Question

I need to parse an MDX with my .Net application. Initially, I used regular expression to do this but the expressions are getting complicated and a regex expert suggested that it will be better if I use parser.

Is there any parser specifically for MDX? I tried Ranet but for some unknown reason it does not install in my machine (does not show any error message).

I need to split the several parts of the MDX into strings. For example, the where clause in one string, from clause in another etc.

Answer 1

The best solution would be to find a parser, but it is always very hard to find a parser for your specific needs. So if you end up with writing a parser Ve Parser is a better tool comparing to regex, because it provides more parsing functionalities, you can generate better output and since you are calling .net methods it implicitly have intellisence for writing your parser.
The downside is that it still is not well-documented, so you may find it difficult for some special scenarios.

Project Link : http://veparser.codeplex.com

NuGet identifier : veparser

If you need to get text for different parts of an MDX here is a partial sample code:

using VeParser;
using System.Linq;
using System.Collections.Generic;
using System;


public class MDXParser : TokenParser
{
    protected override Parser GetRootParser()
    {
        // read the following line as : fill 'select' property of 'current object(which is a statement)' with the 'new value of selectStatement' after facing a sequence of a select statement and then the symbol of ( and then a delemitied list of identierfiers filling the 'fileds' property of 'current selectStatement object' delemitied by ',' and finally expect the sequence to be finished with a symbol of ')'
        var selectStatement = fill("select", create<selectStatment>( seq(expectKeyword_of("select"), expectSymbol_of("("), deleimitedList(expectSymbol_of(","), fill("fields",identifier) ), expectSymbol_of(")"))));
        // read the following line as : fill the from property of 'current object(which is a statement)' with an expected identifier that is after a 'from' keyword
        var fromStatement = seq(expectKeyword_of("from"), fill("from", identifier));
        // the following statement is incomplete, as I just wanted to show a sample bit, If you are interested I can help you complete the parser until the full documentation become available.
        var whereStatement = fill("where", create<whereStatement>(seq(expectKeyword_of("where"))));
        var statement = create<statement>(seq(selectStatement, fromStatement, whereStatement));

        return statement;
    }

    public statement Parse(string code)
    {
        var keywords = new[] { "select", "where", "from" };
        var symbols = new[] { "(",")", ".", "[", "]" };
        var tokenList = Lexer.Parser(code, keywords, symbols, ignoreWhireSpaces : true);
        // Now we have our string input converted into a list of tokens which actually is a list of words but with some additional information about any word, for example a "select" is marked as keyword
        var parseResult = base.Parse(tokenList.tokens);
        if (parseResult == null)
            throw new Exception("Invalid Code, at the moment Ve Parser does not support any error reporting feature.");
        else
            return (statement)parseResult;
    }
}
public class statement
{
    public selectStatment select;
    public string where;
    public identifier from;
}
public class selectStatment
{
    public List<identifier> fields;
}
public class whereStatement
{

}

This code is not complete, I just wanted to demonstrate how to use Ve Parser to write your own parser for MDX. If you liked the library and wanted to use it, I would be happy to provide you with the all descriptions and techniques you need.