rework server startup and cli

This changes the dockerfile as well.

.gitea/workflows/release.yml

@@ -66,8 +66,8 @@ jobs:
           echo "## Installation" >> release-notes.md
           echo "" >> release-notes.md
           echo '```bash' >> release-notes.md
-          echo 'uv sync' >> release-notes.md
-          echo 'uv run python main.py' >> release-notes.md
+          echo 'pip install embeddingbuddy' >> release-notes.md
+          echo 'embeddingbuddy serve' >> release-notes.md
           echo '```' >> release-notes.md
 
       - name: Create Release

CLAUDE.md

@@ -21,29 +21,23 @@ uv sync
 
 **Run the application:**
 
-Development mode (with auto-reload):
+Using the CLI (recommended):
 
 ```bash
-uv run run_dev.py
+# Production mode (no debug, no auto-reload)
+embeddingbuddy serve
+
+# Development mode (debug + auto-reload on code changes)
+embeddingbuddy serve --dev
+
+# Debug logging only (no auto-reload)
+embeddingbuddy serve --debug
+
+# With custom host/port
+embeddingbuddy serve --host 0.0.0.0 --port 8080
 ```
 
-Production mode (with Gunicorn WSGI server):
-
-```bash
-# First install production dependencies
-uv sync --extra prod
-
-# Then run in production mode
-uv run run_prod.py
-```
-
-Legacy mode (basic Dash server):
-
-```bash
-uv run main.py
-```
-
-The app will be available at <http://127.0.0.1:8050>
+The app will be available at <http://127.0.0.1:8050> by default
 
 **Run tests:**
 

Dockerfile

@@ -23,9 +23,6 @@ COPY pyproject.toml uv.lock ./
 
 # Copy source code (needed for editable install)
 COPY src/ src/
-COPY main.py .
-COPY wsgi.py .
-COPY run_prod.py .
 COPY assets/ assets/
 
 # Change ownership of source files before building (lighter I/O)
@@ -59,10 +56,7 @@ RUN chown appuser:appuser /app
 # Copy files from builder with correct ownership
 COPY --from=builder --chown=appuser:appuser /app/.venv /app/.venv
 COPY --from=builder --chown=appuser:appuser /app/src /app/src
-COPY --from=builder --chown=appuser:appuser /app/main.py /app/main.py
 COPY --from=builder --chown=appuser:appuser /app/assets /app/assets
-COPY --from=builder --chown=appuser:appuser /app/wsgi.py /app/wsgi.py
-COPY --from=builder --chown=appuser:appuser /app/run_prod.py /app/run_prod.py
 
 # Switch to non-root user
 USER appuser
@@ -86,5 +80,5 @@ EXPOSE 8050
 HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \
     CMD python -c "import requests; requests.get('http://localhost:8050/', timeout=5)" || exit 1
 
-# Run application with Gunicorn in production
-CMD ["python", "run_prod.py"]
+# Run application in production mode (no debug, no auto-reload)
+CMD ["embeddingbuddy", "serve"]

README.md

@@ -39,7 +39,7 @@ documents and prompts to understand how queries relate to your content.
 uv tool install embeddingbuddy
 
 # Run the application
-embeddingbuddy
+embeddingbuddy serve
 ```
 
 **Option 2: Install with pip/pipx**
@@ -124,26 +124,18 @@ uv sync
 
 2. **Run the application:**
 
-**Development mode** (with auto-reload):
-
 ```bash
-uv run run_dev.py
-```
+# Production mode (no debug, no auto-reload)
+embeddingbuddy serve
 
-**Production mode** (with Gunicorn WSGI server):
+# Development mode (debug + auto-reload on code changes)
+embeddingbuddy serve --dev
 
-```bash
-# Install production dependencies
-uv sync --extra prod
+# Debug logging only (no auto-reload)
+embeddingbuddy serve --debug
 
