Quepid

Quepid makes improving your app's search results a repeatable, reliable engineering process that the whole team can understand. It deals with three issues:

Our collaboration stinks Making holistic progress on search requires deep, cross-functional collaboration. Shooting emails or tracking search requirements in spreadsheets won't cut it.
Search testing is hard Search changes are cross-cutting: most changes will cause problems. Testing is difficult: you can't run hundreds of searches after every relevance change.
Iterations are slow Moving forward seems impossible. To avoid sliding backwards, progress is slow. Many simply give up on search, depriving users of the means to find critical information.

To learn more, please check out the Quepid website and the Quepid wiki.

If you are ready to dive right in, you can use the Hosted Quepid service right now or follow the installation steps to set up your own instance of Quepid.

Below is information related to developing the Quepid open source project, primarily for people interested in extending what Quepid can do!

Development Setup
Elasticsearch
Dev Errata
QA
- Seed Data
Data Map
App Structure
Operating Documentation
Credits

Development Setup

I. System Dependencies

Using Docker Compose

Provisioning from an already built machine takes approximately 3 - 4 minutes. Provisioning from scratch takes approximately 20 minutes.

1. Prerequisites

Make sure you have installed Docker. Go here https://www.docker.com/community-edition#/download for installation instructions. And the Docker app is launched.

To install using brew follow these steps:

brew cask install docker
brew cask install docker-toolbox

NOTE: you may get a warning about trusting Oracle on the first try. Open System Preferences > Security & Privacy, click the Allow Oracle button, and then try again to install docker-toolbox

2. Setup your environment

Run the local Ruby based setup script to setup your Docker images:

bin/setup_docker

Optionally you can seed the database with sample data (the output will print out credentials you can use to login as various sample users):

bin/docker r bin/rake db:seed:sample_users

If you want to create some cases that have 100's and 1000's of queries, then do:

bin/docker r bin/rake db:seed:large_cases

This is useful for stress testing Quepid! Especially the front end application!

3. Running the app

Now fire up Quepid locally at http://localhost:3000:

bin/docker server

It can take up to a minute for the server to respond as it compiles all the front end assets on the first call.

We've created a helper script to run and manage the app through docker that wraps around the docker-compose command. You will need Ruby installed. You can still use docker-compose directly, but for the basic stuff you can use the following:

Start the app: bin/docker server or bin/docker s
Connect to the app container with bash: bin/docker bash or bin/docker ba
Connect to the Rails console: bin/docker console or bin/docker c
Run any command: bin/docker run [COMMAND] or bin/docker r [COMMAND]
Run dev mode as daemon: bin/docker daemon or bin/docker q
Destroy the Docker env: bin/docker destroy or bin/docker d
Run front end unit tests: bin/docker r rails test:frontend
Run back end unit tests: bin/docker r rails test

II. Development Log

While running the app under foreman, you'll only see a request log, for more detailed logging run the following:

tail -f log/development.log

III. Run Tests

There are three types of tests that you can run:

Minitest

These tests run the tests from the Rails side (mainly API controllers, and models):

bin/docker r rails test

Run a single test file via:

bin/docker r rails test test/models/user_test.rb

Or even a single test in a test file by passing in the line number!

bin/docker r rails test test/models/user_test.rb:33

If you need to reset your test database setup then run:

bin/docker r bin/rake db:drop RAILS_ENV=test
bin/docker r bin/rake db:create RAILS_ENV=test

View the logs generated during testing set config.log_level = :debug in test.rb and then tail the log file via:

tail -f log/test.log

JS Lint

To check the JS syntax:

bin/docker r rails test:jshint

Karma

Runs tests for the Angular side. There are two modes for the karma tests:

Single run: bin/docker r rails karma:run
Continuous/watched run: bin/docker r bin/rake karma:start

Note: The karma tests require the assets to be precompiled, which adds a significant amount of time to the test run. If you are only making changes to the test/spec files, then it is recommended you run the tests in watch mode (bin/docker r bin/rake karma:start). The caveat is that any time you make a change to the app files, you will have to restart the process (or use the single run mode).

Rubocop

To check the Ruby syntax:

bin/docker r bundle exec rubocop

Rubocop can often autocorrect many of the lint issues it runs into via --auto-correct-all:

bin/docker r bundle exec rubocop --auto-correct-all

If there is a new "Cop" as they call their rules that we don't like, you can add it to the ./rubocop.yml file.

