RegEx to match formula groups [modified] - C# Discussion Boards

3-Nov-09 4:15

Forgive me if my question is unclear, I'll do my best to clarify it as best as I can.
What I'm trying to do is "Tokenize" nested formula's by using RegEx.

Consider the following formula in LaTeX:

LaTeX Example 1: text before \frac{ \frac{ SubPart 1 }{ SubPart 2 } }{ Part 2 } text after

What I need is a RegEx that can tokenize this LaTeX formula to output the following tokenlist:
Token[0] = text before
Token[1] = \frac{ \frac{ SubPart 1 }{ SubPart 2 } }{ Part 2 }
Token[2] = text after

Looking at Token[1], it has a nested fracture (\frac{...}) in its top-part.
That's the way I need it to be to build an object tree.
In a sense, the fracture in the top-part is a child to it's parent fracture.

Consider the following formula in LaTeX:

LaTeX Example 2: text before \frac{ \left ( SubPart 1 + SubPart 2 \ right ) }{ Part 2 } text after

This should result in the following tokenlist:
Token[0] = text before
Token[1] = \frac{ \left ( SubPart 1 + SubPart 2 \ right ) }{ Part 2 }
Token[2] = text after

Again, looking at Token[1], it has a nested subformula in it.
In this case, the subformula is a child, associated to the toppart of the fracture, to the fracture object.

Final example (and then I assume you catch my drift):

 text before \left ( \frac{ SubPart 1 }{ SubPart 2 } \ right ) + X text after

This should result in the following tokenlist:
Token[0] = text before
Token[1] = \left ( \frac{ SubPart 1 }{ SubPart 2 } \ right )
Token[2] = + X text after

The nested-/sub-formulas will be processed when the parent formula is being constructed, so we need not worry about that part of it here.
What I need, is a RegEx that can handle formula dimensions to the nth degree.

Right now I do this by stepping through the string and seeing if I have a match in the sub-string for a particular item.
When found, I add 1 to a level counter and run an internal loop to find the final closing "}" that brings the level back to 0;
This works fine, but I've been commented upon by using this method and not using tokens, begotten with the use of RegEx.

If anyone has a suggestion of how I should construct my regex, then please let me know.
I've gotten as far as:
Fracture : (\\frac{[\W].*?((}\s)|(}$)))*
Subformula : (\\left $[\W].*?\\right $)*
But as you would guess, this fails miserably when having to deal with nested or multidimensional formulas.
Any help and/or insight on the matter would be greatly appreciated.

modified on Tuesday, November 3, 2009 10:28 AM

PIEBALDconsult3-Nov-09 4:42

PIEBALDconsult

3-Nov-09 4:42

For what purpose? Do your have example input and output?

Björn T.J.M. Spruit3-Nov-09 4:59

Re: RegEx to match formula groups [modified]

3-Nov-09 4:59

The purpose is to create an object tree of the formula.
The examples may be found in my previous post.
These are simplified examples, mind you, but they are representative of the challenge.

Simply put:
Parent object is Formula.
Formula has a collection of child objects of the base type FormulaItem.
FormulaItem[0] could be of type SubFormula (\left (...\right )).
FormulaItem[0], being of the type SubFormula, has one collection of child objects of the base type FormulaItem.

FormulaItem[0].FormulaItem[0] could be of type Fracture (\frac{...}).
FormulaItem[0].FormulaItem[0], being of the type Fracture, has two collections (top-part and bottom-part) of the base type FormulaItem.
Etc...

I hope to have clarified the "why" in this.

PIEBALDconsult3-Nov-09 5:44

PIEBALDconsult

3-Nov-09 5:44

Björn T.J.M. Spruit wrote:
in my previous post.

I didn't see it and I'm not going to go look for it.

Björn T.J.M. Spruit wrote:
I hope to have clarified the "why" in this.

Nope.

modified on Tuesday, November 3, 2009 6:21 PM

Re: RegEx to match formula groups [modified]

Björn T.J.M. Spruit4-Nov-09 5:09

4-Nov-09 5:09

Hmmm, not very friendly then. WTF | :WTF:

No worries, I'm always optimistic and don't mind a 'challenge' when I come across one.
I'll give you a more extended LaTeX formula example:

\left ( Availability~ \right )  \times  \left ( Performance~ \right )  \times  \left ( Quality~ \right )  \times 100 ~ =~  \left (\frac{  I}{II}  \right )  \times  \left (\frac{  III}{ \left ( IV \times I \right ) }  \right )  \times  \left (\frac{  V}{III}  \right )  \times 100

This is an example from how it's actually being used at this very moment.
I'm not sure what more information you need to clarify the "why"?
"Nope" isn't a very articulate way of asking me for the information you require to clarify this to you.
So please let me know what it is you require of me other than what I've told you in order to clarify the "why" more accurately.

Just to recap, I need to objectify a LaTeX formula.
All I need, is a regex that is able to work with nested and multi-dimensional formulas as explained in the previous posts.
The reason I'm exploring this, is because I've been commented upon that I didn't use tokenization by regex to get the formula elements.
If there's anybody who knows how to tokenize a string that is a formula with nested and/or multi-dimensional elements, please let me know, otherwise I'll set this aside as:"Not a viable option, can't be done within a reasonable amount of time."

modified on Wednesday, November 4, 2009 12:25 PM

OriginalGriff3-Nov-09 5:33

OriginalGriff

3-Nov-09 5:33

I'm not going to try and work it out myself, but you may find this handy: Expresso[^] - examines and generates Regular expressions.

