perf: reuse prepared form schemas

src/modules/workflow/Application/BuildFlow/Common/Services/ExcelParsedRow.cs

@@ -0,0 +1,7 @@
+namespace Workflow.Application.BuildFlow.Common.Services;
+
+/// <summary>解析后的 Excel 行。</summary>
+public sealed record ExcelParsedRow(
+    int RowNumber,
+    string JsonObject,
+    IReadOnlyList<ExcelCellError> Errors);

src/modules/workflow/Application/BuildFlow/Common/Services/FormSchemaExcelService.cs

@@ -53,7 +53,7 @@ public class FormSchemaExcelService : IFormSchemaExcelService
     }
 
     /// <inheritdoc />
-    public IReadOnlyList<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes)
+    public IEnumerable<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes)
     {
         var fields = ExtractFields(schemaJson);
         using var ms = new MemoryStream(xlsxBytes);
@@ -62,19 +62,16 @@ public class FormSchemaExcelService : IFormSchemaExcelService
         var lastRow = ws.LastRowUsed();
 
         var headerMap = BuildHeaderMap(ws, fields);
-        var result = new List<ExcelParsedRow>();
         if (lastRow is null)
         {
-            return result;
+            yield break;
         }
 
         for (var r = 2; r <= lastRow.RowNumber(); r++)
         {
             var (json, errors) = BuildRowJson(ws, r, headerMap, fields);
-            result.Add(new ExcelParsedRow(r, json, errors));
+            yield return new ExcelParsedRow(r, json, errors);
         }
-
-        return result;
     }
 
     private static IReadOnlyList<FieldSummary> ExtractFields(string schemaJson)

src/modules/workflow/Application/BuildFlow/Common/Services/IFormSchemaExcelService.cs

@@ -12,12 +12,6 @@ public interface IFormSchemaExcelService
     /// <summary>根据 SchemaJson 字段顺序导出已有 FormData 列表为 xlsx 字节流。</summary>
     byte[] BuildExport(string schemaJson, string formName, IEnumerable<string> dataJsonRows);
 
-    /// <summary>解析上传 xlsx 字节流，按字段顺序返回每行的 JSON 对象字符串。</summary>
+    /// <summary>解析上传 xlsx 字节流，按字段顺序惰性返回每行的 JSON 对象字符串。</summary>
-    IReadOnlyList<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes);
+    IEnumerable<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes);
 }
-
-/// <summary>解析后的 Excel 行。</summary>
-public sealed record ExcelParsedRow(
-    int RowNumber,
-    string JsonObject,
-    IReadOnlyList<ExcelCellError> Errors);

src/modules/workflow/Application/BuildFlow/Imports/Services/ImportJobProcessor.cs

@@ -62,25 +62,24 @@ public class ImportJobProcessor(
                 jobId,
                 leaseOwner,
                 leaseDuration,
-                rows.Count,
                 ct);
             if (nextRowIndex is null) return;
-            if (nextRowIndex.Value > rows.Count)
-            {
-                throw new BusinessException(
-                    ErrorCode.Business,
-                    $"导入任务进度游标 {nextRowIndex.Value} 超过总行数 {rows.Count}");
-            }
 
             var submitHandler = new SubmitFormDataCommandHandler(
                 db,
                 new FormValueDerivationService([], [], db),
                 new FormAggregateRefreshService(db));
 
-            for (var i = nextRowIndex.Value; i < rows.Count; i++)
+            var rowIndex = 0;
+            foreach (var row in rows)
             {
                 ct.ThrowIfCancellationRequested();
-                var row = rows[i];
+                if (rowIndex < nextRowIndex.Value)
+                {
+                    rowIndex++;
+                    continue;
+                }
+
                 if (row.Errors.Count > 0)
                 {
                     var recorded = await CommitFailedRowAsync(
@@ -89,8 +88,10 @@ public class ImportJobProcessor(
                         leaseDuration,
                         row.RowNumber,
                         row.Errors.Select(e => e.Message).ToList(),
+                        rowIndex + 1,
                         ct);
                     if (!recorded) return;
+                    rowIndex++;
                     continue;
                 }
 
