this is working pretty fine:
new dictionary {{0,0}, {1,1}, {3,2}} [3]

this is throwing exception:
new dictionary {{0,0}, {1,1}, {3,2}} [x.Sum()]

System.Reflection.TargetInvocationException : Exception has been thrown by the target of an invocation.
----> System.Exception : The given key was not present in the dictionary.

x is instance of class with object Sum() method, that returns 1

I am upgrading my nugets to .NET Standard, and while upgrading NReco.LambdaParser in my .NET 4.6.1 projects, I get the following error:

Failed to add reference. The package 'NReco.LambdaParser' tried to add a framework reference to 'System.Collections' which was not found in the GAC. This is possibly a bug in the package. Please contact the package owners for assistance.

Is this a known issue, by chance?

Migrate to .NET Standard2.0.

We are getting this Exception on iOs. On Android it works perfectly.

System.PlatformNotSupportedException: Operation is not supported on this platform.
at System.Linq.Expressions.Compiler.DelegateHelpers.MakeNewDelegate (System.Type[] types) <0x102e341e0 + 0x000cc> in <bb5b9967994b42d38a9a568699a96020#cbe1a7eadba32ea3a937327d324ab3f9>:0
at System.Linq.Expressions.Compiler.DelegateHelpers.MakeDelegateType (System.Type[] types) <0x102e33990 + 0x00133> in <bb5b9967994b42d38a9a568699a96020#cbe1a7eadba32ea3a937327d324ab3f9>:0
at System.Linq.Expressions.Expression.Lambda (System.Linq.Expressions.Expression body, System.String name, System.Boolean tailCall, System.Collections.Generic.IEnumerable1[T] parameters) <0x102dcea00 + 0x001bf> in <bb5b9967994b42d38a9a568699a96020#cbe1a7eadba32ea3a937327d324ab3f9>:0 at System.Linq.Expressions.Expression.Lambda (System.Linq.Expressions.Expression body, System.Boolean tailCall, System.Collections.Generic.IEnumerable1[T] parameters) <0x102dce980 + 0x00027> in <bb5b9967994b42d38a9a568699a96020#cbe1a7eadba32ea3a937327d324ab3f9>:0
at System.Linq.Expressions.Expression.Lambda (System.Linq.Expressions.Expression body, System.Linq.Expressions.ParameterExpression[] parameters) <0x102dce950 + 0x0001f> in <bb5b9967994b42d38a9a568699a96020#cbe1a7eadba32ea3a937327d324ab3f9>:0
at NReco.Linq.LambdaParser.Eval (System.String expr, System.Func2[T,TResult] getVarValue) <0x103f20090 + 0x0016b> in <ebdc87cda8e4462a9cdd9c66afdb543c#cbe1a7eadba32ea3a937327d324ab3f9>:0 at NReco.Linq.LambdaParser.Eval (System.String expr, System.Collections.Generic.IDictionary2[TKey,TValue] vars) <0x103f1ff50 + 0x00113> in <ebdc87cda8e4462a9cdd9c66afdb543c#cbe1a7eadba32ea3a937327d324ab3f9>:0
at CPM.Arda.Mobile.Questionnaire.Logic+<>c__DisplayClass0_0`1+<b__0>d[T].MoveNext () <0x103f19220 + 0x0025f> in <3200b1f0bb96405aa11baec7b62f15e6#cbe1a7eadba32ea3a937327d324ab3f9>:0
ExpressionGroudId: 7e7207be-fd8c-40e1-9b90-0ca5bad1a5ec
EvalString: (val1.Contains(val2) || val3.Contains(val4) || val5.Contains(val6) || val7.Contains(val8) || val9.Contains(val10) || val11.Contains(val12) || val13.Contains(val14) || val15.Contains(val16) || val17.Contains(val18) || val19.Contains(val20))
EvalArguments: [val17, ][val16, 8][val5, ][val15, ][val1, ][val14, 7][val13, ][val9, ][val12, 6][val11, ][val10, 5][val4, 2][val8, 4][val7, ][val3, ][val6, 3][val2, 1][val19, ][val20, 10;][val18, 9]

This is the expression string:

(val1.Contains(val2) || val3.Contains(val4) || val5.Contains(val6) || val7.Contains(val8) || val9.Contains(val10) || val11.Contains(val12) || val13.Contains(val14) || val15.Contains(val16) || val17.Contains(val18) || val19.Contains(val20))

And the values:

