pepysdiary

Code for www.pepysdiary.com.

Pushing to main will run the commit through this GitHub Action to run tests. If it passes, it will be deployed automatically to the website.

When changing the python version, it will need to be changed in:

.github/workflows/test.yml
.pre-commit-config.yaml
.python-version (for pyenv)
pyproject.toml (ruff's target-version)
docker/web/Dockerfile

For local development we use Docker. The live site is on an Ubuntu 22 VPS.

Local development setup

1. Create a .env file

Copy .env.dist to .env and alter any necessary settings.

2. Set up a local domain name

Open your /etc/hosts file in a terminal window by doing:

$ sudo vim /etc/hosts

Enter your computer's password. Then add this line somewhere in the file and save:

127.0.0.1 www.pepysdiary.test

3. Build the Docker containers

Download, install and run Docker Desktop.

In same directory as this README, build the containers:

$ docker compose build

Then start up the web, assets and database containers:

$ docker compose up

There are four containers:

pepysdiary_web: the webserver
pepysdiary_db: the postgres server
pepysdiary_assets: the front-end assets builder
pepysdiary_redis: the redis server (for optional caching)

All the repository's code is mirrored in the web and assets containers in the /code/ directory.

4. Set up the database

Once that's running, showing the logs, open another terminal window/tab.

There are two ways we can populate the database. First we'll create an empty one, and second we'll populate it with a dump of data from the live site.

4a. An empty database

The build step will create the database and run the initial Django migrations.

Then create a superuser:

$ ./run manage createsuperuser

(See below for more info on the ./run script.)

4b. Use a dump from the live site

Log into postgres and drop the current (empty) database:

$ ./run psql -d postgres
# DROP DATABASE pepys WITH (FORCE);
# CREATE DATABASE pepys;
# GRANT ALL PRIVILEGES ON DATABASE pepys TO pepys;
# \q

On the VPS, create a backup file of the live site's database:

$ pg_dump -h localhost -U username -d dbname -Fc -b -f ~/dump.sql

Then scp it to your local machine:

$ scp username@your.vps.domain.com:/home/username/dump.sql .

Then copy the dump into Docker and load it into the database:

$ docker cp dump.sql pepys_db:/tmp/
$ docker exec -i pepys_db pg_restore -h localhost -U pepysdiary -d pepysdiary -j 2 /tmp/dump.sql

5. Vist and set up the site

Then go to http://www.pepysdiary.test:8000 and you should see the site.

Log in to the Django Admin, go to the "Sites" section and change the one Site's Domain Name to www.pepysdiary.test:8000 and the Display Name to "The Diary of Samuel Pepys", if it's not already.

Ongoing work

Docker

Whenever you come back to start work you need to start the containers up again by doing this from the project directory:

$ docker compose up

When you want to stop the server, then this from the same directory:

$ docker compose down

You can check if anything's running by doing this, which will list any Docker processes:

$ docker ps

See details on the ./run script below for running things inside the containers.

Python dependencies with virtualenv and pip-tools

Adding and removing python depenencies is most easily done with a virtual environment on your host machine. This also means you can use that environment easily in VS Code.

Set up and activate a virtual environment on your host machine using virtualenv:

$ virtualenv --prompt . venv
$ source venv/bin/activate

We use pip-tools to generate requirements.txt from requirements.in, and install the dependencies. Install the current dependencies into the activated virtual environment:

(venv) $ python -m pip install -r requirements.txt

To add a new depenency, add it to requirements.in and then regenerate requirements.txt:

(venv) $ pip-compile --upgrade --quiet --generate-hashes

And do the pip install step again to install.

To remove a dependency, delete it from requirements.in, run that same pip-compile command, and then:

(venv) $ python -m pip uninstall <module-name>

To update the python dependencies in the Docker container, this should work:

$ ./run pipsync

But you might have to do docker compose build instead?

pre-commit

Install pre-commit to run .pre-commit-config.yml automatically when git commit is done.

Front-end assets

Gulp is used to build the final CSS and JS file(s), and watches for changes in the pepysdiary_assets container. Node packages are installed and upgraded using yarn (see ./run below).

