zszq-trs/YLErpDAL/Modules/RiskEngine/Compile/RuleCompiler.cs

using Microsoft.CodeAnalysis;
using Microsoft.CodeAnalysis.CSharp;
using Microsoft.CodeAnalysis.CSharp.Scripting;
using Microsoft.CodeAnalysis.CSharp.Syntax;
using Microsoft.CodeAnalysis.Scripting;
using Newtonsoft.Json;
using System;
using System.Linq;
using YLErp.BLL;
using YLErp.QdpModule;

namespace YLErp.Modules.RiskEngine
{
    /// <summary>
    /// Roslyn 脚本全局变量容器
    /// </summary>
    public class ScriptGlobals
    {
        /// <summary>
        /// 当前交易ID
        /// </summary>
        public int TradeId { get; set; }

        /// <summary>
        /// 当前触发时点
        /// </summary>
        public string TriggerPoint { get; set; }

        /// <summary>
        /// 数据库上下文，供规则公式直接查询数据库
        /// </summary>
        public YLContext DbContext { get; set; }
    }

    /// <summary>
    /// 规则编译器（Roslyn 版）
    ///
    /// 编译时机：规则定义时（零运行时解析开销）
    /// 执行方式：判风控时直接调用 rule.CompiledScript(context) → bool
    ///
    /// 相比表达式树方案的优势：
    /// - 代码简洁：生成 C# 字符串即可，无需手动拼接 Expression 节点
    /// - 灵活性高：天然支持多条件 AND/OR、复杂运算、未来可扩展任意 C# 语法
    /// - 可调试：生成的脚本代码可直接阅读和理解
    ///
    /// 示例：
    ///   RuleExpr: Convert.ToDecimal(DbContext.trade.First(t => t.id == TradeId).StockEqvNotional) > 100000000m
    ///   编译结果：Func&lt;RiskContext, bool&gt;（内部通过 Roslyn 编译缓存）
    /// </summary>
    public static class RuleCompiler
    {
        private static IYcLogger _logger = LogFactory.GetLogger("RuleCompiler");

        /// <summary>
        /// 高风险方法黑名单。
        /// 这里按方法名做语法层拦截，覆盖数据库写入、原生 SQL、反射、文件、进程、服务定位等入口。
        /// </summary>
        private static readonly HashSet<string> ForbiddenInvocationNames = new HashSet<string>(StringComparer.OrdinalIgnoreCase)
        {
            "SaveChanges",
            "SaveChangesAsync",
            "Add",
            "AddAsync",
            "AddRange",
            "AddRangeAsync",
            "AddOrUpdate",
            "Update",
            "UpdateRange",
            "Remove",
            "RemoveRange",
            "Attach",
            "AttachRange",
            "Set",
            "Entry",
            "ExecuteSqlRaw",
            "ExecuteSqlRawAsync",
            "ExecuteSqlInterpolated",
            "ExecuteSqlInterpolatedAsync",
            "ExecuteSql",
            "ExecuteSqlAsync",
            "ExecuteSqlCommand",
            "ExecuteSqlCommandAsync",
            "ExecuteStoreCommand",
            "ExecuteStoreQuery",
            "ExecuteDelete",
            "ExecuteDeleteAsync",
            "ExecuteUpdate",
            "ExecuteUpdateAsync",
            "FromSqlRaw",
            "FromSqlRawAsync",
            "FromSqlInterpolated",
            "FromSqlInterpolatedAsync",
            "BulkInsert",
            "BulkUpdate",
            "BulkDelete",
            "BulkMerge",
            "GetType",
            "Invoke",
            "InvokeMember",
            "GetMethod",
            "GetProperty",
            "GetField",
            "CreateInstance",
            "Load",
            "LoadFrom",
            "ReadAllText",
            "ReadAllLines",
            "ReadAllBytes",
            "WriteAllText",
            "WriteAllLines",
            "WriteAllBytes",
            "AppendAllText",
            "Delete",
            "Move",
            "Copy",
            "Open",
            "OpenRead",
            "OpenWrite",
            "Create",
            "CreateText",
            "Start",
            "Kill",
            "Exit",
            "GetEnvironmentVariable",
            "SetEnvironmentVariable",
            "GetService",
            "GetRequiredService",
            "CreateScope",
            "Sleep"
        };