@@ -101,7 +102,7 @@ public class ImportJobProcessor(
                         leaseOwner,
                         leaseDuration,
                         job.FormDefinitionId,
-                        i,
+                        rowIndex,
                         row.JsonObject,
                         submitHandler,
                         ct);
@@ -119,12 +120,22 @@ public class ImportJobProcessor(
                         leaseDuration,
                         row.RowNumber,
                         [ex.Message],
+                        rowIndex + 1,
                         ct);
                     if (!recorded) return;
                 }
+
+                rowIndex++;
             }
 
-            await CompleteJobAsync(jobId, leaseOwner, ct);
+            if (nextRowIndex.Value > rowIndex)
+            {
+                throw new BusinessException(
+                    ErrorCode.Business,
+                    $"导入任务进度游标 {nextRowIndex.Value} 超过总行数 {rowIndex}");
+            }
+
+            await CompleteJobAsync(jobId, leaseOwner, rowIndex, ct);
         }
         catch (OperationCanceledException) when (ct.IsCancellationRequested)
         {
@@ -141,7 +152,6 @@ public class ImportJobProcessor(
         Guid jobId,
         string leaseOwner,
         TimeSpan leaseDuration,
-        int totalRows,
         CancellationToken ct)
     {
         db.ChangeTracker.Clear();
@@ -153,7 +163,6 @@ public class ImportJobProcessor(
             return null;
         }
 
-        job.TotalRows = totalRows;
         RenewLease(job, leaseDuration);
         await db.SaveChangesAsync(ct);
         await transaction.CommitAsync(ct);
@@ -184,6 +193,7 @@ public class ImportJobProcessor(
             new SubmitFormDataCommand(formDefinitionId, instanceId, dataJson),
             ct);
         job.SuccessRows++;
+        job.TotalRows = Math.Max(job.TotalRows, rowIndex + 1);
         RenewLease(job, leaseDuration);
         await db.SaveChangesAsync(ct);
         await transaction.CommitAsync(ct);
@@ -196,6 +206,7 @@ public class ImportJobProcessor(
         TimeSpan leaseDuration,
         int rowNumber,
         IReadOnlyList<string> messages,
+        int processedRows,
         CancellationToken ct)
     {
         db.ChangeTracker.Clear();
@@ -210,6 +221,7 @@ public class ImportJobProcessor(
         var errors = DeserializeErrors(job.ErrorJson);
         errors.Add(new ImportRowError(rowNumber, messages));
         job.FailedRows++;
+        job.TotalRows = Math.Max(job.TotalRows, processedRows);
         job.ErrorJson = JsonSerializer.Serialize(errors, JsonOptions);
         RenewLease(job, leaseDuration);
         await db.SaveChangesAsync(ct);
@@ -220,6 +232,7 @@ public class ImportJobProcessor(
     private async Task CompleteJobAsync(
         Guid jobId,
         string leaseOwner,
+        int totalRows,
         CancellationToken ct)
     {
         db.ChangeTracker.Clear();
@@ -237,6 +250,7 @@ public class ImportJobProcessor(
                 ? BuildFlowImportJobStatus.Failed
                 : BuildFlowImportJobStatus.PartialSuccess;
         job.ErrorMessage = null;
+        job.TotalRows = totalRows;
         job.FinishedAt = DateTime.UtcNow;
         job.LastHeartbeatAt = DateTime.UtcNow;
         job.LeaseOwner = null;

src/modules/workflow/Application/Form/Derivation/FormValueDerivationService.cs

@@ -2,6 +2,7 @@ using System.Text.Encodings.Web;
 using System.Text.Json;
 using System.Text.Json.Nodes;
 using Workflow.Application.Form.Relation;
+using Workflow.Application.Form.Schema;
 using Workflow.Domain.Enums;
 using Workflow.Infrastructure.Persistence;
 
