MES-ETL/ConsoleApp2/Program.cs

508 lines
22 KiB
C#
Raw Normal View History

2024-01-12 16:50:37 +08:00
using ConsoleApp2;
using ConsoleApp2.Const;
2023-12-29 16:16:05 +08:00
using ConsoleApp2.HostedServices;
2024-01-04 09:00:44 +08:00
using ConsoleApp2.HostedServices.Abstractions;
2023-12-28 15:18:03 +08:00
using ConsoleApp2.Options;
using ConsoleApp2.Services;
2024-01-15 17:26:44 +08:00
using Microsoft.Extensions.Caching.StackExchangeRedis;
2023-12-28 15:18:03 +08:00
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;
2024-01-04 09:00:44 +08:00
using MySqlConnector;
2023-12-28 15:18:03 +08:00
using Serilog;
using Microsoft.Extensions.Caching.Distributed;
using Serilog.Events;
2024-01-12 16:50:37 +08:00
2023-12-28 15:18:03 +08:00
2023-12-29 16:16:05 +08:00
await RunProgram();
return;
async Task RunProgram()
2023-12-28 15:18:03 +08:00
{
2024-01-12 16:50:37 +08:00
//var inputDir= "D:\\MyDumper";
//ValidateConsole.ValidateInput<string>((_inputDir) =>
//{
// if (Directory.Exists(_inputDir))
// {
// inputDir = _inputDir;
// return true;
// }
// else return false;
//}, "请输入读取csv文件的目录(默认为当前目录下MyDumper文件夹):");
//var maxTask = 16;
//ValidateConsole.ValidateInput<string>((_inputDir) =>
//{
// _ = int.TryParse(_inputDir.ToString(), out var _taskCount);
// if (_taskCount > 0) {
// maxTask = _taskCount;
// return true;
// }
// else return false;
//}, "请输入执行输出的线程数量(默认为16):");
//var flushCount = 2_0000;
//ValidateConsole.ValidateInput<string>((_inputDir) =>
//{
// _ = int.TryParse(_inputDir.ToString(), out var _flashCount);
// if (_flashCount > 0)
// {
// flushCount = _flashCount;
// return true;
// } else return false;
//}, "请输入单次插入的行数(默认为20000):");
2023-12-29 16:16:05 +08:00
ThreadPool.SetMaxThreads(200, 200);
2024-01-12 16:50:37 +08:00
var host = Host.CreateApplicationBuilder(args);
host.Configuration.AddJsonFile(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "appsettings.json"), false, false);
var commandOptions = host.Configuration.GetSection("CmdOptions").Get<CommandOptions>() ??
throw new ApplicationException("未发现配置文件");
2024-01-12 16:50:37 +08:00
Console.WriteLine($"InputDir:{commandOptions?.InputDir}");
2024-01-16 18:00:23 +08:00
var oldestTime = DateTime.ParseExact(commandOptions.OldestTime, "yyyyMM", System.Globalization.DateTimeFormatInfo.InvariantInfo);
2024-01-18 14:36:36 +08:00
//host.Services.Configure<InputTableOptions>(option =>
//{
// option.TableInfoConfig = new Dictionary<string, TableInfo>
// {
2024-01-16 18:00:23 +08:00
2024-01-18 14:36:36 +08:00
// //order_block_plan_item从order_item表查询然后程序插入
// //order_package_item从order_item表查询然后程序插入
// //order_patch_detail生产没有这个表不处理
// {"machine",new TableInfo{SimulaRowCount=14655 }},
// {"order",new TableInfo{SimulaRowCount=5019216 }},
// {"order_block_plan",new TableInfo{SimulaRowCount=2725553 }},//CreateTime < 202301的删除
// {"order_block_plan_result",new TableInfo{SimulaRowCount=1174096 }},
// {"order_box_block",new TableInfo{SimulaRowCount=29755672 }},
// {"order_data_block",new TableInfo{SimulaRowCount=731800334 }},
// {"order_data_goods",new TableInfo{SimulaRowCount=25803671 }},
// {"order_data_parts",new TableInfo{SimulaRowCount=468517543 }},
