public override (List <DataCollection>, List <DataEntity>) GetSchema(DataContainer container)
{
var connection = CassandraConnectionString.Parse(container.ConnectionString);
using (var cluster = Cluster.Builder().AddContactPoint(connection.Address).Build()) {
var tables = cluster.Metadata.GetKeyspace(connection.Keyspace).GetTablesMetadata().ToList();
var dataEntities = tables.SelectMany(t =>
GetSchema(connection.Keyspace, t.Name, cluster, container)).ToList();
var dataCollections = tables.Select(t => new DataCollection(container, t.Name)).ToList();
return(dataCollections, dataEntities);
}
}
public override bool TestConnection(DataContainer container)
{
var connection = CassandraConnectionString.Parse(container.ConnectionString);
try {
using (var cluster = Cluster.Builder().AddContactPoint(connection.Address).Build()) {
cluster.Connect();
}
return(true);
}
catch (Exception)
{
return(false);
}
}
public override List <DataCollectionMetrics> GetDataCollectionMetrics(DataContainer container)
{
var dataCollectionMetrics = new List <DataCollectionMetrics>();
var connection = CassandraConnectionString.Parse(container.ConnectionString);
try
{
using (var cluster = Cluster.Builder().AddContactPoint(connection.Address).Build()) {
using (var session = cluster.Connect()) {
string query = $"select * from system_schema.tables where keyspace_name = '{connection.Keyspace}'";
RowSet res = session.Execute(query);
var rows = res.GetRows();
foreach (Row row in rows)
{
var tableName = row.GetValue <string>("table_name");
string queryCnt = $"select count(*) from {connection.Keyspace}.{tableName}";
RowSet resCnt = session.Execute(queryCnt);
var rowCnt = resCnt.First();
dataCollectionMetrics.Add(new DataCollectionMetrics()
{
Name = tableName,
RowCount = (long)rowCnt[0]
}); //TODO: what about space usage?
}
}
}
}
catch (Exception)
{
return(dataCollectionMetrics);
}
return(dataCollectionMetrics);
}
public override List <string> CollectSample(DataEntity dataEntity, int sampleSize)
{
var result = new List <string>();
var connection = CassandraConnectionString.Parse(dataEntity.Container.ConnectionString);
try
{
using (var cluster = Cluster.Builder().AddContactPoint(connection.Address).Build()) {
using (var session = cluster.Connect()) {
session.Execute($"USE {connection.Keyspace}");
string query = $"SELECT COUNT(*) FROM {dataEntity.Collection.Name}";
RowSet res = session.Execute(query);
var rowCount = res.GetRows().First();
long totalRows = (long)rowCount[0];
double pct = 0.1 + (double)sampleSize / (totalRows <= 0 ? sampleSize : totalRows);
var r = new Random();
//var parts = dataEntity.Name.Split(".");
//query = $"SELECT {parts[0]} FROM {dataEntity.Collection.Name}"; //TODO: support UDT types
query = $"SELECT {dataEntity.Name} FROM {dataEntity.Collection.Name}";
res = session.Execute(query);
var rows = res.GetRows();
foreach (Row row in rows)
{
if (dataEntity.DbDataType.Equals(ColumnTypeCode.List.ToString()))
{
var list = row.GetValue <List <string> >(dataEntity.Name);
foreach (var item in list)
{
if (r.NextDouble() < pct)
{
result.Add(item);
}
if (result.Count >= sampleSize)
{
break;
}
}
}
else if (dataEntity.DbDataType.Equals(ColumnTypeCode.Map.ToString()))
{
var list = row.GetValue <IDictionary <string, int> >(dataEntity.Name); //TODO: only map<string,int> type supported????
foreach (var item in list)
{
if (r.NextDouble() < pct)
{
result.Add(item.Key);
}
if (result.Count >= sampleSize)
{
break;
}
}
}
else
{
if (r.NextDouble() < pct)
{
switch (dataEntity.DataType)
{
case DataType.String:
result.Add(row.GetValue <string>(dataEntity.Name));
break;
case DataType.Int:
result.Add(row.GetValue <int>(dataEntity.Name).ToString());
break;
case DataType.Double:
result.Add(row.GetValue <double>(dataEntity.Name).ToString());
break;
case DataType.DateTime:
result.Add(row.GetValue <DateTime>(dataEntity.Name).ToString());
break;
default:
result.Add(row.GetValue <object>(dataEntity.Name).ToString());
break;
}
}
}
if (result.Count >= sampleSize)
{
break;
}
}
}
}
}
catch (Exception)
{
// Nothing. TODO: rethrow or log
}
return(result);
}