C# 规则引擎RulesEngine

当编写应用程序时，经常性需要花费大量的时间与精力处理业务逻辑，往往业务逻辑的变化需要重构或者增加大量代码，对开发测试人员很不友好。

之前在这篇文章说过，可以使用脚本引擎来将我们需要经常变化的代码进行动态编译执行，自由度非常大，不过对应的需要资源也多。如果只是针对非常具体业务逻辑的变化，可以尝试使用RulesEngine对程序进行操作。

下文使用了官方示例且部分内容翻译自说明文档

简介

RulesEngine是微软推出的规则引擎，规则引擎在很多企业开发中有所应用，是处理经常变动需求的一种优雅的方法。个人任务，规则引擎适用于以下的一些场景：

输入输出类型数量比较固定，但是执行逻辑经常变化；
switch条件经常变化，复杂switch语句的替代；
会变动的，具有多种条件或者规则的业务逻辑；
规则自由度不要求特别高的场景。（这种情况建议使用脚本引擎）

RulesEngine的规则使用JSON进行存储，通过lambda表达式方式表述规则（Rules）。

安装很方便，直接使用nuget进行安装：

Copyinstall-pacakge RulesEngine

规则定义

需要有Rules，有WorkflowName，然后还有一些属性。

Copy[{"WorkflowName": "Discount","Rules": [{"RuleName": "GiveDiscount10","SuccessEvent": "10","ErrorMessage": "One or more adjust rules failed.","ErrorType": "Error","RuleExpressionType": "LambdaExpression","Expression": "input1.country == \"india\" AND input1.loyalityFactor <= 2 AND input1.totalPurchasesToDate >= 5000 AND input2.totalOrders > 2 AND input3.noOfVisitsPerMonth > 2"}]}
]

除了标准的RuleExpressionType，还可以通过定义Rules嵌套多个条件，下面是Or逻辑。

Copy{
"RuleName": "GiveDiscount30NestedOrExample",
"SuccessEvent": "30",
"ErrorMessage": "One or more adjust rules failed.",
"ErrorType": "Error",
"Operator": "OrElse",
"Rules":[{"RuleName": "IsLoyalAndHasGoodSpend","ErrorMessage": "One or more adjust rules failed.","ErrorType": "Error","RuleExpressionType": "LambdaExpression","Expression": "input1.loyalityFactor > 3 AND input1.totalPurchasesToDate >= 50000 AND input1.totalPurchasesToDate <= 100000"},{"RuleName": "OrHasHighNumberOfTotalOrders","ErrorMessage": "One or more adjust rules failed.","ErrorType": "Error","RuleExpressionType": "LambdaExpression","Expression": "input2.totalOrders > 15"}
]
}

示例

可以从官方的代码库中下载示例，定义了上述规则，就可以直接开始用了。示例描述了这么一个应用场景：

根据不同的客户属性，提供不同的折扣。由于销售的情况变化较快，提供折扣的规则也需要经常变动。因此比较适用于规则引擎。

Copypublic void Run()
{Console.WriteLine($"Running {nameof(BasicDemo)}....");//创建输入var basicInfo = "{\"name\": \"hello\",\"email\": \"abcy@xyz.com\",\"creditHistory\": \"good\",\"country\": \"canada\",\"loyalityFactor\": 3,\"totalPurchasesToDate\": 10000}";var orderInfo = "{\"totalOrders\": 5,\"recurringItems\": 2}";var telemetryInfo = "{\"noOfVisitsPerMonth\": 10,\"percentageOfBuyingToVisit\": 15}";var converter = new ExpandoObjectConverter();dynamic input1 = JsonConvert.DeserializeObject<ExpandoObject>(basicInfo, converter);dynamic input2 = JsonConvert.DeserializeObject<ExpandoObject>(orderInfo, converter);dynamic input3 = JsonConvert.DeserializeObject<ExpandoObject>(telemetryInfo, converter);var inputs = new dynamic[]{input1,input2,input3};//加载规则var files = Directory.GetFiles(Directory.GetCurrentDirectory(), "Discount.json", SearchOption.AllDirectories);if (files == null || files.Length == 0)throw new Exception("Rules not found.");var fileData = File.ReadAllText(files[0]);var workflowRules = JsonConvert.DeserializeObject<List<WorkflowRules>>(fileData);//初始化规则引擎var bre = new RulesEngine.RulesEngine(workflowRules.ToArray(), null);string discountOffered = "No discount offered.";//执行规则List<RuleResultTree> resultList = bre.ExecuteAllRulesAsync("Discount", inputs).Result;//处理结果resultList.OnSuccess((eventName) => {discountOffered = $"Discount offered is {eventName} % over MRP.";});resultList.OnFail(() => {discountOffered = "The user is not eligible for any discount.";});Console.WriteLine(discountOffered);
}

输入

输入一般来说是IEnumerable<dynamic>或者是匿名类型，上面实例展示的是由json反序列化形成的dynamic类型，对于程序生成的数据，使用匿名类型更加方便。

Copyvar nestedInput = new {SimpleProp = "simpleProp",NestedProp = new {SimpleProp = "nestedSimpleProp",ListProp = new List<ListItem>{new ListItem{Id = 1,Value = "first"},new ListItem{Id = 2,Value = "second"}}}};

命名空间

和脚本引擎一样，默认规则引擎只能访问System的命名空间。如果需要使用到稍微复杂一些的类型，可以自己定义类型或者函数。比如定义一个这样的函数：

Copypublic static class Utils
{public static bool CheckContains(string check, string valList){if (String.IsNullOrEmpty(check) || String.IsNullOrEmpty(valList))return false;var list = valList.Split(',').ToList();return list.Contains(check);}
}

需要使用的时候，先将类传递给RulesEngine：

Copyvar reSettingsWithCustomTypes = new ReSettings { CustomTypes = new Type[] { typeof(Utils) } };
var engine = new RulesEngine.RulesEngine(workflowRules.ToArray(), null, reSettingsWithCustomTypes);

然后就可以直接在表达式中使用了。

Copy"Expression": "Utils.CheckContains(input1.country, \"india,usa,canada,France\") == true"

规则参数

默认情况下，规则的输入使用的是类似input1 input2这样的形式，如果想直观一点，可以使用RuleParameter来进行封装具体的参数类型。

CopyRuleParameter ruleParameter = new RuleParameter("NIP", nestedInput);
var resultList = bre.ExecuteAllRulesAsync(workflow.WorkflowName, ruleParameter).Result;

本地变量

如果表达式比较复杂的情况下，可以使用本地变量来进行分段处理，这对调试来说会比较方便。

本地变量的关键字为localParams，可以将中间的内容简单理解成var name = expression

Copy{"name": "allow_access_if_all_mandatory_trainings_are_done_or_access_isSecure","errorMessage": "Please complete all your training(s) to get access to this content or access it from a secure domain/location.","errorType": "Error","localParams": [{"name": "completedSecurityTrainings","expression": "MasterSecurityComplainceTrainings.Where(Status.Equals(\"Completed\", StringComparison.InvariantCultureIgnoreCase))"},{"name": "completedProjectTrainings","expression": "MasterProjectComplainceTrainings.Where(Status.Equals(\"Completed\", StringComparison.InvariantCultureIgnoreCase))"},{"name": "isRequestAccessSecured","expression": "UserRequestDetails.Location.Country == \"India\" ? ((UserRequestDetails.Location.City == \"Bangalore\" && UserRequestDetails.Domain=\"xxxx\")? true : false):false"}],"expression": "(completedSecurityTrainings.Any() && completedProjectTrainings.Any()) || isRequestAccessSecured "}