public static async Task <EtlSettings> ReadEtlSampleData(this EtlSettings etlSettings, int lines = 20)
{
etlSettings.Sample = new DataSample();
switch (etlSettings.SourceType)
{
case EtlSourceEnum.SFTP:
{
var sftp = etlSettings.SFTPSource;
var nameRegex = new Regex(sftp.PathRegex);
using (var sftpClient = new SftpClient(sftp.Host, sftp.Username, sftp.Password))
{
sftpClient.Connect();
var files = sftpClient.ListDirectory(sftp.BasePath);
files = files.Where(f => nameRegex.IsMatch(f.FullName)).ToList();
var first = files.FirstOrDefault();
if (first != null)
{
switch (etlSettings.FileType)
{
case EtlFileType.CSV:
{
using (var sftpStream = sftpClient.OpenRead(first.FullName))
{
etlSettings.ReadFromCSVFile(sftpStream, lines);
}
}
break;
}
}
sftpClient.Disconnect();
}
}
break;
case EtlSourceEnum.S3BucketCheck:
{
var s3 = etlSettings.S3CheckSource;
var awsS3API = new AWSS3API(new AWSS3Options()
{
Key = s3.Key,
Secret = s3.Secret,
Bucket = s3.BucketName,
Region = s3.Region,
});
var objects = await awsS3API.ListAllObjectsInBucket(s3.BucketName, s3.Prefix);
var nameRegex = new Regex(s3.PathRegex);
objects = objects.Where(f => nameRegex.IsMatch(f.Key)).ToList();
var first = objects.FirstOrDefault();
if (first != null)
{
switch (etlSettings.FileType)
{
case EtlFileType.CSV:
{
using (var s3Stream = await awsS3API.OpenReadAsync(first.Key, first.BucketName))
{
etlSettings.ReadFromCSVFile(s3Stream, lines);
}
}
break;
}
}
}
break;
case EtlSourceEnum.S3BucketEvent:
{
var s3 = etlSettings.S3EventSource;
var awsS3API = new AWSS3API(new AWSS3Options()
{
Key = s3.Key,
Secret = s3.Secret,
Bucket = s3.BucketName,
Region = s3.Region,
});
if (await awsS3API.FileExists(s3.ExamplePath, s3.BucketName))
{
switch (etlSettings.FileType)
{
case EtlFileType.CSV:
{
using (var s3Stream = await awsS3API.OpenReadAsync(s3.ExamplePath, s3.BucketName))
{
etlSettings.ReadFromCSVFile(s3Stream, lines);
}
}
break;
}
}
}
break;
case EtlSourceEnum.GoogleAnalytics:
{
await etlSettings.GetBigQueryResultSampleByDate(lines);
}
break;
case EtlSourceEnum.AmazonAthena:
{
await etlSettings.GetAthenaQueryResultSampleByDate(lines);
}
break;
case EtlSourceEnum.AmazonAthenaPipes:
{
// need to compile the query
await etlSettings.ParseAthenaQueryPipes();
}
break;
}
// make the sample data smaller
foreach (var row in etlSettings.Sample.Rows.ToList())
{
row.Items = row.Items.Select(item => item.Length < 100 ? item : item.Substring(0, 50) + "..." + item.Substring(item.Length - 50)).ToList();
}
return(etlSettings);
}
