Write a text processing parser to convert a custom syntax into seni-random phrases

We have a custom syntax of the following type: { [ Hello, Greetings, world! planet! ] [ World, Planet, I greet you. I say hello. ] + [ Want a cookie? cup of tea? ] } Which will generate: "Hello, world! Want a cookie?" or "Planet, I say hello. Want a cookie?" This is a job to parse this syntax and output x number of generated results. It should also create mispellings, with transpositions, common keyboard shifts, or doubled or missing characters. It should be a command line program, e.g. generateTexts infile blockNum count mispellingRatio the infile could have many of these blocks (which we call recipes) enclosed in {braces}. So if the file had 100 recipes, you could output 6 result based on the 49th recipe. And on averge, 1 out of 10 words should have a mispelling: generateTexts [login to view URL] 49 6 0.1 I'd prefer this written in ruby, but perl or python are ok. If you have another platform, please justify it in your proposal.