All Tests

If you want to run all of the tests in one go (before you commit and push for example), just run these two commands:

bin/docker r rails test
bin/docker r rails test:frontend

For some reason we can't run both with one command, though we should be able to!.

Performance Testing

If you want to create a LOT of queries for a user for testing, then run

bin/docker r bin/rake db:seed:large_cases

You will have two users, quepid+100sOfQueries@o19s.com and quepid+1000sOfQueries@o19s.com to test with.

IV. Debugging

Debugging Ruby

Debugging ruby usually depends on the situation, the simplest way is to print out the object to the STDOUT:

puts object         # Prints out the .to_s method of the object
puts object.inspect # Inspects the object and prints it out (includes the attributes)
pp object           # Pretty Prints the inspected object (like .inspect but better)

In the Rails application you can use the logger for the output:

Rails.logger object.inspect

If that's not enough and you want to run a debugger, the byebug gem is included for that. Add the line byebug in your Rails code wherever you want a breakpoint and then run the code and you will get an inline REPL. Even in unit tests!

Also, we have the derailed gem available which helps you understand memory issues.

bin/docker r bundle exec derailed bundle:mem

Debugging JS

While running the application, you can debug the javascript using your favorite tool, the way you've always done it.

The javascript files will be concatenated into one file, using the rails asset pipeline.

You can turn that off by toggling the following flag in config/environments/development.rb:

# config.assets.debug = true
config.assets.debug = false

config.assets.debug = true
# config.assets.debug = false

Because there are too many Angular JS files in this application, and in debug mode Rails will try to load every file separately, that slows down the application, and becomes really annoying in development mode to wait for the scripts to load. Which is why it is turned off by default.

PS: Don't forget to restart the server when you change the config.

Also please note that the files secure.js, application.js, and admin.js are used to load all the JavaScript and CSS dependencies via the Rails Asset pipeline. If you are debugging Bootstrap, then you will want individual files. So replace //= require sprockets with //= require bootstrap-sprockets.

Webpacker

To use webpacker, that will compile javascript code into packs and will load changes faster, you need to

bin/rails webpacker:install

Prior to that I had to install:

brew install mysql

Debugging Splainer and other NPM packages

docker-compose.override.yml.example can be copied to docker-compose.override.yml and use it to override environment variables or work with a local copy of the splainer-search JS library during development defined in docker-compose.yml. Example is included. Just update the path to splainer-search with your local checkout! https://docs.docker.com/compose/extends/

Convenience Scripts

This application has two ways of running scripts: rake & thor.

Rake is great for simple tasks that depend on the application environment, and default tasks that come by default with Rails.

Whereas Thor is a more powerful tool for writing scripts that take in args much more nicely than Rake.

Rake

To see what rake tasks are available run:

bin/docker r bin/rake -T

Note: the use of bin/rake makes sure that the version of rake that is running is the one locked to the app's Gemfile.lock (to avoid conflicts with other versions that might be installed on your system). This is equivalent of bundle exec rake.

Common rake tasks that you might use:

# db
bin/docker r bin/rake db:create
bin/docker r bin/rake db:drop
bin/docker r bin/rake db:migrate
bin/docker r bin/rake db:rollback
bin/docker r bin/rake db:schema:load
bin/docker r bin/rake db:seed
bin/docker r bin/rake db:setup

# show routes
bin/docker r bin/rails routes

# tests
bin/docker r rails test
bin/docker r rails test:frontend
bin/docker r bin/rake test:jshint

Thor

The see available tasks:

bin/docker r thor list

Examples include:

case
----
thor case:share CASEID TEAMID  # shares case with an team

ratings
-------
thor ratings:generate SOLRURL FILENAME  # generates random ratings into a .csv file
thor ratings:import CASEID FILENAME     # imports ratings to a case

user
----
thor user:create EMAIL USERNAME PASSWORD    # creates a new user
thor user:grant_administrator EMAIL         # grant administrator privileges to user
thor user:reset_password EMAIL NEWPASSWORD  # resets user's password

To see more details about any of the tasks, run bin/docker r thor help TASKNAME:

thor help user:create
Usage:
  thor user:create EMAIL USERNAME PASSWORD

Options:
  -a, [--administrator], [--no-administrator]  

Description:
  `user:create` creates a new user with the passed in email, name and password.

