/// <summary>
/// Send documents to next worker for Update Fields
/// </summary>
/// <param name="categorizedResultDocuments"></param>
private void SendDocumentsForUpdate(IEnumerable<CategorizedDoc> categorizedResultDocuments)
{
var resultDocuments = new List<ProjectDocumentDetail>();
foreach (var categorizedResultDocument in categorizedResultDocuments)
{
var resultDocument = new ProjectDocumentDetail {DocId = categorizedResultDocument.DocId };
var predictedCategory = string.Empty;
double predictedScore = 0;
if (categorizedResultDocument.CategoryResults != null && categorizedResultDocument.CategoryResults.Any())
{
if (!string.IsNullOrEmpty(categorizedResultDocument.CategoryResults[0].CategoryName))
{
predictedCategory = categorizedResultDocument.CategoryResults[0].CategoryName.Equals("R")
? RelevantCategory
: NotRelevantCategory;
}
predictedScore = categorizedResultDocument.CategoryResults[0].Score;
}
resultDocument.PredictedCategory = predictedCategory;
resultDocument.DocumentScore = predictedScore;
resultDocuments.Add(resultDocument);
if (resultDocuments.Count < _batchSize) continue;
//Send documents to pipeline
Send(resultDocuments);
resultDocuments.Clear();
}
if (resultDocuments.Any())
{
//Send documents to pipeline
Send(resultDocuments);
}
}
/// <summary>
/// Get documents for Rerun job
/// </summary>
/// <param name="documents"></param>
/// <returns></returns>
private List<ProjectDocumentDetail> GetDocumentsForRerunJob(List<DocumentIdentifier> documents)
{
var documentStatusList = _analyticProject.GetProjectDocumentsStausFromProcessSet(_jobParameter.MatterId,
_dataset.CollectionId,
_jobParameter.PrimarySystemJobId,
WorkAssignment.JobId, documents.Select(d => d.ReferenceId).ToList());
var projectDocumentDetailList = new List<ProjectDocumentDetail>();
foreach (var document in documents)
{
var resultdocumentStatus =
documentStatusList.FirstOrDefault(d => d.DocumentReferenceId == document.ReferenceId);
if (resultdocumentStatus != null && resultdocumentStatus.SubSystemStatus)
continue; //Document processed earlier and already succeed, then no need to process again
var projectDocumentDetail = new ProjectDocumentDetail
{
DocumentReferenceId =
document.ReferenceId,
TextFilePath = document.Url
};
projectDocumentDetailList.Add(projectDocumentDetail);
}
return projectDocumentDetailList;
}