        /// <summary>
        /// 高风险成员黑名单。
        /// DbContext.Database、ChangeTracker 等成员会绕过只读查询约束或暴露底层状态，禁止脚本访问。
        /// </summary>
        private static readonly HashSet<string> ForbiddenMemberNames = new HashSet<string>(StringComparer.OrdinalIgnoreCase)
        {
            "Database",
            "ChangeTracker",
            "Assembly",
            "AppDomain"
        };

        /// <summary>
        /// 高风险类型或标识符黑名单。
        /// 用于拦截 File、Process、Environment 等直接作为类型或变量名出现的绕过方式。
        /// </summary>
        private static readonly HashSet<string> ForbiddenTypeOrIdentifierNames = new HashSet<string>(StringComparer.OrdinalIgnoreCase)
        {
            "Activator",
            "Assembly",
            "AppDomain",
            "Environment",
            "File",
            "Directory",
            "Path",
            "Process",
            "HttpClient",
            "WebClient",
            "WebRequest",
            "Socket",
            "Thread",
            "GC"
        };

        /// <summary>
        /// 校验并编译规则表达式。
        /// 当前 RuleExpr 要求是 Roslyn 可直接执行的 bool 表达式。
        /// ruleId 可为空，用于规则未落库时的公式校验。
        /// </summary>
        public static RuleCompileResult ValidateAndCompileFormula(long? ruleId, string formulaExp)
        {
            var ruleIdText = ruleId.HasValue ? ruleId.Value.ToString() : "未落库";
            if (string.IsNullOrWhiteSpace(formulaExp))
            {
                _logger.Error($"规则表达式为空，无法编译 - RuleId: {ruleIdText}");
                return RuleCompileResult.Fail("规则 RuleExpr 不能为空");
            }

            try
            {
                var compiled = CompileScript(ruleId, formulaExp, out string compileErrorMessage);
                if (compiled == null)
                {
                    return RuleCompileResult.Fail(compileErrorMessage ?? "脚本编译失败");
                }

                return RuleCompileResult.Ok(compiled);
            }
            catch (Exception ex)
            {
                _logger.Error($"规则表达式校验异常 - RuleId: {ruleIdText}, Error: {ex.Message}\n脚本代码：{formulaExp}");
                return RuleCompileResult.Fail(ex.Message);
            }
        }

        /// <summary>
        /// 校验变量取值表达式。
        /// 变量表达式只要求能编译为 object，不在保存时执行，避免依赖真实交易和数据库数据。
        /// </summary>
        public static RuleCompileResult ValidateVariableExpression(string variableExpr)
        {
            if (string.IsNullOrWhiteSpace(variableExpr))
                return RuleCompileResult.Fail("变量取值表达式不能为空");

            var compiled = CompileVariableExpression(variableExpr, out var compileErrorMessage);
            if (compiled == null)
                return RuleCompileResult.Fail(compileErrorMessage ?? "变量取值表达式编译失败");

            return RuleCompileResult.Ok(null);
        }

        /// <summary>
        /// 编译变量取值表达式。
        /// 结构化规则执行时使用，表达式返回值由执行器按变量 DataType 统一转换。
        /// </summary>
        public static Func<RiskContext, object> CompileValueExpression(string variableExpr, out string errorMessage)
        {
            if (string.IsNullOrWhiteSpace(variableExpr))
            {
                errorMessage = "变量取值表达式不能为空";
                return null;
            }

            return CompileVariableExpression(variableExpr, out errorMessage);
        }

        /// <summary>
        /// 编译变量取值表达式。优先兼容原有单表达式写法；失败后再按多语句脚本编译。
        /// 多语句脚本支持用 return 表达最终值，编译前会转换为 Roslyn Script 稳定支持的末尾表达式。
        /// </summary>
        private static Func<RiskContext, object> CompileVariableExpression(string variableExpr, out string errorMessage)
        {
            if (IsLikelyVariableScript(variableExpr))
            {
                var normalizedScript = NormalizeVariableScript(variableExpr);
                var scriptCompiled = CompileScript<object>(null, normalizedScript, out errorMessage);
                return scriptCompiled;
            }

            // 原有变量表达式大多是单个查询表达式，包成 object 后可直接作为脚本返回值。
            var expressionScript = $"(object)({variableExpr})";
            var compiled = CompileScript<object>(null, expressionScript, out var expressionErrorMessage);
            if (compiled != null)
            {
                errorMessage = null;
                return compiled;
            }

