Zeeguu-API is an open API that allows tracking and modeling the progress of a learner in a foreign language with the goal of recommending paths to accelerate vocabulary acquisition.
The API is also currently deployed as the backend for the zeeguu.org website.
The API offers translations for the words that a learner encounters in their readings. The history of the translated words and their context is saved and used to build a dynamic model of the user knowledge. The context is used to extract the words the user knows and their topics of interest.
A teacher agent recommends the most important words to be studied next in order for the learner to accelerate his vocabulary retention. This information can be used as input by language exercise applications, for example interactive games.
A text recommender agent crawls websites of interest to the user and recommend materials to read which are in the zone of proximal development.
To read more about the API see the article published about Zeeguu in the CHI'18 conference.
Once you clone the repo, please run:
git config --local core.hooksPath .githooks/
This will make the rules in the .githooks/rules folder be run before every commit. The rules check for well-known bugs and code conventions.
docker
on your machine. For Ubuntu you can run the following:sudo apt-get install docker.io -y
This is useful for MacOS machines (M1 and later) on which MySQL does not seem to be running within docker
Build the zeguu_api_dev
development image
docker build -f Dockerfile.development -t zeeguu_api_dev .
Run the _playground.py to ensure that you have something in the DB
docker-compose up dev_play
To ensure that the changes to the files on your local dev machine are reflected inside the container try to modify
something in the tools\_playground.py
file and rerun this command. Do you see the changes? That's good.
Run the development server inside of the container
docker-compose up dev_server
to test it open http://localhost:9001/available_languages in your browser and you should be able to see a list of language codes that are supported by the system
Test the deployment
docker-compose up dev_test
Running from a docker image, at least on my M1 Max from 2021, is terribly slow. The _playground.py
script takes 1s
natively and 6s in Docker. Tests natively are 22s and in Docker are 280s!
So for running the development server this is ok, but for actual development, this might be quite annoying :(
mkdir /Users/mircea/zeeguu-data
envsubst
installed (i.e. brew install gettext
)default_env
to a newly created .env
filegenerate-configs.sh
; verify that api.cfg
and fmd.cfg
have meaningful values insidedocker compose up
localhost:8080/available_languages
: if you see an array like ["de", "es", "fr", "nl", "en"]
you have the API working.Once you make changes to the code you have to restart the apache2ctl inside the container. To test this do the following:
available_languages
in system_languages.py
and then
run docker exec -it api-zapi-1 apache2ctl restart
That's all. Go have fun!
(Mircea, Feb 2024)
On Mac, if you want to run mysql locally, and not from within Docker, you need to install mysql-client with brew:
brew install mysql-client
Mircea: On my M2 mac the pip instal mysqlclient
(called indirectly via pip install -r requirements
) still fails till
I define the following:
export MYSQLCLIENT_CFLAGS="-I/opt/homebrew/opt/mysql-client/include/mysql/"
export MYSQLCLIENT_LDFLAGS="-L/opt/homebrew/opt/mysql-client/lib -lmysqlclient"