能否通过Pydantic内置功能将模型类（非实例）导出为可跨语言识别的JSON？

你想要解析Pydantic模型类本身的结构并转换成其他语言可理解的格式，Pydantic确实提供了内置方法来获取模型的元数据结构，核心是利用模型的结构描述API，下面分版本说明实现方式：

1. Pydantic v2 版本 #

在v2中，使用model_json_schema()方法可以获取模型的JSON Schema格式结构，我们可以基于这个结果转换成你需要的嵌套结构：

示例代码 #

from datetime import datetime
from pydantic import BaseModel

class BarModel(BaseModel):
    whatever: int

class FooBarModel(BaseModel):
    foo: datetime
    bar: BarModel

def convert_schema(model, schema_cache=None):
    if schema_cache is None:
        schema_cache = {}
    model_name = model.__name__
    if model_name in schema_cache:
        return {model_name: schema_cache[model_name]}
    
    schema = model.model_json_schema()
    fields = {}
    for field_name, field_info in schema['properties'].items():
        # 处理嵌套模型
        if '$ref' in field_info:
            ref_name = field_info['$ref'].split('/')[-1]
            ref_model = model.model_registry[ref_name]
            fields[field_name] = convert_schema(ref_model, schema_cache)
        else:
            # 处理基础类型，适配你需要的格式
            field_type = field_info['type']
            if field_type == 'string' and 'format' in field_info:
                field_type = field_info['format']
            # 映射成示例中的类型名称
            type_mapping = {'integer': 'int', 'date-time': 'datetime'}
            field_type = type_mapping.get(field_type, field_type)
            fields[field_name] = field_type
    
    schema_cache[model_name] = fields
    return {model_name: fields}

# 转换BarModel
print(convert_schema(BarModel))
# 输出: {'BarModel': {'whatever': 'int'}}

# 转换FooBarModel
print(convert_schema(FooBarModel))
# 输出: {'FooBarModel': {'foo': 'datetime', 'bar': {'BarModel': {'whatever': 'int'}}}}

2. Pydantic v1 版本 #

在v1中，使用schema()方法获取模型结构，处理逻辑类似：

示例代码 #

from datetime import datetime
from pydantic import BaseModel

class BarModel(BaseModel):
    whatever: int

class FooBarModel(BaseModel):
    foo: datetime
    bar: BarModel

def convert_schema_v1(model, schema_cache=None):
    if schema_cache is None:
        schema_cache = {}
    model_name = model.__name__
    if model_name in schema_cache:
        return {model_name: schema_cache[model_name]}
    
    schema = model.schema()
    fields = {}
    for field_name, field_info in schema['properties'].items():
        if '$ref' in field_info:
            ref_name = field_info['$ref'].split('/')[-1]
            ref_model = next(m for m in BaseModel.__subclasses__() if m.__name__ == ref_name)
            fields[field_name] = convert_schema_v1(ref_model, schema_cache)
        else:
            field_type = field_info['type']
            if field_type == 'string' and 'format' in field_info:
                field_type = field_info['format']
            # 映射成示例中的类型名称
            type_mapping = {'integer': 'int', 'date-time': 'datetime'}
            field_type = type_mapping.get(field_type, field_type)
            fields[field_name] = field_type
    
    schema_cache[model_name] = fields
    return {model_name: fields}

# 转换示例
print(convert_schema_v1(BarModel))
# 输出: {'BarModel': {'whatever': 'int'}}
print(convert_schema_v1(FooBarModel))
# 输出: {'FooBarModel': {'foo': 'datetime', 'bar': {'BarModel': {'whatever': 'int'}}}}

说明 #

• 转换函数会递归处理嵌套模型，自动解析引用的子模型结构
• 通过type_mapping可以灵活调整类型名称，完全匹配你预期的输出格式
• 底层依赖Pydantic内置的模型结构描述功能，不需要额外第三方库

内容的提问来源于stack exchange，提问作者jbuddy_13