Best bit is it break it down and explains it in English!

No trees were harmed in the sending of this message; however, a significant number of electrons were slightly inconvenienced.

This message is made of fully recyclable Zeros and Ones

Björn T.J.M. Spruit4-Nov-09 23:02

C# default interface method?

4-Nov-09 23:02

Thank you for your advise.
The application is a good one and I would certainly advise it to anyone who's going to work with regular expressions.

Limited, controlled recursions of a finite count, can be tackled with regular expressions, though it makes the regular expressions cumbersome.
Infinite and intuitive recursions with regular expressions can't be done in .NET as of yet.
http://badassery.blogspot.com/2006/03/regex-recursion-without-balancing.html[^]

I'm still investigating this and will post my findings if there's anything interesting to report.

creating a report

JollyMansArt3-Nov-09 4:07

JollyMansArt

3-Nov-09 4:07

I am wanting to create a crystal report. In doing so I want to point the report to a xsd data object. If there is a better way please let me know.
The issue here is their may be 50 to 200 columns on the fly in this report and it may increase over time. I don't want to keep having to modify the application and recompiling to generate the report.

So I thought that I would find a way to dynamically create the xsd data object. But I don't know how to do this. Any Suggestions?

Mark McArthey3-Nov-09 4:05

Mark McArthey

3-Nov-09 4:05

I'm trying to implement the idea of a default interface. Maybe this isn't possible, and if there's another manner in which I can implement this please let me know.
The reason for this is that I have an ErrorLog class that I'd like others to use while allowing them to customize the actual writing of the data (if they choose). I have written a default WriteLog method that will write some basic information to a general database table but it obviously won't apply in all cases, therefore I would like to provide the option to override this.

My code is similar to the following. I've left out some syntax and details, but it should be close enough to understand. Please ask for any clarification.

Both of these classes are in the same project:

ErrorLog.cs:

namespace Library {
    public interface IErrorLog {
        void WriteLog(ErrorLog el)
    }

    public class ErrorLog
    {
        public void Log() {
            WriteLog(this);
        }
    }
}

WriteLog.cs:

namespace Library {
    public class Write : IErrorLog {
        public void WriteLog(ErrorLog el) {
            // implementation
        }
    }
}

I'd like other applications/projects to implement the IErrorLog interface which would require them to create a WriteLog() method. If they'd like to use the default they can simply refer to the Write.WriteLog() method.
Something is off with my organization or implementation, though. It seems I can't include the WriteLog() "default" in the same project because when I add a reference to Library.ErrorLog in the other application it doesn't require me to implement WriteLog(). I suppose that's because it's already been defined in the project. Do I need to split these into two projects? If so, how do I reference the "default" method?

Thanks for any help since I've thoroughly confused myself.

Ghydo3-Nov-09 5:07

Ghydo

3-Nov-09 5:07

Maybe I didn't understand your question clearly, but your ErrorLog class should contain a reference to an IErrorLog interface or you can't access any WriteLog() method.

Example:

namespace Library {
	public interface IErrorLog {
		void WriteLog(ErrorLog el) 
	}
		
	public class ErrorLog
	{
		private IErrorLog errorLog;

		public ErrorLog(IErrorLog errorLog) {
			if (errorLog == null)
				this.errorLog = new Write();
			else
				this.errorLog = errorLog;
		}

		public void Log() {
			errorLog.WriteLog(this);
		}
	}
}

In the example you can pass an IErrorLog to the constructor if you wish to use a custom IErrorLog, or null if you want to use the "default" Write class.

I don't think that there is a "default" interface that you can force to implement in an assembly.

Mark McArthey3-Nov-09 6:43

Mark McArthey

3-Nov-09 6:43

Thank you. I've been able to get this working by then calling the aggregated interface so it looks like all is good. I am assuming that you meant to assign Log() to the interface, though?
Thanks for your time!

Ghydo3-Nov-09 7:36

Ghydo

3-Nov-09 7:36

You're welcome!

PIEBALDconsult3-Nov-09 7:40

PIEBALDconsult

3-Nov-09 7:40

That's very confusing. What is ErrorLog supposed to be? An error to be logged or something that should log an error? It appears in your code that it is both, and that is likely to be a rather poor design.

In an application, an error occurs and you want to log it. You need a class to represent that error and you need a class to handle logging the error.

In most cases, your error class should derive from System.Exception. Your error logging class should then handle such errors.

namespace Library
{
    interface IErrorLog
    {
        void WriteLog ( System.Exception Error ) ;
    }

    public class ErrorLog : IErrorLog
    {
        // Constructor

        void WriteLog
        (
            System.Exception Error
        )
        {
            // Perhaps write to the Console
        }
    }
}

Other ErrorLogs can then be derived to log to other destinations.

You can then instantiate an ErrorLog and call el.WriteLog ( someexception ) ;

But I'm wondering whether or not you are trying to write someexception.Log ( el ) ;
If so, you could write an Extension Method to accomplish that:

public static void
Log
(
    this System.Exception Error
,
    IErrorLog             ErrorLog
)
{
    ErrorLog.WriteLog ( Error ) ;
}

(None of this code has been tested.)