2025迁移版本，多项规则修改

MesETL.App/Helpers/DumpDataHelper.cs

@@ -1,5 +1,7 @@
 using System.Text.Json;
 using System.Text.RegularExpressions;
+using MesETL.App.HostedServices;
+using Serilog;
 using ZstdSharp;
 
 namespace MesETL.App.Helpers;
@@ -27,16 +29,13 @@ public static partial class DumpDataHelper
         string[] ParseHeader(ReadOnlySpan<char> headerStr)
         {
             headerStr = headerStr[1..^1];
-            Span<Range> ranges = stackalloc Range[50];
-            var count = headerStr.Split(ranges, ',');
-            var arr = new string[count];
-
-            for (var i = 0; i < count; i++)
+            var headers = new List<string>();
+            foreach (var range in headerStr.Split(','))
             {
-                arr[i] = headerStr[ranges[i]].Trim("@`").ToString(); // 消除列名的反引号，如果是变量则消除@
+                headers.Add(headerStr[range].Trim("@`").ToString()); // 消除列名的反引号，如果是变量则消除@
             }
-        
-            return arr;
+
+            return headers.ToArray();
         }
     }
     
@@ -45,6 +44,7 @@ public static partial class DumpDataHelper
     /// </summary>
     /// <param name="filePath"></param>
     /// <returns></returns>
+    [Obsolete("用ParseMyDumperFile替代")]
     public static string GetTableNameFromCsvFileName(ReadOnlySpan<char> filePath)
     {
         filePath = filePath[(filePath.LastIndexOf('\\') + 1)..];
@@ -68,6 +68,30 @@ public static partial class DumpDataHelper
 
         return filePath[(firstDotIdx+1)..secondDotIdx].ToString();
     }
+    
+    public enum MyDumperFileType { Dat, Sql }
+
+    public record MyDumperFileMeta(string Path, string Database, string TableName, int Index, MyDumperFileType Type);
+
+    public static MyDumperFileMeta ParseMyDumperFile(ReadOnlySpan<char> path)
+    {
+        try
+        {
+            var fileName = Path.GetFileName(path).ToString();
+            var parts = fileName.Split('.');
+            var type = parts[3] switch
+            {
+                "dat" => MyDumperFileType.Dat,
+                "sql" => MyDumperFileType.Sql,
+                _ => throw new ArgumentException("不支持的MyDumper文件类型", nameof(path))
+            };
+            return new MyDumperFileMeta(path.ToString(), parts[0], parts[1], int.Parse(parts[2]), type);
+        }
+        catch (Exception e)
+        {
+            throw new ArgumentException($"此文件不是MyDumper导出的文件 {path}", nameof(path), e);
+        }
+    }
 
     /// <summary>
     /// 从MyDumper导出的SQL文件内容中读取CSV文件名
@@ -122,17 +146,46 @@ public static partial class DumpDataHelper
         var reader = new StreamReader(ds);
         return await reader.ReadToEndAsync();
     }
-    
-    public static bool IsJson(string str)
+
+    /// <summary>
+    /// 适用于文件输入服务以及MyDumper Zst导出目录的文件元数据构建函数
+    /// </summary>
+    /// <param name="filePath"></param>
+    /// <returns></returns>
+    /// <exception cref="ApplicationException"></exception>
+    public static FileInputInfo? MyDumperFileInputMetaBuilder(string filePath)
     {
+        // 只查找后缀为.dat.zst的文件
+        if (!filePath.EndsWith(".dat.zst")) return null;
+            
+        var fileMeta = ParseMyDumperFile(filePath);
+        var inputDir = Path.GetDirectoryName(filePath);
+        string[]? headers;
         try
         {
-            JsonDocument.Parse(str);
-            return true;
+            // 查找同目录下同表的SQL文件
+            var sqlFile = Directory.GetFiles(inputDir!)
+                .SingleOrDefault(f => f.Equals(filePath.Replace(".dat.zst", ".sql.zst")));
+            if (sqlFile is null)
+            {
+                Log.Debug("{TableName}表的SQL文件不存在", fileMeta.TableName);
+                return null;
+            }
+            headers = GetCsvHeadersFromSqlFile(
+                DecompressZstAsStringAsync(File.OpenRead(sqlFile)).Result);
         }
-        catch (JsonException)
+        catch (InvalidOperationException e)
         {
-            return false;
+            throw new ApplicationException($"目录下不止一个{fileMeta.TableName}表的SQL文件", e);
         }
+
+        return new FileInputInfo
+        {
+            FileName = filePath,
+            TableName = fileMeta.TableName,
+            Headers = headers,
+            Database = fileMeta.Database,
+            Part = fileMeta.Index
+        };
     }
 }