如何確定在方法中引用哪些自我屬性？

Question

考慮：

class A(object): 
    def m(self): 
    print self.x + self.y + self.z() 

有沒有一種方法來查詢：

self_attributes_referenced(A.m) #=> ['x', 'y'] 

沒有做反編譯或解析？

是否有相應的方法來調用自我方法？

self_methods_referenced(A.m) #=> ['z'] 

Answer 1

像這樣的東西應該工作：

import inspect 


class Referee(object): 
    def __getattribute__(self, name): 
     attr = object.__getattribute__(self, name) 
     if inspect.ismethod(attr): 
      try: 
       meths_referenced = object.__getattribute__(
        self, 'meths_referenced') 
      except AttributeError: 
       meths_referenced = set() 
       object.__setattr__(self, 'meths_referenced', meths_referenced) 

      meths_referenced.add(name) 
     else: 
      try: 
       attrs_referenced = object.__getattribute__(
        self, 'attrs_referenced') 
      except AttributeError: 
       attrs_referenced = set() 
       object.__setattr__(self, 'attrs_referenced', attrs_referenced) 

      attrs_referenced.add(name) 

     return attr 


def attributes_referenced(obj): 
    try: 
     return object.__getattribute__(obj, 'attrs_referenced') 
    except AttributeError: 
     return set() 


def methods_referenced(obj): 
    try: 
     return object.__getattribute__(obj, 'meths_referenced') 
    except AttributeError: 
     return set() 


class A(Referee): 
    def __init__(self): 
     self.x = 1 
     self.y = 2 

    def z(self): 
     return 3 

    def m(self): 
     print self.x + self.y + self.z() 


a = A() 
a.m() 
print attributes_referenced(a) 
print methods_referenced(a) 

此打印：

6 
set(['y', 'x']) 
set(['z', 'm']) 

繼承Referee，每當你想提供此功能。 爲什麼你需要這個功能超出了我的理解。 =）

Answer 2

在這個問題中，您將解析排除爲解決方案，但您是否意識到使用ast module解析和修改Python源代碼是多麼容易？

首先定義一個NodeTransformer：

import ast 

class RecordReferencesTransformer(ast.NodeTransformer): 
    def visit_FunctionDef(self, node): 
     def is_attr_on_self(n): 
      return type(n) is ast.Attribute and type(n.value) is ast.Name and n.value.id == 'self' 

     targets = [ast.Attribute(value=ast.Name(id=node.name, ctx=ast.Load()), attr='self_attributes_referenced', ctx=ast.Store())] 
     attrs = [n.attr for n in ast.walk(node) if is_attr_on_self(n)] 
     value = ast.Tuple(elts=[ast.Str(s=attr) for attr in attrs], ctx=ast.Load()) 
     return [node, ast.Assign(targets=targets, value=value)] 

然後而不是直接使用import加載我們的模塊，我們需要一點額外的樣板在我們NodeTransformer勻場：

filename = 'program.py' 
parsed = ast.parse(open(filename).read()) 
with_references = RecordReferencesTransformer().visit(parsed) 
ast.fix_missing_locations(with_references) 
exec compile(with_references, filename, 'exec') 

print A.m.self_attributes_referenced 
a = A() 
a.m() 

輸出：

('x', 'y') 
3 

實施self_methods_referenced作爲練習留給讀者。

最後，program.py來源：

class A(object): 
    x, y = 1, 2 
    def m(self): 
     print self.x + self.y