Add CI

.github/actions/start-services/action.yaml

@@ -0,0 +1,35 @@
+name: Start Services
+description: Start Docker Compose services
+
+inputs:
+  profiles:
+    description: "Comma-separated list of profiles to start (e.g. 'all' or 'rest-api,minio')"
+    required: false
+    default: "all"
+  services:
+    description: "Comma-separated list of specific services to start (e.g. 'database,elasticsearch')"
+    required: false
+    default: ""
+
+runs:
+  using: composite
+  steps:
+    - name: Set permission for mount paths
+      shell: bash
+      run: |
+        chmod -R 777 ./data/php
+        chmod -R 777 ./logs
+    - name: Start services
+      shell: bash
+      run: |
+        PROFILE_FLAGS=""
+        for profile in $(echo "${{ inputs.profiles }}" | tr ',' ' '); do
+          PROFILE_FLAGS="$PROFILE_FLAGS --profile $profile"
+        done
+        SERVICES=$(echo "${{ inputs.services }}" | tr ',' ' ')
+        docker compose $PROFILE_FLAGS up -d --build --wait $SERVICES
+
+    - name: Show service status
+      if: always()
+      shell: bash
+      run: docker compose ps

.github/workflows/integration-test.yaml

@@ -0,0 +1,90 @@
+name: Integration Test
+
+on:
+  pull_request:
+    paths:
+      - "docker-compose.yaml"
+      - "scripts/**"
+      - "config/**"
+
+jobs:
+  dataset-processing:
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Start services
+        uses: ./.github/actions/start-services
+        with:
+          profiles: "rest-api,evaluation-engine"
+
+      - name: Create dataset
+        id: dataset
+        run: |
+          DATASET_ID=$(bash scripts/upload_dataset.sh)
+          echo "Dataset ID: $DATASET_ID"
+          echo "id=$DATASET_ID" >> "$GITHUB_OUTPUT"
+
+      - name: Dataset Processing
+        run: bash scripts/wait_for_dataset.sh "${{ steps.dataset.outputs.id }}" 180
+
+      - name: Show logs on failure
+        if: failure()
+        run: docker compose logs --tail=50
+
+  task-processing:
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Start services
+        uses: ./.github/actions/start-services
+        with:
+          profiles: "rest-api,evaluation-engine"
+
+      - name: Create task
+        id: task
+        run: |
+          TASK_ID=$(bash scripts/create_task.sh 129 1)
+          echo "Task ID: $TASK_ID"
+          echo "id=$TASK_ID" >> "$GITHUB_OUTPUT"
+
+      - name: Task Processing
+        run: bash scripts/wait_for_task.sh "${{ steps.task.outputs.id }}" 180
+
+      - name: Show logs on failure
+        if: failure()
+        run: docker compose logs --tail=50
+
+  run-processing:
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Start services
+        uses: ./.github/actions/start-services
+        with:
+          profiles: "rest-api,evaluation-engine"
+
+      - name: Upload run
+        id: run
+        run: |
+          RUN_ID=$(bash scripts/upload_run.sh)
+          echo "Run ID: $RUN_ID"
+          echo "id=$RUN_ID" >> "$GITHUB_OUTPUT"
+
+      - name: Run Processing
+        run: bash scripts/wait_for_run.sh "${{ steps.run.outputs.id }}" 180
+
+      - name: Show logs on failure
+        if: failure()
+        run: docker compose logs --tail=50

README.md

@@ -114,6 +114,20 @@ be slightly altered to use them in the test setup: change https to http.
 
 ## Development
 
+### Services
+When working on the `services` repository itself, the `scripts` directory provides a small number of integration tests to catch the most glaring issues. They are currently limited to REST API and evaluation engine functionality and also run automatically in CI.
+
+For example, after starting services, to test dataset upload run:
+```bash
+./scripts/upload_dataset.sh
+```
+If successful, this will also print the dataset id assigned to the new dataset.
+To test if dataset processing works, we can then run:
+```bash
+./scripts/wait_for_dataset.sh DATASET_ID
+```
+where `DATASET_ID` is replaced by the identifier printed in the previous step.
+
 ### PHP, Parquet and Croissant converter
 If you want to do local development on containers that are part of the docker-compose, you want those containers to change based on your code. You should have the relevant code somewhere on your system. You will need to tell the docker-compose where to find it. You can do so by setting environment variables.
 

scripts/create_task.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+# Creates a task for a given dataset and estimation procedure, prints the task ID to stdout.
+# Usage: create_task.sh DATASET_ID ESTIMATION_PROCEDURE_ID [TARGET_FEATURE]
+set -euo pipefail
+
+DATASET_ID="${1:?Usage: create_task.sh DATASET_ID ESTIMATION_PROCEDURE_ID [TARGET_FEATURE]}"
+ESTIMATION_PROCEDURE_ID="${2:?Usage: create_task.sh DATASET_ID ESTIMATION_PROCEDURE_ID [TARGET_FEATURE]}"
+TARGET_FEATURE="${3:-class}"
+
+TMPFILE=$(mktemp /tmp/task_desc.XXXXXX.xml)
+trap 'rm -f "$TMPFILE"' EXIT
+
+cat > "$TMPFILE" << EOF
+<oml:task_inputs xmlns:oml="http://openml.org/openml">
+  <oml:task_type_id>1</oml:task_type_id>
+  <oml:input name="source_data">${DATASET_ID}</oml:input>
+  <oml:input name="estimation_procedure">${ESTIMATION_PROCEDURE_ID}</oml:input>
+  <oml:input name="target_feature">${TARGET_FEATURE}</oml:input>
+</oml:task_inputs>
+EOF
+
+RESPONSE=$(curl -sS -X POST "http://localhost:8000/api/v1/xml/task?api_key=normaluser" \
+  -F "description=@${TMPFILE};type=text/xml")
+
+TASK_ID=$(echo "$RESPONSE" | sed -n 's/.*<oml:id>\([0-9]*\)<\/oml:id>.*/\1/p')
+
+if [ -z "$TASK_ID" ]; then
+  echo "Server response:" >&2
+  echo "$RESPONSE" >&2
+  exit 1
+fi
+
+echo "$TASK_ID"