Parse complex formula string with nested parenthesis in C#

I am looking for a way to parse complex VFP calculations strings.

An example of one of these formulas would be :

ABS((ABS(-48) - ABS(245 + 34 - 98))) + ABS(ABS(10 + 9)) > 100

As you can see, I have already substituted all variable patterns with there respective values - my issue is in parsing out the ABS() calls so I can substitute them out with C# Math.Abs().

In my case, I want to receive a list of string:

• ABS(-48)
• ABS(245 + 34 -98)
• ABS(10 + 9)

I thought using regex pattern matching would be the ticket, but I cannot figure out how to define a pattern that would give me (with 100% accuracy) something like "ABS(245 + 34 -98)" out of the original formula.

I would love to either know the pattern I need to use, or see someone else's example of parsing something like this successfully (I have tried many SO solutions that never parse it correctly....), or even better a suggested library that can evaluate (returns true or false) that calculation in as it is (I thought I had it with Datatable.Compute() - but it returns ABS(x) as null....).

Also, so everyone does not comment about it, I did not write these calculations...I am only trying to parse them out in C# wihtout having to change the string formulas that already exist.

Thanks in advance!

If you want to match inmost calls only (not a proper abstract syntax tree):

>I want to receive a list of string:
> - ABS(-48)
> - ABS(245 + 34 -98)
>- ABS(10 + 9)

you can try regular expression with a pattern like this:

[A-Za-z]+\s*\([0-9\+\*\-\s]+\)

C# code will be

using System.Text.RegularExpressions;
using System.Linq;

...

string text = "ABS((ABS(-48) - ABS(245 + 34 - 98))) + ABS(ABS(10 + 9)) > 100";

List<string> result = Regex
  .Matches(text, @"[A-Za-z]+\s*\([0-9\/\+\/\*\-\s]+\)")
  .Cast<Match>()
  .Select(match => match.Value)
  .ToList();

Console.WriteLine(string.Join(Environment.NewLine, result));

Output:

ABS(-48)
ABS(245 + 34 - 98)
ABS(10 + 9)

Fiddle

There is no trick to this and explaining how to do it is out of the scope of a stackoverflow question. It's a complex task and you'll just need to work through it.

A far simpler approach would be to use an existing tool such as my SoftCircuits.ExpressionEvaluator.

This library supports all the expressions in your question except the boolean operations. It's also open source, so you can review exactly how it parses it.

SoftCircuits.Silk is another one of my libraries. This implements an interpreted programming language and fully supports boolean expressions. Obviously, this library is more complex. But, again, it's all open source if you want to review how it's done.

Since we are chill with recommending tools (for what ever reason).

I have been using NCalc for ages now. https://github.com/ncalc/ncalc

Though, I would recommend looking for a new and activly mainted library eg. support for the latest .net type as System.Int128 would be nice.

There for searching on github seems your best bet
https://github.com/search?q=math+formula+language%3AC%23&type=repositories

and as was already mentioned, Regular expression isn't the right tool for the job.

Here a running demo:

https://dotnetfiddle.net/0vPKkd

Notice:

I needed to change ABS to Abs, I would assume you can specify to ignore-case, but I am sure a big boy like you will figure it out.