[QUESTION] How to transform data

First check

• I used the GitHub search to find a similar issue and didn’t find it.
• I already searched in Google “How to X in FastAPI” and didn’t find any information.
• I searched the FastAPI documentation, with the integrated search.

Description

How can I transform data received from ORM when passing to Pydantic?

For example: We have a table in PostgresQL named objects, which has 4 columns: id, title, lat, lng and we created a model to it named ObjectModel:

class ObjectModel(Base):
    __tablename__ = 'objects'

    id = Column(Integer, primary_key=True)
    title = Column(String)
    latitude = Column(Float, name='lat')
    longitude = Column(Float, name='lng')

When we make a request (db.query(models.ObjectModel).all()) to receive all content of the table we will get something like this:

[
  {
    "id": 1,
    "title": "object 1",
    "latitude": 12.34,
    "longitude": 12.34
  },
  {
    "id": 2,
    "title": "object 2",
    "latitude": 23.45,
    "longitude": 23.45
  }
]

How to transform it with Pydantic to something like this:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Point",
        "coordinates": [
          "12.34",
          "12.34"
        ]
      },
      "properties": {
        "id": 1,
        "title": "object 1"
      }
    },
    {
      "type": "Feature",
      "geometry": {
        "type": "Point",
        "coordinates": [
          "23.45",
          "23.45"
        ]
      },
      "properties": {
        "id": 2,
        "title": "object 2"
      }
    }
  ]
}

Or how to return objects instead of fields: Instead of

{
  "id": 1,
  "title": "object 1"
}

return

[
  {
    "key": "Object number",
    "value": 1
  },
  {
    "key": "Object title",
    "value": "object 1"
  }
]

Additional context

Add any other context or screenshots about the feature request here.

[tiangolo]

Thanks for the help here @fullonic ! 🙇 👏

@Andreigr0 as you are not really returning the data as it comes from the DB but applying some transformations, I think that’s extra logic that would probably belong to your specific app. So it would be best to first modify the data to have the shape that you need in your code, and then return it, and have the Pydantic model just for the data with the final shape.

What I mean is, it will be a lot easier to not try and do the transformation of the data in a Pydantic model passed to response_model but rather in your own code. It will be a lot easier to debug, understand, etc. And as it is part of your custom business logic, it would probably be better living in your function and not in a Pydantic model.

[github-actions[bot]]

Assuming the original issue was solved, it will be automatically closed now. But feel free to add more comments or create new issues.