// {"order_item",new TableInfo{SimulaRowCount=1345520079 }},
// {"order_module",new TableInfo{SimulaRowCount=103325385 }},
// {"order_module_extra",new TableInfo{SimulaRowCount=54361321 }},
// {"order_module_item",new TableInfo{SimulaRowCount=69173339 }},
// {"order_package",new TableInfo{SimulaRowCount=16196195 }},
// {"order_process",new TableInfo{SimulaRowCount=3892685 }},//orderNo < 202301的
// {"order_process_step",new TableInfo{SimulaRowCount=8050349 }},//orderNo < 202301的删除
// {"order_process_step_item",new TableInfo{SimulaRowCount=14538058 }},//orderNo < 202301的删除
// {"order_scrap_board",new TableInfo{SimulaRowCount=123998 }},
// {"process_group",new TableInfo{SimulaRowCount=1253 }},
// {"process_info",new TableInfo{SimulaRowCount=7839 }},
// {"process_item_exp",new TableInfo{SimulaRowCount=28 }},
// {"process_schdule_capacity",new TableInfo{SimulaRowCount=39736 }},
// {"process_step_efficiency",new TableInfo{SimulaRowCount=8 }},
// {"report_template",new TableInfo{SimulaRowCount=7337 }},
// {"simple_package",new TableInfo{SimulaRowCount=130436 }},//orderNo < 202301的删除
// {"simple_plan_order",new TableInfo{SimulaRowCount=351470 }},//CreateTime < 202301的删除
// {"sys_config",new TableInfo{SimulaRowCount=2296 }},
// {"work_calendar",new TableInfo{SimulaRowCount=11 }},
// {"work_shift",new TableInfo{SimulaRowCount=59 }},
// {"work_time",new TableInfo{SimulaRowCount=62 }},
// };
//});
2023-12-29 16:16:05 +08:00
host.Services.Configure<CsvOptions>(option =>
{
2024-01-04 09:00:44 +08:00
option.Delimiter = ",";
2023-12-29 16:16:05 +08:00
option.QuoteChar = '"';
});
2024-01-12 16:50:37 +08:00
host.Services.Configure<DataInputOptions>(options =>
{
options.InputDir = commandOptions.InputDir;
var _csvOptions = new CsvOptions { Delimiter = ",", QuoteChar = '"' };
options.CreateSource = (string tableName) =>
{
var source = new ZstSource(commandOptions.InputDir, tableName, _csvOptions.Delimiter, _csvOptions.QuoteChar);
return source;
};
});
2023-12-29 16:16:05 +08:00
host.Services.Configure<DataTransformOptions>(options =>
{
var tenantDbOptions = host.Configuration.GetRequiredSection("TenantDb").Get<TenantDbOptions>()
?? throw new ApplicationException("分库配置项不存在");
options.DatabaseFilter = record =>
{
var companyId = int.Parse(record[tenantDbOptions.TenantKey]); // 每个实体都应存在CompanyID否则异常
return tenantDbOptions.GetDbNameByTenantKeyValue(companyId);
};
2024-01-12 16:50:37 +08:00
2024-01-16 18:00:23 +08:00
options.TransformBinary = field => commandOptions != null && commandOptions.Isutf8mb4 ? $"_utf8mb4 0x{field}" : $"0x{field}";
//数据过滤
options.RecordFilter = async (record, cache) =>
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
//var index = Array.IndexOf(record.Headers, "ShardKey");
if (record.TryGetField("ShardKey", out var skStr))
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
short.TryParse(skStr, out var sk);
if (sk < commandOptions.OldestShardKey) return false;
2024-01-12 16:50:37 +08:00
}
2024-01-16 18:00:23 +08:00
if (record.TryGetField("CreateTime", out var createTime))
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
_ = DateTime.TryParse(createTime.Replace("\"", ""), out var time);
if (time < oldestTime) return false;
