Contents
Schemepy front-end will define a unified interface. It will be independent to the back-end used.
Using Schemepy to embed Scheme in the Python program involves 6 phases:
If you've installed Schemepy properly in your sys.path, loading can be done simply by
import schemepy as scheme
The configuration is totally optional because usually the default values can serve you well. But if you do want some customization, the rule of thumb is to do it before you start using Schemepy.
By default, Schemepy will search your system automatically and use the first supported Scheme implementation as the back-end it found. If no one found, it will use the build-in pure-Python Scheme implementation. You can customize this process if you would prefer one back-end to the other.
FIXME: describe how it is customized.
FIXME: describe how it is customized.
To create a new VM is simple:
vm = scheme.VM()
This VM will contain a basic environment where Scheme code can run (FIXME: shall we define a standard environment? It it necessary? Or is it feasible? This would have something to do with sandboxing).
The VM can be extended by adding global variables or primitives to the environment.
Adding global variables to the VM can be done through define or set, the have similar semantic to define and set! in Scheme.
vm.define("var", val)
The value should be a Scheme value. One can get a Scheme value by calling scheme.toscheme(val) from a Python value val.
Functions can also be added to the VM by the same way as normal objects:
def myadd(a, b):
a = vm.fromscheme(a)
b = vm.fromscheme(b)
return vm.toscheme(a+b)
vm.define("myadd", vm.toscheme(myadd, shallow=True))
Alternatively, you can omit the shallow=True to let Schemepy do the convertion of the function parameters and return values for you automatically.
def myadd2(a, b):
return a+b
vm.define("myadd2", vm.toscheme(myadd2)
Here's the map between Scheme type and Python type, more detailed description can be found in the type mapping document:
| Scheme Type | Python Type |
|---|---|
| bool | bool |
| int | int |
| float | float |
| complex | complex |
| symbol | schemepy.symbol |
| cons | schemepy.cons |
| vector | list |
| hash | dict |
| python data | Normal Object |
| object | Scheme Object |
| primitive function | callable |
| lambda | callable |
Schemepy compiles Scheme source code into a compiled form. It can be executed multiple times in different environments.
compiled = vm.compile(source)
When an error occured during parsing, a CompileException will be thrown. It has the following attributes:
To run a piece of compiled source code, just call eval method of the VM:
result = vm.eval(compiled)
Like install_function, the result can be automatically converted. This behavior can be controlled by a global configuration or through the autoconvert keyword parameter of eval.