public void CreateExpressionTest1()
{
// Arrange
Expression parameterExpression = Expression.Parameter(typeof(Employee), "sample");
Expression expressionProperty = Expression.Property(parameterExpression, nameof(Employee.Subordinates));
Expression expressionConstant = Expression.Constant(2);
MethodInfo methodInfo = typeof(Enumerable).GetMethods()
.First(method => method.Name == "Count" && method.GetParameters().Length == 1)
.MakeGenericMethod(typeof(object));
var expressionCount = Expression.Call(methodInfo, expressionProperty);
Expression expectedExpression = Expression.Equal(expressionCount, expressionConstant);
var _operator = new CountOperator();
// Act
Expression resultExpression = _operator.CreateExpression(expressionProperty, expressionConstant);
// Assert
Assert.AreEqual(expectedExpression.ToString(), resultExpression.ToString(), "Expression not correct");
}
public IActionResult Execute([FromBody] string logicalPlanJson)
{
if (host == null)
{
host = SiloWrapper.Instance.host;
}
if (client == null)
{
client = ClientWrapper.Instance.client;
}
Stream req = Request.Body;
req.Seek(0, System.IO.SeekOrigin.Begin);
string json = new StreamReader(req).ReadToEnd();
//apply TPC-H Query-1
//json="{\"logicalPlan\":{\"operators\":[{\"tableName\":\"<file>\",\"operatorID\":\"operator-3bc05014-357d-45f5-a053-9baf1a62bd27\",\"operatorType\":\"ScanSource\"},{\"attributeName\":\"_c10\",\"attributeType\":\"date\",\"comparisonType\":\">\",\"compareTo\":\"1991-01-01\",\"operatorID\":\"operator-345a8b3d-2b1b-485e-b7c1-1cd9f579ce4f\",\"operatorType\":\"Comparison\"},{\"groupByAttribute\":\"_c8\",\"aggregationAttribute\":\"_c4\",\"aggregationFunction\":\"sum\",\"operatorID\":\"operator-89854b72-7c28-436b-b162-ead3daa75f72\",\"operatorType\":\"GroupBy\"},{\"attributeName\":\"_c0\",\"attributeType\":\"string\",\"operatorID\":\"operator-c7d7e79c-49ca-46a4-8420-490c25cd052d\",\"operatorType\":\"InsertionSort\"}],\"links\":[{\"origin\":\"operator-3bc05014-357d-45f5-a053-9baf1a62bd27\",\"destination\":\"operator-345a8b3d-2b1b-485e-b7c1-1cd9f579ce4f\"},{\"origin\":\"operator-345a8b3d-2b1b-485e-b7c1-1cd9f579ce4f\",\"destination\":\"operator-89854b72-7c28-436b-b162-ead3daa75f72\"},{\"origin\":\"operator-89854b72-7c28-436b-b162-ead3daa75f72\",\"destination\":\"operator-c7d7e79c-49ca-46a4-8420-490c25cd052d\"}]},\"workflowID\":\"texera-workflow-824ec494-8f6c-41a3-a3c0-29ca6dc7fe97\"}";
Console.WriteLine("JSON BODY = " + json);
Dictionary <string, Operator> map = new Dictionary <string, Operator>();
JObject o = JObject.Parse(json);
JArray operators = (JArray)o["logicalPlan"]["operators"];
Guid workflowID;
//remove "texera-workflow-" at the begining of workflowID to make it parsable
if (!Guid.TryParse(o["workflowID"].ToString().Substring(16), out workflowID))
{
throw new Exception($"Parse workflowID failed! For {o["workflowID"].ToString().Substring(16)}");
}
Workflow workflow = new Workflow(workflowID);
int table_id = 0;
foreach (JObject operator1 in operators)
{
Operator op = null;
if ((string)operator1["operatorType"] == "ScanSource")
{
//example path to HDFS through WebHDFS API: "http://localhost:50070/webhdfs/v1/input/very_large_input.csv"
ScanPredicate scanPredicate = new ScanPredicate((string)operator1["tableName"], table_id++);
op = new ScanOperator(scanPredicate);
}
else if ((string)operator1["operatorType"] == "KeywordMatcher")
{
KeywordPredicate keywordPredicate = new KeywordPredicate(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["keyword"] != null?operator1["keyword"].ToString():"");
