在.net中使用aquiles访问Cassandra(二)

上文中我们已经建立了项目的基本结构，今天实现数据的修改。在NoSQL中，通常添加和修改都认为是对数据的一种Mutation。

 

1.建立描述修改Row的实体。

    public class RowMutation
    {
        public string Key { get; set; }

        public IList<CellMutation> Mutations { get; set; }

        public RowMutation()
        {

        }

        public RowMutation(string key)
        {
            Key = key;
        }

        public RowMutation(string key, IList<CellMutation> mutations)
        {
            Key = key;
            Mutations = mutations;
        }
    }

    public class CellMutation
    {
        public string ColumnName { get; set; }

        public string DataValue { get; set; }

        public CellMutation()
        {

        }

        public CellMutation(string name)
        {
            ColumnName = name;
        }

        public CellMutation(string name, string value)
        {
            ColumnName = name;
            DataValue = value;
        }
    }

2.引用aquiles命名空间。

using Aquiles.Helpers.Encoders;
using Aquiles.Cassandra10;
using Aquiles.Core.Cluster;

3.初始化成员变量

        private string _clusterName { get; set; }
        private ConsistencyLevel _consistencyLevel { get; set; }
        private string _keyspaceName { get; set; }
        private ICluster _cluster = null;

        public AquilesDemo()
        {
            _clusterName = "xxxx";
            _consistencyLevel = ConsistencyLevel.LOCAL_QUORUM;
           _keyspaceName = "xxxx";

            _cluster = AquilesHelper.RetrieveCluster(_clusterName);
        }

4.建立一个通用的修改多行多列的方法，其它改变方式可以在此基础上完成调用，如修改一多一列，一行多列等。

       public void Mutate(string columnFamily, IList<RowMutation> rowMutations)
        {
            if (string.IsNullOrWhiteSpace(columnFamily)) throw new ArgumentNullException("columnFamily");

            rowMutations = PrepareMutationList(rowMutations);
            Dictionary<byte[], Dictionary<string, List<Apache.Cassandra.Mutation>>> mutation_map = new Dictionary<byte[], Dictionary<string, List<Apache.Cassandra.Mutation>>>();

            foreach (var rowMutation in rowMutations)
            {
                byte[] key = ByteEncoderHelper.UTF8Encoder.ToByteArray(rowMutation.Key);

                Dictionary<string, List<Apache.Cassandra.Mutation>> cfMutation = new Dictionary<string, List<Apache.Cassandra.Mutation>>();
                List<Apache.Cassandra.Mutation> mutationList = new List<Apache.Cassandra.Mutation>();

                foreach (CellMutation cellMutation in rowMutation.Mutations)
                {
                    if (cellMutation.DataValue == null) continue;

                    Apache.Cassandra.Mutation mutation = new Apache.Cassandra.Mutation()
                    {
                        Column_or_supercolumn = new ColumnOrSuperColumn()
                        {
                            Column = new Column()
                            {
                                Name = ByteEncoderHelper.UTF8Encoder.ToByteArray(cellMutation.ColumnName),
                                Timestamp = DateTime.Now.ToUnixTimestamp(),
                                Value = ByteEncoderHelper.UTF8Encoder.ToByteArray(cellMutation.DataValue),
                            },
                        },
                    };
                    mutationList.Add(mutation);
                }

                if (mutationList.Count > 0)
                {
                    cfMutation.Add(columnFamily, mutationList);
                    mutation_map.Add(key, cfMutation);
                }
            }

            if (mutation_map.Count == 0) return;
            _cluster.Execute(new ExecutionBlock(delegate(Apache.Cassandra.Cassandra.Client client)
            {
                client.batch_mutate(mutation_map, _consistencyLevel);
                return null;
            }), _keyspaceName);

        }

 5. 由于调用端可能会将同一行的修改分为多个RowMutation传入，所以上面的代码使用PrepareMutationList方法根据RowKey作了聚合处理。

        /// <summary>
        /// 聚合列表中相同rowkey的项为同一Mutation。
        /// </summary>
        /// <param name="rowMutations"></param>
        /// <returns></returns>
        private IList<RowMutation> PrepareMutationList(IList<RowMutation> rowMutations)
        {
            if (rowMutations == null) return null;

            //按rowkey分组
            var rowMutationGroups = rowMutations.GroupBy(rm => rm.Key);

            //分组后的总数量相同，则认为rowkey在列表中是唯一的，不用做聚合处理
            if (rowMutations.Count == rowMutationGroups.Count()) return rowMutations;

            //遍历分组结果
            IList<RowMutation> result = new List<RowMutation>();
            foreach (var rmg in rowMutationGroups)
            {
                RowMutation rowMutation = new RowMutation();
                rowMutation.Key = rmg.Key;

                //将同一分组的所有CellMutation放在一个列表中
                List<CellMutation> cellMutations = new List<CellMutation>();
                foreach (var rm in rmg)
                {
                    cellMutations.AddRange(rm.Mutations);
                }

                if (cellMutations.Count() > 0)
                {
                    rowMutation.Mutations = cellMutations;
                    result.Add(rowMutation);
                }
            }
            return result;
        }