A library to enable making simple domain specific languages
This is a library to aid developers in creating neat little embedded DSLs when using python, without having to do any complex parsing or anything like that. The resulting DSLs look a bit more like english than python does.
The idea for this was inspired by some ruby stuff that I’ve been using lately. I’ve been using ruby quite a bit lately, and while I am still not a huge fan of the language, I do like the idea of easy to code up DSLs. Since I wanted a DSL for a python project I was working on, I played with a few ideas and ended up with this.
First, create some very basic python objects, and call the parse_dsl function or parse_dsl_file function with the file to process, and an object that the DSL is executed with. Here’s a sample recipe DSL for baking:
author = 'Mom' level = 'Easy' with Instructions('Yummy cookies'): preheat_oven = '375F' with Ingredients: add += '2 1/4 cups', 'flour' add += '1 tsp', 'baking soda' add += '1 tsp', 'salt' add += '1 cup', 'butter' add += '3/4 cup', 'sugar' add += '3/4 cup', 'brown sugar' add += '1 tsp', 'vanilla extract' add += '2', 'eggs' add += '2 cups', 'chocolate chips' add += '1 cup', 'nuts'
And the implementation to parse that dsl is super simple:
import dsltool class Ingredients(object): def __init__(self): self.add = dsltool.ItemList() class Instructions(object): preheat_oven = None ingredients = Ingredients steps = Steps def __init__(self, name): self.name = name class Recipe(object): author = None level = None def __init__(self): self.instructions = Instructions if __name__ == '__main__': obj = dsltool.parse_dsl_file('cookies.pydsl', Recipe)
There is a working example in the ‘examples’ directory. Check it out.
No idea. I’m sure it’s not that bad. I imagine typical usage of a DSL isn’t going to be in a performance-dependent part of your application anyways.
If the user writes the DSL incorrectly, there are lots of possible gotchas. If enough people use this thing and file bug reports, together we can remove many of them. I’ll try to list them as I find them.
Do not expect 0.1.x releases to have a stable API. I’m still exploring the possibilities of this, so as better ways to do things emerge newer releases will have those instead.
Uses py.test for testing. Still need better tests, but it catches a lot of common mistakes I’ve made so far. ;)
It’s all on github, so file your bug reports there. If you have a patch, the best way to do it is just create a fork and send me a pull request. If you don’t include a test case for your patch (these are really easy to create!), then it probably won’t be accepted.