@@ -33,15 +34,22 @@ public sealed class FormValueDerivationService : IFormValueDerivationService
         _db = db;
     }
 
-    public async Task<FormDerivationResult> DeriveAsync(
+    public Task<FormDerivationResult> DeriveAsync(
         string schemaJson,
         string dataJson,
         FormDerivationContext context,
         CancellationToken ct)
+        => DeriveAsync(PreparedFormSchema.Create(schemaJson), dataJson, context, ct);
+
+    public async Task<FormDerivationResult> DeriveAsync(
+        PreparedFormSchema preparedSchema,
+        string dataJson,
+        FormDerivationContext context,
+        CancellationToken ct)
     {
         ct.ThrowIfCancellationRequested();
 
-        var schema = JsonNode.Parse(schemaJson) as JsonObject ?? new JsonObject();
+        var schema = preparedSchema.Schema.DeepClone().AsObject();
         var data = JsonNode.Parse(dataJson) as JsonObject ?? new JsonObject();
         var derivedPaths = new List<string>();
 

src/modules/workflow/Application/Form/Derivation/IFormValueDerivationService.cs

@@ -1,3 +1,5 @@
+using Workflow.Application.Form.Schema;
+
 namespace Workflow.Application.Form.Derivation;
 
 /// <summary>统一表单提交前派生值管线：计算字段、关联快照、统计字段都在这里收口。</summary>
@@ -8,4 +10,11 @@ public interface IFormValueDerivationService
         string dataJson,
         FormDerivationContext context,
         CancellationToken ct);
+
+    /// <summary>使用预解析 Schema 执行派生，供批量场景复用字段元数据。</summary>
+    Task<FormDerivationResult> DeriveAsync(
+        PreparedFormSchema schema,
+        string dataJson,
+        FormDerivationContext context,
+        CancellationToken ct);
 }

src/modules/workflow/Application/Form/Schema/FormDataValidator.cs

@@ -33,6 +33,31 @@ public static class FormDataValidator
         IReadOnlyDictionary<string, string>? nodeFieldPermissions = null)
     {
         var schemaValidation = SchemaValidator.Validate(schemaJson, allowedComponents);
+        return Validate(
+            schemaValidation,
+            dataJson,
+            currentNodeKey,
+            nodeFieldPermissions);
+    }
+
+    /// <summary>使用预解析 Schema 校验表单数据，避免批量场景逐行重复解析。</summary>
+    public static FormDataValidationResult Validate(
+        PreparedFormSchema schema,
+        string dataJson,
+        string? currentNodeKey = null,
+        IReadOnlyDictionary<string, string>? nodeFieldPermissions = null)
+        => Validate(
+            schema.Validation,
+            dataJson,
+            currentNodeKey,
+            nodeFieldPermissions);
+
+    private static FormDataValidationResult Validate(
+        SchemaValidationResult schemaValidation,
+        string dataJson,
+        string? currentNodeKey,
+        IReadOnlyDictionary<string, string>? nodeFieldPermissions)
+    {
         if (!schemaValidation.IsValid)
         {
             return FormDataValidationResult.Invalid(schemaValidation.Errors);

