Youtube Autonomous FastAPI Docker API Module

The module that is handled by a docker container to isolate the environment we need for each purpose, exposed by a FastAPI that allows asking for the specific resources.

This project is using poetry to handle the dependencies and the virtual environment, and docker to manage the specific python version and all the code with the app to be executed.

This is the entry point API for all the functionality our Youtube Autonomous system is providing. This API will delegate in other API-containers to get what is needed. Only this API is exposed, and all the other containers have another one (for internal use only).

Steps to do

Considering that we have a docker-compose.yml that includes the different services, we have to copy the VibeVoice voices to the demo/voices folder to let the internal container detect them to be used. We can have any custom voice we want, that will be autodetected, such as demo/voices/fr-Eva_woman.wav.

Endpoints

Youtube

GET

• /youtube/videos/{id_video}
• /youtube/videos/{id_video}/data
• /youtube/videos/{id_video}/download-best-quality
• /youtube/videos/{id_video}/download-lowest-quality
• /youtube/videos/{id_video}/download-1080

You can replace the {id_video} with this id e_YBmIxeyUg for testing.

POST

No endpoints available

Llama.cpp

GET

No endpoints available

POST

• /llamacpp/gemma4
  • prompt - The prompt we want to give
  • output_schema - The schema we want as the output

VibeVoice

GET

• /vibevoice/speech
  • text - The text to narrate
  • voice - The voice we want to use to narrate. Choose one voice in demo/voices or copy your own.
• /vibevoice/audio
  • text - The text to narrate
  • voice - The voice we want to use to narrate (en-Alice_woman, en-Carter_man, fr-Eva_woman, etc.) Choose one voice in demo/voices or copy your own.

POST

No endpoints available

Proxy

GET

• /proxy/image
  • url - The url to download the image from.

POST

No endpoints available

Other Instructions

Generate the image

We need to generate the docker image with all the things we need to be able to run the app inside.

• Use the $ docker build --no-cache -t {LIBRARY_NAME} . command to generate the Docker image by using the dockerfile file and ignoring the caché. This will download the python 3.12.x version, install the dependencies and copy the code. Use this if any dependency has changed since the last time. If you are just updating the code, you can use the following instead.
• Use the $ docker build -t {LIBRARY_NAME} . command to generate the Docker image by using the dockerfile file. It will download the python 3.12.x version, install the dependencies and copy the code, each of these steps only if needed (the cache will make it be ignore if it didn't change).

Run the container

We need to run the container, so it will be mounted and the app will be runing and ready to use.

• Use the command $ docker rm -f {LIBRARY_NAME} 2>nul to remove the previous container if existing, so we are able to mount it again from zero.
• Use the command $ docker run -d -p "%port%":8000 --name {LIBRARY_NAME} {LIBRARY_NAME} to run and mount the container and make the app be ready to use.

You can also execute the run_server_docker.bat to do all together using the caché (faster).

The docker will run uvicorn in the port 8000 internally, but our navigator will redirect the 8001 to that one, so we can have different docker containers working at the same time to provide different services using the same base (uvicorn + fastapi).

Other details

• To run the project locally, execute the run_server.bat file or use $ poetry run uvicorn yta_fastapi_docker_base.app.main:app --reload directly.