Add application insights integration to pygeoapi (#38)

* Add Application Insight integration

* Add Application Insight integration

---------

Co-authored-by: Olesia Solonko <olesia.solonko@ngi.no>

Author: solonkoolesya

README.md

@@ -29,6 +29,7 @@ When deploying in different environments (local, Kubernetes, etc.), the followin
 - `PYGEOAPI_SERVER_URL`: The external URL where pygeoapi will be accessible. This is used to generate correct links in the API responses.
   - For local development: `http://localhost:5000`
   - For Kubernetes/production: Your domain, e.g., `https://your-domain.com/api`
+- `APPLICATIONINSIGHTS_CONNECTION_STRING`: Connection string to Azure Application Insights. If not specifies, telemetry disabled, no user statistic gathered.
 
 #### Frontend:
 

docker-compose.yml

@@ -41,6 +41,7 @@ services:
       - CONTAINER_HOST=0.0.0.0
       - CONTAINER_PORT=5000
       - PYGEOAPI_SERVER_URL=http://localhost:5000
+      - APPLICATIONINSIGHTS_CONNECTION_STRING=""
     profiles:
       - local
 

pygeoapi/docker/Dockerfile.pygeoapi

@@ -30,6 +30,9 @@ RUN apt-get update \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/*
 
+# Install Application Insights for Python
+RUN pip install opencensus-ext-azure opencensus-ext-flask opencensus-ext-requests
+
 # Build version string
 ARG REPO_VERSION
 RUN PY_VER=$(python3 -c 'import pkg_resources;print(pkg_resources.get_distribution("pygeoapi").version)') \
@@ -38,6 +41,9 @@ RUN PY_VER=$(python3 -c 'import pkg_resources;print(pkg_resources.get_distributi
 # Create readiness indicator directory
 RUN mkdir -p /tmp/health
 
+# Add Application Insights instrumentation script
+COPY pygeoapi/docker/appinsights_init.py /appinsights_init.py
+
 # Add init script
 COPY pygeoapi/docker/init.sh /init.sh
 RUN chmod +x /init.sh
@@ -46,6 +52,9 @@ RUN chmod +x /init.sh
 COPY pygeoapi/docker/healthcheck.sh /healthcheck.sh
 RUN chmod +x /healthcheck.sh
 
+# Environment variable for Application Insights
+ENV APPLICATIONINSIGHTS_CONNECTION_STRING=""
+
 HEALTHCHECK --interval=15s --timeout=10s --start-period=180s --retries=10 CMD ["/healthcheck.sh"]
 
 ENTRYPOINT ["/init.sh"]

pygeoapi/docker/appinsights_init.py

@@ -0,0 +1,67 @@
+# appinsights_init.py - Lightweight Application Insights setup for pygeoapi
+import os
+import logging
+from opencensus.ext.azure.log_exporter import AzureLogHandler
+from opencensus.ext.azure.trace_exporter import AzureExporter
+from opencensus.ext.flask.flask_middleware import FlaskMiddleware
+from opencensus.trace.samplers import ProbabilitySampler
+
+
+def setup_app_insights(app):
+    """
+    Setup lightweight Application Insights for pygeoapi
+    Only tracks user interactions and basic telemetry
+    """
+    connection_string = os.environ.get('APPLICATIONINSIGHTS_CONNECTION_STRING')
+
+    if not connection_string:
+        print("Application Insights not configured - APPLICATIONINSIGHTS_CONNECTION_STRING not set")
+        return
+
+    try:
+        # Setup request tracking with low sampling rate for lightweight monitoring
+        middleware = FlaskMiddleware(
+            app,
+            exporter=AzureExporter(connection_string=connection_string),
+            sampler=ProbabilitySampler(rate=0.1)  # Sample 10% of requests for lightweight monitoring
+        )
+
+        # Setup basic logging for user interactions
+        logger = logging.getLogger(__name__)
+        logger.addHandler(AzureLogHandler(connection_string=connection_string))
+        logger.setLevel(logging.INFO)
+
+        # Add custom properties for better analytics
+        @app.before_request
+        def before_request():
+            from flask import request
+            from opencensus.trace import execution_context
+
+            tracer = execution_context.get_opencensus_tracer()
+            if tracer:
+                span = tracer.current_span()
+                if span:
+                    # Add custom properties for analytics
+                    span.add_attribute('user_agent', request.headers.get('User-Agent', ''))
+                    span.add_attribute('referrer', request.headers.get('Referer', ''))
+                    span.add_attribute('api_endpoint', request.endpoint or 'unknown')
+
+        print("Application Insights configured successfully")
+
+    except Exception as e:
+        print(f"Failed to setup Application Insights: {e}")
+
+
+# Custom logging function for user interФactions
+def log_user_interaction(action, details=None):
+    """Log user interactions for analytics"""
+    connection_string = os.environ.get('APPLICATIONINSIGHTS_CONNECTION_STRING')
+    if connection_string:
+        logger = logging.getLogger('user_analytics')
+        logger.addHandler(AzureLogHandler(connection_string=connection_string))
+
+        log_data = {'action': action}
+        if details:
+            log_data.update(details)
+
+        logger.info(f"User interaction: {action}", extra={'custom_dimensions': log_data})