            // 单表达式编译失败时，再按多语句脚本兜底，兼容包含局部变量、if、throw 的复杂变量。
            var normalizedFallbackScript = NormalizeVariableScript(variableExpr);
            compiled = CompileScript<object>(null, normalizedFallbackScript, out var scriptErrorMessage);
            if (compiled != null)
            {
                errorMessage = null;
                return compiled;
            }

            errorMessage = $"单表达式编译失败：{expressionErrorMessage}；多语句脚本编译失败：{scriptErrorMessage}";
            return null;
        }

        /// <summary>
        /// 粗略判断变量表达式是否更像多语句脚本。
        /// 多语句脚本不能再包成 (object)(...)，需要直接按 C# Script 编译。
        /// </summary>
        private static bool IsLikelyVariableScript(string variableExpr)
        {
            return variableExpr.IndexOf(';') >= 0 ||
                variableExpr.IndexOf("return", StringComparison.OrdinalIgnoreCase) >= 0 ||
                variableExpr.IndexOf("throw", StringComparison.OrdinalIgnoreCase) >= 0 ||
                variableExpr.IndexOf("if", StringComparison.OrdinalIgnoreCase) >= 0;
        }

        /// <summary>
        /// 规范化多语句变量脚本。
        /// Roslyn Script 更稳定的返回方式是“最后一行表达式”，这里把末尾 return 表达式转换为末尾表达式。
        /// </summary>
        private static string NormalizeVariableScript(string variableExpr)
        {
            var syntaxTree = CSharpSyntaxTree.ParseText(variableExpr, new CSharpParseOptions(kind: SourceCodeKind.Script));
            var root = syntaxTree.GetCompilationUnitRoot();
            var lastStatement = root.Members
                .OfType<GlobalStatementSyntax>()
                .Select(statement => statement.Statement)
                .LastOrDefault();

            if (lastStatement is ReturnStatementSyntax returnStatement && returnStatement.Expression != null)
            {
                var returnText = returnStatement.ToFullString();
                var expressionText = returnStatement.Expression.ToFullString();
                var index = variableExpr.LastIndexOf(returnText, StringComparison.Ordinal);
                if (index >= 0)
                    return variableExpr.Substring(0, index) + expressionText;
            }

            return variableExpr;
        }

        /// <summary>
        /// 校验并编译规则，编译成功后同时写入规则对象和内存缓存。
        /// </summary>
        public static RuleCompileResult ValidateAndCompileRule(RiskRule rule)
        {
            if (rule == null)
            {
                _logger.Error("规则对象为空，无法编译 - RuleId: 0");
                return RuleCompileResult.Fail("规则不能为空");
            }

            if (rule.Id <= 0)
            {
                _logger.Error($"规则 Id 非法，无法编译 - RuleId: {rule.Id}");
                return RuleCompileResult.Fail("规则 Id 不能为空");
            }
            if (rule.Status != RiskRuleStatus.Active)
            {
                _logger.Error($"规则状态非启用，跳过编译 - RuleId: {rule.Id}, Status: {rule.Status}");
                return RuleCompileResult.Fail("规则 状态为未启动");
            }

            var result = ValidateAndCompileFormula(rule.Id, rule.RuleExpr);
            if (!result.Success)
            {
                _logger.Error($"规则校验失败 - RuleId: {rule.Id}, Error: {result.ErrorMessage}");
                rule.CompiledScript = null;
                return RuleCompileResult.Fail($"规则[{rule.Id}]编译失败：{result.ErrorMessage}");
            }

            rule.CompiledScript = result.CompiledScript;
            RuleCompiledCache.Set(rule.Id.ToString(), result.CompiledScript);
            return result;
        }

        /// <summary>
        /// 用 Roslyn 编译 C# 布尔表达式为可执行委托。
        /// </summary>
        private static Func<RiskContext, bool> CompileScript(long? ruleId, string scriptCode, out string errorMessage)
        {
            return CompileScript<bool>(ruleId, scriptCode, out errorMessage);
        }

        /// <summary>
        /// 用 Roslyn 编译 C# 脚本代码为指定返回类型的可执行委托。
        /// 规则公式编译为 bool，变量取值表达式编译为 object。
        /// </summary>
        private static Func<RiskContext, T> CompileScript<T>(long? ruleId, string scriptCode, out string errorMessage)
        {
            errorMessage = null;
            var ruleIdText = ruleId.HasValue ? ruleId.Value.ToString() : "未落库";

            if (!ValidateScriptSafety(scriptCode, out errorMessage))
            {
                _logger.Error($"规则脚本安全校验失败 - RuleId: {ruleIdText}, Error: {errorMessage}\n脚本代码：{scriptCode}");
                return null;
            }

