Click here to Skip to main content
15,867,686 members
Articles / Operating Systems / Windows
Article

The expression evaluator revisited (Eval function in 100% managed .NET)

Rate me:
Please Sign up or sign in to vote.
4.98/5 (166 votes)
18 May 2007CPOL10 min read 621.1K   6.9K   204   200
This second article about evaluation in .NET introduces a parser which pre-compiles the expressions

Sample image

Introduction

Most applications need to evaluate a formula at run-time. The .NET Framework, despite allowing advanced compilation support, does offer a quick and light-weight eval function. This article introduces a usable eval function with some rarely available functionalities:

  • Fast, single-pass parser
  • Highly extensible; you can add your own variables and functions without having to change the library itself
  • Common priorities are supported : 2+3*5 returns 2+(3*5)=17
  • Boolean operations are supported, i.e. "and," "not," "or"
  • Comparison operators are supported, i.e. <=, >=, <>
  • Supports numbers, dates, strings, and objects
  • Supports calls to object properties, fields and methods
  • Runs an expression multiple times without needing reparsing
  • Can automatically detect when an expression needs to be re-evaluated
  • The expression syntax is completely checked before starting the evaluation
  • Fully human-readable code -- not generated using Lex/Yacc tools -- and therefore permits amendments to the core syntax of the evaluator, if needed

This article also attempts to explain how the whole thing works.

Why an interpreter?

People often tell me that there is no place in their application for an evaluator because it is too complicated for their users. I do not agree with this vision. An evaluator is a cheap way to HIDE the complexity for the average user and provide powerful features for the advanced user. Lets take an example.

In your application, you let the users choose the title of the window. This is convenient and simple; it's just a textbox where they can type what they want. The difficulty comes when some users want more. Let's say that they want to see their User IDs or the time. Then you have 3 alternatives:

  1. You add a form in your program and give your user more options on what they can show.
  2. You don't do it.
  3. You use an evaluator (i.e. mine).

The first option requires lots of work on your side and can potentially confuse the more basic users. The second option won't confuse your users, but might lose you the more advanced ones. The third option is ideal because you can keep your textbox and let the powerful user type what they want.

Title of the window for user : %[USERID], the time is %[NOW]

And you're done. The interface is still using a regular textbox and is not complicated. On the coding side, it is really not much to add. In terms of power, you can add a new variable every day and as long as you document it all, your users stay satisfied.

Why not use the .NET Framework built-in compiler?

Using the .NET Framework compilation capabilities seem to be the most obvious way to make an evaluator. However, in practice this technique has a nasty side effect. It looks like it creates a new DLL in memory each time you evaluate your function and it seems nearly impossible to unload the DLL. You can refer to remarks at the end of the article Evaluating Mathematical Expressions by Compiling C# Code at Runtime for more details.

Using other engines or application domains is an option if you want a full VBScript or C# syntax. If you need to write classes and loops, this is probably the way to go. This evaluator is neither using CodeDOM nor trying to compile VB source. It parses an expression character-by-character and evaluates its value without using any third party DLL.

Using the code

The evaluator can be run with just two lines of code:

In VB

VB
Dim ev As New Eval3.Evaluator
MsgBox(ev.Parse("1+2+3").value)

In C#

VB
Eval3.Evaluator ev = new Eval3.Evaluator(
    Eval3.eParserSyntax.c,/*caseSensitive*/ false);
MessageBox.Show(ev.Parse("1+2+3").value.ToString());

Providing variables or functions to the evaluator

By default, the evaluator does not define any function or variable anymore. This way, you can really decide which function you want your evaluator to understand. To extend the evaluator, you need to create a class. Below is a VB Sample; a C# version is available in the Zip file.

Public Class class1
    Public field1 As Double = 2.3

    Public Function method2() As Double
        Return 3.4
    End Function

    Public ReadOnly Property prop3() As Integer
        Get
            Return 4.5
        End Get
    End Property
End Class


Note that only public members will be visible.

Usage

Dim ev As New Eval3.Evaluator
ev.AddEnvironmentFunctions(New class1)
MsgBox(ev.Parse("field1*method2*prop3").value.ToString)


You can also use a more dynamic version. I don't really like this method, but it can be useful. Note that the value of an extention can change once parsed, but the type should not.