[val20, 10;]
[val17, 2;4;5;7;8;]
[val12, 6]
[val7, 2;4;5;7;8;]
[val2, 1]
[val19, 2;4;5;7;8;]
[val13, 2;4;5;7;8;]
[val8, 4]
[val3, 2;4;5;7;8;]
[val1, 2;4;5;7;8;]
[val5, 2;4;5;7;8;]
[val16, 8]
[val18, 9]
[val14, 7]
[val4, 2]
[val11, 2;4;5;7;8;]
[val9, 2;4;5;7;8;]
[val10, 5]
[val15, 2;4;5;7;8;]
[val6, 3]

For example, a=b and a==b, the effect on both is the same.

I just discovred your library and its awesome.
I am just wondering if it would be possible to do this kind of operations :
NOT(NOT(true))
or
NOT(1==1)
I tried it and obtained this exception :
System.InvalidOperationException: 'An expression of type 'System.Boolean' cannot be used to initialize an array of type 'System.Object''

The idea behind those simple exemples is to allow the possibility to chain operations.

Regards,

Eric

Right now all invocations are performed through reflection (InvokeMethod / InvokeDelegate / InvokeIndexer / InvokePropertyOrField). At the same time, evaluation performance is acceptable for most applications: 10,000 evals take about 20-30ms (depending on CPU).

It is possible to improve evaluation performance in cases if invocation is not ambiguous (only one method overload is available according to number of parameters); this might be useful if LambdaParser is used for in-memory filtering of large dataset (say, >100k rows) by user-defined expression.

If someone needs this improvement please leave a comment or vote for this issue.

Hi,

after a lot of searching I came across this package which seemed perfect for what I needed.

I'm basically trying to filter a DataTable with a LINQ query that's been generated in runtime.

Imagine I have the DataTable dtToFilter, I'd do something like this: (From row In dtToFilter.AsEnumerable Where row("Name").ToString.StartsWith("A") OR row("Name").ToString.StartsWith("B") AND row("Age").ToString = "18" AND row("Occupation").ToString.Contains("rpa") Select row).CopyToDataTable

In my case, the above code is generated an therefore available in a string variable (filterString). My idea was to use your package like this:

Dim lambdaParser As NReco.Linq.LambdaParser = New NReco.Linq.LambdaParser()
Dim context As New Dictionary(Of String, Object)
context.Add("dtToFilter", dtToFilter)
output = lambdaParser.Eval(filterString, context)

Unfortunately it throws the following error:

 NReco.Linq.LambdaParserException: Expected ')' at 5: (From row In dtToFilter.AsEnumerable Where row("Name").ToString.StartsWith("A") OR row("Name").ToString.StartsWith("B") AND row("Age").ToString = "18" AND row("Occupation").ToString.Contains("rpa") Select row).CopyToDataTable 
   at NReco.Linq.LambdaParser.ParseValue(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParsePrimary(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseUnary(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseMultiplicative(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseAdditive(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseEq(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseAnd(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseOr(String expr, Int32 start)
   at NReco.Linq.LambdaParser.ParseConditional(String expr, Int32 start)
   at NReco.Linq.LambdaParser.Parse(String expr)
   at NReco.Linq.LambdaParser.Eval(String expr, Func`2 getVarValue)
   at NReco.Linq.LambdaParser.Eval(String expr, IDictionary`2 vars)
   at UiPathCodeRunner_3e7f94b226e64e9ab40aa7833cf4c058.Run(String filterString, DataTable dtToFilter, Object& output)

Do you have any tips on how to solve this or is this not possible?

Edit: Here's a simple fiddle: https://dotnetfiddle.net/uK5Mbe

parsing fails with string in expression

test string:

var exp = "true ? \"<br><br><span style=\\\"color:red;\\\">hello</span> world\" : \"\""
            
var result = SUT.Eval(exp, new Dictionary<string, object>());

fails with

NReco.Linq.LambdaParserException : Expected ':' at 30: true ? "<br><br><span style=\"color:red;\">hello</span> world" : ""
   at NReco.Linq.LambdaParser.ParseConditional(String expr, Int32 start)
   at NReco.Linq.LambdaParser.Parse(String expr)
   at NReco.Linq.LambdaParser.Eval(String expr, Func`2 getVarValue)
   at NReco.Linq.LambdaParser.Eval(String expr, IDictionary`2 vars)

var res = lambdaParser.Eval("Convert.ToInt32("111")", varContext)

this is not work well. how to support more system namespace class and functions?

