예제 #1
0
        /// <inheritdoc cref="IBulkDeduplicationManager.GetDuplicateEntitiesGroups"/>
        public BulkDuplicatesGroupResponse GetDuplicateEntitiesGroups(string entityName, string[] columns,
                                                                      int clientOffset, int clientCount)
        {
            var response = new BulkDuplicatesGroupResponse {
                Groups = new Collection <BulkEntityClientDuplicatesGroup>()
            };
            int totalCountFromBpm = 0;
            var offset            = clientOffset;
            var esq = GetDuplicateEntitiesEsq(entityName);
            var serverToClientColumnNameMap = GetServerToClientColumnNameMap(columns, esq);

            for (int i = 0; i < MaxFetchPages; i++)
            {
                var duplicateRecordGroups = FetchDuplicatesFromService(entityName, offset, clientCount);
                if (duplicateRecordGroups == null)
                {
                    break;
                }
                var duplicatesCount = duplicateRecordGroups.Sum(x => x.Count());
                offset += duplicatesCount;
                foreach (var duplicateRecords in duplicateRecordGroups)
                {
                    Guid groupId           = duplicateRecords.Key;
                    var  duplicateEntities = GetDuplicateEntities(esq, serverToClientColumnNameMap, duplicateRecords.ToList());
                    var  group             = response.Groups.Find(x => x.GroupId == groupId);
                    if (group == null)
                    {
                        group = new BulkEntityClientDuplicatesGroup {
                            GroupId = groupId,
                            Rows    = new EntityCollection()
                        };
                        response.Groups.Add(group);
                    }
                    int missingRecords = clientCount - totalCountFromBpm;
                    int skippedRecords = duplicateEntities.Count - missingRecords;
                    totalCountFromBpm  += duplicateEntities.Count;
                    response.NextOffset = offset - skippedRecords;
                    group.Rows.AddRange(duplicateEntities.Take(missingRecords));
                }
                if (duplicatesCount < clientCount || totalCountFromBpm >= clientCount)
                {
                    break;
                }
            }
            if (response.Groups.IsEmpty())
            {
                return(null);
            }
            AddIsUniqueColumn(esq);
            response.RowConfig         = QueryExtension.GetColumnConfig(esq, serverToClientColumnNameMap);
            response.TotalCountRecords = response.Groups.Sum(x => x.Rows.Count);
            return(response);
        }
예제 #2
0
        /// <inheritdoc cref="IBulkDeduplicationManager.GetGroupsOfDuplicates"/>
        public BulkDuplicatesGroupResponse GetGroupsOfDuplicates(string entityName, string[] columns,
                                                                 int offset, int count, int topDuplicatesPerGroup)
        {
            var response = new BulkDuplicatesGroupResponse {
                Groups = new Collection <BulkEntityClientDuplicatesGroup>()
            };
            var esq = GetDuplicateEntitiesEsq(entityName);
            var serverToClientColumnNameMap = GetServerToClientColumnNameMap(columns, esq);
            var serviceGroupOffset          = offset;
            var serviceGroupCount           = count;

            for (int i = 0; i < MaxFetchPages; i++)
            {
                var duplicateGroupsInfo = FetchGroupOfDuplicatesFromService(entityName, serviceGroupOffset, serviceGroupCount, topDuplicatesPerGroup);
                foreach (var duplicateGroup in duplicateGroupsInfo)
                {
                    var duplicateGroupResponse = GetDuplicatesGroupResponse(duplicateGroup, esq, serverToClientColumnNameMap);
                    if (duplicateGroupResponse == null)
                    {
                        continue;
                    }
                    response.Groups.Add(duplicateGroupResponse);
                }
                var allRowsFetchedFromService = duplicateGroupsInfo.Count < serviceGroupCount;
                if (allRowsFetchedFromService)
                {
                    serviceGroupOffset = -1;
                    break;
                }
                serviceGroupOffset += serviceGroupCount;
                serviceGroupCount   = count - response.Groups.Count;
                var requestedRowsCountFetched = response.Groups.Count >= count;
                if (requestedRowsCountFetched)
                {
                    break;
                }
            }
            AddIsUniqueColumn(esq);
            response.RowConfig  = QueryExtension.GetColumnConfig(esq, serverToClientColumnNameMap);
            response.NextOffset = serviceGroupOffset;
            return(response);
        }