Subj : Re: Parser generators for a simple language form. How?
To   : comp.programming
From : Rob Thorpe
Date : Tue Aug 02 2005 01:14 pm

Jon Ripley wrote:
> This has been bugging me for a while and I'd appreciate some input.
>
> I have a number of simple scripting languages of similiar specifications
> that I use for rapid application development that I would like to make
> available to a wider audience. The wider audience use whichever
> operating system they want to on whatever platform they purchased.
>
> Currently they are all implemented in a non-portable language and whilst
> they suit my current needs perfectly they are of limited use to others
> and need to be recoded in a portable language to be compiled for
> whatever OS/platform any given user may want to use them on. My first
> choice for the portable language is C and these will all be command line
> tools.
>
> I would like to use a tool to generate the parser for each language,
> generally the only difference in the languages are the names of the
> functions and the parameters that they take. The format looks similiar
> to the C language - I decided that this would be the most transparent to
>   the largest number of users.
>
> The general format is as follows:
>
> /* Multiline comment */
> mylang( <number> )
> {
>    Function1( <string>, <number> );
>    Function1( <string>, <number> );
>    Function2(); // Line comment
>    Function3([<number|string>])
>    {
>      // can use data1; here if the last two parameters are
>      // not-relevant, NULL or use default settings
>      data1,"data1",flags;
>      data2,"data2",flags;
>      data3,"data3",flags;
>    }
>    Function4( <string> ) [
> Data here is passed straight to output as formatted
> ]
> }
>
> When FunctionX is encountered the values or pointers to values are
> stored in a structure specific to that function and are later used to
> create the desired output. A function may be used multiple times to
> declare different 'things' of the same type. Here's a working example
> made up on the spot that could be useful for rapid development of
> command line tools. It accepts a number of useful parameters including
> and when compiled would create a complete program shell in C for the
> described application.
>
> cligen(0)
> {
>
>    CLIName("MyProg");
>    Version("1.00","01-01-1900");
>    Usage("ValidateOptions();","[-i] <input> [[-o] <outfile>] [-h]");
>    Parameters()
>    {
>      "o","output",Optional,1,"OutFile";
>      "i","input",Required,1,"InFile";
>      "h","help",Optional,0,,"Help();";
>    }
>    // Whitespace is only significant inside [] blocks
>    Inline() [
> void Help(void) {
>    printf("Hello, World!\n");
> }
> ]
> }
>
> Any suggestions for tools that I can use which would be appropriate for
> generating parsers for languages of this type? They should not be too
> difficult to learn as I want to spend the time porting the code rather
> than spending months working out how to use the parser generator. I have
> looked into yacc and bison but these don't seem to be appropriate and
> appear to have a huge learning curve.
>
> The worst case scenario and what I am currently facing is that I will
> need to write a scripting language of the above type that creates the C
> source code of scripting languages of the above type.
>
> Help is muchly appreciated as I am currently deadlocked,

You should be able to do what you need with a recursive descent parser.
 Search the web for that phrase.  Also search the web for cfoogol - a
nice demonstration of the principle.

Yacc and Bison aren't really all that hairy (except for their error
messages).  Look at the little languages Eric S Raymond has written on
his website, they give you a good idea of how to do it.

Also see my article on the subject:
http://realworldtech.com/page.cfm?ArticleID=RWT041902173146

.