src/modules/workflow/Application/Form/Schema/PreparedFormSchema.cs

@@ -0,0 +1,43 @@
+using System.Text.Json.Nodes;
+
+namespace Workflow.Application.Form.Schema;
+
+/// <summary>
+/// 预解析表单 Schema。批量用例复用字段校验结果和 JSON 结构，避免逐行重复解析。
+/// </summary>
+public sealed class PreparedFormSchema
+{
+    private PreparedFormSchema(
+        string schemaJson,
+        JsonObject schema,
+        SchemaValidationResult validation)
+    {
+        SchemaJson = schemaJson;
+        Schema = schema;
+        Validation = validation;
+    }
+
+    /// <summary>原始 Schema JSON。</summary>
+    public string SchemaJson { get; }
+
+    /// <summary>解析后的 Schema JSON 对象。</summary>
+    public JsonObject Schema { get; }
+
+    /// <summary>已提取字段元数据的 Schema 校验结果。</summary>
+    public SchemaValidationResult Validation { get; }
+
+    /// <summary>校验并准备一个可复用的表单 Schema。</summary>
+    public static PreparedFormSchema Create(string schemaJson)
+    {
+        var validation = SchemaValidator.Validate(schemaJson);
+        if (!validation.IsValid)
+        {
+            throw new BusinessException(
+                ErrorCode.InvalidParam,
+                $"表单 Schema 无效: {string.Join("; ", validation.Errors)}");
+        }
+
+        var schema = JsonNode.Parse(schemaJson) as JsonObject ?? new JsonObject();
+        return new PreparedFormSchema(schemaJson, schema, validation);
+    }
+}

tests/BuildFlow.Workflow.Tests/BuildFlow/FormSchemaExcelServiceTests.cs

@@ -49,7 +49,7 @@ public class FormSchemaExcelServiceTests
         using var outMs = new MemoryStream();
         wb.SaveAs(outMs);
 
-        var rows = _service.Parse(SampleSchema, outMs.ToArray());
+        var rows = _service.Parse(SampleSchema, outMs.ToArray()).ToList();
 
         rows.Should().HaveCount(1);
         var row = rows[0];
@@ -73,7 +73,7 @@ public class FormSchemaExcelServiceTests
         using var outMs = new MemoryStream();
         wb.SaveAs(outMs);
 
-        var rows = _service.Parse(SampleSchema, outMs.ToArray());
+        var rows = _service.Parse(SampleSchema, outMs.ToArray()).ToList();
 
         rows.Should().HaveCount(1);
         rows[0].Errors.Should().ContainSingle(e => e.Message.Contains("必填"));

tests/BuildFlow.Workflow.Tests/Performance/LazyExcelRowTests.cs

@@ -0,0 +1,42 @@
+using ClosedXML.Excel;
+using FluentAssertions;
+using Workflow.Application.BuildFlow.Common.Services;
+using Xunit;
+
+namespace BuildFlow.Workflow.Tests.Performance;
+
+/// <summary>
+/// Excel 导入行惰性枚举契约测试。
+/// </summary>
+public class LazyExcelRowTests
+{
+    [Fact]
+    public void ParseReturnsLazyEnumerableInsteadOfMaterializedList()
+    {
+        var service = new FormSchemaExcelService();
+        var template = service.BuildTemplate(Schema, "惰性导入");
+        using var input = new MemoryStream(template);
+        using var workbook = new XLWorkbook(input);
+        var worksheet = workbook.Worksheets.First();
+        worksheet.Cell(2, 1).Value = "first";
+        worksheet.Cell(3, 1).Value = "second";
+        worksheet.Cell(4, 1).Value = "third";
+        using var output = new MemoryStream();
+        workbook.SaveAs(output);
+
+        var rows = service.Parse(Schema, output.ToArray());
+
+        rows.Should().NotBeAssignableTo<IReadOnlyList<ExcelParsedRow>>();
+        rows.Take(1).Should().ContainSingle(x =>
+            x.RowNumber == 2 && x.JsonObject.Contains("first", StringComparison.Ordinal));
+    }
+
+    private const string Schema = """
+    {
+      "type": "object",
+      "properties": {
+        "name": { "type": "string", "title": "名称", "required": true, "x-component": "Input" }
+      }
+    }
+    """;
+}

