public async Task GivenABlobFile_WhenExecutorWithoutAnonymize_DataShouldBeSame(string connectionString, string containerName, string blobName)
{
string targetContainerName = Guid.NewGuid().ToString("N");
string targetBlobName = Guid.NewGuid().ToString("N");
BlobContainerClient containerClient = new BlobContainerClient(connectionString, targetContainerName);
await containerClient.CreateIfNotExistsAsync();
try
{
BlobClient sourceBlobClient = new BlobClient(connectionString, containerName, blobName, DataFactoryCustomActivity.BlobClientOptions.Value);
BlockBlobClient targetBlobClient = new BlockBlobClient(connectionString, targetContainerName, targetBlobName, DataFactoryCustomActivity.BlobClientOptions.Value);
using FhirBlobDataStream stream = new FhirBlobDataStream(sourceBlobClient);
using FhirStreamReader reader = new FhirStreamReader(stream);
FhirBlobConsumer consumer = new FhirBlobConsumer(targetBlobClient);
var executor = new FhirPartitionedExecutor <string, string>(reader, consumer, content => content);
await executor.ExecuteAsync(CancellationToken.None).ConfigureAwait(false);
Assert.Equal(sourceBlobClient.GetProperties().Value.ContentLength, targetBlobClient.GetProperties().Value.ContentLength);
}
finally
{
await containerClient.DeleteIfExistsAsync().ConfigureAwait(false);
}
}
public void AnonymizeBulkDataFolder(string inputFolder, string outputFolder, bool isRecursive)
{
var directorySearchOption = isRecursive ? SearchOption.AllDirectories : SearchOption.TopDirectoryOnly;
var bulkResourceFileList = Directory.EnumerateFiles(inputFolder, "*.ndjson", directorySearchOption).ToList();
Console.WriteLine($"Find {bulkResourceFileList.Count()} bulk data resource files in '{inputFolder}'.");
foreach (var bulkResourceFileName in bulkResourceFileList)
{
Console.WriteLine($"Processing {bulkResourceFileName}");
var bulkResourceOutputFileName = GetResourceOutputFileName(bulkResourceFileName, inputFolder, outputFolder);
if (isRecursive)
{
var resourceOutputFolder = Path.GetDirectoryName(bulkResourceOutputFileName);
Directory.CreateDirectory(resourceOutputFolder);
}
int completedCount = 0;
int failedCount = 0;
int consumeCompletedCount = 0;
using (FileStream inputStream = new FileStream(bulkResourceFileName, FileMode.Open))
using (FileStream outputStream = new FileStream(bulkResourceOutputFileName, FileMode.Create))
{
using FhirStreamReader reader = new FhirStreamReader(inputStream);
using FhirStreamConsumer consumer = new FhirStreamConsumer(outputStream);
Func <string, string> anonymizeFunction = (content) => _engine.AnonymizeJson(content);
Stopwatch stopWatch = new Stopwatch();
stopWatch.Start();
FhirPartitionedExecutor executor = new FhirPartitionedExecutor(reader, consumer, anonymizeFunction);
executor.PartitionCount = Environment.ProcessorCount;
Progress <BatchAnonymizeProgressDetail> progress = new Progress <BatchAnonymizeProgressDetail>();
progress.ProgressChanged += (obj, args) =>
{
Interlocked.Add(ref completedCount, args.ProcessCompleted);
Interlocked.Add(ref failedCount, args.ProcessFailed);
Interlocked.Add(ref consumeCompletedCount, args.ConsumeCompleted);
Console.WriteLine($"[{stopWatch.Elapsed.ToString()}]: {completedCount} Process completed. {failedCount} Process failed. {consumeCompletedCount} Consume completed.");
};
executor.ExecuteAsync(CancellationToken.None, false, progress).Wait();
}
Console.WriteLine($"Finished processing '{bulkResourceFileName}'!");
}
}
public async Task GivenAFhirStreamReader_WhenLoadData_ShouldLoadAllDataFromStream()
{
using MemoryStream inputStream = new MemoryStream();
using StreamWriter writer = new StreamWriter(inputStream);
await writer.WriteLineAsync("abc");
await writer.WriteLineAsync("bcd");
await writer.WriteLineAsync("");
writer.Flush();
inputStream.Position = 0;
using FhirStreamReader reader = new FhirStreamReader(inputStream);
Assert.Equal("abc", await reader.NextAsync());
Assert.Equal("bcd", await reader.NextAsync());
Assert.Equal("", await reader.NextAsync());
Assert.Null(await reader.NextAsync());
}
private async Task AnonymizeSingleBlobInNdJsonFormatAsync(BlobClient inputBlobClient, BlockBlobClient outputBlobClient, string blobName, string inputFolderPrefix)