-# Run in production mode
-uv run run_prod.py
-```
-
-**Legacy mode** (basic Dash server):
-
-```bash
-uv run main.py
+# Custom host/port
+embeddingbuddy serve --host 0.0.0.0 --port 8080
 ```
 
 3. **Open your browser** to <http://127.0.0.1:8050>
@@ -231,10 +223,8 @@ src/embeddingbuddy/
 │       └── interactions.py    # User interaction callbacks
 └── utils/                     # Utility functions
 
-main.py            # Application runner (at project root)
-main.py            # Application runner (at project root)
-run_dev.py         # Development server runner
-run_prod.py        # Production server runner
+# CLI entry point
+embeddingbuddy serve    # Main CLI command to start the server
 ```
 
 ### Testing

main.py

@@ -1,10 +0,0 @@
-from src.embeddingbuddy.app import create_app, run_app
-
-
-def main():
-    app = create_app()
-    run_app(app)
-
-
-if __name__ == "__main__":
-    main()

pyproject.toml

@@ -18,7 +18,8 @@ dependencies = [
 ]
 
 [project.scripts]
-embeddingbuddy = "embeddingbuddy.app:main"
+embeddingbuddy = "embeddingbuddy.cli:main"
+embeddingbuddy-serve = "embeddingbuddy.app:serve"
 
 [project.optional-dependencies]
 test = [

run_dev.py

@@ -1,32 +0,0 @@
-#!/usr/bin/env python3
-"""
-Development runner with auto-reload enabled.
-This runs the Dash development server with hot reloading.
-"""
-import os
-from src.embeddingbuddy.app import create_app, run_app
-
-def main():
-    """Run the application in development mode with auto-reload."""
-    # Force development settings
-    os.environ["EMBEDDINGBUDDY_ENV"] = "development"
-    os.environ["EMBEDDINGBUDDY_DEBUG"] = "true"
-
-    # Check for OpenSearch disable flag (optional for testing)
-    # Set EMBEDDINGBUDDY_OPENSEARCH_ENABLED=false to test without OpenSearch
-    opensearch_status = os.getenv("EMBEDDINGBUDDY_OPENSEARCH_ENABLED", "true")
-    opensearch_enabled = opensearch_status.lower() == "true"
-    
-    print("🚀 Starting EmbeddingBuddy in development mode...")
-    print("📁 Auto-reload enabled - changes will trigger restart")
-    print("🌐 Server will be available at http://127.0.0.1:8050")
-    print(f"🔍 OpenSearch: {'Enabled' if opensearch_enabled else 'Disabled'}")
-    print("⏹️  Press Ctrl+C to stop")
-    
-    app = create_app()
-    
-    # Run with development server (includes auto-reload when debug=True)
-    run_app(app, debug=True)
-
-if __name__ == "__main__":
-    main()

run_prod.py

@@ -1,52 +0,0 @@
-#!/usr/bin/env python3
-"""
-Production runner using Gunicorn WSGI server.
-This provides better performance and stability for production deployments.
-"""
-import os
-import subprocess
-import sys
-from src.embeddingbuddy.config.settings import AppSettings
-
-def main():
-    """Run the application in production mode with Gunicorn."""
-    # Force production settings
-    os.environ["EMBEDDINGBUDDY_ENV"] = "production"
-    os.environ["EMBEDDINGBUDDY_DEBUG"] = "false"
-    # Disable OpenSearch by default in production (can be overridden by setting env var)
-    if "EMBEDDINGBUDDY_OPENSEARCH_ENABLED" not in os.environ:
-        os.environ["EMBEDDINGBUDDY_OPENSEARCH_ENABLED"] = "false"
-    
-    print("🚀 Starting EmbeddingBuddy in production mode...")
-    print(f"⚙️  Workers: {AppSettings.GUNICORN_WORKERS}")
-    print(f"🌐 Server will be available at http://{AppSettings.GUNICORN_BIND}")
-    print("⏹️  Press Ctrl+C to stop")
-    
-    # Gunicorn command
-    cmd = [
-        "gunicorn",
-        "--workers", str(AppSettings.GUNICORN_WORKERS),
-        "--bind", AppSettings.GUNICORN_BIND,
-        "--timeout", str(AppSettings.GUNICORN_TIMEOUT),
-        "--keep-alive", str(AppSettings.GUNICORN_KEEPALIVE),
-        "--access-logfile", "-",
-        "--error-logfile", "-",
-        "--log-level", "info",
-        "wsgi:application"
-    ]
-    
-    try:
-        subprocess.run(cmd, check=True)
-    except KeyboardInterrupt:
-        print("\n🛑 Shutting down...")
-        sys.exit(0)
-    except subprocess.CalledProcessError as e:
-        print(f"❌ Error running Gunicorn: {e}")
-        sys.exit(1)
-    except FileNotFoundError:
-        print("❌ Gunicorn not found. Install it with: uv add gunicorn")
-        print("💡 Or run in development mode with: python run_dev.py")
-        sys.exit(1)
-
-if __name__ == "__main__":
-    main()