pygeoapi/docker/init.sh

@@ -20,6 +20,14 @@ fi
 PYGEOAPI_VERSION=$(pygeoapi --version 2>&1)
 echo "$(get_timestamp) [init] PyGeoAPI version: $PYGEOAPI_VERSION"
 
+# Check Application Insights configuration
+if [ -n "$APPLICATIONINSIGHTS_CONNECTION_STRING" ]; then
+    echo "$(get_timestamp) [init] Application Insights connection string found, enabling telemetry"
+    export PYTHONPATH="/appinsights_init.py:$PYTHONPATH"
+else
+    echo "$(get_timestamp) [init] No Application Insights connection string found, running without telemetry"
+fi
+
 # Remove any existing readiness indicator
 rm -f /tmp/health/init_complete
 echo "$(get_timestamp) [init] Removed any existing readiness indicators"
@@ -44,6 +52,57 @@ mkdir -p /tmp/health
 touch /tmp/health/init_complete
 echo "$(get_timestamp) [init] Initialization complete"
 
+# Start pygeoapi with Application Insights if configured
+if [ -n "$APPLICATIONINSIGHTS_CONNECTION_STRING" ]; then
+    echo "$(get_timestamp) [init] Starting pygeoapi service with Application Insights"
+
+    # First, generate the OpenAPI document that pygeoapi needs
+    echo "$(get_timestamp) [init] Generating OpenAPI document"
+    pygeoapi openapi generate /pygeoapi/local.config.yml --output-file /tmp/openapi.yml
+
+    # Create a startup script that integrates Application Insights
+    cat > /tmp/start_with_insights.py << 'EOF'
+import sys
+import os
+
+# Add the root directory to Python path to find appinsights_init
+sys.path.insert(0, '/')
+
+try:
+    # Import the Application Insights setup
+    import appinsights_init
+    print("[init] Application Insights module loaded successfully")
+
+    # Import pygeoapi app
+    from pygeoapi.flask_app import APP as app
+
+    # Setup Application Insights
+    appinsights_init.setup_app_insights(app)
+    print("[init] Application Insights integrated successfully")
+
+except ImportError as e:
+    print(f"[init] Application Insights setup skipped: {e}")
+    # Continue without Application Insights
+    pass
+except Exception as e:
+    print(f"[init] Application Insights setup failed: {e}")
+    # Continue without Application Insights
+    pass
+
+print("[init] Application Insights setup completed")
+EOF
+
+    # Execute the Application Insights setup (but don't start the app yet)
+    python3 /tmp/start_with_insights.py
+
+else
+    echo "$(get_timestamp) [init] Starting pygeoapi service without Application Insights"
+    # Still need to generate OpenAPI document
+    echo "$(get_timestamp) [init] Generating OpenAPI document"
+    pygeoapi openapi generate /pygeoapi/local.config.yml --output-file /tmp/openapi.yml
+fi
+
+
 # Start the pygeoapi service
 echo "$(get_timestamp) [init] Starting pygeoapi service"
 exec /entrypoint.sh "$@"