2024-01-12 16:50:37 +08:00
}
2024-01-16 18:00:23 +08:00
if (record.TryGetField("OrderNo", out var orderNo))
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
try
{
var yearMonth = orderNo.Substring(0, 6);
var dt = DateTime.ParseExact(yearMonth, "yyyyMM", System.Globalization.DateTimeFormatInfo.InvariantInfo);
if (dt < oldestTime) return false;
2024-01-12 16:50:37 +08:00
2024-01-16 18:00:23 +08:00
}
catch (Exception ex)
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
return false;//订单号转换失败,跳过
2024-01-12 16:50:37 +08:00
}
2024-01-16 18:00:23 +08:00
}
2024-01-12 16:50:37 +08:00
2024-01-16 18:00:23 +08:00
if (record.TableName == "order_package")
{
if (record.TryGetField("PakageNo", out var pkNo))
{
if (pkNo.Length <= 2) return false;
}
}
if (record.TableName == "order_block_plan")
{
if (record.TryGetField("OrderNos", out var nos))
{
if (nos.Length <= 2) return false;
}
}
if (record.TableName == "order_process_step" || record.TableName == "order_process_step_item")
{
//如果缓存中不存在OrderProcessID,则丢弃
2024-01-18 14:36:36 +08:00
//if(record.TryGetField("OrderProcessID",out var orderProcessID))
var value = await cache.GetStringAsync(record.GetCacheKey("OrderProcessID"));
if (string.IsNullOrEmpty(value))return false;
2024-01-18 14:36:36 +08:00
2024-01-16 18:00:23 +08:00
}
if (record.TableName == "order_block_plan_result" )
{
//如果缓存中不存在ID,则丢弃(ID 对应order_block_plan中的ID)
2024-01-18 14:36:36 +08:00
var value = await cache.GetStringAsync(record.GetCacheKey("ID"));
if (string.IsNullOrEmpty(value)) return false;
2024-01-18 14:36:36 +08:00
2024-01-12 16:50:37 +08:00
}
2024-01-16 18:00:23 +08:00
return true;
2024-01-12 16:50:37 +08:00
2024-01-16 18:00:23 +08:00
};
//数据修改
options.RecordModify = (record) =>
{
2024-01-17 11:40:16 +08:00
if (record.TableName == "order_block_plan")
{
if (record.TryGetField("OrderNos", out var nos))
{
2024-01-18 14:36:36 +08:00
if (nos.Length <= 2) record.SetField("OrderNos", "");
2024-01-17 11:40:16 +08:00
}
}
2024-01-16 18:00:23 +08:00
if (record.TableName == "order_process")//修改order_process.NextStepID的默认值为0
{
if (record.TryGetField("NextStepID", out var idStr))
{
2024-01-15 17:26:44 +08:00
2024-01-16 15:35:54 +08:00
if (idStr == "\\N")
2024-01-16 18:00:23 +08:00
{
record.SetField("NextStepID", "0");
}
}
}
2024-01-12 16:50:37 +08:00
2024-01-16 18:00:23 +08:00
};
//数据缓存
options.RecordCache = async (record, cache) =>
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
if (record.TableName == "order")
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
if (record.TryGetField("OrderNo", out var orderNo))
{
if (record.TryGetField("CompanyID", out var companyid))
{
2024-01-18 14:36:36 +08:00
await cache.SetStringAsync(record.GetCacheKey("OrderNo"), companyid);
2024-01-16 18:00:23 +08:00
}
}
2024-01-12 16:50:37 +08:00
}
2024-01-16 18:00:23 +08:00
if (record.TableName == "order_process")
{
if (record.TryGetField("OrderNo", out var orderNo))
{
var yearMonth = orderNo.Substring(2, 4);
var sk = yearMonth + "0";
if( record.TryGetField("ID", out var id))
{
2024-01-18 14:36:36 +08:00
await cache.SetStringAsync(record.GetCacheKey("ID"), sk);
2024-01-16 18:00:23 +08:00
}
}
}
if (record.TableName == "order_block_plan")
