This commit is contained in:
2023-12-29 16:16:05 +08:00
parent 6b88f5bd40
commit c53d2927bb
24 changed files with 909 additions and 386 deletions

View File

@@ -0,0 +1,70 @@
using ConsoleApp2.Helpers;
using ConsoleApp2.Options;
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Options;
namespace ConsoleApp2.HostedServices;
public class CsvInputService : BackgroundService
{
private readonly ILogger _logger;
private readonly IOptions<CsvOptions> _csvOptions;
private readonly TaskManager _taskManager; // TBD
private readonly DataRecordQueue _producerQueue;
private readonly ProcessContext _context;
public CsvInputService(ILogger<CsvInputService> logger,
IOptions<CsvOptions> csvOptions,
[FromKeyedServices(ProcessStep.Producer)]TaskManager taskManager,
[FromKeyedServices(ProcessStep.Producer)]DataRecordQueue producerQueue,
ProcessContext context)
{
_logger = logger;
_csvOptions = csvOptions;
_taskManager = taskManager;
_producerQueue = producerQueue;
_context = context;
}
protected override async Task ExecuteAsync(CancellationToken cancellationToken)
{
var inputDir = _csvOptions.Value.InputDir;
_logger.LogInformation("***** Csv input service start, working dir: {InputDir}, thread id: {ThreadId} *****", inputDir, Environment.CurrentManagedThreadId);
var files = Directory.GetFiles(inputDir).Where(s => s.EndsWith(".sql") && !s.Contains("schema")).ToArray();
if (files.Length == 0)
{
_logger.LogInformation("No sql files found in {InputDir}", inputDir);
return;
}
foreach (var sqlPath in files)
{
_logger.LogInformation("Working sql file: {SqlPath}", sqlPath);
var headers = await DumpDataHelper.GetCsvHeadersFromSqlFileAsync(sqlPath);
var csvFiles = await DumpDataHelper.GetCsvFileNamesFromSqlFileAsync(sqlPath);
foreach (var csvFile in csvFiles)
{
var csvPath = Path.Combine(inputDir, csvFile);
// var source = new JsvSource(csvPath, headers, _logger);
var source = new NewCsvSource(csvPath, headers, logger: _logger);
while (await source.ReadAsync())
{
_context.AddInput();
_producerQueue.Enqueue(source.Current);
if (cancellationToken.IsCancellationRequested)
return;
}
}
_logger.LogInformation("File '{File}' input completed", Path.GetFileName(sqlPath));
}
_context.CompleteInput();
_logger.LogInformation("***** Csv input service completed *****");
}
}

View File

@@ -0,0 +1,76 @@
using ConsoleApp2.Helpers;
using ConsoleApp2.Options;
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Options;
namespace ConsoleApp2.HostedServices;
public class DataTransformService : BackgroundService
{
private readonly ILogger _logger;
private readonly IOptions<DataTransformOptions> _options;
private readonly DataRecordQueue _producerQueue;
private readonly DataRecordQueue _consumerQueue;
private readonly ProcessContext _context;
public DataTransformService(ILogger<DataTransformService> logger,
IOptions<DataTransformOptions> options, // TBD: database filter
[FromKeyedServices(ProcessStep.Producer)]DataRecordQueue producerQueue,
[FromKeyedServices(ProcessStep.Consumer)]DataRecordQueue consumerQueue,
ProcessContext context)
{
_logger = logger;
// _taskManager = taskManager;
_options = options;
_producerQueue = producerQueue;
_consumerQueue = consumerQueue;
_context = context;
}
protected override async Task ExecuteAsync(CancellationToken cancellationToken)
{
_logger.LogInformation("***** Data transform service started, thread id: {ThreadId} *****", Environment.CurrentManagedThreadId);
while (!_context.IsInputCompleted || _producerQueue.Count > 0)
{
// var dbOptions = _options.Value.DatabaseFilter(record);
if (!_producerQueue.TryDequeue(out var record)) continue;
record.Database = _options.Value.DatabaseFilter?.Invoke(record);
for (var i = 0; i < record.Fields.Length; i++)
{
var field = record[i];
if (field == "\\N")
{
field = "NULL";
goto Escape;
}
// else if(DumpDataHelper.CheckHexField(field))
// field = $"0x{field}";
switch (_options.Value.GetColumnType(record.TableName, record.Headers[i]))
{
case ColumnType.Blob or ColumnType.Text:
field = string.IsNullOrEmpty(field) ? "''" : $"0x{field}";
break;
default:
field = field;
break;
}
Escape:
record[i] = field;
}
_consumerQueue.Enqueue(record);
_context.AddTransform();
}
_context.CompleteTransform();
_logger.LogInformation("***** Data transformation service completed *****");
}
}

