public void TidyExtractor_InvalidFilePaths_ThrowsArgumentException()
{
// Arrange
TidyDataCsvExtractor sut;
// Act and Assert
Assert.Throws <ArgumentException>(() => sut = new TidyDataCsvExtractor(
@"Nodes/Manual/Assets/foo.csv",
pathToFileWithValidDictionaryV1));
}
public void Extract_Nulls_ReturnsAllValues()
{
// Arrange
TidyDataCsvExtractor sut = new TidyDataCsvExtractor(
pathToFileWithNullsV1,
pathToFileWithValidDictionaryV1);
int expectedDocs = 10;
// Act
TidyData actual = sut.Extract <HandHarvestYieldV1>();
// Assert
Assert.Equal(expectedDocs, actual.Observations.Count);
}
public void Extract_ValidData_ReturnsExpected()
{
// Arrange
TidyDataCsvExtractor sut = new TidyDataCsvExtractor(
pathToFileWithValidDataSlimV1,
pathToFileWithValidDictionaryV1);
TidyData expected = ManualArranger
.GetTidyDataDerivedFromActualDataV1();
// Act
TidyData actual = sut.Extract <HandHarvestYieldV1>();
// Assert
Assert.Equal(expected, actual);
}
public async Task SoilGridPointSurveyV1ToCosmos_ActualData_CreatesExpectedRecords()
{
// Arrange
var extractor = new TidyDataCsvExtractor(
pathToFileWithValidSoilGridPointSurveyV1Data,
pathToFileWithValidSoilGridPointSurveyV1Dictionary);
EtlEvent etlEvent = new EtlEvent(
"EtlEvent",
"LocalProcess",
"http://files.cafltar.org/data/schema/documentDb/v2/etlEvent.json",
"CookEastSoilGridPointSurvey",
"0.1",
"",
DateTime.UtcNow);
var transformer = new CosmosDBSqlApiSampleV2Transformer
<SoilGridPointSurveyV1, SoilSample>(
new MapFromSoilGridPointSurveyToSoilSample(),
"http://files.cafltar.org/data/schema/documentDb/v2/sample.json",
etlEvent.Id,
"CookEastSoilGridPointSurvey",
"CookEast",
"SoilSample");
var loader = new DocumentLoader(
client,
"cafdb",
"items");
// Act
TidyData extracted = extractor.Extract <SoilGridPointSurveyV1>();
List <SoilSample> transformed = transformer.Transform(extracted);
StoredProcedureResponse <bool>[] results = await loader.LoadBulk(transformed);
Assert.Equal(30, transformed.Count);
Assert.NotEmpty(results);
}
static async Task MainAsync()
{
EtlEvent etlEvent = new EtlEvent(
"EtlEvent",
"LocalProcess",
"http://files.cafltar.org/data/schema/documentDb/v2/etlEvent.json",
"CookEastSoilGridPointSurvey",
"1.0",
"CookEastSoilGridPointSurvey_DotNet_SoilGridPointToCosmosDB",
DateTime.UtcNow);
var builder = new ConfigurationBuilder()
.SetBasePath(Directory.GetCurrentDirectory())
.AddJsonFile("appsettings.json");
var configuration = builder.Build();
JsonSerializerSettings serializerSettings = new JsonSerializerSettings
{
NullValueHandling = NullValueHandling.Ignore
};
string data = configuration["PathToData"];
string dict = configuration["PathToDictionary"];
if (!File.Exists(data) |
!File.Exists(dict))
{
throw new FileNotFoundException();
}
etlEvent.Inputs.Add(data);
etlEvent.Inputs.Add(dict);
DocumentClient client;
try
{
client = new DocumentClient(
new Uri(
configuration["CosmosServiceEndpoint"]),
configuration["CosmosAuthKey"],
serializerSettings);
}
catch (Exception e)
{
etlEvent.Logs.Add(
$"Error creating DocumentClient: {e.Message}");
throw new Exception("Error creating DocumentClient", e);
}
var extractor = new TidyDataCsvExtractor(
configuration["PathToData"],
configuration["PathToDictionary"]);
var transformer = new CosmosDBSqlApiSampleV2Transformer
<SoilGridPointSurveyV1, SoilSample>(
new MapFromSoilGridPointSurveyToSoilSample(),
"http://files.cafltar.org/data/schema/documentDb/v2/sample.json",
etlEvent.Id,
"CookEastSoilGridPointSurvey",
"CookEast",
"SoilSample");
var loader = new DocumentLoader(
client,
"cafdb",
"items");
try
{
TidyData extracted = extractor.Extract <SoilGridPointSurveyV1>();
etlEvent.Logs.Add(
$"Extracted TidyData with {extracted.Observations.Count} observations");
List <SoilSample> transformed = transformer.Transform(extracted);
etlEvent.Logs.Add(
$"Transformed TidyData to {transformed.Count} SoilSamples");
int docsLoaded = 0;
int docsError = 0;
foreach (SoilSample sample in transformed)
{
ResourceResponse <Document> result =
await loader.LoadNoReplace(sample);
if (result.StatusCode == HttpStatusCode.Created)
{
etlEvent.Outputs.Add(result.Resource.Id);
docsLoaded++;
}
else
{
docsError++;
}
// Notify data written then sleep to conserve Cosmos RU
Console.Write(".");
Thread.Sleep(40);
}
etlEvent.Logs.Add(
$"Loaded {docsLoaded.ToString()} SoilSamples");
etlEvent.Logs.Add(
$"Error loading {docsError.ToString()} SoilSamples");
}
catch (Exception e)
{
etlEvent.Logs.Add(
$"Error in ETL pipeline: {e.Message}");
throw new Exception("Error in ETL pipeline", e);
}
finally
{
etlEvent.DateTimeEnd = DateTime.UtcNow;
ResourceResponse <Document> result = await loader.LoadNoReplace(etlEvent);
}
}