{
if (record.TryGetField("CompanyID", out var companyid))
{
record.TryGetField("ID", out var id);
2024-01-18 14:36:36 +08:00
await cache.SetStringAsync(record.GetCacheKey("ID"), companyid);
2024-01-16 18:00:23 +08:00
}
}
};
//数据替换
options.RecordReplace = async (record, cache) =>
2024-01-12 16:50:37 +08:00
{
2024-01-16 18:00:23 +08:00
//删除数据源里simple_plan_order.ProcessState 字段和值
if (record.TableName == "simple_plan_order")//修改order_process.NextStepID的默认值为0
{
var nextStepIdIndex = Array.IndexOf(record.Headers, "ProcessState");
if (nextStepIdIndex > -1)
{
var headers = record.Headers.Where(t => t != "ProcessState").ToArray();
var fs = record.Fields.ToList();
fs.RemoveAt(nextStepIdIndex);
var fields = fs.ToArray();
return new DataRecord(fields, record.TableName, headers, record.CompanyID);
}
}
if (record.TableName == "order")//修改order_process.NextStepID的默认值为0
{
var nextStepIdIndex = Array.IndexOf(record.Headers, "IsBatch");
if (nextStepIdIndex > -1)
{
var headers = record.Headers.Where(t => t != "IsBatch").ToArray();
var fs = record.Fields.ToList();
fs.RemoveAt(nextStepIdIndex);
var fields = fs.ToArray();
return new DataRecord(fields, record.TableName, headers, record.CompanyID);
}
}
if (record.TableName == "order_block_plan_result")//修改order_process.NextStepID的默认值为0
{
if (record.TryGetField("ID", out var id))
{
var headers = new List<string>(record.Headers);
var fields =new List<string>(record.Fields);
headers.Add("CompanyID");
2024-01-18 14:36:36 +08:00
var companyidResult =await cache.GetStringAsync(record.GetCacheKey("ID"));
2024-01-17 15:39:38 +08:00
_ = int.TryParse(companyidResult, out var companyid);
2024-01-16 18:00:23 +08:00
fields.Add(companyid.ToString());
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), companyid);
}
}
if(record.TableName == "order_box_block")
{
if (!record.TryGetField("CompanyID", out var companyid))
{
if (record.TryGetField("OrderNo", out var orderNo))
{
var headers = new List<string>(record.Headers);
var fields = new List<string>(record.Fields);
headers.Add("CompanyID");
2024-01-18 14:36:36 +08:00
var companyidResult = await cache.GetStringAsync(record.GetCacheKey("OrderNo"));
_ = int.TryParse(companyidResult, out var cpid);
2024-01-16 18:00:23 +08:00
fields.Add(cpid.ToString());
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), cpid);
}
}
}
if (record.TableName == "order_module")
{
if (record.TryGetField("ViewFileName",out var value))
{
var index=Array.IndexOf(record.Headers, "ViewFileName");
var headers = new List<string>(record.Headers);
headers.RemoveAt(index);
var fields = new List<string>(record.Fields);
fields.RemoveAt(index);
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), record.CompanyID);
}
if (!record.TryGetField("ShardKey", out var skStr))
{
if (record.TryGetField("OrderNo", out var orderNo))
{
var yearMonth = orderNo.Substring(2, 4);
var sk = yearMonth + "0";
var headers = new List<string>(record.Headers);
var fields = new List<string>(record.Fields);
headers.Add("ShardKey");
fields.Add(sk);
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), record.CompanyID);
}
}
2024-01-16 18:00:23 +08:00
}
if (record.TableName == "order_process")
{