op = new KeywordOperator(keywordPredicate);
}
else if ((string)operator1["operatorType"] == "Aggregation")
{
CountPredicate countPredicate = new CountPredicate();
op = new CountOperator(countPredicate);
}
else if ((string)operator1["operatorType"] == "Comparison")
{
FilterPredicate filterPredicate = new FilterPredicate(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["compareTo"].ToString(), operator1["comparisonType"].ToString());
switch (operator1["attributeType"].ToString())
{
case "int":
op = new FilterOperator <int>(filterPredicate);
break;
case "double":
op = new FilterOperator <double>(filterPredicate);
break;
case "date":
op = new FilterOperator <DateTime>(filterPredicate);
break;
case "string":
op = new FilterOperator <string>(filterPredicate);
break;
}
}
else if ((string)operator1["operatorType"] == "CrossRippleJoin")
{
int inputLimit = operator1["batchingLimit"] == null?1000:int.Parse(operator1["batchingLimit"].ToString());
CrossRippleJoinPredicate crossRippleJoinPredicate = new CrossRippleJoinPredicate(table_id++, inputLimit);
op = new CrossRippleJoinOperator(crossRippleJoinPredicate);
}
else if ((string)operator1["operatorType"] == "HashRippleJoin")
{
HashRippleJoinPredicate hashRippleJoinPredicate = new HashRippleJoinPredicate(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), table_id++);
op = new HashRippleJoinOperator(hashRippleJoinPredicate);
}
else if ((string)operator1["operatorType"] == "InsertionSort")
{
SortPredicate sortPredicate = new SortPredicate(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")));
switch (operator1["attributeType"].ToString())
{
case "int":
op = new SortOperator <int>(sortPredicate);
break;
case "double":
op = new SortOperator <double>(sortPredicate);
break;
case "date":
op = new SortOperator <DateTime>(sortPredicate);
break;
case "string":
op = new SortOperator <string>(sortPredicate);
break;
}
}
else if ((string)operator1["operatorType"] == "GroupBy")
{
int groupByIndex = int.Parse(operator1["groupByAttribute"].ToString().Replace("_c", ""));
int aggregationIndex = int.Parse(operator1["aggregationAttribute"].ToString().Replace("_c", ""));
GroupByPredicate groupByPredicate = new GroupByPredicate(groupByIndex, aggregationIndex, operator1["aggregationFunction"].ToString());
op = new GroupByOperator(groupByPredicate);
}
else if ((string)operator1["operatorType"] == "Projection")
{
List <int> projectionIndexs = operator1["projectionAttributes"].ToString().Split(",").Select(x => int.Parse(x.Replace("_c", ""))).ToList();
ProjectionPredicate projectionPredicate = new ProjectionPredicate(projectionIndexs);
op = new ProjectionOperator(projectionPredicate);
}
else if ((string)operator1["operatorType"] == "HashJoin")
{
HashJoinPredicate hashJoinPredicate = new HashJoinPredicate(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), table_id++);
op = new HashJoinOperator(hashJoinPredicate);
}
if (op != null)
{
map.Add((string)operator1["operatorID"], op);
}
}
JArray links = (JArray)o["logicalPlan"]["links"];
foreach (JObject link in links)
{
Operator origin = map[(string)link["origin"]];
Operator dest = map[(string)link["destination"]];
origin.AddOutOperator(dest);
dest.AddInOperator(origin);
}
workflow.InitializeOperatorSet(new HashSet <Operator>(map.Values));
List <TexeraTuple> results = ClientWrapper.Instance.DoClientWork(client, workflow).Result;
if (results == null)
{
results = new List <TexeraTuple>();
}
