git clone https://github.com/guerard/ce-placemats-flask.git && cd ce-placemats-flask The directory you're
in now is what's considered the project root.
Install MongoDB and run a local instance.
PyCharm is the best IDE for writing python.
In python projects we isolate our dependencies to a given project using a tool called virtualenv.
Inside the project root you should create your virtual environment by running:
virtualenv -p python3 .venv
To actually use the virtual environment you'll need to prepare your shell so that all python-related commands take effect there (and only there). Here is how to 'activate' your shell. All commands should be entered from a shell that has been activated.
pip install -r requirements.txt from the project root. If there are errors you may be missing
some system dependencies; Google the error message, package that's failing, and OS to figure out
what you need to install.
Open PyCharm and then choose to import a project. Select the project root directory. PyCharm should detect
your virtual environment, but if not you should select the python interpreter (it's a binary) from within
the .venv directory.
Fix GeoText
In geotext.py (one of our library dependencies), line 45, add encoding='utf-8' as a kwarg to open
e.g. change that line to:
with open(filename, 'r', encoding='utf-8') as f:PyCharm should detect the runConfigurations from within the .idea directory. These contain the commands
needed to run both the http server ('flask run') and the task consumer ('consumer'). The run commands are
selected in the top-right of the IDE near the Play and Debug buttons. After choosing a run command from the
drop down, run it using the Play button.
flask run runs the HTTP server on 127.0.0.1:5000
consumer starts the task consumer
Best thing about python is being able to run code in a shell-like environment a.k.a. "the REPL".
PyCharm's 'Python Console' will open a python shell associated with the project's virtual environment.
From there you can import code just as you would in source code e.g.
from app.placemats.data.ncbi_client import * would run the code in that package, and import all the defined
classes and functions into the current shell's scope (similar to how importing a package works when writing
python source files). Try running the previous command in the REPL during setup.
To test that it worked, run configure_client() and then run a search on pubmed using
pubmed_search('alopecia areata treatment'). The REPL is a useful tool whenever you need to double-check
the value of an expression or explore an API.
Both commands are run from the project root:
To run the HTTP server:
FLASK_ENV=development FLASK_APP=app/main.py python3 -m flask run
And the consumer:
python3 -m app.placemats.consumer.widgets_task_consumer
Whenever you add a new dependency, you have to make sure you update both requirements.txt
and re-run the docker image locally before deploying to make sure the container runs OK
still. To update requirements.txt, first pip install whatever packages you want to add
and then run pip freeze > requirements.txt. Caution: b.c.o. the way pip works, this will
snapshot all packages currently installed in your virtual environment. Be sure not to add
packages that are not actually used in code.
Run docker-compose build && docker-compose up from the project root directory.
The API will be served on 127.0.0.1:8080
All list endpoints support pagination via the limit and skip query parameters. limit is
the max. count to return, and skip determines the starting index.
GET /layouts/<search terms>
GET /layouts/ -- Max. limit: 50
GET /widgets/<widget_id>
GET /widgets/ -- Max. limit: 10
You can choose how to host the MongoDB instance. In this case, I've chosen to deploy it as a docker container similar to the steps outlined below except using a mongo image instead. You'll just need to make sure that instance is accessible to the ce-placemats-flask app, and know the IP address/hostname to reach it.
The app is packaged as a minimal alpine image with only the necessary dependencies installed. Supervisor is used to run two daemon processes: the flask http server and the queue consumer.
The below commands (bash script aliases) are what I use to deploy the docker image. They should be run from the project's root directory, and do not require you to be in an activated virtual environment shell. You'll need to make sure you're running relatively recent versions of gcloud and docker, and that you're authorized to push to the conceptualeyes-169807 project's image repository (gcr).
alias gcpDeploy='buildCePlacemats && gcpTagCePlacemats && gcpCePush'
alias gcpCePush='gcloud docker -- push us.gcr.io/conceptualeyes-169807/ce-placemats-flask:latest'
alias gcpTagCePlacemats='docker tag ce-placemats us.gcr.io/conceptualeyes-169807/ce-placemats-flask:latest'
alias buildCePlacemats='docker build -t ce-placemats .'
alias runCePlacemats='docker run -p 8080:80 -e NCBI_EMAIL=robot@gmail.com -e MONGO_URL=mongodb://host.docker.internal:27017 -e SKIP_AUTH0=TRUE ce-placemats'
alias buildAndRunCePlacemats='buildCePlacemats && runCePlacemats'Once the image is uploaded, then go to the GCP console and set up a new VM. Select the container-optimized
VM image, and select the ce-placemats-flaks image. You'll need to make sure you're passing the correct value
of MONGO_URL and NCBI_EMAIL as environment variables to the VM (configurable via the console). This can
also be done via CLI too; here are
the docs
Currently there's a VM already running the placemats image, and it's configured to pull the latest, default
image (:latest). Check out
the docs for more details on
updating or changing the image the container is running.
Once you build, tag and push the local image to google cloud repository, you will need to 'bounce' (restart) all the instances in the instance group 'ce-placemats-2'.