  EXAMPLES:

  $ thor user:create foo@example.com "Eric Pugh" mysuperstrongpassword

  With -a option, will mark the user as Administrator

  EXAMPLES:

  $ thor user:create -a admin@example.com Administrator mysuperstrongpassword

Elasticsearch

You will need to configure Elasticsearch to accept requests from the browser using CORS. To enable CORS, add the following to elasticsearch's config file. Usually, this file is located near the elasticsearch executable at config/elasticsearch.yml.

http.cors:
  enabled: true
  allow-origin: /https?:\/\/localhost(:[0-9]+)?/

See more details on the wiki at https://github.com/o19s/quepid/wiki/Troubleshooting-Elasticsearch-and-Quepid

Dev Errata

I'd like to use a new Node module, or update a existing one

Typically you would simply do:

bin/docker r yarn add foobar

bin/docker r yarn upgrade foobar

which will install/upgrade the Node module, and then save that dependency to package.json.

Then check in the updated package.json and yarn.lock files.

Use bin/docker r yarn outdated to see what packages you can update!!!!

I'd like to use a new Ruby Gem, or update a existing one

Typically you would simply do:

bin/docker r bundle add foobar

which will install the new Gem, and then save that dependency to Gemfile.

You can also upgrade a gem that doesn't have a specific version in Gemfile via:

bin/docker r bundle update foobar

You can remove a gem via:

bin/docker r bundle remove foobar --install

Then check in the updated Gemfile and Gemfile.lock files. For good measure run the bin/setup_docker.

To understand if you have gems that are out of date run:

bin/docker r bundle outdated --groups

I'd like to test SSL

There's a directory .ssl that contains they key and cert files used for SSL. This is a self signed generated certificate for use in development ONLY!

The key/cert were generated using the following command:

openssl req -new -newkey rsa:2048 -sha1 -days 365 -nodes -x509 -keyout .ssl/localhost.key -out .ssl/localhost.crt

PS: It is not necessary to do that again.

The docker-compose.yml file contains an nginx reverse proxy that uses these certificates. You can access Quepid at https://localhost or http://localhost. (Quepid will still be available over http on port 80.)

I'd like to test OpenID Auth

Add dev docs here!

Modifying the database

Here is an example of generating a migration:

bin/docker r bundle exec bin/rails g migration FixCuratorVariablesTriesForeignKeyName

Followed by bin/docker r bundle exec rake db:migrate

Updating RubyGems

Modify the file Gemfile and then run:

bin/docker r bundle install

You will see a updated Gemfile.lock, go ahead and check it and Gemfile into Git.

How does the Frontend work?

We use Angular 1 for the front end, and as part of that we use the angular-ui-bootstrap package for all our UI components. This package is tied to Bootstrap version 3. We import the Bootstrap 3 CSS directly via the file bootstrap.css.

For the various Admin pages, we actually are using Bootstrap 5! That is included via the package.json using NPM. See admin.js for the line //= require bootstrap/dist/js/bootstrap.bundle which is where we are including.

We currently use Rails Sprockets to compile everything, but do have dreams of moving the JavaScript over to Webpacker.

QA

There is a code deployment pipeline to the http://quepid-staging.herokuapp.com site that is run on successful commits to master.

If you have pending migrations you will need to run them via:

heroku run bin/rake db:migrate -a quepid-staging
heroku restart -a quepid-staging

Seed Data

The following accounts are created through the seeds. They all follow the following format:

email: quepid+[type]@o19s.com
password: password

where type is one of the following:

admin: An admin account
1case: A user with 1 case
2case: A user with 2 cases
solr: A user with a Solr case
es: A user with a ES case
realisticActivity: A user with a Solr case that has 10s of queries and 30 tries
100sOfQueries: A user with a Solr case that has 100s of queries (usually disabled)
1000sOfQueries: A user with a Solr case that has 1000s of queries (usually disabled)
oscOwner: A user who owns the team 'OSC'
oscMember: A user who is a member of the team 'OSC'
CustomScorer: A user who has a custom scorer
CustomScorerDefault: A user who has a custom scorer that is set as their default

Data Map

Check out the Data Mapping file for more info about the data structure of the app.

Rebuild the ERD via bin/docker r bundle exec rake erd:image

App Structure

Check out the App Structure file for more info on how Quepid is structured.

Operating Documentation

Check out the Operating Documentation file for more informations how Quepid can be operated and configured for your company.

Thank You's

Quepid would not be possible without the contributions from many individuals and organizations.

Specifically we would like to thank Erik Bugge and the folks at Kobler for funding the Only Rated feature released in Quepid 6.4.0.

Quepid wasn't always open source! Check out the credits for a list of contributors to the project.

If you would like to fund development of a new feature for Quepid do get in touch!

mcalavera81/quepid