The ./run script

The ./run script makes it easier to run things that are within the Docker containers. This will list the commands available, which are outlined below:

$ ./run

`./run cmd`

Run any command in the web container. e.g.

$ ./run cmd ls -al

`./run sh`

Starts a Shell session in the web container.

`./run manage`

Run the Django manage.py file with any of the usual commands, within the pipenv virtual environment. e.g.

$ ./run manage makemigrations

The development environment has django-extensions installed so you can use its shell_plus and other commands. e.g.:

$ ./run manage shell_plus
$ ./run manage show_urls

`./run tests`

Runs all the Django tests. If it complains you might need to do ./run manage collecstatic first.

Run a folder, file, or class of tests, or a single test, something like this:

$ ./run tests tests.core
$ ./run tests tests.core.test_views
$ ./run tests tests.core.test_views.HomeViewTestCase
$ ./run tests tests.core.test_views.HomeViewTestCase.test_response_200

`./run coverage`

Run all the tests with coverage. The HTML report files will be at htmlcov/index.html.

`./run ruff`

Run ruff check . over the code to check Python formatting:

$ ./run ruff
$ ./run ruff --fix

`./run psql`

Conects to PosgreSQL with psql. Add any required arguments on the end. Uses the hines database unless you specify another like:

$ ./run psql -d databasename

`./run pipsync`

Update the installed python depenencies depending on the contents of requirements.txt.

`./run yarn:outdated`

List any installed Node packages (used for building front end assets) that are outdated.

`./run yarn:upgrade`

Update any installed Node packages that are outdated.

Removing users and annotations from the local database

What I did to create a version of the database without personal information. After exporting the live database and importing into the local Docker database:

$ ./run psql
pepysdiary=# TRUNCATE django_comments, django_comment_flags, auth_user, annotations_annotation, membership_person;
pepysdiary=# UPDATE diary_entry SET comment_count=0;
pepysdiary=# UPDATE encyclopedia_topic SET comment_count=0;
pepysdiary=# UPDATE indepth_article SET comment_count=0;
pepysdiary=# UPDATE letters_letter SET comment_count=0;
pepysdiary=# UPDATE news_post SET comment_count=0;
pepysdiary=# \q
$ ./run manage createsuperuser

Then logged into Django Admin and changed the Site "Domain name" to www.pepysdiary.test:8000.

Then dumped that modified database:

$ docker exec -i pepys_db pg_dump pepysdiary -U pepysdiary -h localhost | gzip > pepys_dump.gz

VPS set-up

The complete set-up of an Ubuntu VPS is beyond the scope of this README. Requirements:

Local postgresql
Local redis (for caching)
pipx, virtualenv and pyenv
gunicorn
nginx
systemd
cron

1. Create a database

username$ sudo su - postgres
postgres$ createuser --interactive -P
postgres$ createdb --owner pepys pepys
postgres$ exit

2. Create a directory for the code

username$ sudo mkdir -p /webapps/pepys/
username$ sudo chown username:username /webapps/pepys/
username$ mkdir /webapps/pepys/logs/
username$ cd /webapps/pepys/
username$ git clone git@github.com:philgyford/pepysdiary.git code

3. ## Install python version, set up virtualenv, install python dependencies

username$ pyenv install --list  # All those available to install
username$ pyenv versions        # All those already installed and available
username$ pyenv install 3.10.8  # Whatever version we're using

Make the virtual environment and install pip-tools:

username$ cd /webapps/pepys/code
username$ virtualenv --prompt pepys venv -p $(pyenv which python)
username$ source venv/bin/activate
(pepys) username$ python -m pip install pip-tools

Install dependencies from requirements.txt:

(pepys) username$ pip-sync

4. Create `.env` file

(pepys) username$ cp .env.dist .env

Then fill it out as required.

5. Set up database

Either do ./manage.py migrate and ./manage.py createsuperuser to create a new database, or import an existing database dumbp.

