Text Game Commands Parsing?

Question

I'm interested in writing a parsing engine such as the ones in Inform, which is very robust.

I've tried a couple of different approach:

• regex. For regex, I can only match {verb} {noun1 optional-noun2} {optional-preposition} {indirect-noun1 optional-indirect-noun2} This doesn't work that well.
• list of predefined verbs and their matched preposition and nouns. This approach is too complex, since there has to be A LOT of predefined data.
• splitting into different words and run them through a dictionary maps. Again this is too complex and not practical due to the amiguity of certain words such as "look" or "put (on, out)"

I know about language processing though I have actually no idea how to start and some of the stuff are too complicated. (No CS background, I'm just learning on my own right now).

Some ideas I have:

• As I've said, natural language processing.
• Design a parser that parses something similar to SQL.

Is there any other ideas? How would I go about design and implementing them?

Note I'm using Python as my language. and in case you missed it, I'm not design a programming language, I'm just parsing user command such as "go north" or complex stuff, like "put the fire out with the blanket"

Answer 1

Besides the mentioned pyparsing, which looks very appropriate (I've only used it a tiny bit), you should look into using nltk. It's a great collection of tools for natural language processing, and there's even a book.