Mythic Logos

See a primitive working parser at http://mythiclogos.com/parsing/version01 Does arithmetic, does not handle ill-formed errors (missing “)” ) gracefully (does nothing actually–if you check in firebug, just gives warning).

So I have spent the past week or so creating the foundations for a parser. And reacquainting myself with Javascript and jQuery. I love both the language and the library. I hope to write more on the joys of javascripting.

But for now, I am writing about parsers. First, let me say that I have always just wanted to master the art of parsing. And that most parsing and grammar specifications and so forth have just left me scratching my head. Not that I tried to hard. For some reason, most computer science speak nauseates me. I suppose like most abstract math presentations. In both cases, the subjects of interest are incredibly useful and beautiful and wondrous, but the presentations and formulations leave much to be desired.

Alright, I have very little patience with other people’s stuff. I like to do stuff myself.

So that is why parsers have always held a flame in my eye. To make languages for whatever I am doing, that is power. Language and thought are entwined in each other. To think it, one must be able to express it. Or it is lost, just another emotion floating in the vast mental sea. But language is like a boat, nay,  an arc, that carries thoughts and communities around the mental landscape. Not every thought works with every language, or at least not easily. So the ability to build a variety of small boats to carry domains of thought, that is power.

And that is parsing. I have been studying Pratt’s Parsing technique, mainly from Douglas Crockford’s site: http://javascript.crockford.com/tdop/tdop.html

He[Pratt] claimed the technique is simple to understand, trivial to implement, easy to use, extremely efficient, and very flexible. It is dynamic, providing support for truly extensible languages.

So I have been working on creating a general framework where I can just plug in some syntax and get some results. Yesterday, I achieved arithmetic. Parentheses are partially done. But I think I need to refactor.

I was looking at a book on Domain Specific Languages (that’s it title), and it makes the good point of decoupling the behavior from the parsing of the language. So I want to see if I can accomplish that better since right now, I feel that I am to close to the parsing level. But I am having a hard time seeing how I am not just doing the same thing again. I think I need to dig deeper into understanding and see how each kind of behavior could be established.

But at least I was successful with creating an arithmetic parser.

With luck, I will soon create versions to parse simplified javascript, css, html, tex, asciimath, a variety of drawing languages (using canvas to draw), a gamebook language, wikicreole and other markups, and anything else I have a mind for. Creating languages I hope will become very easy. And one thing I want to make sure these all can do is to interoperate with each other. In particular, being able to markup a page for regular text with places to run math and code would be very powerful.

One dream project is to also have a web version of textmate, to an extent. I like syntax highlighting, brace matching, code folding, line numbering, find/replace with regular expressions, and a nice save/check/run system would be nice. All keyboard.

Another use for parsing might be voice dictation. I am still waiting to get the new dragon dictate for mac, but if it works as I hope, then maybe I can dictate this material. Having a parser will enable all of that to be translated fairly easily.

But of course the main target is my own version of GeoGebra. There is a lot of potential to make something utterly incredible and useful. That is my dream. It is getting closer.