test_integration_cli.yml

name: CLI integration tests

on:
  pull_request:
    types: [assigned, opened, synchronize, reopened]
    paths:
      - "projects/adapter/**"
      - ".github/actions/setup-local-fal/**"
      - ".github/workflows/test_integration_cli.yml"

  push:
    branches: [main]
    paths:
      - "projects/adapter/**"

  schedule:
    # every monday
    - cron: "0 0 * * 1"

  workflow_dispatch:
    inputs:
      adapter:
        description: dbt Adapter to test with
        required: false
        default: "<ALL>"
        type: choice
        options:
          - "<ALL>"
          - postgres
          - bigquery
          - snowflake
          - duckdb
          - fal

      python:
        description: Python version to test with
        required: false
        default: "3.8"
        type: choice
        options:
          - "<ALL>"
          - "3.8"
          - "3.9"
          - "3.10"
          - "3.11"

      dbt:
        description: dbt version to test with
        required: false
        default: "latest"
        type: choice
        options:
          - "<ALL>"
          - "latest"
          - "1.5.*"

jobs:
  matrix-adapter:
    runs-on: ubuntu-latest
    outputs:
      list: ${{ steps.matrix-step.outputs.list }}
    steps:
      - id: matrix-step
        shell: python
        run: |
          OPTIONS = [
            'postgres',
            'bigquery',
            'snowflake',
            'fal'
          ]
          EXTRA_OPTIONS = [
            'duckdb',
          ]
          OUTPUT = OPTIONS

          if '${{ github.event_name }}' == 'pull_request':
            import re

            PR_TITLE = '${{ github.event.pull_request.title }}'.lower()
            PR_BRANCH = '${{ github.head_ref }}'.lower()
            PR_DESCRIPTION = '''${{ github.event.pull_request.body }}'''.lower()
            PR_DESCRIPTION = re.sub("<!--.*?-->", "", PR_DESCRIPTION, flags=re.DOTALL)

            # Only test adapters mentioned in the pull request title or branch.
            # We always test postgres and fal adapter as a sanity check.
            OUTPUT = [
              a for a in OPTIONS + EXTRA_OPTIONS
              if a == 'postgres' or a == 'fal' or
                a in PR_TITLE or
                a in PR_BRANCH or
                a in PR_DESCRIPTION
            ]

          elif '${{ github.event_name }}' == 'push':
            OUTPUT = ['postgres']

          elif '${{ github.event_name }}' == 'workflow_dispatch':
            INPUT_CHOICE = '${{ github.event.inputs.adapter }}'
            if INPUT_CHOICE == '<ALL>':
              OUTPUT = OPTIONS + EXTRA_OPTIONS
            else:
              OUTPUT = [INPUT_CHOICE]

          import json
          print("::set-output name=list::" + json.dumps(OUTPUT))

  matrix-python:
    runs-on: ubuntu-latest
    outputs:
      list: ${{ steps.matrix-step.outputs.list }}
    steps:
      - id: matrix-step
        shell: python
        run: |
          OPTIONS = [
            "3.8",
            "3.9",
            "3.10",
            "3.11",
          ]
          OUTPUT = ["3.8"]

          if '${{ github.event_name }}' == 'pull_request':
            import re

            PR_TITLE = '${{ github.event.pull_request.title }}'.lower()
            PR_BRANCH = '${{ github.head_ref }}'.lower()
            PR_DESCRIPTION = '''${{ github.event.pull_request.body }}'''.lower()
            PR_DESCRIPTION = re.sub("<!--.*?-->", "", PR_DESCRIPTION, flags=re.DOTALL)

            # Test version mentioned in the pull request title or branch.
            OUTPUT = [
              v for v in OPTIONS
              if v in PR_TITLE or
                v in PR_BRANCH or
                v in PR_DESCRIPTION
            ]

            if not OUTPUT:
              # If none were found in PR info
              OUTPUT=["3.8"]

          elif '${{ github.event_name }}' in ('schedule', 'push'):
            OUTPUT=OPTIONS

          elif '${{ github.event_name }}' == 'workflow_dispatch':
            INPUT_CHOICE = '${{ github.event.inputs.python }}'
            if INPUT_CHOICE == '<ALL>':
              OUTPUT = OPTIONS
            else:
              OUTPUT = [INPUT_CHOICE]

          import json
          print("::set-output name=list::" + json.dumps(OUTPUT))

  matrix-dbt:
    runs-on: ubuntu-latest
    outputs:
      list: ${{ steps.matrix-step.outputs.list }}
    steps:
      - id: matrix-step
        shell: python
        run: |
          OPTIONS = [
            "1.5.*",
          ]
          OUTPUT = OPTIONS

          if '${{ github.event_name }}' == 'workflow_dispatch':
            INPUT_CHOICE = '${{ github.event.inputs.dbt }}'
            if INPUT_CHOICE == '<ALL>':
              OUTPUT = OPTIONS + EXTRA_OPTIONS
            else:
              OUTPUT = [INPUT_CHOICE]

          import json
          print("::set-output name=list::" + json.dumps(OUTPUT))

  run:
    needs:
      - matrix-adapter
      - matrix-dbt
      - matrix-python
    runs-on: ubuntu-latest
    strategy:
      fail-fast: false
      matrix:
        profile: ${{ fromJSON(needs.matrix-adapter.outputs.list) }}
        dbt: ${{ fromJSON(needs.matrix-dbt.outputs.list) }}
        python: ${{ fromJSON(needs.matrix-python.outputs.list) }}

