This is a web map that helps explore Indigenous language data. This README file includes new materials added in Milestones 3 and 2 of this project. See Milestone 1 deliverables here.
Clone the project.
git clone https://github.com/First-Peoples-Cultural-Council/fplm.git
Install Docker and docker-compose.
Copy the local settings template. They're all in one file, dc.dev.yml. docker-compose.override.yml is gitignored so you can store secrets there if needed.
cp dc.dev.yml docker-compose.override.yml
Spin up the project.
docker-compose up
Your Django app is served at http://localhost/api
Your Vue app is served at http://localhost
. The front-end won't work properly unless you have a realistic dataset. In this project, the database is quite small, we suggest using a production snapshot for development, because this gives better dev/prod parity for easier development. The other option is to populate tables using a script (an example is provided for migrating out of Druapl) or create your data manually in the Django admin.
Acquire a database dump. If the file is db.sql
in your repo root, do:
./docs/restore-pg
To get all languages:
curl http://maps.fpcc.ca/api/language/
Details or a specific language:
curl http://maps.fpcc.ca/api/language/18/
To get all communities:
curl http://maps.fpcc.ca/api/community/
Details or a specific community:
curl http://maps.fpcc.ca/api/community/18/
API Documentation is available at localhost/api/docs
Three endpoints are available to update directly, via API: /api/language/
, /api/community
, and api/stats
.
First, you should authenticate your API client as an FPCC admin user. For example using curl
:
curl --request POST --header "Content-Type: application/json" \
--data '{"username": "admin", "password": "********"}' http://maps.fpcc.ca/api-token-auth/
This will return a token, such as {"token":"cfc2b213a4adfbae02332fbbfb45ec09e56413a4"}
Then, you can call the API using the returned token in your authorization header. For example, first load your data using the public API to ensure you're changing the right objects, as in the above section and update some field using a PATCH request. This request will not clear other unspecified fields so it's useful for making corrections.
curl --request PATCH --header "Content-Type: application/json" \
--header "Authorization: Token cfc2b213a4adfbae02332fbbfb45ec09e56413a4" \
--data '{"regions": "Kootenays"}' http://maps.fpcc.ca/api/language/18/
To create a new object, use POST, for example a new Community:
curl --request POST --header "Content-Type: application/json" --header "Authorization: Token cfc2b213a4adfbae02332fbbfb45ec09e56413a4" \
--data '{"name":"Heiltsuk Nation New","champion_ids": [22], "language_ids":[27],"sleeping":false,"other_names":"Heiltsuk,Bella Bella,Heiltsuk-Oweekala","regions":"","audio_file":null, "english_name":"","other_names":"Heiltsuk Band","internet_speed":"","population":0,"point":{"type":"Point","coordinates":[-128.145551,52.160363]},"email":"admin@example.net","website":"http://www.bellabella.net","phone":"","alt_phone":"(250) 999-9999","fax":"(250) 999-9999"}' http://maps.fpcc.ca/api/community/
To add some stats:
curl --request POST --header "Content-Type: application/json" --header "Authorization: Token cfc2b213a4adfbae02332fbbfb45ec09e56413a4" --data '{ "fluent_speakers": 2, "semi_speakers": 3, "active_learners": 4, "language": 18, "community":255}' http://maps.fpcc.ca/api/stats/
To create a new Recording, use POST:
curl --header "Authorization: Token a213aeb9e9fc723511c37096e69f72822391aae4" --request POST -sS http://maps.fpcc.ca/api/recording/ -H 'Content-Type: multipart/form-data;' -F 'speaker=speaker name' -F 'date_recorded=2010-10-10' -F 'recorder=recorder name' -F 'audio_file=@/home/denis/demonyms.txt'
To add an existing Recording as a Language audio, use PATCH:
curl --header "Authorization: Token a213aeb9e9fc723511c37096e69f72822391aae4" --request PATCH -sS http://maps.fpcc.ca/api/language/26/add_language_audio/ --header "Content-Type: application/json" --data '{ "recording_id": 41 }'