if (!record.TryGetField("ShardKey", out var skStr))
{
if(record.TryGetField("OrderNo", out var orderNo))
{
var yearMonth = orderNo.Substring(2, 4);
var sk = yearMonth + "0";
var headers = new List<string>(record.Headers);
var fields = new List<string>(record.Fields);
headers.Add("ShardKey");
fields.Add(sk);
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), record.CompanyID);
}
}
}
if(record.TableName == "order_process_step"|| record.TableName == "order_process_step_item")
{
if (!record.TryGetField("ShardKey",out var sk))
{
if (record.TryGetField("OrderProcessID",out var processID))
{
2024-01-18 14:36:36 +08:00
var shardKey =await cache.GetStringAsync(record.GetCacheKey("OrderProcessID"));
2024-01-16 18:00:23 +08:00
var headers = new List<string>(record.Headers);
var fields = new List<string>(record.Fields);
headers.Add("ShardKey");
2024-01-18 14:36:36 +08:00
fields.Add(shardKey??"0");
2024-01-16 18:00:23 +08:00
return new DataRecord(fields.ToArray(), record.TableName, headers.ToArray(), record.CompanyID);
}
}
}
// 旧数据库order_data_block.CompanyID数据有误重新计算
if (record.TableName == "order_data_block")
2024-01-16 18:00:23 +08:00
{
var companyID = await cache.GetStringAsync($"order_{record["OrderNo"]}");
if (string.IsNullOrEmpty(companyID))
// throw new InvalidDataException($"没有获取到OrderNo={record["OrderNo"]}的CompanyID");
return record;
record["CompanyID"] = companyID;
return record;
2024-01-16 18:00:23 +08:00
}
return null;
};
//数据生成
options.RecordAdd = (record) =>
2024-01-15 17:26:44 +08:00
{
2024-01-16 18:00:23 +08:00
var resultList = new List<DataRecord>();
if (record.TableName == "order_item")
2024-01-15 17:26:44 +08:00
{
2024-01-16 18:00:23 +08:00
record.TryGetField("ID", out var itemID);
record.TryGetField("ShardKey", out var shardKey);
record.TryGetField("PlanID", out var planID);
record.TryGetField("PackageID", out var packageID);
record.TryGetField("CompanyID", out var companyID);
_=int.TryParse(planID, out var pid);
if (pid > 0)
{
resultList.Add(new DataRecord(new[] { itemID, shardKey, planID, companyID },
"order_block_plan_item",
new[] { "ItemID", "ShardKey", "PlanID", "CompanyID" }));
}
_ = int.TryParse(packageID, out var pkid);
if(pkid > 0)
2024-01-15 17:26:44 +08:00
{
2024-01-16 18:00:23 +08:00
resultList.Add(new DataRecord(new[] { itemID, shardKey, packageID, companyID },
"order_package_item",
new[] { "ItemID", "ShardKey", "PackageID", "CompanyID" }
));
2024-01-15 17:26:44 +08:00
}
}
2024-01-16 18:00:23 +08:00
return resultList;
2024-01-15 17:26:44 +08:00
};
2023-12-29 16:16:05 +08:00
options.ColumnTypeConfig = new()
{
{ "simple_plan_order.PlaceData", ColumnType.Blob },
{ "order_block_plan_result.PlaceData", ColumnType.Blob },
{ "order_box_block.Data", ColumnType.Blob },
{ "order_data_goods.ExtraProp", ColumnType.Text },
{ "order_module_extra.JsonStr", ColumnType.Text },
{ "process_info.Users", ColumnType.Text },
{ "order_process_schdule.CustomOrderNo", ColumnType.Text },
{ "order_process_schdule.OrderProcessStepName", ColumnType.Text },
{ "order_process_schdule.AreaName", ColumnType.Text },
{ "order_process_schdule.ConsigneeAddress", ColumnType.Text },
{ "order_process_schdule.ConsigneePhone", ColumnType.Text },