Public Class class2
    Implements iVariableBag

    Public Function GetVariable(
        ByVal varname As String) As Eval3.iEvalTypedValue _
           Implements Eval3.iVariableBag.GetVariable
        Select Case varname
            Case "dyn1"
                Return New Eval3.EvalVariable("dyn1", 1.1, _
                     "Not used yet", GetType(Double))
        End Select
    End Function
End Class

    ev.AddEnvironmentFunctions(New class2)
    Dim code As opCode = ev.Parse("dyn1*field1")
    MsgBox(code.value & " " & code.value)

Recognized types

The evaluator can work with any object, but it will allow common operators (+ - * / and or) only on usual types. Internally, I use these types:

enum evalType
    number    will convert from integer, double, single, byte, int16... 
    boolean   
    string
    date      equivalent to datetime
    object    anything else

There is a shared function in the evaluator to return all those types as string:

Evaluator.ConvertToString(res)
This function will return every type using a default format.

How does this all work?

If you just want to use the library, please refer to the 'Using the code' section. The following sections are just for curious people who want to know it works. The techniques I used are rather traditional and can, I hope, be a good introduction to the compilation theory.

The evaluator is made of a classic Tokenizer followed by a classic Parser. I wrote both of them in VB, without using any Lex or Bisons tools. The aim was readability over speed. Tokenizing, parsing and execution are all done in one pass. This is elegant and, at the same time, quite efficient because the evaluator never looks ahead or backwards more than one character.

The tokenization

The first thing the evaluator needs to do is split up the string you provide into a set of Tokens. This operation is called tokenization and in my library it is done by a class called tokenizer

Sample image Sample image Sample image

The tokenizer reads the characters one by one and changes its state according to the characters it encounters. When it recognizes one of the Token types, it returns it to the parser. If it does not recognize a character, it will raise a syntax error exception. Once the class is created with this command,

tokenizer = new Tokenizer("1+2*3+V1")

...the evaluator will just access tokenizer.type to read the type of the first token of the string. The type returned is one of those listed in the chart below. Note that the tokenizer is not reading the entire string. To improve performance, it will only read a single token at a time and return its type. To access the next token, the evaluator will call the method tokenizer.nextToken(). When the tokenizer reaches the end of the string, it returns a special token end_of_formula.

Recognized token types
enum eTokenType
 operator_plus                +
 operator_minus             -
 operator_mul               *
 operator_div               /
 operator_percent           %
 open_parenthesis           (
 comma                      ,
 dot                        .
 close_parenthesis          )
 operator_ne                <>
 operator_gt                <=
 operator_ge                >=
 operator_eq                =
 operator_le                <=
 operator_lt                <
 operator_and               AND
 operator_or                OR
 operator_not               NOT
 operator_concat            & 
any word starting with a letter or _ 
 value_identifier       
 value_true                 TRUE 
 value_false                FALSE 
any number starting 0-9 or . 
 value_number           
any string starting ' or " 
 value_string           
 open_bracket               [ 
 close_bracket              ]
Initial state 

 none                   
State once the last 
character is reached
 end_of_formula          

The parser

Sample image Sample imageSample image

The parser has been completely rewritten in this version. The parser is using the information provided by the tokenizer (the big brown box) to build a set of objects out of it (the stack on the right). In my library, each of these objects is called an OpCode. Each OpCode returns a value and can have parameters or not.

Opcode 1
Opcode 2
Opcode 3
Opcode *
Two Opcode +
and Opcode +

The OpCodes + and * have two parameters. The rest of the OpCodes have none. One of the more complicated concepts of the parser is that of priorities. In our expression...

1 + 2 * 3 + v1

...the evaluator has to understand that what we really mean is:

1 + (2 * 3) + v1

In other words, we need to do the multiplication first. So, how can this be done in one pass? At any time, the parser knows its level of priority:

VB
enum ePriority
  none = 0
  concat = 1
  or = 2
  and = 3
  not = 4
  equality = 5
  plusminus = 6
  muldiv = 7
  percent = 8
  unaryminus = 9

When the parser encounters an operator, it will recursively call the parser to get the right part. When the parser returns the right part, the operator can apply its operation (for example, +) and the parsing continues. The interesting part is that while calculating the right part, the Tokenizer already knows its current level of priority. Therefore, while parsing the right part, if it detects an operator with more priority it will continue its parsing and return only the resulting value.

The interpretation

The last part of the evaluation process, is the interpretation. This part is now running a lot faster thanks to the OpCode.


