Type based creator

A django application which creates dynamic tables/models.

One particular cool feature of this application is that besided the model ID there is no other model/table defined in the source code. Things are trully dynamic :)

Another cool feature of this application is that it preserves the ability to access the models even after application restart. I've spent several hours figuring out this feature alone. Django normally loads it's models to the registry from the source code, so upon restart the information about dynamic models is usally lost. I use a small dirty trick to bypass this limitation - I regenrate the models on the fly by quering the fields/columns. This allows me to preserve the models even if the application is restarted.

How to run things

To run this locally you will need poetry.

Simply execute:

curl -sSL https://install.python-poetry.org | python - --version "1.4.2"

Once the poetry is installed you can install the application with:

make setup

You can run tests with:

make test

And you can start the application for manual testing with:

make server

API endpoints and how to make requests

Table Enpoint

POST http://localhost:8005/table - create dynamic model.

The request body should be in format {name:type}:

{
    "field_1": "number",
    "field_2": "boolean"
}

Accepted types:

• number
• boolean
• string

Example cURL:

curl --request POST \
  --url http://localhost:8005/table \
  --header 'Content-Type: application/json' \
  --data '{
  "field_1": "number",
  "some_field": "boolean"
}'

---

GET http://localhost:8005/table/{id} - get dynamic model.

Example cURL:

curl --request GET \
  --url http://localhost:8005/table/1 \
  --header 'Content-Type: application/json'

PUT http://localhost:8005/table/{id} - update dynamic model. Sets new fields to the model and updates the types of the existing fields if provided in payload.

Example cURL:

curl --request PUT \
  --url http://localhost:8005/table/1 \
  --header 'Content-Type: application/json' \
  --data '{
  "field_1": "boolean",
  "some_field": "string",
  "new_field": "number"
}'

Cool Table Endnpoint

Cool table is pretty similar with the key difference that the expected format of the request is: {title:value}

Title can be any string and value should be of the type int/float/boolean/string. The application will determine the type of the value and will created fields named according to title and with the type corresponding to the value type. Since we are not creating instances of the table, the actual value will be ignored and only used for the type extraction.

POST http://localhost:8005/cool-table - create dynamic model.

Example body:

{
    "field_1": "some string",
    "field_2": true,
    "filed_3": 14
}

Example cURL:

curl --request POST \
  --url http://localhost:8005/cool-table \
  --header 'Content-Type: application/json' \
  --data '{
  "field_1": true
}'

---

PUT http://localhost:8005/cool-table/1 - create dynamic model.

Example cURL:

curl --request PUT \
  --url http://localhost:8005/cool-table/1 \
  --header 'Content-Type: application/json' \
  --data '{
  "field_1": -3,
  "some_interesting_field": "meaningless_text"
}'

NOTE: there is a problem with updating boolean to postgres-specific double precision field, so it is recommend to only use intergers.

Row Endnpoint

POST http://localhost:8005/table/row - create a row of a dynamic model.

NOTE: the request must contain the fields with the same names and types as the current dynamic model, overwise you'll get back a bad request.

Example cURL:

curl --request POST \
  --url http://localhost:8005/table/1/row \
  --header 'Content-Type: application/json' \
  --data '{
  "field_1": 12,
  "new_field": false,
  "some_field": "dfgdfsd",
  "some_interesting_field": "dfsdfdsf"
}'

---

GET http://localhost:8005/table/rows - gets all rows of a dynamic model.

Example cURL:

curl --request GET \
  --url http://localhost:8005/table/1/rows \
  --header 'Content-Type: application/json'