To add an existing Recording as a Language greeting audio, use PATCH:
curl --header "Authorization: Token a213aeb9e9fc723511c37096e69f72822391aae4" --request PATCH -sS http://maps.fpcc.ca/api/language/26/add_greeting_audio/ --header "Content-Type: application/json" --data '{ "recording_id": 41 }'
To add an existing Recording as a Community audio, use PATCH:
curl --header "Authorization: Token a213aeb9e9fc723511c37096e69f72822391aae4" --request PATCH -sS http://maps.fpcc.ca/api/community/206/add_audio/ --header "Content-Type: application/json" --data '{ "recording_id": 41 }'
There is an important difference between the objects you can write and read from APIs. Related object collections are referenced as a list of IDs, with the _ids
suffix. So, to set the language of a community, you would do:
curl --request PATCH --header "Content-Type: application/json" \
--header "Authorization: Token cfc2b213a4adfbae02332fbbfb45ec09e56413a4" \
--data '{"language_ids": [18]}' http://maps.fpcc.ca/api/community/255/
Even though what is returned includes the entire language object inline, not just its ID:
{"id":253,"name":"Halfway River First Nations","languages":[{"name":"Dakelh (ᑕᗸᒡ)","id":18,"color":"RGB(0, 208, 104)","bbox"... }]}
Lastly, to upload an audio file to a language or other object, make a separate PATCH request, not using JSON, but just the default raw for encoding:
curl --header "Authorization: Token cfc2b213a4adfbae02332fbbfb45ec09e56413a4" --request PATCH -sS http://localhost/api/language/18/ -F 'audio_file=@./test.mp3'
The API writes objects "atomically", meaning only one database row can be edited or added per request. This is to help make the API simple and predicable (simple consistent CRUD for each table), as writing inline objects (while convenient) can lead to nontrivial edge cases. (For example, we need conventions on whether to assume anything not included in a PATCH is to be deleted from the set, modified if it includes updates, and should those modifications follow PATCH conventions as well...). For a small single-purpose writable API that wasn't part of our project focus, the atomic method is predictable and simple, allowing our focus to be on other scope.
To upload new grants, use the load_grants
command. FPCC will provide a .xlsx file containing a list of grants. The file should have the following columns in the first sheet: Grant, Language, Year, Recipient, Community/Affiliation, Title, Project Brief, Amount, Address, City, Province, Postal Code
docker-compose exec web python manage.py load_grants --file_name=/link/to/your/file
You may also provide a --test flag in order to see the grants being imported and if there are any errors.
docker-compose exec web python manage.py load_grants --file_name=/link/to/your/file --test=1
After successfully uploading the grants, we will use the migrate_manytomany_fields
command in order to link these grants to their Language/Community/Artist/Public Art.
docker-compose exec web python manage.py migrate_manytomany_fields
To work on a feature locally, configure your editor to use the black
code style for Python, and the prettier
style for Javascript, HTML and CSS. Use the local .pretterrc
file.
If you ever have coding style problems, you can fix them by running:
docker-compose exec frontend yarn lint --fix
These are the Vscode settings for automatic linting
"eslint.validate": [
{
"language": "vue",
"autoFix": true
},
{
"language": "javascript",
"autoFix": true
},
{
"language": "javascriptreact",
"autoFix": true
}
],
"eslint.autoFixOnSave": true,
"editor.formatOnSave": false,
"vetur.validation.template": false,
"editor.fontSize": 16,
"terminal.integrated.scrollback": 50000
We use pylint to detect linting errors in Python. Disclaimer: pylint is only used to detect errors an does not automatically fix them. Linting fixes have to be manually applied by the developer.