{
var processedCount = 0;
int skippedCount = 0;
var consumedCount = 0;
using FhirBlobDataStream inputStream = new FhirBlobDataStream(inputBlobClient);
FhirStreamReader reader = new FhirStreamReader(inputStream);
FhirBlobConsumer consumer = new FhirBlobConsumer(outputBlobClient);
var engine = AnonymizerEngine.CreateWithFileContext(_configFile, blobName, inputFolderPrefix);
Func <string, string> anonymizerFunction = (item) =>
{
try
{
return(engine.AnonymizeJson(item));
}
catch (Exception ex)
{
Console.WriteLine($"[{blobName}]: Anonymize partial failed, you can find detail error message in stderr.txt.");
Console.Error.WriteLine($"[{blobName}]: Resource: {item}\nErrorMessage: {ex.Message}\n Details: {ex.ToString()}\nStackTrace: {ex.StackTrace}");
throw;
}
};
Stopwatch stopWatch = Stopwatch.StartNew();
FhirPartitionedExecutor <string, string> executor = new FhirPartitionedExecutor <string, string>(reader, consumer, anonymizerFunction);
executor.PartitionCount = Environment.ProcessorCount * 2;
Progress <BatchAnonymizeProgressDetail> progress = new Progress <BatchAnonymizeProgressDetail>();
progress.ProgressChanged += (obj, args) =>
{
Interlocked.Add(ref processedCount, args.ProcessCompleted);
Interlocked.Add(ref skippedCount, args.ProcessSkipped);
Interlocked.Add(ref consumedCount, args.ConsumeCompleted);
Console.WriteLine($"[{stopWatch.Elapsed.ToString()}][tid:{args.CurrentThreadId}]: {processedCount} Completed. {skippedCount} Skipped. {consumedCount} consume completed.");
};
await executor.ExecuteAsync(CancellationToken.None, progress).ConfigureAwait(false);
}
public async Task AnonymizeAsync()
{
var directorySearchOption = _isRecursive ? SearchOption.AllDirectories : SearchOption.TopDirectoryOnly;
var bulkResourceFileList = Directory.EnumerateFiles(_inputFolder, "*.ndjson", directorySearchOption).ToList();
Console.WriteLine($"Find {bulkResourceFileList.Count()} bulk data resource files in '{_inputFolder}'.");
foreach (var bulkResourceFileName in bulkResourceFileList)
{
Console.WriteLine($"Processing {bulkResourceFileName}");
var bulkResourceOutputFileName = GetResourceOutputFileName(bulkResourceFileName, _inputFolder, _outputFolder);
if (_isRecursive)
{
var resourceOutputFolder = Path.GetDirectoryName(bulkResourceOutputFileName);
Directory.CreateDirectory(resourceOutputFolder);
}
int completedCount = 0;
int failedCount = 0;
int consumeCompletedCount = 0;
using (FileStream inputStream = new FileStream(bulkResourceFileName, FileMode.Open))
using (FileStream outputStream = new FileStream(bulkResourceOutputFileName, FileMode.Create))
{
using FhirStreamReader reader = new FhirStreamReader(inputStream);
using FhirStreamConsumer consumer = new FhirStreamConsumer(outputStream);
Func <string, string> anonymizeFunction = (content) =>
{
try
{
var engine = AnonymizerEngine.CreateWithFileContext(_configFilePath, bulkResourceFileName, _inputFolder);
var settings = new AnonymizerSettings()
{
IsPrettyOutput = false,
ValidateInput = _validateInput,
ValidateOutput = _validateOutput
};
return(engine.AnonymizeJson(content, settings));
}
catch (Exception ex)
{
Console.Error.WriteLine($"Error:\nResource: {content}\nErrorMessage: {ex.ToString()}");
throw;
}
};
Stopwatch stopWatch = new Stopwatch();
stopWatch.Start();
FhirPartitionedExecutor <string, string> executor = new FhirPartitionedExecutor <string, string>(reader, consumer, anonymizeFunction);
executor.PartitionCount = Environment.ProcessorCount * 2;
Progress <BatchAnonymizeProgressDetail> progress = new Progress <BatchAnonymizeProgressDetail>();
progress.ProgressChanged += (obj, args) =>
{
Interlocked.Add(ref completedCount, args.ProcessCompleted);
Interlocked.Add(ref failedCount, args.ProcessFailed);
Interlocked.Add(ref consumeCompletedCount, args.ConsumeCompleted);
Console.WriteLine($"[{stopWatch.Elapsed.ToString()}][tid:{args.CurrentThreadId}]: {completedCount} Process completed. {failedCount} Process failed. {consumeCompletedCount} Consume completed.");
};
await executor.ExecuteAsync(CancellationToken.None, false, progress).ConfigureAwait(false);
}
Console.WriteLine($"Finished processing '{bulkResourceFileName}'!");
}
}