Hi,

first of all thank you for you project. It is helping a lot in a project I am working on.

There is an issue when I use null to evaluate a boolean expression, for example:

20 == null

In this case, null is considered a variable and it is expected on the values dictionary.

The workaround I am using is:
values["null"] = null;

Thanks again and regards,

Hey guys, I'm trying to perform the expression below like Sum(x=>x.UnitQuantity) or Select(x=>x...) but I'm unable to parse it properly and I am getting this exception error below. Am I missing something with my expression?

NReco.Linq.LambdaParserException: 'Expected value at 22: order.OrderLines.Sum(d => d.UnitPrice * d.UnitQuantity)'

`

       var lambdaParser = new LambdaParser();
        var order = new Order()
        {
            Id = 1,
            Paid = true,
            OrderReference = "Animal",
            OrderLines = new List<OrderLine>
                {
                    new OrderLine
                    {
                        Id = 1,
                        UnitQuantity = 3,
                        Category = "CAT",
                        UnitPrice = 1.1M
                    },
                    new OrderLine
                    {
                        Id = 2,
                        UnitQuantity = 7,
                        Category = "DOG",
                        UnitPrice = 3.1M
                    }
                }
        };

        var context = new Dictionary<string, object>
        {
            ["order"] = order
        };

        //var expression = "order.Paid ? order.OrderLines.Sum(d=>d.UnitPrice * d.UnitQuantity) : order.OrderLines.Sum(d=>d.UnitPrice * d.UnitQuantity) * 2.0M";
					
  var expression = "order.OrderLines.Sum(d => d.UnitPrice * d.UnitQuantity)";
       var result = lambdaParser.Eval(expr, context);

`

I get this error when i try to add the latest version of this nuget package to my Xamarin.Forms project.

Failed to add reference. The package 'NReco.LambdaParser' tried to add a framework reference to 'System.Collections' which was not found in the GAC. This is possibly a bug in the package. Please contact the package owners for assistance.

Hello,

I have been trying to figure out why a more complex expression using vars was failing, but after working it down to a simple example, I still get an unexpected failure:

var result = parser.Eval("200 / 150 >= .75", vars);

Expected value at 12: 200 / 150 >= .75

I have tried using parentheses also, to no avail. It looks from the unit tests that all of the basic operators and such that I'm using are supported (even in the more complex example), so I'm scratching my head hoping it's not something so blindingly obvious that I'm going to feel like an idiot :)

I have used this parser for more complex expressions with no issue, and the only difference in this case that I can discern is that this is a nested operation (whereby 200 / 150 needs to be evaluated, and then the result used for the comparison). But, some of the unit tests seem to indicate this is possible, for example:

Assert.Equal(true, lambdaParser.Eval(" (1+testObj.IntProp)==2 ? testObj.FldTrue : false ", varContext));

So, I'm at a loss! What have I missed?

Motivation: in complex expressions where the same calculation result may be used more than once it expression may become monstrous because of copy-pastes of the same sub-expression, and also this causes multiple evaluations of the same value.

To prevent that let's add support of local variables that may be defined in that way:

var a = 5; var b = fromContext*2;  b>a ? b : a 

To keep full backward compatibility variables syntax should be controlled via LambdaParser.AllowVars property (false by default).

Expression of "values.Count" is working.
But values.Max or values.Min is not working...
can you fix it?

Not an issue but a question - company I work for is interested in using the library but wants to know what security constraints are imposed within the parser context.
Specifically things that help avoid equivalent of sql injection attacks.

Currently underscore '_' is allowed only as first letter of names ( ReadLexem method ).

Current comparison logic handles nulls in the following way:

• if a=null then "a==null" return true
• if a not null then "a>null" returns true

In some cases SQL-like behavior is desired when any comparison with null returns false.
For this purpose

• lets enhance API to add ability to specify custom comparer
• add "NullBehaviour" property to ValueComparer implementation. By default current logic is preserved ("MinValue") and lets add "Sql" option that returns false if any operand is null.

Hi, found a problem when evaluating a function call on an externally supplied function - it only works for parameters that implement IConvertible. Simple value types have it but custom classes don't - then the eval fails with InvalidCastException. Test code provided below, stack trace too for the failing case.