View File

@@ -0,0 +1,85 @@
using ConsoleApp2.Entities;
using ConsoleApp2.Options;
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Options;
using MySqlConnector;
namespace ConsoleApp2.HostedServices;
public class MysqlOutputService : BackgroundService
{
private readonly ILogger _logger;
private readonly DataRecordQueue _consumerQueue;
private readonly IOptions<DatabaseOptions> _options;
private readonly ProcessContext _context;
public MysqlOutputService(ILogger<MysqlOutputService> logger,
[FromKeyedServices(ProcessStep.Consumer)]DataRecordQueue consumerQueue,
IOptions<DatabaseOptions> options,
ProcessContext context)
{
_logger = logger;
_consumerQueue = consumerQueue;
_options = options;
_context = context;
}
protected override async Task ExecuteAsync(CancellationToken stoppingToken)
{
_logger.LogInformation("***** Mysql output service started *****");
var tasks = new List<Task>();
var records = new List<DataRecord>();
while (!_context.IsTransformCompleted || _consumerQueue.Count > 0)
{
if (!_consumerQueue.TryDequeue(out var record)) continue;
records.Add(record);
if (records.Count >= 200)
{
var recordsCopy = records;
tasks.Add(Task.Run(async () => await FlushAsync(recordsCopy), stoppingToken));
records = [];
}
if (tasks.Count >= 10)
{
await Task.WhenAll(tasks);
tasks.Clear();
}
}
await Task.WhenAll(tasks);
await FlushAsync(records);
_context.CompleteOutput();
_logger.LogInformation("***** Mysql output service completed *****");
}
private async Task FlushAsync(IEnumerable<DataRecord> records)
{
var count = 0;
await using var output = new MySqlDestination(new MySqlConnectionStringBuilder
{
Server = _options.Value.Host,
Port = _options.Value.Port,
Database = _options.Value.Database,
UserID = _options.Value.User,
Password = _options.Value.Password,
ConnectionTimeout = 180,
}.ConnectionString, _logger, true);
foreach (var record in records)
{
await output.WriteRecordAsync(record);
count++;
}
await output.FlushAsync();
_context.AddOutput(count);
}
}

View File

@@ -0,0 +1,64 @@
using ConsoleApp2.Entities;
using ConsoleApp2.Helpers;
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
namespace ConsoleApp2.HostedServices;
public class SqlFileOutputService : BackgroundService
{
private readonly string _outputFile = "D:/DumpOutput/cferp_test_1.sql"; //
private readonly DataRecordQueue _consumerQueue;
private readonly ILogger _logger;
private readonly ProcessContext _context;
public SqlFileOutputService(
ILogger<SqlFileOutputService> logger,
[FromKeyedServices(ProcessStep.Consumer)]
DataRecordQueue consumerQueue,
ProcessContext context)
{
_logger = logger;
_consumerQueue = consumerQueue;
_context = context;
}
protected override async Task ExecuteAsync(CancellationToken stoppingToken)
{
_logger.LogInformation("***** Sql file output service started, thread id: {ThreadId} *****", Environment.CurrentManagedThreadId);
var count = 0;
var tableRecords = new Dictionary<string, IList<DataRecord>>();
while (!_context.IsTransformCompleted || _consumerQueue.Count > 0)
{
if (!_consumerQueue.TryDequeue(out var record)) continue;
tableRecords.AddOrUpdate(record.TableName, [record], (key, value) =>
{
value.Add(record);
return value;
});
++count;
if (count >= 200)
{
await File.AppendAllTextAsync(_outputFile,
MySqlDestination.SerializeRecords(tableRecords), stoppingToken);
tableRecords.Clear();
_context.AddOutput(count);
count = 0;
}
}
await File.AppendAllTextAsync(_outputFile,
MySqlDestination.SerializeRecords(tableRecords), stoppingToken);
tableRecords.Clear();
_context.AddOutput(count);
_context.CompleteOutput();
_logger.LogInformation("***** Sql file output service completed *****");
}
}

