docker.yaml

name: Docker

on:
  pull_request:
  push:
    branches:
      - master
jobs:
  tests:
    name: Docker tests
    runs-on: ubuntu-latest
    services:
      postgres:
        image: postgres:11.5
        env:
          POSTGRES_MULTIPLE_DATABASES: searchneu_dev
          POSTGRES_USER: postgres
        ports:
          - 5432:5432
      elasticsearch:
        image: docker.elastic.co/elasticsearch/elasticsearch:7.10.0
        env:
          discovery.type: single-node
        ports:
          - 9200:9200
    env:
      TWILIO_ACCOUNT_SID: AC_dummy_value_so_tests_dont_error_out
      TWILIO_AUTH_TOKEN: 123
      elasticURL: "http://localhost:9200"
    steps:
      - uses: actions/checkout@v2

      - name: Build the Docker container
        run: docker build --no-cache -t cca-test:latest .

      - name: Run the newly-built Docker container
        run: |-
          docker run -e TWILIO_ACCOUNT_SID=AC_dummy_value_so_tests_dont_error_out -e TWILIO_AUTH_TOKEN=123 \
            --network=host -p 4000:4000 -d --name test cca-test:latest

      - name: Wait until the GraphQL server is ready
        run: |-
          while (docker inspect -f '{{.State.Status}}' test | grep running) && ! ( docker logs test | grep "ready at http://localhost:4000/"); \
          do sleep 2; \
          done

      - name: Docker logs
        run: docker logs test

      - name: Check that the Docker container is still running
        run: docker inspect -f '{{.State.Status}}' test | grep running

      - name: Ensure the GQL server is ready
        run: docker logs test | grep "ready at http://localhost:4000/"

      - name: Check the GraphQL server healthcheck to test network
        run: curl localhost:4000/.well-known/apollo/server-health

      - name: Run a custom scrape (shouldn't take more than ~5 minutes)
        run: docker exec test sh -c "LOG_LEVEL=VERBOSE CUSTOM_SCRAPE=true TERMS_TO_SCRAPE=202140 yarn prod:scrape"

      #### If these start failing, it might not be our fault
      ## Since this test uses live data, there is a chance it could change
      # EG. NEU could remove the data for old semesters
      # In that case, feel free to update these tests to a new semester/prof/course
      - name: Check that CS3500 was populated in Postgres, and that Amit is its professor
        run: |-
          curl 'http://localhost:4000' \
            -X POST -H 'content-type: application/json' \
            --data '{"query":"query { search(termId: \"202140\", query:\"CS3500\") { nodes { ... on ClassOccurrence { name sections { profs } } } } }"}' \
            -v --silent --stderr - | grep "Amit Shesh"

      - name: Check that CS3500 was populated in Elasticsearch (ie. don't use the course code)
        run: |-
          curl 'http://localhost:4000' \
            -X POST -H 'content-type: application/json' \
            --data '{ "query": "query { search(termId: \"202140\", query:\"Object-Oriented Design\") { nodes { ... on ClassOccurrence { name sections { profs } } } } }" }' \
            -v --silent --stderr - | grep "Object-Oriented Design"

      # The employees need a few seconds to populate, otherwise, the below query fails
      - name: sleep for a few seconds to allow the Employees to populate
        run: sleep 10

      - name: Check that Amit Shesh is a professor
        run: |-
          curl 'http://localhost:4000' \
            -X POST -H 'content-type: application/json' \
            --data '{ "query": "query { search(termId: \"202140\", query:\"Amit Shesh\") { nodes { ... on Employee { name } } } }" }' \
            -v --silent --stderr - | grep "Amit Shesh"

      - name: Delete the CS3500 course
        run: |-
          psql -U postgres -h localhost -p 5432 -d searchneu_dev -c \
              "DELETE FROM courses WHERE subject = 'CS' and class_id = '3500';"

      # Basically an inverse grep - it tries SELECT-ing the CS3500 class, and expects the name ("Object") to NOT be present
      - name: Ensure CS3500 no longer exists in our database
        run: |-
          ! psql -U postgres -h localhost -p 5432 -d searchneu_dev -c \
              "SELECT name FROM courses WHERE subject = 'CS' and class_id = '3500';" | grep "Object"

      - name: Run the updater ONLY ONCE, so that it scrapes missing classes
        run: docker exec test sh -c "LOG_LEVEL=VERBOSE CUSTOM_SCRAPE=true UPDATE_ONLY_ONCE=true TERMS_TO_SCRAPE=202140 yarn prod:updater"
        # This should take 5 minutes MAX.
        # In the scrape step, we used `CUSTOM_SCRAPE=true`. That limits what courses are being scraped - see `scrapers/filter` for more details.
        #   We need to use that here as well.
        #   Since we ran a custom scrape, we're missing a lot of classes. If we ran the updater without the custom restrictions, it'd
        #     attempt to scrape ALL of the missing classes. Instead, we'll only be scraping one.
        #   SO - this'll only scrape ~800 sections and ~1 class.
        timeout-minutes: 5

      - name: Ensure CS3500 is back in our database
        run: |-
          psql -U postgres -h localhost -p 5432 -d searchneu_dev -c \
              "SELECT name FROM courses WHERE subject = 'CS' and class_id = '3500';" | grep "Object"