GSoC'25 idea: Migrate IHR API

[dpgiakatos]

All results displayed on the Internet Health Report website are accessible via the IHR REST API. The current codebase is developed on an outdated version of Django (2.2.27), which is no longer supported, and we are not utilizing all of Django's features. Therefore, instead of just upgrading the Django version, we plan to migrate the codebase from the complex Django framework to a simpler framework, such as FastAPI. Additionally, the current Django framework includes code for initializing/managing the database. This time, we want to implement Bash scripts for initializing/managing the database, which will result in better maintenance.

Expected results:

• Migrate all Django endpoints to FastAPI
• Convert the database code from Django to Bash scripts
• Dockerize the application (excluding the Bash scripts, which will be run as is)

Knowledge Prerequisite:

• Django (to understand the current codebase)
• FastAPI
• SQL
• Docker

Resources:

• https://github.com/InternetHealthReport/ihr-django
• https://docs.djangoproject.com/en/2.2/
• https://fastapi.tiangolo.com/
• https://learnscripting.org/streamlining-database-operations-running-sql-queries-from-a-shell-script/

Old discussion page:
https://github.com/orgs/InternetHealthReport/discussions/40

[EbotFabien]

Hello @dpgiakatos & @romain-fontugne,

I've been exploring the Django codebase for the Migrate IHR API project and analyzing the best approach for transitioning to FastAPI. While reviewing the architecture, I have a few questions:

1️⃣ What are the key challenges you anticipate during the migration?
2️⃣ Should API endpoints remain 100% compatible, or are optimizations encouraged?
3️⃣ Are there any planned database schema modifications, or should it remain as-is?

Looking forward to your insights!

[dpgiakatos]

Hi, I am answering your questions:

1. To understand the current codebase
2. Yes, all API endpoints should remain 100% compatible (no changes)
3. No, the database will remain as it is

[EbotFabien]

Thanks for your feedback

[EbotFabien]

Hi @Romain, @Dimitrios,

I’ve successfully deployed the IHR-Django Backend on my localhost and am working through the setup. Could you please provide some guidance on how to obtain the login credentials or a valid token to access the API? I’d like to get familiar with the database models and test the functionality, but the README doesn't include any login details.

Additionally, are there any other resources or materials, apart from the documentation, that could help me make proper requests or better understand how to interact with the API?

Any help would be greatly appreciated!

[dpgiakatos]

You have to follow the documentation for creating a local instance of our database. We will not provide any token for direct access to our deployed database for security reasons.

[EbotFabien]

Just did thanks,i'm able to navigate through the API documentation.

[neoandmatrix]

Hi @dpgiakatos i was going through the current codebase and analyzing the things that might be required to be modified in order to achieve a successful migration to FastApi and had a few doubts regarding the same.

1. Is there a recommended/reference codebase structure that i should follow when migrating the project from Django to FastAPI, or is it better to decide based on the project's specific needs as django has an opinionated codebase structure.

2. Although during migration many packages will be added or removed but a prominent package that might be used most will be orm. Since django has a built in orm which is being used in the project, but FastApi dosen't have a in built orm , so is something preferred or it can be decided based on personal choice?

3. If any other frameworks or packages are preferred can you please let me know.

Looking forward to your views on above. Will be very help helpful to me if you could provide some guidance for above questions.

Thnaks.

[dpgiakatos]

Hi, I am answering your questions:

1. No, there isn't a recommended codebase structure, you will have to propose one in your proposal that will be easy to maintain
2. Yes, we will have to use an ORM, but I prefer to have SQL code instead of using the ORM. This can be discussed with other mentors.
3. The goal is to successfully migrate to FastAPI. So, FastAPI is the main framework. Also, some packages that have been used in the Django framework should also be reused (e.g. the one used during analysis). If we want to use a specific package, we will let you know.

[neoandmatrix]

Thanks for clarification.

[neoandmatrix]

@dpgiakatos The existing project uses just pip as it's package manager but now there are some better package managers like uv can we also consider using that also during this migration.

Please Guide.
Thanks.

[dpgiakatos]

No, pip will be the package manager. Thanks.

[RathQd]

@dpgiakatos I would like to contribute to this project. I would like to know 1 things.

1. Why are we thinking to use bash script instead of third party good migration libraries which works fine with Fastapi.

[dpgiakatos]

Hi, I am answering your question:

1. With the bash scripts, we will handle some CRUD operations that happen only once. The plan is to minimize the dependencies needed.

[Amrish-Sharma]

@dpgiakatos : I am interested in contributing to this project. Could you please clarify the requirements on below queries

1. How do you envision the Bash scripts managing the database? Will they handle migrations, backups, or both? Are there any specific database operations that need to be handled by these scripts?
2. What is the expected deployment environment for the Dockerized FastAPI application, and are there any specific Docker or container orchestration tools that you plan to use for deployment?

[dpgiakatos]

Hi, I am answering your questions:

1. In the Django framework, there are the use cases that we are looking for in the Bash scripts.
2. We will use only Docker.

[Amrish-Sharma]

thanks for your quick response, @dpgiakatos !

[pkkulk]

@dpgiakatos I have some questions
Should we follow a phased migration approach (running both Django & FastAPI in parallel for a while), or do you prefer a complete switch at once?

Would you like the Bash scripts to handle database seeding (adding initial data), or will that be managed separately?

[dpgiakatos]

Hello, I am answering your question:

1. Probably the switch will be made at once. I think this is something that does not need to be included in your proposal.
2. Only the database seeding.

Thanks.

[rjalamanac]

Hi everyone,

I'm currently drafting a GSoC proposal for migrating the IHR API from Django to FastAPI. I wanted to clarify whether OAuth support is needed as part of the authentication and authorization mechanisms.

Should we implement OAuth2, or would a simpler authentication method (e.g., JWT or session-based auth) be preferable? Also, are there any existing authentication flows that need to be preserved in the new API?

Looking forward to your feedback!

[dpgiakatos]

Hello, our API is free for everyone. We do not have a registration system and we will not have one. The endpoints that you can see in the current implementation, which contain registration, are deprecated (we will not include any endpoint which is comment in). We will not include them during the migration.
You can view the endpoints that we want to migrate in this page:
https://www.ihr.live/en/api

Thanks.

[Saarthak-Crib]

[dpgiakatos]

Hello,

I am answering your questions below:

1. We want to use bash scripts only for the initialization of the database.
2. There is no need for something that would only run once.
3. You will have to explore this using the current codebase.
4. We do not need an API testing process. We (mentors) will configure the GitHub Actions.

Thanks.