Sample imageSample image Sample image

To get the result out of the stack of OpCodes, you just need to call the root OpCode value. In our sample, the root OpCode is a + operator. The property Value will in turn call the value of each of the operands and the result will be added and returned. As you can see from this picture, the speed of evaluation is now quite acceptable. The program below needs 3 full expression evaluations for every single pixel in the image. For this image, it required 196,608 evaluations and, despite that, it returned in less than a second.

Sample image

The class at the core of this new project is the OpCode class. The key property in the opCode class is the property 'value'.

Public MustInherit Class opCode
    Public Overridable ReadOnly Property value(
        ) As Object Implements iEvalValue.value
    MustOverride ReadOnly Property ReturnType(
        ) As evalType Implements iEvalValue.evalType
    ...
End Class

Each OpCode returns its value through it. For the operator +, the value is calculated this way:

Return DirectCast(mParam1.value, Double) + DirectCast(mParam2.value, Double)

Is that really faster?

It is faster if you need to evaluate the functions more than once. If you need to evaluate the function only once, you might not care about speed anyway. So, I would recommend this new version in either case. As you can see from the picture above, 3 formulas are evaluated for every pixel of the image. The image being 256x256 pixels, the evaluator had to calculate 196,608 expressions. So, simple expressions are returned in less than 5 microseconds. I think this is acceptable for most applications.

Dynamic variables

Dynamic variables are an interesting concept. The idea is that if you use several formulas in your application, you don't want to recalculate all the formulas when a variable changes. The evaluator as a built-in ability to do that. On this page, the program uses the dynamic ability:

Sample image

To use this ability once you have parsed your expression:

mFormula3 = ev.Parse(tbExpression3.Text)

You only have to wait for the event mFormula3.ValueChanged

Private Sub mFormula3_ValueChanged( _
          ByVal Sender As Object, _
          ByVal e As System.EventArgs) _
          Handles mFormula3.ValueChanged
    Dim v As String = Evaluator.ConvertToString(mFormula3.value)
    lblResults3.Text = v
    LogBox3.AppendText(Now.ToLongTimeString() & ": " & v & vbCrLf)
End Sub

You said it supports objects?

Yes, the evaluator supports the . operator. If you enter the expression theForm.text then the evaluator will return the title of the form. If you enter the expression theForm.left, it will return its runtime left position. This feature is only experimental and has not been tested yet. That is why I have put this code here, hoping that others will find its features valuable and submit their improvements.

How does this work?

In fact, the object came free. I used System.Reflection to evaluate the custom functions. The same code is used to access the object's methods and properties. When the parser encounters an identifier that is a keyword without any meaning to it, it will try to reflect the CurrentObject to see if it can find a method or a property with the same name.

VB
mi = CurrentObject.GetType().GetMethod(func, _
 _Reflection.BindingFlags.IgnoreCase _
 Or Reflection.BindingFlags.Public _ Or Reflection.BindingFlags.Instance)

If a method or a property is found, it will feed its parameters.

valueleft = mi.Invoke(CurrentObject, _
  _ System.Reflection.BindingFlags.Default, Nothing, 
  _ DirectCast(parameters.ToArray(GetType(Object)), Object()), Nothing)

Are there any known bugs or requests?

The following are requests/bugs from the original project:

Someone reported that you need the option 'Compare Text' for the evaluator to work properly. I think this is fixed now. If you want the evaluator to be case-sensitive you can ask for it in the evaluator constructor.

Someone also reported that the evaluator did not like having a comma as a decimal point in the windows international settings. This is fixed, too, I believe.

My request: If you find this library useful or interesting, don't forget to vote for me. :-)

Points of interest

Speed Tests: I wish I could have the time to compare various eval methods. If someone wants to help, please contact me. To my knowledge, this is the only formula evaluator available on CodeProject with a separate Tokenizer, Parser and Interpretor. The extensibility is extremely easy due to internal use of System.Reflection.

History

18th May 2007

  • Article edited and posted to the main CodeProject.com article base.

4th May 2006

  • Fix a bug introduced in the last version where the functions were not recognized properly.
  • Add a few more samples in both C# and VB sample programs using Arrays and default members (Controls.Item).

27th April 2006

  • Implements Array
  • Starts differencing C# and Vb

20th April 2006

  • Try to Improve the article with more pictures.