List <JObject> resultJson = new List <JObject>();
foreach (TexeraTuple tuple in results)
{
JObject jsonTuple = new JObject();
jsonTuple.Add("TableID", tuple.TableID);
for (int i = 0; i < tuple.FieldList.Length; ++i)
{
jsonTuple.Add("_c" + i, tuple.FieldList[i]);
}
resultJson.Add(jsonTuple);
}
TexeraResult texeraResult = new TexeraResult();
texeraResult.code = 0;
texeraResult.result = resultJson;
texeraResult.resultID = Guid.NewGuid();
return(Json(texeraResult));
}
private void ApplyLogicalPlan(JObject plan)
{
Console.WriteLine(plan);
JArray operators = (JArray)plan["operators"];
int batchSize = Constants.BatchSize;
JArray jsonLinks = (JArray)plan["links"];
foreach (JObject link in jsonLinks)
{
Guid origin = Guid.Parse(link["origin"].ToString().Substring(9));
Guid dest = Guid.Parse(link["destination"].ToString().Substring(9));
if (forwardLinks.ContainsKey(origin))
{
forwardLinks[origin].Add(dest);
}
else
{
forwardLinks[origin] = new List <Guid> {
dest
};
}
if (backwardLinks.ContainsKey(dest))
{
backwardLinks[dest].Add(origin);
}
else
{
backwardLinks[dest] = new List <Guid> {
origin
};
}
}
foreach (JObject operator1 in operators)
{
Operator op = null;
Guid operatorID = Guid.Parse(operator1["operatorID"].ToString().Substring(9));
if ((string)operator1["operatorType"] == "ScanSource")
{
//example path to HDFS through WebHDFS API: "http://localhost:50070/webhdfs/v1/input/very_large_input.csv"
HashSet <int> idxes = null;
if (((string)operator1["tableName"]).Contains("customer"))
{
idxes = new HashSet <int> {
0
};
}
else if (((string)operator1["tableName"]).Contains("orders"))
{
idxes = new HashSet <int> {
0, 1
};
}
else if (((string)operator1["tableName"]).Contains("lineitem"))
{
idxes = new HashSet <int> {
4, 8, 10
};
}
else if (((string)operator1["tableName"]).Contains("large_input"))
{
idxes = new HashSet <int> {
0, 1
};
}
op = new ScanOperator((string)operator1["tableName"], idxes);
}
else if ((string)operator1["operatorType"] == "LocalScanSource")
{
op = new LocalFileScanOperator((string)operator1["fileName"]);
}
else if ((string)operator1["operatorType"] == "LocalMaterializer")
{
op = new LocalMaterializerOperator(operatorID);
}
else if ((string)operator1["operatorType"] == "KeywordMatcher")
{
op = new KeywordOperator(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["keyword"] != null?operator1["keyword"].ToString():"");
}
else if ((string)operator1["operatorType"] == "Aggregation")
{
op = new CountOperator();
}
else if ((string)operator1["operatorType"] == "Comparison")
{
switch (operator1["attributeType"].ToString())
{
case "int":
op = new FilterOperator <int>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["compareTo"].ToString(), operator1["comparisonType"].ToString());
break;
case "double":
op = new FilterOperator <double>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["compareTo"].ToString(), operator1["comparisonType"].ToString());
break;
case "date":
op = new FilterOperator <DateTime>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["compareTo"].ToString(), operator1["comparisonType"].ToString());
break;
case "string":
op = new FilterOperator <string>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")), operator1["compareTo"].ToString(), operator1["comparisonType"].ToString());
break;
}
}
else if ((string)operator1["operatorType"] == "CrossRippleJoin")
{
int innerIndex = int.Parse(operator1["innerTableAttribute"].ToString().Replace("_c", ""));
int outerIndex = int.Parse(operator1["outerTableAttribute"].ToString().Replace("_c", ""));