src/embeddingbuddy/app.py

@@ -1,15 +1,21 @@
+"""
+EmbeddingBuddy application factory and server functions.
+
+This module contains the main application creation logic with imports
+moved inside functions to avoid loading heavy dependencies at module level.
+"""
+
+
+def create_app():
+    """Create and configure the Dash application instance."""
+    import os
     import dash
     import dash_bootstrap_components as dbc
-from .config.settings import AppSettings
     from .ui.layout import AppLayout
     from .ui.callbacks.data_processing import DataProcessingCallbacks
     from .ui.callbacks.visualization import VisualizationCallbacks
     from .ui.callbacks.interactions import InteractionCallbacks
 
-
-def create_app():
-    import os
-
     # Get the project root directory (two levels up from this file)
     project_root = os.path.dirname(os.path.dirname(os.path.dirname(__file__)))
     assets_path = os.path.join(project_root, "assets")
@@ -124,6 +130,9 @@ def _register_client_side_callbacks(app):
 
 
 def run_app(app=None, debug=None, host=None, port=None):
+    """Run the Dash application with specified settings."""
+    from .config.settings import AppSettings
+
     if app is None:
         app = create_app()
 
@@ -134,10 +143,69 @@ def run_app(app=None, debug=None, host=None, port=None):
     )
 
 
-def main():
-    """Main entry point for the embeddingbuddy CLI command."""
+def serve(host=None, port=None, dev=False, debug=False):
+    """Start the EmbeddingBuddy web server.
+
+    Args:
+        host: Host to bind to (default: 127.0.0.1)
+        port: Port to bind to (default: 8050)
+        dev: Development mode - enable debug logging and auto-reload (default: False)
+        debug: Enable debug logging only, no auto-reload (default: False)
+    """
+    import os
+    from .config.settings import AppSettings
+
+    # Determine actual values to use
+    actual_host = host if host is not None else AppSettings.HOST
+    actual_port = port if port is not None else AppSettings.PORT
+
+    # Determine mode
+    # --dev takes precedence and enables both debug and auto-reload
+    # --debug enables only debug logging
+    # No flags = production mode (no debug, no auto-reload)
+    use_reloader = dev
+    use_debug = dev or debug
+
+    # Only print startup messages in main process (not in Flask reloader)
+    if not os.environ.get('WERKZEUG_RUN_MAIN'):
+        mode = "development" if dev else ("debug" if debug else "production")
+        print(f"Starting EmbeddingBuddy in {mode} mode...")
+        print("Loading dependencies (this may take a few seconds)...")
+        print(f"Server will start at http://{actual_host}:{actual_port}")
+        if use_reloader:
+            print("Auto-reload enabled - server will restart on code changes")
+
     app = create_app()
-    run_app(app)
+
+    # Suppress Flask development server warning in production mode
+    if not use_debug and not use_reloader:
+        import warnings
+        import logging
+
+        # Suppress the werkzeug warning
+        warnings.filterwarnings('ignore', message='.*development server.*')
+
+        # Set werkzeug logger to ERROR level to suppress the warning
+        werkzeug_logger = logging.getLogger('werkzeug')
+        werkzeug_logger.setLevel(logging.ERROR)
+
+    # Use Flask's built-in server with appropriate settings
+    app.run(
+        debug=use_debug,
+        host=actual_host,
+        port=actual_port,
+        use_reloader=use_reloader
+    )
+
+
+def main():
+    """Legacy entry point - redirects to cli module.
+
+    This is kept for backward compatibility but the main CLI
+    is now in embeddingbuddy.cli for faster startup.
+    """
+    from .cli import main as cli_main
+    cli_main()
 