Check linting for the entire backend project
docker-compose exec web sh pylint.sh
Check linting for an entire folder
docker-compose exec web sh pylint.sh <folder_name>
Check linting for a specific file
docker-compose exec web sh pylint.sh <folder_name>/<file_name>/
Open one of the models.py
, and add your field according to the docs, ie) (new line marked with +
)
class Language(BaseModel)
- flavour=CharField(default='Strawberry', max_length=31)
other_names = models.TextField(default="", blank=True)
fv_archive_link = models.URLField(max_length=255, blank=True, default="")
Then create, and apply the migration. Make sure you add the generated migration to GIT.
docker-compose exec web python manage.py makemigrations
docker-compose exec web python manage.py migrate
git add .
If you want this field to be editable in the admin, this will happen by default. Depending on the previous rules, you may need to edit (admin.py)[https://docs.djangoproject.com/en/2.2/ref/contrib/admin/]
Done via GitHub Actions
.github/workflows/cd-prod.yml
- The master
branch is deployed by GitHub Actions to production, maps.fpcc.ca
by default.
Only the https://github.com/First-Peoples-Cultural-Council/maps
has GH Action Secrets for this deployment set up, as follows.
PROD_ENVS= # populate the local.env.template file provided.
KNOWN_HOSTS= # `ssh-keyscan <production server>` output
PROD_DEPLOY_KEY= # private key of producion server
This project was originally ported from a Drupal database, and we have a somewhat generic way of getting things out of Drupal the first time. Doing this requires populating the old database secrets in your docker-compose.override.yml
docker-compose exec web python manage.py bootstrap
to get languages.
docker-compose exec web python manage.py get_sleeping
to import an old KML source for languageion region geometry (included in repo).
docker-compose exec web python manage.py load_arts
to get arts.
To import categories from the csv file found at /web/fixtures/categories.csv run the following command:
docker-compose exec web python manage.py get_categories
The docker container is by default on sleep. Need to comment out command: sleep 1000000
on docker-compose.override.yml
then restart the container.
The test container is dependant on the frontend and the web container, and make sure these are running
docker-compose up test
For backend testing, we are using Django and Django Rest Framework's built-in testing modules. The test files are either named tests.py
or tests_<name>.py
.
Examples:
Running all tests:
docker-compose exec web sh test.sh
Testing a specific app
docker-compose exec web sh test.sh language
Testing a specific file
docker-compose exec web sh test.sh language.tests.tests_language
Testing a specific class with multiple tests
docker-compose exec web sh test.sh language.tests.tests_language.LanguageAPITests
Testing a specific test case
docker-compose exec web sh test.sh language.tests.tests_language.LanguageAPITests.test_language_detail_route_exists
Coverage test (specifying app/file/class/test also applies)
docker-compose exec web sh test-cov.sh
Running a coverage test will create a coverage.xml
file which indicates which lines were executed and which lines were not. This will also generate a report in the terminal which shows a percentage of the total coverage, and the coverage per file, based on the tests executed.
For more information about tests, run docker-compose exec web python manage.py help test
The system sends users notifications weekly, including:
- new contributions (places and media) in their profile's selected language/community.
- when other users "favourite" their contributions.
- requests to approve/verify contributions (only language admins receive this)
The user is only notified of each event once, as determined by their last_notified
attribute (see web/users/models.py).
To send a test notification email:
docker-compose exec web python manage.py test_notifications --email <email of user> --days <number of days>
Specifying a number days (integer) will always force-send updates the specified number of days of updates, regardless of whether those updates have already been sent.
- Fully Dockerized, and configured with docker-compose.
- Uses PostgreSQL and PostGIS.
- API-Driven Django. We don't use Django's templates for anything.
- Uses Nuxt.js for SEO-friendly modern templates.
- Proxies all ports through port 80, the default, including websockets, so there's no need to worry about the port of anything when developing.
- Authentication - Cognito JWT
- APIs - Django Rest Framework
- GIS - GeoJSON
- Linting - Pylint
- Testing Coverage - Coverage
- API documentation - DRF YASG (Swagger)