如何在Python加载YAML文件时解析内部变量引用？

默认的yaml.safe_load不会处理${xxx}格式的变量引用，以下两种方法可以实现变量解析：

方法一：手动递归替换变量 #

无需额外依赖，通过递归遍历YAML数据，识别并替换变量引用。

import yaml
from typing import Dict, Any

def resolve_vars(data: Dict[str, Any], root: Dict[str, Any]) -> Dict[str, Any]:
    resolved = {}
    for key, val in data.items():
        if isinstance(val, dict):
            resolved[key] = resolve_vars(val, root)
        elif isinstance(val, str) and val.startswith("${") and val.endswith("}"):
            # 拆分变量路径，如"download.input_data_dir"转为列表
            path_parts = val[2:-1].split(".")
            current = root
            try:
                for part in path_parts:
                    current = current[part]
                resolved[key] = current
            except KeyError:
                # 变量不存在时保留原始值，也可抛出异常
                resolved[key] = val
        else:
            resolved[key] = val
    return resolved

# 加载原始YAML
with open("../params.yaml", "r") as f:
    raw_params = yaml.safe_load(f)

# 解析变量
params = resolve_vars(raw_params, raw_params)

# 验证结果
print(params['prepare']['input_dir'])  # 输出: ./data/input
print(params['process']['output_dir'])  # 输出: ./output/1

方法二：使用Jinja2模板引擎预渲染 #

借助Jinja2的模板渲染能力，先替换变量再加载YAML，适合复杂场景。

首先安装依赖：

pip install jinja2

代码示例：

import yaml
from jinja2 import Environment, BaseLoader

# 读取YAML文件内容
with open("../params.yaml", "r") as f:
    yaml_content = f.read()

# 加载原始数据作为渲染上下文
with open("../params.yaml", "r") as f:
    raw_params = yaml.safe_load(f)

# 配置Jinja2使用${}作为变量分隔符
env = Environment(loader=BaseLoader(), variable_start_string="${", variable_end_string="}")
template = env.from_string(yaml_content)
# 渲染替换变量
rendered_content = template.render(**raw_params)

# 加载渲染后的YAML
params = yaml.safe_load(rendered_content)

# 验证结果
print(params['prepare']['input_dir'])  # 输出: ./data/input
print(params['process']['output_dir'])  # 输出: ./output/1

内容的提问来源于stack exchange，提问作者raj