View File

@@ -0,0 +1,118 @@
using System.Diagnostics;
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
namespace ConsoleApp2.HostedServices;
public class TaskMonitorService : BackgroundService
{
private readonly IHostApplicationLifetime _lifetime;
private readonly ILogger<TaskMonitorService> _logger;
private readonly ProcessContext _context;
private readonly DataRecordQueue _producerQueue;
private readonly DataRecordQueue _consumerQueue;
public TaskMonitorService(IHostApplicationLifetime lifetime,
ILogger<TaskMonitorService> logger,
ProcessContext context,
[FromKeyedServices(ProcessStep.Producer)]
DataRecordQueue producerQueue,
[FromKeyedServices(ProcessStep.Consumer)]
DataRecordQueue consumerQueue)
{
_lifetime = lifetime;
_logger = logger;
_context = context;
_producerQueue = producerQueue;
_consumerQueue = consumerQueue;
}
protected override async Task ExecuteAsync(CancellationToken stoppingToken)
{
await Task.Factory.StartNew(Monitor, stoppingToken);
}
private async Task Monitor()
{
var sw = Stopwatch.StartNew();
var lastTime = sw.ElapsedMilliseconds;
var lastInputCount = _context.InputCount;
var lastTransformCount = _context.TransformCount;
var lastOutputCount = _context.OutputCount;
bool endCheck = false;
while (true)
{
EndCheck:
// var running = 0;
// var error = 0;
// var completed = 0;
// var canceled = 0;
// foreach (var task in _taskManager.Tasks)
// {
// switch (task.Status)
// {
// case TaskStatus.Canceled:
// canceled++;
// break;
// case TaskStatus.Faulted:
// error++;
// break;
// case TaskStatus.RanToCompletion:
// completed++;
// break;
// default:
// running++;
// break;
// }
// }
var time = sw.ElapsedMilliseconds;
var inputCount = _context.InputCount;
var transformCount = _context.TransformCount;
var outputCount = _context.OutputCount;
var elapseTime = (time - lastTime) / 1000f;
var inputSpeed = (inputCount - lastInputCount) / elapseTime;
var transformSpeed = (transformCount - lastTransformCount) / elapseTime;
var outputSpeed = (outputCount - lastOutputCount) / elapseTime;
// _logger.LogInformation(
// "Task monitor: running: {Running}, error: {Error}, completed: {Completed}, canceled: {Canceled}, outputSpeed: {Speed} records/s",
// running, error, completed, canceled, outputSpeed);
_logger.LogInformation(
"Process monitor: input: {inputStatus}, transform: {transformStatus}, output: {outputStatus}\nInput: {InputCount}, Transform: {TransformCount}, Output: {OutputCount}",
_context.IsInputCompleted ? "completed" : $"running {inputSpeed:F2} records/s",
_context.IsTransformCompleted ? "completed" : $"running {transformSpeed:F2} records/s",
_context.IsOutputCompleted ? "completed" : $"running {outputSpeed:F2} records/s",
inputCount,
transformCount,
outputCount);
_logger.LogInformation("Queue monitor: producer queue: {ProducerQueue}, consumer queue: {ConsumerQueue}",
_producerQueue.Count, _consumerQueue.Count);
await Task.Delay(2000);
lastTime = time;
lastInputCount = inputCount;
lastTransformCount = transformCount;
lastOutputCount = outputCount;
if (_context is { IsInputCompleted: true, IsTransformCompleted: true, IsOutputCompleted: true })
{
if (!endCheck)
{
endCheck = true;
goto EndCheck;
}
break;
}
}
_logger.LogInformation("***** All tasks completed *****");
_logger.LogInformation("***** ElapseTime: {Time}", (sw.ElapsedMilliseconds / 1000f).ToString("F3"));
// _lifetime.StopApplication();
}
}

View File

@@ -0,0 +1,35 @@
using ConsoleApp2.Services;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
namespace ConsoleApp2.HostedServices;
public class VoidOutputService : BackgroundService
{
private readonly ILogger _logger;
private readonly DataRecordQueue _consumerQueue;
private readonly ProcessContext _context;
public VoidOutputService([FromKeyedServices(ProcessStep.Consumer)] DataRecordQueue consumerQueue,
ProcessContext context, ILogger<VoidOutputService> logger)
{
_consumerQueue = consumerQueue;
_context = context;
_logger = logger;
}
protected override Task ExecuteAsync(CancellationToken stoppingToken)
{
_logger.LogInformation("***** Void output service started, thread id: {ThreadId} *****", Environment.CurrentManagedThreadId);
while (!_context.IsTransformCompleted || _consumerQueue.Count > 0)
{
if (_consumerQueue.TryDequeue(out var record))
_context.AddOutput();
}
_context.CompleteOutput();
_logger.LogInformation("***** Void output service completed *****");
return Task.CompletedTask;
}
}