test-split-backend.yml

name: Split-Backend E2E Tests

on:
  push:
    branches: [ main ]
  pull_request:
    branches: [ main ]
  workflow_dispatch:

jobs:
  e2e-split-backend:
    runs-on: ubuntu-latest
    timeout-minutes: 15

    services:
      postgres:
        image: pgvector/pgvector:pg16
        env:
          POSTGRES_DB: rag_vector_db
          POSTGRES_USER: rag_user
          POSTGRES_PASSWORD: rag_password
        ports:
          - 5432:5432
        options: >-
          --health-cmd "pg_isready -U rag_user -d rag_vector_db"
          --health-interval 10s
          --health-timeout 5s
          --health-retries 5

    steps:
    - uses: actions/checkout@v4

    - name: Set up Python 3.10
      uses: actions/setup-python@v5
      with:
        python-version: "3.10"
        cache: 'pip'

    - name: Cache HuggingFace embedding model
      uses: actions/cache@v4
      with:
        path: ~/.cache/huggingface/hub
        key: hf-model-all-MiniLM-L6-v2

    - name: Install dependencies
      run: |
        python -m pip install --upgrade pip
        pip install -r requirements.txt
        pip install pytest requests

    - name: Bootstrap API key, start server, run E2E tests
      env:
        DB_HOST: localhost
        DB_PORT: 5432
        DB_NAME: rag_vector_db
        DB_USER: rag_user
        DB_PASSWORD: rag_password
        API_HOST: 127.0.0.1
        API_PORT: 9000
        API_REQUIRE_AUTH: "true"
        API_AUTH_FORCE_ALL: "true"
      run: |
        set -euo pipefail

        # ── Run database migrations ────────────────────────────────
        echo "::group::Run database migrations"
        python -c "
        import sys
        sys.path.insert(0, '.')
        from alembic.config import Config
        from alembic import command
        cfg = Config('alembic.ini')
        command.upgrade(cfg, 'head')
        print('Migrations complete')
        "
        echo "::endgroup::"

        # ── Bootstrap API key ───────────────────────────────────────
        echo "::group::Bootstrap API key"
        API_KEY=$(python -c "
        import sys, os
        sys.path.insert(0, '.')
        os.environ.setdefault('DB_HOST', 'localhost')
        from auth import create_api_key_record
        record = create_api_key_record('e2e-test')
        print(record['key'])
        ")
        echo "API key created (prefix: ${API_KEY:0:12}...)"
        echo "::endgroup::"

        # ── Start API server ────────────────────────────────────────
        echo "::group::Start API server"
        nohup python -m uvicorn api:app \
          --host 127.0.0.1 \
          --port 9000 \
          --log-level info \
          > server.log 2>&1 &
        SERVER_PID=$!
        echo "Server started (PID: $SERVER_PID)"
        echo "::endgroup::"

        # Cleanup on exit
        trap "echo 'Stopping server (PID: $SERVER_PID)...'; kill $SERVER_PID 2>/dev/null || true" EXIT

        # ── Wait for healthy ────────────────────────────────────────
        echo "::group::Wait for server health"
        HEALTHY=false
        for i in $(seq 1 60); do
          # Check if server process died
          if ! kill -0 $SERVER_PID 2>/dev/null; then
            echo "ERROR: Server process died!"
            echo "=== server.log ==="
            cat server.log
            exit 1
          fi

          STATUS=$(curl -s http://127.0.0.1:9000/health | python -c "import sys,json; print(json.load(sys.stdin).get('status',''))" 2>/dev/null || echo "")
          if [ "$STATUS" = "healthy" ]; then
            HEALTHY=true
            echo "Server healthy after $((i * 2))s"
            break
          fi
          echo "Waiting... ($i/60, status=$STATUS)"
          sleep 2
        done

        # Post-loop guard: fail if health was never reached
        if [ "$HEALTHY" != "true" ]; then
          echo "ERROR: Server did not become healthy within 120s"
          echo "=== server.log ==="
          cat server.log
          exit 1
        fi
        echo "::endgroup::"

        # ── Warm up embedding model ──────────────────────────────
        # The model loads lazily on first embedding request.  Fire a search
        # now (before pytest) so the 120s test timeout is never at risk.
        echo "::group::Warm up embedding model"
        echo "Sending warm-up search to trigger model load (up to 300s)..."
        curl -s -X POST http://127.0.0.1:9000/api/v1/search \
          -H "X-API-Key: $API_KEY" \
          -H "Content-Type: application/json" \
          -d '{"query": "warmup", "top_k": 1}' \
          --max-time 300 > /dev/null || echo "Warm-up request timed out (model may still be loading)"
        echo "Warm-up complete"
        echo "::endgroup::"

        # ── Run E2E tests ───────────────────────────────────────────
        echo "::group::Run E2E tests"
        E2E_API_KEY="$API_KEY" E2E_BASE_URL="http://127.0.0.1:9000" \
          python -m pytest tests/test_e2e_split_backend.py -v --tb=short
        echo "::endgroup::"

        echo "All E2E tests passed!"

    - name: Upload server log on failure
      if: failure()
      uses: actions/upload-artifact@v4
      with:
        name: server-log
        path: server.log
        if-no-files-found: ignore
        retention-days: 7