tests/BuildFlow.Workflow.Tests/Performance/PreparedFormSchemaTests.cs

@@ -0,0 +1,89 @@
+using BuildFlow.Shared.Domain.Errors;
+using BuildFlow.Shared.Domain.Exceptions;
+using FluentAssertions;
+using System.Text.Json.Nodes;
+using Workflow.Application.Form.Derivation;
+using Workflow.Application.Form.Schema;
+using Xunit;
+
+namespace BuildFlow.Workflow.Tests.Performance;
+
+/// <summary>
+/// 预解析表单 Schema 在批量校验和派生中的复用测试。
+/// </summary>
+public class PreparedFormSchemaTests
+{
+    [Fact]
+    public async Task PreparedSchemaCanBeReusedForValidationAndDerivation()
+    {
+        var prepared = PreparedFormSchema.Create(ComputedSchema);
+        var derivation = new FormValueDerivationService([], []);
+
+        var firstValidation = FormDataValidator.Validate(
+            prepared,
+            """{"quantity":2,"unitPrice":10}""");
+        var first = await derivation.DeriveAsync(
+            prepared,
+            """{"quantity":2,"unitPrice":10}""",
+            new FormDerivationContext(Guid.NewGuid(), Guid.NewGuid(), null),
+            CancellationToken.None);
+        var second = await derivation.DeriveAsync(
+            prepared,
+            """{"quantity":3,"unitPrice":5}""",
+            new FormDerivationContext(Guid.NewGuid(), Guid.NewGuid(), null),
+            CancellationToken.None);
+
+        firstValidation.IsValid.Should().BeTrue();
+        ReadTotal(first.DataJson).Should().Be(20);
+        ReadTotal(second.DataJson).Should().Be(15);
+        prepared.Validation.Fields.Should().HaveCount(3);
+    }
+
+    [Fact]
+    public void InvalidSchemaCannotBePrepared()
+    {
+        var act = () => PreparedFormSchema.Create("""
+            {
+              "type": "object",
+              "properties": {
+                "supplier": {
+                  "type": "object",
+                  "x-component": "RelationPicker",
+                  "x-field-kind": "relation",
+                  "x-relation": { "idField": "id" }
+                }
+              }
+            }
+            """);
+
+        var exception = act.Should().Throw<BusinessException>().Which;
+        exception.Code.Should().Be(ErrorCode.InvalidParam);
+    }
+
+    private static decimal ReadTotal(string dataJson)
+        => JsonNode.Parse(dataJson)!["total"]!.GetValue<decimal>();
+
+    private const string ComputedSchema = """
+    {
+      "type": "object",
+      "properties": {
+        "quantity": { "type": "number", "x-component": "InputNumber" },
+        "unitPrice": { "type": "number", "x-component": "InputNumber" },
+        "total": {
+          "type": "number",
+          "x-component": "InputNumber",
+          "x-field-kind": "computed",
+          "x-derived": {
+            "mode": "computed",
+            "writePolicy": "overwrite",
+            "expression": {
+              "operator": "multiply",
+              "left": { "type": "field", "path": "quantity" },
+              "right": { "type": "field", "path": "unitPrice" }
+            }
+          }
+        }
+      }
+    }
+    """;
+}

tests/BuildFlow.Workflow.Tests/Reliability/ImportJobRecoveryTests.cs

@@ -201,7 +201,7 @@ public class ImportJobRecoveryTests(PostgresWorkflowFixture fixture)
             string formName,
             IEnumerable<string> dataJsonRows) => [];
 
-        public IReadOnlyList<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes) => rows;
+        public IEnumerable<ExcelParsedRow> Parse(string schemaJson, byte[] xlsxBytes) => rows;
     }
 
     private sealed class CancellingRows(
@@ -220,7 +220,14 @@ public class ImportJobRecoveryTests(PostgresWorkflowFixture fixture)
             }
         }
 
-        public IEnumerator<ExcelParsedRow> GetEnumerator() => rows.GetEnumerator();
+        public IEnumerator<ExcelParsedRow> GetEnumerator()
+        {
+            for (var index = 0; index < rows.Count; index++)
+            {
+                if (index == cancelAtIndex) cancel();
+                yield return rows[index];
+            }
+        }
 
         System.Collections.IEnumerator System.Collections.IEnumerable.GetEnumerator()
             => GetEnumerator();