如何在API应用与业务逻辑应用间共享Swagger生成的数据模型？

Great question—separating your API layer from business logic while reusing models is a smart move, and there are a few solid approaches tailored to your Flask/Connexion + Swagger setup. Let’s break down the best options:

方案1：抽离模型为独立共享Python包（最推荐） #

This is the cleanest, most scalable approach—it fully decouples your API and BL-App, centering all model definitions in a single source of truth. Here’s how to set it up:

• Create a standalone model package

  1. Spin up a new Python project (e.g., pet_core_models) with a standard structure:

     pet_core_models/
     ├── pyproject.toml  # or setup.py for older projects
     └── pet_core_models/
         ├── __init__.py
         ├── dog.py
         └── [other model files]
     

  2. Copy the model classes generated by Swagger Codegen (from python-flask/swagger_server/models/) into this package. For example, take the Dog class from dog.py and paste it into pet_core_models/dog.py. Whenever you update your Swagger YAML, regenerate the models and sync them to this package.

• Update your API project to use the shared package

  1. Add the shared package to your API project’s dependencies. For local development, use an editable install:

     pip install -e ../pet_core_models
     

     For production, publish it to a private PyPI server or include it as a git submodule.
  2. Replace the auto-generated model files in swagger_server/models/ with simple imports from the shared package. For swagger_server/models/dog.py:

     from pet_core_models.dog import Dog
     __all__ = ['Dog']
     

     This lets Connexion/Flask find the models it needs while pulling the actual definition from your shared package.

• Update your BL-App to use the shared package
  Add pet_core_models to your BL-App’s dependencies, then import models directly:

  from pet_core_models.dog import Dog
  
  # Use the Dog class in your business logic, e.g.:
  def get_all_dogs():
      return [Dog(name="Hasso", age=21), Dog(name="Lassy", age=15)]
  

Why this works: #

• Models are maintained once, eliminating duplication.
• API and BL-App are fully decoupled—changes to one don’t break the other (as long as model contracts stay compatible).
• Easy to version control the model package, so you can roll back or test changes safely.

方案2：让BL-App直接依赖API生成的模型（最简单） #

If your Swagger YAML is the single source of truth for your data structures, you can skip the shared package and have your BL-App use the models generated by Swagger Codegen directly.

• Link the API models to your BL-App
  Add the API project’s swagger_server directory as a dependency in your BL-App’s requirements.txt:

  -e ../python-flask/swagger_server
  

• Import models in your BL-App
  Use the same imports as the API layer:

  from swagger_server.models.dog import Dog
  
  # Your business logic here
  

Pros & Cons: #

• Pros: No extra setup—reuse the auto-generated models directly.
• Cons: Creates tight coupling between your BL-App and API project. If you restructure the API project (e.g., move models), you’ll have to update imports in your BL-App.

方案3：双向生成模型（for strict sync needs） #

If you want to define models in one place and auto-generate code for both the API and BL-App, use code generation tools to keep everything in sync:

• If Swagger YAML is your source:
  Use openapi-generator to generate API models, and datamodel-code-generator to generate pydantic-style models for your BL-App from the same YAML. You can script this in a CI pipeline to auto-update models whenever your Swagger spec changes.
• If BL-App models are your source:
  Define models in your BL-App using pydantic, then use pydantic2openapi to generate a Swagger YAML from those models. Use that YAML to generate your API code with Swagger Codegen.

Why this works: #

• Ensures 100% consistency between API and BL-App models—no manual syncing needed.
• Best for teams where model definitions change frequently.

Quick Tips #

• Make sure serialization/deserialization logic aligns between both layers. Swagger-generated models often include to_dict() and from_dict() methods—use those in your BL-App to keep data handling consistent.
• If you go the shared package route, add version numbers to avoid unexpected breaking changes between API and BL-App.

内容的提问来源于stack exchange，提问作者Maus