    # Run only the latest commit pushed to PR
    concurrency:
      group: "${{ github.head_ref || github.run_id }}-${{ github.workflow }}-${{ matrix.profile }}-${{ matrix.dbt }}-${{ matrix.python }}"
      cancel-in-progress: true

    steps:
      - uses: actions/checkout@v2

      - name: Setup local fal
        uses: ./.github/actions/setup-local-fal
        with:
          python: ${{ matrix.python }}
          dbt: ${{ matrix.dbt }}
          adapter: ${{ matrix.profile }}

      - name: Start Docker database
        working-directory: projects/adapter/cli_tests
        if: contains(fromJSON('["postgres", "fal"]'), matrix.profile)
        run: docker-compose up -d

      - name: Install conda
        uses: s-weigand/setup-conda@v1
        with:
          activate-conda: false
          python-version: ${{ matrix.python }}
          # PyJokes is available on conda-forge
          conda-channels: anaconda, conda-forge

      - name: Setup behave
        working-directory: projects/adapter/cli_tests
        run: pip install behave ipython

      - name: Run tests
        id: test_run
        working-directory: projects/adapter/cli_tests
        env:
          FAL_STATS_ENABLED: false
          # BigQuery
          KEYFILE: ${{ secrets.GCP_SA_KEY }}
          GCLOUD_PROJECT: ${{ secrets.GCP_PROJECT_ID }}
          BQ_DATASET: ${{ secrets.BQ_DATASET }}
          # Snowflake
          SF_ACCOUNT: ${{ secrets.SF_ACCOUNT }}
          SF_USER: ${{ secrets.SF_USER }}
          SF_PASSWORD: ${{ secrets.SF_PASSWORD }}
          SF_ROLE: ${{ secrets.SF_ROLE }}
          SF_DATABASE: ${{ secrets.SF_DATABASE }}
          SF_WAREHOUSE: ${{ secrets.SF_WAREHOUSE }}
          SF_SCHEMA: ${{ secrets.SF_SCHEMA }}
          # Duckdb
          DB_PATH: ${{ github.workspace }}/duck.db
        run: |
          # Database and schema setup for sources
          if [[ '${{ matrix.profile }}' == "bigquery" ]]
          then
            export DBT_DATABASE="$GCLOUD_PROJECT" DBT_SCHEMA="$BQ_DATASET"
          fi
          if [[ '${{ matrix.profile }}' == "snowflake" ]]
          then
            export DBT_DATABASE="$SF_DATABASE" DBT_SCHEMA="$SF_SCHEMA"
          fi
          if [[ '${{ matrix.profile }}' == "duckdb" ]]
          then
            # TODO: which to use for sources? Example:
            #   database: "{{ env_var('DBT_DATABASE', 'test') }}"
            #   schema: "{{ env_var('DBT_SCHEMA', 'dbt_fal') }}"
            export DBT_DATABASE="" DBT_SCHEMA=""
          fi

          if [[ '${{ matrix.profile }}' == "bigquery" ]]
          then
            echo $KEYFILE > $HOME/keyfile.json
            ls -la $HOME/keyfile.json
            export KEYFILE_DIR=$HOME
            echo 'keyfile is ready'
          fi

          # Could not get the real job_id easily from context
          UUID=$(uuidgen | head -c8)
          export DB_NAMESPACE="${{ github.run_id }}_${UUID}"

          BEHAVE_TAGS="--tags=-TODO-${{ matrix.profile }}"

          if [[ '${{ matrix.profile }}' != 'postgres' ]] && [[ '${{ matrix.profile }}' != 'fal' ]]
          then
            # 'broken_profile' tests only works for postgres and postgres+fal right now
            BEHAVE_TAGS="$BEHAVE_TAGS --tags=-broken_profile"
          fi

          behave $BEHAVE_TAGS -fplain -D profile=${{ matrix.profile }}

      - name: Send custom JSON data to Slack workflow
        if: false
        # if: (failure() || cancelled()) && github.event_name == 'schedule'
        id: slack
        uses: slackapi/slack-github-action@v1.18.0
        with:
          # For posting a rich message using Block Kit
          payload: |
            {
              "text": "Integration tests failed for dbt-${{ matrix.profile }}@${{ matrix.dbt }} (Python ${{ matrix.python }})\nhttps://github.com/fal-ai/fal/actions/runs/${{ github.run_id }}"
            }
        env:
          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
          SLACK_WEBHOOK_TYPE: INCOMING_WEBHOOK