通过查询内联替换AST节点

问题描述

给出如下输入:

query = ("class_name","function_name","arg_name")

如何用其他提供的node替换找到的内容?

之前阶段的解析示例:

class Foo(object):
    def f(self,g: str = "foo"): pass

后阶段的示例:

class Foo(object):
    def f(self,g: int = 5): pass

对假设函数进行以下调用:

replace_ast_node(
    query=("Foo","f","g"),node=ast.parse("class Foo(object):\n    def f(self,g: str = 'foo'): pass"),# Use `AnnAssign` over `arg`; as `defaults` is higher in the `FunctionDef`
    replace_with=AnnAssign(
        annotation=Name(ctx=Load(),id="int"),simple=1,target=Name(ctx=Store(),id="g"),value=Constant(kind=None,value=5),),)

我在一起a simple solution for finding a node with the query list进行了黑客攻击,这具有为("Foo","g")以及{{3}中的a parser/emitter }至arg。但是我无法弄清楚这个阶段。 AnnAssign是否按顺序遍历? -因此,我应该不断遍历,附加当前名称并检查当前位置是否为完整查询字符串? -我觉得我缺少一些干净的解决方案……

解决方法

我决定将其分为两个问题。首先,让每个节点了解其在宇宙中的位置:

def annotate_ancestry(node):
    """
    Look to your roots. Find the child; find the parent.
    Sets _location attribute to every child node.

    :param node: AST node. Will be annotated in-place.
    :type node: ```ast.AST```
    """
    node._location = [node.name] if hasattr(node,'name') else []
    parent_location = []
    for _node in ast.walk(node):
        name = [_node.name] if hasattr(_node,'name') else []
        for child in ast.iter_child_nodes(_node):
            if hasattr(child,'name'):
                child._location = name + [child.name]
                parent_location = child._location
            elif isinstance(child,ast.arg):
                child._location = parent_location + [child.arg]

然后实现上述ast.NodeTransformer的一种方法:

class RewriteAtQuery(ast.NodeTransformer):
    """
    Replace the node at query with given node

    :ivar search: Search query,e.g.,['class_name','method_name','arg_name']
    :ivar replacement_node: Node to replace this search
    """

    def __init__(self,search,replacement_node):
        """
        :param search: Search query
        :type search: ```List[str]```

        :param replacement_node: Node to replace this search
        :type replacement_node: ```ast.AST```
        """
        self.search = search
        self.replacement_node = replacement_node
        self.replaced = False

    def generic_visit(self,node):
        """
        Visit every node,replace once,and only if found

        :param node: AST node
        :type node: ```ast.AST```

        :returns: AST node,potentially edited
        :rtype: ```ast.AST```
        """
        if not self.replaced and hasattr(node,'_location') \
           and node._location == self.search:
            node = self.replacement_node
            self.replaced = True
        return ast.NodeTransformer.generic_visit(self,node)

您已经完成=)


用法/测试:

parsed_ast = ast.parse(class_with_method_and_body_types_str)
annotate_ancestry(parsed_ast)
rewrite_at_query = RewriteAtQuery(
    search="C.method_name.dataset_name".split("."),replacement_node=arg(
        annotation=Name(ctx=Load(),id="int"),arg="dataset_name",type_comment=None,),).visit(parsed_ast)
self.assertTrue(rewrite_at_query.replaced,True)
# Additional test to compare AST produced with desired AST [see repo]

相关问答

依赖报错 idea导入项目后依赖报错,解决方案:https://blog....
错误1:代码生成器依赖和mybatis依赖冲突 启动项目时报错如下...
错误1:gradle项目控制台输出为乱码 # 解决方案:https://bl...
错误还原:在查询的过程中,传入的workType为0时,该条件不起...
报错如下,gcc版本太低 ^ server.c:5346:31: 错误:‘struct...