileNameWithoutExtension(file)));
dictionaryPaths.AddRange(fileList);
}
var sizeInBytes = dictionaryPaths.Sum(f => new FileInfo(f).Length);
if (freeMemInBytes > 0 && freeMemInBytes < sizeInBytes * Constants.DictionaryInMemoryMultiplier)
{
throw new Common.Exceptions.OutOfResourceException(ServiceResources.NotEnoughResourceToActivateService);
}
var service = serviceQuery.Get(settings.ServiceId);
service.Status = (int)ServiceStatusEnum.Busy;
serviceQuery.Update(service.Id, service);
var lockObject = new object();
var counter = 0;
var allCount = dictionaryPaths.Count;
var deserializedDics = new ConcurrentBag <DictionaryProtoBuf>();
Parallel.ForEach(dictionaryPaths, parallelService.ParallelOptions(), (path, loopState) => {
token.ThrowIfCancellationRequested();
deserializedDics.Add(BaseProtoBuf.DeSerialize <DictionaryProtoBuf>(path));
lock (lockObject)
{
if (++counter % 15 == 0)
{
processHandler.Changed(processId, Math.Round(counter / (double)allCount * 100, 2));
}
}
});
var globalStoreClassifier = new GlobalStoreClassifier();
if (deserializedDics.Any())
{
var scorersDic = deserializedDics.GroupBy(d => d.Id).ToDictionary(d => d.Key, d => new Cerebellum.Scorer.PeSScorer(d.ToDictionary(di => di.NGram, di => di.Dictionary)));
globalStoreClassifier.ClassifierScorers = scorersDic;
}
var tagsDics = settings.Tags.ToDictionary(t => t.Id, t => t);
var tagsDic = settings.Tags.ToDictionary(
t => t.Id,
t => tagService.GetTagModel(settings.DataSetName, t.Id, false, tagsDics)
);
var analyzeQuery = queryFactory.GetAnalyzeQuery(settings.DataSetName);
var emphasizedTagsWords = new Dictionary <string, List <string> >();
foreach (var tagId in settings.EmphasizedTagIdList)
{
var tokens = analyzeQuery.Analyze(tagsDic[tagId].Name, 1).ToList();
emphasizedTagsWords.Add(tagId, tokens);
}
globalStoreClassifier.ClassifierEmphasizedTagIds = emphasizedTagsWords;
globalStoreClassifier.ClassifiersSettings = settings;
globalStoreClassifier.ClassifierTags = tagsDic;
globalStoreClassifier.ClassifierParentTagIds = tagsDic.SelectMany(td => td.Value.Properties.Paths
.Select(p => p.Id)).Distinct().ToDictionary(p => p, p => p);
GlobalStore.ActivatedClassifiers.Add(settings.ServiceId, globalStoreClassifier);
processHandler.Finished(processId, string.Format(ServiceResources.SuccessfullyActivated_0_Service_1, ServiceTypeEnum.Classifier, service.Name));
service.Status = (int)ServiceStatusEnum.Active;
serviceQuery.Update(service.Id, service);
}
catch (Exception ex)
{
var service = serviceQuery.Get(settings.ServiceId);
service.Status = (int)ServiceStatusEnum.Prepared;
serviceQuery.Update(service.Id, service);
if (GlobalStore.ActivatedClassifiers.IsExist(settings.ServiceId))
{
GlobalStore.ActivatedClassifiers.Remove(settings.ServiceId);
}
if (ex.InnerException != null && ex.InnerException is OperationCanceledException)
{
processHandler.Cancelled(processId);
}
else
{
processHandler.Interrupted(processId, ex);
}
GC.Collect();
}
}
public void Add(string id, GlobalStoreClassifier classifier)
{
ClassifierDictionary.Add(id, classifier);
}