            // 配置编译选项：引用必要的程序集
            var options = ScriptOptions.Default
                .WithReferences(
                    typeof(RiskContext).Assembly,
                    typeof(RiskCalendarHelper).Assembly,
                    typeof(YLContext).Assembly,
                    typeof(YLErp.DBModels.trade).Assembly,
                    typeof(QdpCalendarHelper).Assembly,
                    typeof(JsonConvert).Assembly,
                    typeof(Microsoft.EntityFrameworkCore.DbContext).Assembly,
                    typeof(Queryable).Assembly
                )
                .WithImports("System", "System.Linq", "Newtonsoft.Json", "YLErp.DBModels", "YLErp.QdpModule", "YLErp.Modules.RiskEngine");

            // 创建脚本（尚未执行，仅编译）
            var script = CSharpScript.Create<T>(scriptCode, options, globalsType: typeof(ScriptGlobals));

            // 编译（提前发现语法错误）
            //后续编译需提供接口，返回前端编译信息，包含编译错误列表
            var compilation = script.GetCompilation();
            var diagnostics = compilation.GetDiagnostics();
            var errors = diagnostics.Where(d => d.Severity == Microsoft.CodeAnalysis.DiagnosticSeverity.Error).ToList();
            if (errors.Any())
            {
                errorMessage = string.Join("; ", errors.Select(e => e.GetMessage()));
                _logger.Error($"规则编译失败 - RuleId: {ruleIdText}, Error: {errorMessage}\n脚本代码：{scriptCode}");
                return null;
            }

            // 生成可调用委托
            var runner = script.CreateDelegate();

            // 包装为同步的 Func<RiskContext, T>
            return ctx =>
            {
                try
                {
                    var globals = new ScriptGlobals
                    {
                        TradeId = ctx.TradeId,
                        TriggerPoint = ctx.TriggerPoint,
                        DbContext = ctx.DbContext
                    };
                    return runner(globals).GetAwaiter().GetResult();
                }
                catch (Exception ex)
                {
                    // 脚本执行异常（如空引用、类型转换失败）交由执行层按阻断处理
                    _logger.Error($"规则执行异常 - RuleId: {ruleIdText}, Error: {ex.Message}\n脚本代码：{scriptCode}");
                    throw;
                }
            };
        }

        /// <summary>
        /// 脚本安全校验。
        /// 在 Roslyn 编译前先扫描语法树，禁止写库、反射、文件、网络、进程等高风险语法入口。
        /// </summary>
        private static bool ValidateScriptSafety(string scriptCode, out string errorMessage)
        {
            var tree = CSharpSyntaxTree.ParseText(scriptCode, new CSharpParseOptions(kind: SourceCodeKind.Script));
            var root = tree.GetRoot();

            // 允许变量声明初始化（如 decimal a = ...），但禁止后续赋值或复合赋值，避免脚本修改对象状态。
            if (root.DescendantNodes().OfType<AssignmentExpressionSyntax>().Any())
            {
                errorMessage = "规则表达式不允许包含赋值语句";
                return false;
            }

            if (root.DescendantNodes().OfType<PrefixUnaryExpressionSyntax>().Any(IsIncrementOrDecrement) ||
                root.DescendantNodes().OfType<PostfixUnaryExpressionSyntax>().Any(IsIncrementOrDecrement))
            {
                errorMessage = "规则表达式不允许包含自增或自减语句";
                return false;
            }

            foreach (var memberAccess in root.DescendantNodes().OfType<MemberAccessExpressionSyntax>())
            {
                var memberName = memberAccess.Name.Identifier.ValueText;
                if (!string.IsNullOrWhiteSpace(memberName) && ForbiddenMemberNames.Contains(memberName))
                {
                    errorMessage = $"规则表达式不允许访问高风险成员：{memberName}";
                    return false;
                }
            }