6. Set up gunicorn with systemd

Symlink the files in this repo to correct location for systemd:

username$ sudo ln -s /webapps/pepys/code/conf/systemd_gunicorn.socket /etc/systemd/system/gunicorn_pepys.socket
username$ sudo ln -s /webapps/pepys/code/conf/systemd_gunicorn.service /etc/systemd/system/gunicorn_pepys.service

Start the socket:

username$ sudo systemctl start gunicorn_pepys.socket
username$ sudo systemctl enable gunicorn_pepys.socket

Check the socket status:

username$ sudo systemctl status gunicorn_pepys.socket

Start the service:

username$ sudo systemctl start gunicorn_pepys

5. Set up nginx

Symlink the file in this repo to correct location:

username$ sudo ln -s /webapps/pepys/code/conf/nginx.conf /etc/nginx/sites-available/pepys

Enable this site:

username$ sudo ln -s /etc/nginx/sites-available/pepys /etc/nginx/sites-enabled/pepys

Remove the default site if it's not already:

username$ sudo rm /etc/nginx/sites-enabled/default

Check configuration before (re)starting nginx:

username$ sudo nginx -t

Start nginx:

 username$ sudo service nginx start

6. Set up cron for management commands

username$ crontab -e

Add this:

# pepys - fetch some content from Wikipedia
10 3,4,5,6 * * * /webapps/pepys/code/venv/bin/python /webapps/pepys/code/manage.py fetch_wikipedia --num=30 > /dev/null 2>&1

Other stuff

Allow service restarts without a password, so that GitHub Actions autodeploy works
Set up Let's Encrypt for the domain

Bootstrap

W're using Boostrap SASS to generate a custom set of Bootstrap CSS. Comment/uncomment lines in assets/sass/_bootstrap_custom.scss to change which parts of Bootstrap's CSS are included in the generated CSS file.

Bootstrap's JavaScript

We must manually download a custom version of Bootstrap's JavaScript file.

Instead of steps 2 and 3 you can hopefully upload the assets/js/vendor/bootstrap_config.json file.

Go to https://getbootstrap.com/docs/3.4/customize/
Toggle off all of the checkboxes.
Under the "jQuery plugins" section check the boxes next to these plugins:
- Linked to components
  - Alert dismissal
  - Dropdowns
  - Tooltips
  - Popovers
  - Togglable tabs
- Magic
  - Collapse
Scroll to the bottom of the page and click the "Compile and Download" button.
Copy the two .js files into assets/js/vendor/, replacing the existing files.

Wikipedia content

To fetch content for all Encyclopedia Topics which have matching Wikipedia pages, run this:

./manage.py fetch_wikipedia --all --verbosity=2

It might take some time. See encyclopedia/management/commands/fetch_wikipedia.py for more options.

Media files

Whether in local dev or Heroku, we need an S3 bucket to store Media files in (Static files are served using Whitenoise).

Go to the IAM service, Users, and 'Add User'.
Enter a name and check 'Programmatic access'.
'Attach existing policies directly', and select 'AmazonS3FullAccess'.
Create user.
Save the Access key and Secret key.
On the list of Users, click the user you just made and note the User ARN.
Go to the S3 service and 'Create Bucket'. Name it, select the region, and click through to create the bucket.
Click the bucket just created and then the 'Permissions' tab. Add this policy, replacing BUCKET-NAME and USER-ARN with yours:

{
  "Statement": [
    {
      "Sid": "PublicReadForGetBucketObjects",
      "Effect": "Allow",
      "Principal": {
        "AWS": "*"
      },
      "Action": ["s3:GetObject"],
      "Resource": ["arn:aws:s3:::BUCKET-NAME/*"]
    },
    {
      "Action": "s3:*",
      "Effect": "Allow",
      "Resource": ["arn:aws:s3:::BUCKET-NAME", "arn:aws:s3:::BUCKET-NAME/*"],
      "Principal": {
        "AWS": ["USER-ARN"]
      }
    }
  ]
}

Click on 'CORS configuration' and add this:

<CORSConfiguration>
<CORSRule>
    <AllowedOrigin>*</AllowedOrigin>
    <AllowedMethod>GET</AllowedMethod>
    <MaxAgeSeconds>3000</MaxAgeSeconds>
    <AllowedHeader>Authorization</AllowedHeader>
</CORSRule>
</CORSConfiguration>

Upload all the files to the bucket in the required location.
Update the server's environment variables for AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_STORAGE_BUCKET_NAME.

philgyford / pepysdiary

readme