internal static void DFTextFileLoadDataFrameSample()
{
var requestsSchema = StructType.CreateStructType(
new List <StructField>
{
StructField.CreateStructField("guid", "string", false),
StructField.CreateStructField("datacenter", "string", false),
StructField.CreateStructField("abtestid", "string", false),
StructField.CreateStructField("traffictype", "string", false),
}
);
var requestsDateFrame = GetSqlContext().TextFile(SparkCLRSamples.Configuration.GetInputDataPath(RequestsLog), requestsSchema);
requestsDateFrame.RegisterTempTable("requests");
var guidFilteredDataFrame = GetSqlContext().Sql("SELECT guid, datacenter FROM requests where guid = '4628deca-139d-4121-b540-8341b9c05c2a'");
guidFilteredDataFrame.Show();
requestsDateFrame.ShowSchema();
requestsDateFrame.Show();
var count = requestsDateFrame.Count();
guidFilteredDataFrame.ShowSchema();
guidFilteredDataFrame.Show();
var filteredCount = guidFilteredDataFrame.Count();
if (SparkCLRSamples.Configuration.IsValidationEnabled)
{
Assert.AreEqual(10, count);
Assert.AreEqual(1, filteredCount);
}
}
private static DataFrame GetMetricsDataFrame()
{
var metricsSchema = StructType.CreateStructType(
new List <StructField>
{
StructField.CreateStructField("unknown", "string", false),
StructField.CreateStructField("date", "string", false),
StructField.CreateStructField("time", "string", false),
StructField.CreateStructField("guid", "string", false),
StructField.CreateStructField("lang", "string", false),
StructField.CreateStructField("country", "string", false),
StructField.CreateStructField("latency", "integer", false)
}
);
return
(GetSqlContext()
.TextFile(SparkCLRSamples.Configuration.GetInputDataPath(MetricsLog), metricsSchema));
}
