Developer Resources

General Notes


We use precommit commits for formatting. Set it up locally with

pre-commit install

Flake Settings

Flake is used to statically verify code syntax. If the developer is running flake from the command line, they should ignore the following checks in order to emulate the same checks as the CI machine.




module level import not at top of file


line too long (82 characters) or max-line = 100


do not assign a lambda expression, use a def


line break occurred before a binary operator


line break occurred after a binary operator

To run flake locally call:

tox -e flake8

Python Type Hints

Python type hints are beginning to be added to the SEED codebase. The benefits are eliminating some accidental typing mistakes to prevent bugs as well as a better IDE experience.


SEED does not require exhaustive type annotations, but it is recommended you add them if you create any new functions or refactor any existing code where it might be beneficial (e.g. types that appear ambiguous or that the IDE can’t determine) and not require a ton of additional effort.

When applicable, we recommend you use built-in collection types such as list, dict or tuple instead of the capitalized types from the typing module. You can also use TypedDict and NotRequired from the typing_extensions package to specify the types of required/optional keys of dictionaries.

Common gotchas:

  • If trying to annotate a class method with the class itself, import from __future__ import annotations

  • If you’re getting warnings about runtime errors due to a type name, make sure your IDE is set up to point to an environment with python 3.9

  • If you’re wasting time trying to please the type checker, feel free to throw # type: ignore on the problematic line (or at the top of the file to ignore all issues for that file)

Type Checking

CI currently runs static type checking on the codebase using mypy. For your own IDE, we recommend the following extensions:

  • VSCode: Pylance (uses Microsoft’s Pyright type checking)

To run the same typechecking applied in CI (i.e., using mypy) you can run the following

tox -e mypy

Django Notes

Adding New Fields to Database

Adding new fields to SEED can be complicated since SEED has a mix of typed fields (database fields) and extra data fields. Follow the steps below to add new fields to the SEED database:

  1. Add the field to the PropertyState or the TaxLotState model. Adding fields to the Property or TaxLot models is more complicated and not documented yet.

  2. Add field to list in the following locations:

  • models/ Column.DATABASE_COLUMNS

  • TaxLotState.coparent or PropertyState.coparent: SQL query and keep_fields

  1. Run ./ makemigrations

  2. Add in a Python script in the new migration to add in the new column into every organizations list of columns. Note that the new_db_fields will be the same as the data in the Column.DATABASE_COLUMNS that were added.

    def forwards(apps, schema_editor):
        Column = apps.get_model("seed", "Column")
        Organization = apps.get_model("orgs", "Organization")
        new_db_fields = [
                'column_name': 'geocoding_confidence',
                'table_name': 'PropertyState',
                'display_name': 'Geocoding Confidence',
                'column_description': 'Geocoding Confidence',
                'data_type': 'number',
            }, {
                'column_name': 'geocoding_confidence',
                'table_name': 'TaxLotState',
                'display_name': 'Geocoding Confidence',
                'column_description': 'Geocoding Confidence',
                'data_type': 'number',
        # Go through all the organizations
        for org in Organization.objects.all():
            for new_db_field in new_db_fields:
                columns = Column.objects.filter(
                if not columns.count():
                    new_db_field['organization_id'] =
                elif columns.count() == 1:
                    # If the column exists, then update the display_name and data_type if empty
                    c = columns.first()
                    if c.display_name is None or c.display_name == '':
                        c.display_name = new_db_field['display_name']
                    if c.data_type is None or c.data_type == '' or c.data_type == 'None':
                        c.data_type = new_db_field['data_type']
                            for col in columns:
                    # If the column exists, then update the column_description if empty
                    if c.column_description is None or c.column_description == '':
                        c.column_description = new_db_field['column_description']
                    print("  More than one column returned")
    class Migration(migrations.Migration):
        dependencies = [
            ('seed', '0090_auto_20180425_1154'),
        operations = [
            ... existing db migrations ...,
  3. Run migrations ./ migrate

  4. Run unit tests, fix failures. Below is a list of files that need to be fixed (this is not an exhaustive list)




  1. (Optional) Update example files to include new fields

  2. Test import workflow with mapping to new fields


Toggle maintenance mode to display a maintenance page and prevent access to all site resources including API endpoints:

docker exec seed_web ./docker/ on
docker exec seed_web ./docker/ off

AngularJS Integration Notes

Template Tags

Angular and Django both use {{ and }} as variable delimiters, and thus the AngularJS variable delimiters are renamed {$ and $}.

window.BE.apps.seed = angular.module('BE.seed', ['$interpolateProvider', ($interpolateProvider) => {

