Update: Similar setup but for CircleCI 2.0 is HERE
Deploying to CircleCI as easy as creating yaml file (sometimes you even do not need it as Circle will automatically recognize structure and behavior). But usually it is better to setup at least basic configuration. The problem begins with more complicated functionality and CI documentation is minimal, or to say ‘very hacker friendly’.
My problem was how to use build-in PostgreSQL database with Python application. This is quick tip how I solved it.
I created simple Python app for this post purpose. All files are inside my repository, which I connected to Github with version 1.0 of CircleCI.
CircleCI settings are all default except BUILD SETTINGS: Environment Variables, where I added variable
Name: DB_NAME
Value: circleci
Create circle.yml file, and put inside root of repository
machine:
timezone:
Europe/Warsaw
python:
version: 3.5.2
environment:
DATABASE_URL: postgres://ubuntu:@127.0.0.1:5432/circle_test
dependencies:
pre:
- pip install pytest
- pip install psycopg2
database:
override:
- psql -U ubuntu -d circle_test -a -f schema.sql
test:
override:
- py.test --junitxml=$CIRCLE_TEST_REPORTS/output.xml
The most time it took me to find correct configuration how to override database, and it turned out to be very simple.
- psql -U ubuntu -d circle_test -a -f schema.sql
This line deciphered using psql-docs
- -U ubuntu: username
- -d circleci: database name
- -a echo all
- -f schema.sql: filename
If you look inside my repository at the test_database.py file. You will find line which says:
DATABASE = os.environ["DB_NAME"]
Here I am using that environmental variable, added in the beginning.
That is it. Now push to Github, and enjoy having automatic testing. If it still does not work, try to fork my repo and play with it.
Links: