/// <summary>
/// This method helps to get all the Duplicate counts in a collection whose hash value count>0 (which is combination of original docs+ duplicate docs)
/// </summary>
private List <DocumentHashMapEntity> GetDocumentHashValues(IEnumerable <DeduplicationDatasetBEO> lsDataset,
string algorithm, string compareType)
{
List <DocumentHashMapEntity> lsDocumentHash = null;
try
{
var lsCollectionId = new List <string>();
foreach (var datasetInfo in lsDataset)
{
lsCollectionId.Add(datasetInfo.CollectionId);
if (_matterId == string.Empty)
{
_matterId = datasetInfo.MatterId.ToString(CultureInfo.InvariantCulture);
}
}
var hashTypeId = GetHashTypeId(algorithm, compareType);
_vault = VaultRepository.CreateRepository(Convert.ToInt64(_matterId));
var lsCollectionTemp = lsCollectionId.Select(collection => new Guid(collection)).ToList();
if (lsCollectionId.Count > 1)
{
lsDocumentHash = _vault.GetDocumentDuplicatesAcrossCollections(lsCollectionTemp, hashTypeId);
}
else
{
DataSetCollectionId = lsCollectionTemp[0];
lsDocumentHash = _vault.GetDocumentDuplicatesByHashValue(lsCollectionTemp[0], hashTypeId,
string.Empty);
}
GetOriginalDocuments(lsCollectionId);
}
catch (Exception ex)
{
EvLog.WriteEntry(_jobid + Constants.EVENT_GENERATE_TASK_KEY,
Constants.ErrorForDocumentHashValues + ex.Message);
LogException(_jobid, ex, Constants.ErrorForOriginalDocuments, LogCategory.Job, string.Empty,
ErrorCodes.ProblemInGenerateTasks);
}
return(lsDocumentHash);
}