What is interesting - when I implement the IConvertible interface it works OK without calling any methods from IConvertible - so it looks like runtime doesn't use it at all, it's required just to be there. But unfortunately i dont have the option to implement IConvertible on all objects passed to evaluator. Can you please verify this and provide some fix? Unfortunately, i'm not able to do the VS builds myself for all .Net versions.

(edit) this is probably caused by always calling Convert.ChangeType on method params in LambdaParameterWrapper which always requires IConvertible. Instead, it should cast if possible, and only do Convert.ChangeType when casting wouldn't work.

Thanks
RG

Test Name:	NReco_TestFunctionCall
Test FullName:	Test1.EvalTests.NReco_TestFunctionCall
Test Source:	D:\rafal\facile\FacileWeb\Test1\EvalTests.cs : line 357
Test Outcome:	Failed
Test Duration:	0:00:29.5014222

Result StackTrace:	
at System.Convert.ChangeType(Object value, Type conversionType, IFormatProvider provider)
   at NReco.Linq.LambdaParameterWrapper.InvokeDelegate(Object obj, Object[] args)
   at lambda_method(Closure , LambdaParameterWrapper , LambdaParameterWrapper )
 --- End of inner exception stack trace ---
    at System.RuntimeMethodHandle.InvokeMethod(Object target, Object[] arguments, Signature sig, Boolean constructor)
   at System.Reflection.RuntimeMethodInfo.UnsafeInvokeInternal(Object obj, Object[] parameters, Object[] arguments)
   at System.Delegate.DynamicInvokeImpl(Object[] args)
   at NReco.Linq.LambdaParser.Eval(String expr, Func`2 getVarValue)
   at NReco.Linq.LambdaParser.Eval(String expr, IDictionary`2 vars)
   at Test1.EvalTests.NReco_TestFunctionCall() in D:\rafal\facile\FacileWeb\Test1\EvalTests.cs:line 371
Result Message:	
Test method Test1.EvalTests.NReco_TestFunctionCall threw exception: 
System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. ---> System.InvalidCastException: Object must implement IConvertible.

public class LambdaTestLine
        {
            public int Id { get; set; }
            public decimal Amount { get; set; }
            public string Label { get; set; }
        }

        [TestMethod]
        public void NReco_TestFunctionCall()
        {
            var p = new NReco.Linq.LambdaParser();

            var vars = new Dictionary<string, object>
            {
                {"funfun", (Func<object, string>) (x => "AAA" + x + "BBB") },
                {"someValue", "ding" },
                {"someObject", new LambdaTestLine { Id=150, Amount=22, Label = "xxxx" } }
            };

            var x0 = p.Eval("funfun(someValue)", vars);
            Assert.AreEqual("AAAdingBBB", x0);

            var x1 = p.Eval("funfun(someObject)", vars); //this explodes
            
        }

Hello

I´ve found that the parser raises an exception in cases like this:
not ( foo = bar)
anyway **not** works as expected in expressions like
((foo and foo1 = bar1 ) or not (foo2 = bar2))
by replacing **not** by ! you get the expression working always.

I think this could be maybe an edge case of the issue #30

Thanks for the good job.

Jose

This evaluates to true: lambdaParser.Eval("true == 2", new Dictionary<string, object>())
This evaluates false: lambdaParser.Eval("2 == true", new Dictionary<string, object>())

In the first instance, LambdaParser will try to convert 2 to a bool, and the expression is reduced to true == true.
In the second instance true is converted to an int32, and the expression becomes 2 == 0.

The issue lies in the way LambdaParser attempts to compare the two parameters and the implementation of the Convert.ToBoolean and Convert.ToInt32 methods.

• ToBoolean(int32) will return true if the value provided is anything but 0, otherwise false.
• ToInt32(bool) will return 1 if the value provided is true, otherwise 0.

In other words, the methods are not inverse of each other.

I'm not saying this is a bug with LambdaParser but it is an implementation detail that is worth flagging so that people don't get caught out by it.

A different approach would be to allow the caller to specify whether they wish LambdaParser to attempt to convert parameter types. If the caller sets this to false, in the example above, bool and int32 are not comparable (without conversion) and the expression would always evaluate to false.

Is it possible to support nested expression parsing like this:
lambdaParser.Parse("(List<string> list) => list.Where(o => !string.IsNullOrEmpty(o))");

i need it ...

Would it be possible to support operations with types that support these operations but are not IConvertible?