Django CSRF Token and AJAX Requests

For ease of making angular $http requests, we automatically add the CSRF token to all $http requests as recommended by$http, $cookies) => {
  $http.defaults.headers.common['X-CSRFToken'] = $cookies['csrftoken'];

Routes and Partials or Views

Routes in static/seed/js/seed.js (the normal angularjs app.js)

SEED_app.config(['stateHelperProvider', '$urlRouterProvider', '$locationProvider', (stateHelperProvider, $urlRouterProvider, $locationProvider) => {
      name: 'home',
      url: '/',
      templateUrl: static_url + 'seed/partials/home.html'
      name: 'profile',
      url: '/profile',
      templateUrl: static_url + 'seed/partials/profile.html',
      controller: 'profile_controller',
      resolve: {
        auth_payload: ['auth_service', '$q', 'user_service', function (auth_service, $q, user_service) {
          var organization_id = user_service.get_organization().id;
          return auth_service.is_authorized(organization_id, ['requires_superuser']);
        user_profile_payload: ['user_service', function (user_service) {
          return user_service.get_user_profile();

HTML partials in static/seed/partials/


Information about error logging can be found here -

Below is a standard set of error messages from Django.

A logger is configured to have a log level. This log level describes the severity of the messages that the logger will handle. Python defines the following log levels:

DEBUG: Low level system information for debugging purposes
INFO: General system information
WARNING: Information describing a minor problem that has occurred.
ERROR: Information describing a major problem that has occurred.
CRITICAL: Information describing a critical problem that has occurred.

Each message that is written to the logger is a Log Record. The log record is stored in the web server & Celery

BEDES Compliance and Managing Columns

Columns that do not represent hardcoded fields in the application are represented using a Django database model defined in the seed.models module. The goal of adding new columns to the database is to create seed.models.Column records in the database for each column to import. Currently, the list of Columns is dynamically populated by importing data.

There are default mappings for ESPM are located here:

Resetting the Database

This is a brief description of how to drop and re-create the database for the seed application.

The first two commands below are commands distributed with the Postgres database, and are not part of the SEED application. The third command below will create the required database tables for SEED and setup initial data that the application expects (e.g. initial columns for BEDES). The last command below (spanning multiple lines) will create a new superuser and organization that you can use to login to the application, and from there create any other users or organizations that you require.

Below are the commands for resetting the database and creating a new user:

createuser -U seed seeduser

psql -d postgres -U seeduser -c 'DROP DATABASE seed;'
psql -d postgres -U seeduser -c 'CREATE DATABASE seed;'
psql -d seed -U seeduser -c 'CREATE EXTENSION IF NOT EXISTS postgis;'
psql -d seed -U seeduser -c 'CREATE EXTENSION IF NOT EXISTS timescaledb;'

./ migrate
./ create_default_user \
    --username=[email protected] \
    --password=password \

Restoring a Database Dump

psql -d postgres -U seeduser -c 'DROP DATABASE seed;'
psql -d postgres -U seeduser -c 'CREATE DATABASE seed;'
psql -d seed -U seeduser -c 'CREATE EXTENSION IF NOT EXISTS postgis;'
psql -d seed -U seeduser -c 'CREATE EXTENSION IF NOT EXISTS timescaledb;'
psql -d seed -U seeduser -c 'SELECT timescaledb_pre_restore();'

# restore a previous database dump (must be pg_restore 12+)
pg_restore -d seed -U seeduser /backups/prod-backups/prod_20191203_000002.dump
# if any errors appear during the pg_restore process check that the `installed_version` of the timescaledb extension where the database was dumped matches the extension version where it's being restored
# `SELECT default_version, installed_version FROM pg_available_extensions WHERE name = 'timescaledb';`

psql -d seed -U seeduser -c 'SELECT timescaledb_post_restore();'

./ migrate

# if needed add a user to the database
./ create_default_user \
    --username=[email protected] \
    --password=password \

If restoring a production backup to a different deployment update the site settings for password reset emails, and disable celerybeat Salesforce updates/emails:

./ shell

from django.contrib.sites.models import Site
site = Site.objects.first()
site.domain = '' = 'SEED Dev1'

from seed.models import Organization

from django_celery_beat.models import PeriodicTask, PeriodicTasks
PeriodicTask.objects.filter(enabled=True, name__startswith='salesforce_sync_org-').update(enabled=False)

Migrating the Database

Migrations are handles through Django; however, various versions have customs actions for the migrations. See the migrations page for more information based on the version of SEED.


JS tests can be run with Jasmine at the url /angular_js_tests/.