{ "report_source.Sql", ColumnType.Text },
{ "report_source.KeyValue", ColumnType.Text },
{ "report_source.Setting", ColumnType.Text },
{ "order_data_block.RemarkJson", ColumnType.Text },
{ "order_patch_detail.BlockDetail", ColumnType.Text },
{ "order_scrap_board.OutLineJson", ColumnType.Text },
{ "simple_package.Items", ColumnType.Text },
{ "order_batch_pack_config.Setting", ColumnType.Text },
{ "machine.Settings", ColumnType.Text },
{ "sys_config.Value", ColumnType.Text },
{ "sys_config.JsonStr", ColumnType.Text },
{ "process_item_exp.ItemJson", ColumnType.Text },
{ "report_template.Template", ColumnType.Text },
{ "report_template.SourceConfig", ColumnType.Text },
2024-01-18 14:36:36 +08:00
{ "order_block_plan.OrderNos", ColumnType.Json },
2023-12-29 16:16:05 +08:00
{ "order_block_plan.BlockInfo", ColumnType.Text },
};
});
2024-01-04 09:00:44 +08:00
host.Services.Configure<DatabaseOutputOptions>(options =>
2023-12-29 16:16:05 +08:00
{
2024-01-17 11:40:16 +08:00
//options.ConnectionString = new MySqlConnectionStringBuilder
//{
// Server = "127.0.0.1",
// Port = 33309,
// Database = "cferp_test",
// UserID = "root",
// Password = "123456",
// MaximumPoolSize = 50, // 这个值应当小于 max_connections
//}.ConnectionString;
options.ConnectionString = new MySqlConnectionStringBuilder(host.Configuration.GetConnectionString("MySqlMaster"))
2024-01-04 09:00:44 +08:00
{
2024-01-17 11:40:16 +08:00
CharacterSet = "utf8",
AllowUserVariables = true,
IgnoreCommandTransaction = true,
TreatTinyAsBoolean = false,
MaximumPoolSize = 50
2024-01-04 09:00:44 +08:00
}.ConnectionString;
2023-12-29 16:16:05 +08:00
});
2023-12-29 16:16:05 +08:00
host.Services.AddLogging(builder =>
{
builder.ClearProviders();
builder.AddSerilog(new LoggerConfiguration()
.WriteTo.Console()
.WriteTo.File("./log/error.log", restrictedToMinimumLevel:LogEventLevel.Error)
// .WriteTo.File("./log/info.log", restrictedToMinimumLevel:LogEventLevel.Information) //性能考虑暂不使用
.CreateLogger()
);
2023-12-29 16:16:05 +08:00
});
2024-01-18 14:36:36 +08:00
host.Services.AddScoped<ProcessContext>();
2023-12-29 16:16:05 +08:00
host.Services.AddKeyedSingleton<DataRecordQueue>(ProcessStep.Producer);
host.Services.AddKeyedSingleton<DataRecordQueue>(ProcessStep.Consumer);
2024-01-04 09:00:44 +08:00
host.Services.AddTransient<TaskManager>();
host.Services.AddSingleton<ErrorRecorder>();
2023-12-29 16:16:05 +08:00
2024-01-04 09:00:44 +08:00
host.Services.AddHostedService<MainHostedService>();
2023-12-29 16:16:05 +08:00
host.Services.AddHostedService<TaskMonitorService>();
2024-01-16 18:00:23 +08:00
host.Services.AddSingleton<IInputService, InputService>();
2024-01-04 09:00:44 +08:00
host.Services.AddSingleton<ITransformService, TransformService>();
host.Services.AddSingleton<IOutputService, OutputService>();
2024-01-16 18:00:23 +08:00
var redisOptions = host.Configuration.GetSection("RedisCacheOptions").Get<RedisCacheOptions>() ?? new RedisCacheOptions();
host.Services.AddStackExchangeRedisCache(options =>
{
options.Configuration = redisOptions.Configuration;
2024-01-17 15:10:11 +08:00
options.InstanceName = redisOptions.InstanceName;
});
2023-12-29 16:16:05 +08:00
var app = host.Build();
await app.RunAsync();
}