AddStartDenpendency(GetSource(backwardLinks[operatorID][1]), GetSource(backwardLinks[operatorID][0]));
op = new CrossRippleJoinOperator(innerIndex, outerIndex, backwardLinks[operatorID][0]);
}
else if ((string)operator1["operatorType"] == "HashRippleJoin")
{
int innerIndex = int.Parse(operator1["innerTableAttribute"].ToString().Replace("_c", ""));
int outerIndex = int.Parse(operator1["outerTableAttribute"].ToString().Replace("_c", ""));
AddStartDenpendency(GetSource(backwardLinks[operatorID][1]), GetSource(backwardLinks[operatorID][0]));
op = new HashRippleJoinOperator(innerIndex, outerIndex, backwardLinks[operatorID][0]);
}
else if ((string)operator1["operatorType"] == "InsertionSort")
{
switch (operator1["attributeType"].ToString())
{
case "int":
op = new SortOperator <int>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")));
break;
case "double":
op = new SortOperator <double>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")));
break;
case "date":
op = new SortOperator <DateTime>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")));
break;
case "string":
op = new SortOperator <string>(int.Parse(operator1["attributeName"].ToString().Replace("_c", "")));
break;
}
}
else if ((string)operator1["operatorType"] == "GroupBy")
{
int groupByIndex = int.Parse(operator1["groupByAttribute"].ToString().Replace("_c", ""));
int aggregationIndex = int.Parse(operator1["aggregationAttribute"].ToString().Replace("_c", ""));
op = new GroupByOperator(groupByIndex, aggregationIndex, operator1["aggregationFunction"].ToString());
}
else if ((string)operator1["operatorType"] == "GroupByFinal")
{
op = new GroupByFinalOperator(operator1["aggregationFunction"].ToString());
}
else if ((string)operator1["operatorType"] == "Projection")
{
List <int> projectionIndexs = operator1["projectionAttributes"].ToString().Split(",").Select(x => int.Parse(x.Replace("_c", ""))).ToList();
op = new ProjectionOperator(projectionIndexs);
}
else if ((string)operator1["operatorType"] == "HashJoin")
{
int innerIndex = int.Parse(operator1["innerTableAttribute"].ToString().Replace("_c", ""));
int outerIndex = int.Parse(operator1["outerTableAttribute"].ToString().Replace("_c", ""));
AddStartDenpendency(GetSource(backwardLinks[operatorID][1]), GetSource(backwardLinks[operatorID][0]));
op = new HashJoinOperator(innerIndex, outerIndex, backwardLinks[operatorID][0]);
}
else if ((string)operator1["operatorType"] == "SentimentAnalysis")
{
int predictIndex = int.Parse(operator1["targetAttribute"].ToString().Replace("_c", ""));
op = new SentimentAnalysisOperator(predictIndex);
}
else if ((string)operator1["operatorType"] == "HashBasedMaterializer")
{
Guid hasherID = Guid.Parse(operator1["hasherID"].ToString().Substring(9));
Guid inputID = Guid.Parse(operator1["inputID"].ToString().Substring(9));
string hashFunc = nodeMetadata[hasherID].GetHashFunctionAsString(inputID);
op = new HashBasedMaterializerOperator(operatorID, Constants.DefaultNumGrainsInOneLayer, hashFunc);
}
else if ((string)operator1["operatorType"] == "HashBasedFolderScan")
{
op = new HashBasedFolderScanOperator(Constants.WebHDFSEntry + "/amber-tmp/" + operator1["folderRoot"], Constants.DefaultNumGrainsInOneLayer);
}
if (op != null)
{
nodeMetadata.Add(operatorID, op);
}
else
{
Console.WriteLine("operator implementation for " + (string)operator1["operatorType"] + " not found");
}
}
}
public override T VisitCountOperator(CountOperator node)
{
throw new NotImplementedException();
}