Python unit tests are run with

python test --settings=config.settings.test
Note on geocode-related testing:

Most of these tests use and cassettes to capture and reuse recordings of HTTP requests and responses. Given that, unless you want to make changes and/or refresh the cassettes/recordings, there isn’t anything needed to run the geocode tests.

In the case that the geocoding logic/code is changed or you’d like to the verify the MapQuest API is still working as expected, you’ll need to run the tests with a small change. Namely, you’ll want to provide the tests with an API key via an environment variable called “TESTING_MAPQUEST_API_KEY” or within your file with that same variable name.

In order to refresh the actual cassettes, you’ll just need to delete or move the old ones which can be found at “.seed/tests/data/vcr_cassettes”. The API key should be hidden within the cassettes, so these new cassettes can and should be pushed to GitHub.

Run coverage using

coverage run test --settings=config.settings.test
coverage report --fail-under=83

Python compliance uses PEP8 with flake8

# or
tox -e flake8

JS Compliance uses ESLint

npm run lint
npm run lint:fix

Building Documentation

Older versions of the source code documentation are (still) on readthedocs; however, newer versions are built and pushed to the seed-website repository manually. To build the documentation follow the script below:

cd docs
rm -rf htmlout
sphinx-build -b html source htmlout

For releasing, copy the htmlout directory into the seed-platform’s website repository under docs/code_documentation/<new_version>. Make sure to add the new documentation to the table in the docs/

Contribution Instructions / Best Practices

If this is the first time contributing and you are outside of the DOE National Lab system, then you will need to review and fill out the contribution agreement which is found in SEED’s Contribution Agreement in the GitHub repository

The desired workflow for development and submitting changes is the following:

  1. Fork the repository on GitHub if you do not have access to the repository, otherwise, work within the repository.

  2. Ensure there is a ticket/issue created for the work you are doing. Verify that the ticket is assigned to you and that it is part of the latest project board on the GitHub site (

  3. Move the ticket/issue to ‘In Progress’ in the GitHub project tracker when you begin work

  4. Create a branch off of develop (unless it is a hotfix, then branch of the appropriate tag). The recommended naming convention is <issue_id>-short-descriptive-name.

  5. Make changes and write a test for the code added.

  6. Make sure tests pass locally. Most branches created and pushed to GitHub will be tested automatically.

  7. Upon completion of the work, create a pull request (PR) against the develop branch (or hotfix branch if applicable). In the PR description fill out the requested information and include the issue number (e.g., #1234).

  8. Assign one label to the PR (not the ticket/issue) in order to auto-populate change logs (e.g., Bug, Feature, Maintenance, Performance, DoNotPublish) This is required and CI will fail if not present.
    • Bug (these will appear as “Bug Fixes” in the change log)

    • Feature (features will appear as “New Features” item in the change log)

    • Enhancement (these will appear as “Improvements” in the change log)

    • Maintenance (these will appear under “Maintenance” in the change log)

    • Performance (these will appear under “Maintenance” in the change log)

    • Documentation (these will appear under “Maintenance” in the change log)

    • Do not publish (these will no appear in the change log)

  9. Ensure all tests pass.

  10. Assign a reviewer to the PR.

  11. If the reviewer requests changes, then addresses changes and re-assign the reviewer as needed.

  12. Once approved, merge the PR!

  13. Move the related ticket(s)/issue(s) to the ‘Ready to Deploy’ column in the GitHub project tracker.

Release Instructions

To make a release do the following:

  1. Create a branch from develop to prepare the updates (e.g., 2.21.0-release-prep).

  2. Update the root package.json file with the release version number, and then run npm install. Always use MAJOR.MINOR.RELEASE.

  3. Update the docs/sources/migrations.rst file with any required actions.

  4. Commit the changes and push the release prep branch to GitHub, then go to the Releases page to draft a new release which will generate the changelog.

  5. Copy the GitHub changelog results into Cleanup the formatting and items as needed (make sure the spelling is correct, starts with a capital letter, if any PRs were missing the Do not publish label, etc.) and push the changelog update.

  6. Make sure that any new UI needing localization has been tagged for translation, and that any new translation keys exist in the project. (see translation documentation).

  7. Create PR for release preparation and merge after tests/reviews pass.

  8. Create a new Release using the develop branch and new release number as the tag ( Include list of changes since previous release (e.g., the additions to

  9. Locally, merge the develop branch into the main branch and push.

  10. Verify that the Docker versions are built and pushed to Docker Hub (

  11. Publish the new documentation in the seed-platform website repository (see instructions above under Building Documentation).