            // 只拦截独立标识符，避免把实体字段 p.File、p.Path 误判为高风险类型名。
            foreach (var identifier in root.DescendantNodes().OfType<IdentifierNameSyntax>())
            {
                var identifierName = identifier.Identifier.ValueText;
                if (!IsMemberAccessName(identifier) && !string.IsNullOrWhiteSpace(identifierName) && ForbiddenTypeOrIdentifierNames.Contains(identifierName))
                {
                    errorMessage = $"规则表达式不允许访问高风险类型或标识符：{identifierName}";
                    return false;
                }
            }

            foreach (var objectCreation in root.DescendantNodes().OfType<ObjectCreationExpressionSyntax>())
            {
                var typeName = GetTypeName(objectCreation.Type);
                if (!string.IsNullOrWhiteSpace(typeName) && ForbiddenTypeOrIdentifierNames.Contains(typeName))
                {
                    errorMessage = $"规则表达式不允许创建高风险类型：{typeName}";
                    return false;
                }
            }

            foreach (var invocation in root.DescendantNodes().OfType<InvocationExpressionSyntax>())
            {
                var methodName = GetInvocationName(invocation.Expression);
                if (!string.IsNullOrWhiteSpace(methodName) && ForbiddenInvocationNames.Contains(methodName))
                {
                    errorMessage = $"规则表达式不允许调用高风险方法：{methodName}";
                    return false;
                }
            }

            errorMessage = null;
            return true;
        }

        /// <summary>
        /// 判断前缀一元表达式是否为自增/自减。
        /// </summary>
        private static bool IsIncrementOrDecrement(PrefixUnaryExpressionSyntax expression)
        {
            return expression.IsKind(SyntaxKind.PreIncrementExpression) || expression.IsKind(SyntaxKind.PreDecrementExpression);
        }

        /// <summary>
        /// 判断后缀一元表达式是否为自增/自减。
        /// </summary>
        private static bool IsIncrementOrDecrement(PostfixUnaryExpressionSyntax expression)
        {
            return expression.IsKind(SyntaxKind.PostIncrementExpression) || expression.IsKind(SyntaxKind.PostDecrementExpression);
        }

        /// <summary>
        /// 判断标识符是否是成员访问右侧名称，例如 p.Path 中的 Path。
        /// </summary>
        private static bool IsMemberAccessName(IdentifierNameSyntax identifier)
        {
            return identifier.Parent is MemberAccessExpressionSyntax memberAccess && memberAccess.Name == identifier;
        }

        /// <summary>
        /// 从调用表达式中提取方法名，兼容 SaveChanges() 和 DbContext.SaveChanges() 两种写法。
        /// </summary>
        private static string GetInvocationName(ExpressionSyntax expression)
        {
            return expression switch
            {
                MemberAccessExpressionSyntax memberAccess => memberAccess.Name.Identifier.ValueText,
                IdentifierNameSyntax identifier => identifier.Identifier.ValueText,
                _ => null
            };
        }

        /// <summary>
        /// 从类型语法中提取类型短名，用于匹配高风险类型黑名单。
        /// </summary>
        private static string GetTypeName(TypeSyntax type)
        {
            return type switch
            {
                IdentifierNameSyntax identifier => identifier.Identifier.ValueText,
                QualifiedNameSyntax qualifiedName => qualifiedName.Right.Identifier.ValueText,
                AliasQualifiedNameSyntax aliasQualifiedName => aliasQualifiedName.Name.Identifier.ValueText,
                _ => type?.ToString()
            };
        }
    }

    public class BuildMemberAccessResult
    {
        public bool Success { get; set; }

        public string Expression { get; set; }

        public string ErrorMessage { get; set; }

        public static BuildMemberAccessResult Ok(string expression)
        {
            return new BuildMemberAccessResult
            {
                Success = true,
                Expression = expression
            };
        }

        /// <summary>
        /// 将操作符统一为 C# 操作符
        /// </summary>
        private static string FormatOperator(string op)
        {
            return op?.Trim() switch
            {
                // 数值型 C# 符号
                "=" or "==" => "==",
                "!=" or "≠" => "!=",
                ">" => ">",
                "<" => "<",
                ">=" => ">=",
                "<=" => "<=",
                // 日期型中文语义 → C# 符号
                "早于" => "<",
                "晚于" => ">",
                "等于" => "==",
                "不早于" => ">=",
                "不晚于" => "<=",
                _ => throw new ArgumentException($"不支持的操作符：{op}")
            };
        }
        public static BuildMemberAccessResult Fail(string errorMessage)
        {
            return new BuildMemberAccessResult
            {
                Success = false,
                ErrorMessage = errorMessage
            };
        }
    }
}