Example:

varContext["complexZero"] = new System.Numerics.Complex();
varContext["complexOne"] = new System.Numerics.Complex(1d, 1d);

lambdaParser.Eval("complexOne * complexZero", varContext)

I am using LambdaParser in a SL5 application, every thing is working correctly on Windows clients, but formulas are not evaluated on MacOs. The application is running with elevated privileges (signed and OOB).
Maybe need to reference some extra assemblies ?

Hello,

when I try the following code

var value = "testString";
var env = new Dictionary<string, object>();
env.Add("value", value);
var parser = new LambdaParser();
string res = (string)parser.Eval("value != null && value.Contains(\"t\") ? \"contains\" : \"does not contain\"", env);

I get this exception: System.InvalidOperationException: "The binary operator AndAlso is not defined for the types 'System.Boolean' and 'NReco.Linq.LambdaParameterWrapper'."

There seems to be a problem combining a check for null with another condition. If I do only the null check there is no problem.

We're exploring a possible use case here:
ExtendedXmlSerializer/home#382

And the thought arose to check out your parser, which has a Parse method. :)

However, I am having some trouble using it and feel I may be using it incorrectly. Unfortunately I do not see any tests around this method, so asking here.

Is it possible to use this method to parse the following into a expression (of Func<int>)?

"() => 123"

Thank you for any guidance you can provide. 👍

Is it both possible and ready to use ?

Or is there a possibility to register external functions to extent feature set?

ref to dotnet math: (https://docs.microsoft.com/de-de/dotnet/api/system.math.pow?view=netcore-3.1)

I used the simple code but it resulted in a null reference exception. Is this supported?

var varContext = new Dictionary<string, object>();
var lambdaParser = new NReco.Linq.LambdaParser();
Console.WriteLine(lambdaParser.Eval("System.DateTime.Now", varContext));
Console.Read();

It is useful to support method calls with variable number of parameters (in c# marked as 'params').
Good example is String.Format.

I was testing Lambdaparser.eval() to handle user expression, where values are added to an expandoObject. This way I could create a dynamic object that contains properties which can be used in the expression. But it seems that expandoObjects and dynamic objects are not supported. Here is a snippet of my unit test with the ExpandoObject:

var lambdaParser = new LambdaParser();
var varContext = new Dictionary<string, object>(); 

dynamic dynamicDiplomas = new ExpandoObject();
dynamicDiplomas.ODB = true; /* this should come from a DB */
dynamicDiplomas.IDB = null;
varContext["diplomas"] = dynamicDiplomas;

expression = "diplomas.ODB || diplomas.IDB != null";

// MissingMemberException
Assert.Throws<System.Reflection.TargetInvocationException>(() => lambdaParser.Eval(expression, varContext));

Is this a known limitation?

I have a requirement to convert predicates issued as string to lambda used for EF queries. Is it possible with lambdaparser? Code snippet will be helpful. Thanks

Example:
using (var context = new EntityContext())
{
var customersList = context.Customers
.Where(c => "c.Invoices.Count > 2")
.ToList();
}

When I wrote some code expressions, I wanted to check the syntax, such as whether the properties and methods were correct. I think using type reflection can check, but I wrote an incorrect property and did not prompt me for any errors. Why do I need this feature? Because I have defined some types of variables that may not have been assigned values, even though I have written code expressions, using Eval() function to check in real time will definitely result in errors. So, I wonder if you can implement pre checking for some code expressions, just to check if the syntax is correct. I think it should be possible in theory, but I don't know how to do it

var varContext = new Dictionary<string, object>();
varContext["dt"] = datatable;

Expression expr = lambdaParser.Parse("dt.Rows1[0]");
var exprParams = LambdaParser.GetExpressionParameters(expr);

dt is a System.DataTable type variable,Rows1 is an incorrect property,But it didn't give me any exceptions,i need some checks,how to do?

I can see from the examples that the expression can invoke native methods, like ToUpper. Is that safe to use when expressions are entered by a web application users?
Actually I'm looking for a simple math parser with conditional support ("if...else"), but all the solutions I found (yet) are using code generation in one way or another, which in my case is both overkill and security risk.

Currently ternary operator requires extra brackets to get 'switch-like' evaluation flow, for example:

a<0 ? 0 : (a>100 ? 100 : -a)

In case when there are many 'elseif' this requires a lot of nested brackets. To avoid that it makes sense to allow next ternary operator for part (which is currently not allowed), this makes possible to use expressions like that:

v<0 ? "red"
: v<25 ? "white" 
: v<50 ? "lightgreen" 
: "green"