 
 if __name__ == "__main__":

src/embeddingbuddy/cli.py

@@ -0,0 +1,74 @@
+"""
+Lightweight CLI entry point for EmbeddingBuddy.
+
+This module provides a fast command-line interface that only imports
+heavy dependencies when actually needed by subcommands.
+"""
+import argparse
+import sys
+
+
+def main():
+    """Main CLI entry point with minimal imports for fast help text."""
+    parser = argparse.ArgumentParser(
+        prog="embeddingbuddy",
+        description="EmbeddingBuddy - Interactive embedding visualization tool",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  embeddingbuddy serve                    # Production mode (no debug, no auto-reload)
+  embeddingbuddy serve --dev              # Development mode (debug + auto-reload)
+  embeddingbuddy serve --debug            # Debug logging only (no auto-reload)
+  embeddingbuddy serve --port 8080        # Custom port
+  embeddingbuddy serve --host 0.0.0.0     # Bind to all interfaces
+        """
+    )
+
+    subparsers = parser.add_subparsers(
+        dest="command",
+        help="Available commands",
+        metavar="<command>"
+    )
+
+    # Serve subcommand
+    serve_parser = subparsers.add_parser(
+        "serve",
+        help="Start the web server",
+        description="Start the EmbeddingBuddy web server for interactive visualization"
+    )
+    serve_parser.add_argument(
+        "--host",
+        default=None,
+        help="Host to bind to (default: 127.0.0.1)"
+    )
+    serve_parser.add_argument(
+        "--port",
+        type=int,
+        default=None,
+        help="Port to bind to (default: 8050)"
+    )
+    serve_parser.add_argument(
+        "--dev",
+        action="store_true",
+        help="Development mode: enable debug logging and auto-reload"
+    )
+    serve_parser.add_argument(
+        "--debug",
+        action="store_true",
+        help="Enable debug logging (no auto-reload)"
+    )
+
+    args = parser.parse_args()
+
+    if args.command == "serve":
+        # Only import heavy dependencies when actually running serve
+        from embeddingbuddy.app import serve
+        serve(host=args.host, port=args.port, dev=args.dev, debug=args.debug)
+    else:
+        # No command specified, show help
+        parser.print_help()
+        sys.exit(0)
+
+
+if __name__ == "__main__":
+    main()

src/embeddingbuddy/config/settings.py

@@ -69,7 +69,7 @@ class AppSettings:
     TEXT_PREVIEW_LENGTH = 100
 
     # App Configuration
-    DEBUG = os.getenv("EMBEDDINGBUDDY_DEBUG", "True").lower() == "true"
+    DEBUG = os.getenv("EMBEDDINGBUDDY_DEBUG", "False").lower() == "true"
     HOST = os.getenv("EMBEDDINGBUDDY_HOST", "127.0.0.1")
     PORT = int(os.getenv("EMBEDDINGBUDDY_PORT", "8050"))
 

src/embeddingbuddy/wsgi.py

@@ -0,0 +1,11 @@
+"""
+WSGI entry point for production deployment.
+Use this with a production WSGI server like Gunicorn.
+"""
+from embeddingbuddy.app import create_app
+
+# Create the application instance
+application = create_app()
+
+# For compatibility with different WSGI servers
+app = application

uv.lock

@@ -412,7 +412,7 @@ wheels = [
 
 [[package]]
 name = "embeddingbuddy"
-version = "0.5.1"
+version = "0.6.3"
 source = { editable = "." }
 dependencies = [
     { name = "dash" },

wsgi.py

@@ -1,20 +0,0 @@
-"""
-WSGI entry point for production deployment.
-Use this with a production WSGI server like Gunicorn.
-"""
-from src.embeddingbuddy.app import create_app
-
-# Create the application instance
-application = create_app()
-
-# For compatibility with different WSGI servers
-app = application
-
-if __name__ == "__main__":
-    # This won't be used in production, but useful for testing
-    from src.embeddingbuddy.config.settings import AppSettings
-    application.run(
-        host=AppSettings.HOST,
-        port=AppSettings.PORT,
-        debug=False
-    )