19rd April 2006

  • C# compatibility (a few variables and members were renamed to avoid c# keywords conflicts).
  • C# sample
  • Move the core evaluator within a DLL
  • Allow 'on the fly variable' through a new interface called 'iVariableBag'

13rd April 2006

  • New article (as the original was not editable).
  • Entirely new parser using OpCodes
  • 10th Feb 2005
  • Greatly increased the length and detail of the article

7th Feb 2005

  • First release.

License

This article, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)


Written By
Software Developer (Senior)
France France
I am a French programmer.
These days I spend most of my time with the .NET framework, JavaScript and html.

Comments and Discussions

 
QuestionIf there is any interest in an updated/enhanced version Pin
Member 109560108-Oct-21 3:44
Member 109560108-Oct-21 3:44 
I have used this project for years as a way to avoid Reflection when doing math and logic evaluations and have been keeping this code updated and enhanced to function with string equality, etc.

Currently, I have a portion of the project (the evaluator class itself) 100% usable in VS2019

It's not perfect since there was some coaxing of the code to get things like complex logic comparisons to work with different types (1 > 2) == ("text" == "test")
GeneralMy vote of 5 Pin
Member 1407812811-Apr-19 19:54
Member 1407812811-Apr-19 19:54 
Questioniif to if then else syntax Pin
RoarkDude29-Mar-17 7:57
RoarkDude29-Mar-17 7:57 
QuestionNice Library Pin
Mark Hofland15-Dec-16 2:09
Mark Hofland15-Dec-16 2:09 
QuestionIf Statement Pin
Deepshikha Kaneria24-Aug-16 2:52
Deepshikha Kaneria24-Aug-16 2:52 
QuestionIf Syntax Pin
Deepshikha Kaneria24-Aug-16 2:52
Deepshikha Kaneria24-Aug-16 2:52 
QuestionRegarding function having optional parameter Pin
Deepshikha Kaneria24-Aug-16 0:28
Deepshikha Kaneria24-Aug-16 0:28 
QuestionCan't get it to work in 2016 Pin
Member 1248928227-Apr-16 10:56
Member 1248928227-Apr-16 10:56 
AnswerRe: Can't get it to work in 2016 Pin
Member 1248928228-Apr-16 3:14
Member 1248928228-Apr-16 3:14 
Praise5 Stars Pin
Mr Guy1247-Apr-16 4:32
Mr Guy1247-Apr-16 4:32 
GeneralRe: 5 Stars Pin
Pascal Ganaye26-Apr-16 7:38
Pascal Ganaye26-Apr-16 7:38 
QuestionGreat stuff just what i need. Thx for shearing. Pin
crodex31-Dec-15 5:30
crodex31-Dec-15 5:30 
Questionvba for Excel Pin
Member 1178457722-Jun-15 14:20
Member 1178457722-Jun-15 14:20 
QuestionNeed Support Pin
Andrew Hutama Indrajaya24-Feb-15 15:40
Andrew Hutama Indrajaya24-Feb-15 15:40 
GeneralExcellent project. Thank you for sharing Pin
miteleda8-May-14 15:33
miteleda8-May-14 15:33 
GeneralI can't download this code.Please send to my mailbox. Pin
netvan18-Mar-14 19:13
professionalnetvan18-Mar-14 19:13 
SuggestionVery good! Pin
po4teda18-Jan-14 6:29
po4teda18-Jan-14 6:29 
QuestionDynamic Variable at runtime Pin
Arpit Gandhi12-Sep-13 19:45
Arpit Gandhi12-Sep-13 19:45 
QuestionGet Object in my form Pin
bach20009-Sep-13 4:43
bach20009-Sep-13 4:43 
QuestionTrunc example implementation sucks Pin
teejay877-Aug-13 0:33
teejay877-Aug-13 0:33 
QuestionHelp needed - How to trap event when value of a variable changes Pin
Zahid Hayat5-Jul-13 22:30
Zahid Hayat5-Jul-13 22:30 
AnswerRe: Help needed - How to trap event when value of a variable changes Pin
Pascal Ganaye5-Jul-13 22:59
Pascal Ganaye5-Jul-13 22:59 
GeneralRe: Help needed - How to trap event when value of a variable changes Pin
Zahid Hayat6-Jul-13 10:33
Zahid Hayat6-Jul-13 10:33 
QuestionBoolean return Value Pin
Member 78239519-Dec-12 0:52
Member 78239519-Dec-12 0:52 
AnswerRe: Boolean return Value Pin
Daniel Balogh12-Apr-16 2:41
Daniel Balogh12-Apr-16 2:41 

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Praise Praise    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.