As of January 23rd, 2018, the MNP main codebase is now open source, under Gnu Public License (GPL). Visit the live site at: https://www.mondaynightpinball.com
On a fresh machine, this is the minimum to allow the MNP web site to run.
This assumes Node.js is installed - which is beyond the scope of these installation steps.
If no data folders exist, setupMNPFolders.sh will create them and populate with a minimal sample 'season-0' with a single match.
git clone https://github.com/mondaynightpinball/main.git
source ./setupMNPFolders.sh
node http.js
To change the season, you can either edit the .env file (in the root project folder - created by setupMNPFolders.sh. Alternately, you can specify a different CURRENT_SEASON environment variable on the command line.
CURRENT_SEASON=season-14
node http.js
git clone https://github.com/mondaynightpinball/main.git
npm i
# Run from project home
mkdir data
mkdir data/ifpa
mkdir data/matches
mkdir data/players
mkdir data/sessions
mkdir data/stats
#starting with season 15 we are storing stats under the current season folder.
# As a backup and log of events, we save all non-auth posts to a directory.
mkdir data/posts
# All image data is currently just saved on the server.
# TODO: Save image data to a scalable storage service like AWS/S3
mkdir data/uploads
# You need season-# directories for any season you want to load.
# See model/seasons.js
mkdir data/season-8
The code needs a few files setup to get things going.
Many of the config files that were used for seasons 6, 7, and 8 are available, as well as all the match data, schedule, and rosters.
Visit the data archive
TODO: It would be awesome to have the code just point to a DATA_HOME .env var. Then people could clone the data archive and point the server code at it (until a better solution becomes available).
The venues.json is not a per-season database. It represents a realtime view of the venues. New venues are added via the web interface by the superuser.
data/venues.json has the form:
{
"ABC": {
"name": "ABC Venue",
"machines": [ array_of_machine_keys ],
"key": "ABC",
"address": "[Optional] 123 Fake St, Springfield, WA",
"neighborhood", "[Optional] Neighborhood"
},
...
}
data/machines.json defines all the machines that are available to
be used in a match. People can still enter machines that are not on
this list (much to our dismay), but this is the source for machine keys and names.
{
"MachineKey": {
"key": "MachineKey",
"name": "Machine Name"
},
...
}
TODO: Probably don't need to include the key in the machine object, and we could probably also just store a map of key -> name.
data/ifpa_num.csv is in the form of:
player_name,ifpa_num
...
Where the ifpa_num is the player's official IFPA number.
See IFPA Player Search
In each of the data/season-{n} directories, you need the following files:
matches.csv defines the schedule and should have the following columns:
week_num,date,away_key,home_key,venue_key
A special note that playoff weeks are denoted as week_num = 90 + playoff_week, so the first week of playoffs is week 91.
teams.csv defines teams and has columns:
team_key,venue_key,team_name
playerdb.csv is the master roster file and has columns:
player_lookup,player_name,team_key,role_on_team
group.csv contains a list of groups and the teams in each group
group_identifier, group_name, team_name, team_name, ...
player_lookup has most recently been player_name.trim().toLowerCase().
TODO: It would be much better to have unique keys for every player.
Currently, there may be an issue that if you don't have a gmail OAuth token, you might run into issues when you start the server.
TODO: Add more details/resources on how to setup the gmail token.
If you want to run a secure server, you will need to setup a certificate file.
.credentials/https.opts.mnp.json
TODO: Add more details/resources on how to setup the certificate.
Once you have all the season files above setup, you need to run the script that parses all the CSV files and generates a season.json (currently written to standard out, so I send that to a file on the command line).
node importers/import-season > data/season-{n}/season.json
The system doesn't filter which matches are visible, so it is usually advised to generate matches a few hours before they start.
This is one of the areas of the site that could be very much improved once we move to a proper database. That project exists and has a fair amount of test coverage. The goal is to integrate that backend code with this full stack project to replace the models.
Eventually stats would likely be real time, but for now we run a script to generate player stats.
node run util/compute-stats
To get things running, you can run one of the following commands:
# Start just HTTP (you will need to have sudo/root access)
node http.js
# Start just in HTTPS (you will need to have sudo/root
# as well as the necessary certificate config)
node https.js
# Start the service on both 80 and 443, with a redirect to 443
# This is what the website uses (after some service wrapping).
node app.js
If you are interested in how to wrap up the app to run as a service, we use forever and forever-service, both npm modules and not too hard to setup.
However, we don't feel like this is the best long term solution.
There are more solutions, and layers like nginx that can help clean up our solution.
We very much want developers to be able to help make the MNP site more reliable, secure, beautiful, and responsive. Feel free to submit PRs, and check out our issues.