Implementation:Langgenius Dify Create App

Knowledge Sources	Dify
Domains	Backend, Architecture, Flask, Deployment
Last Updated	2026-02-12 00:00 GMT

Overview

Concrete tool for constructing a fully initialized Dify Flask application instance with all 26 extensions loaded in the correct order, provided by the create_app() factory function in api/app_factory.py.

Description

The app_factory.py module at api/app_factory.py provides two factory functions:

create_app() (line 63) -- Creates a full-featured DifyApp instance used by Gunicorn in the API container and by the Celery worker. Initializes all 26 extensions in sequence.
create_migrations_app() (line 152) -- Creates a minimal app with only ext_database and ext_migrate initialized. Used exclusively for running Flask-Migrate database migrations (e.g., flask db upgrade).

The initialize_extensions() function (line 73) iterates through a list of extension modules. Each module must expose an init_app(app) function. Optionally, modules may provide an is_enabled() function to conditionally disable themselves.

The initialization sequence is:

extensions = [
    ext_timezone,
    ext_logging,
    ext_warnings,
    ext_import_modules,
    ext_orjson,
    ext_forward_refs,
    ext_set_secretkey,
    ext_compress,
    ext_code_based_extension,
    ext_database,
    ext_app_metrics,
    ext_migrate,
    ext_redis,
    ext_storage,
    ext_logstore,
    ext_celery,
    ext_login,
    ext_mail,
    ext_hosting_provider,
    ext_sentry,
    ext_proxy_fix,
    ext_blueprints,
    ext_commands,
    ext_fastopenapi,
    ext_otel,
    ext_request_logging,
    ext_session_factory,
]

In the Docker deployment, the API container runs Gunicorn which calls create_app() in each worker process. The worker container invokes the same factory through Celery. The MIGRATION_ENABLED=true environment variable controls whether the API container runs database migrations at startup.

Usage

Import and call create_app() when bootstrapping the Dify API server or Celery worker. Use create_migrations_app() when running database schema migrations.

Code Reference

Source Location: api/app_factory.py, lines 63-70 (create_app), lines 73-149 (initialize_extensions), lines 152-160 (create_migrations_app)

Signature:

def create_app() -> DifyApp:
    start_time = time.perf_counter()
    app = create_flask_app_with_configs()
    initialize_extensions(app)
    end_time = time.perf_counter()
    if dify_config.DEBUG:
        logger.info("Finished create_app (%s ms)", round((end_time - start_time) * 1000, 2))
    return app


def initialize_extensions(app: DifyApp):
    # Imports 26 extension modules from extensions package
    # Initializes Flask context capture for workflow execution
    from context.flask_app_context import init_flask_context
    init_flask_context()

    extensions = [ext_timezone, ext_logging, ..., ext_session_factory]
    for ext in extensions:
        short_name = ext.__name__.split(".")[-1]
        is_enabled = ext.is_enabled() if hasattr(ext, "is_enabled") else True
        if not is_enabled:
            continue
        start_time = time.perf_counter()
        ext.init_app(app)
        # Logs timing in DEBUG mode


def create_migrations_app() -> DifyApp:
    app = create_flask_app_with_configs()
    ext_database.init_app(app)
    ext_migrate.init_app(app)
    return app

Import statement:

from app_factory import create_app, create_migrations_app

I/O Contract

Inputs

Parameter	Type	Required	Default	Description
Environment variables	os.environ	Yes	From .env / dify_config	All configuration is read from environment variables via the `dify_config` Pydantic model
Flask config	dict	Auto	dify_config.model_dump()	Configuration dictionary loaded into `app.config`

Outputs

Output	Type	Description
DifyApp instance	DifyApp	Fully configured Flask application with all extensions initialized, ready to serve requests
DifyApp (migrations)	DifyApp	Minimal Flask application with only database and migration extensions, for schema management

Usage Examples

API server startup via Gunicorn (Docker entrypoint):

# In the Gunicorn configuration or WSGI entry point:
from app_factory import create_app

app = create_app()
# Gunicorn binds to 0.0.0.0:5001 with gevent workers

Celery worker initialization:

# The Celery extension (ext_celery) patches the Flask app context
# The worker container uses MODE=worker which invokes Celery
# Celery internally calls create_app() to establish the Flask context

Database migration (using create_migrations_app):

# Used by Flask-Migrate CLI commands:
# flask db upgrade
# This creates a lightweight app with only database + migration extensions
from app_factory import create_migrations_app

app = create_migrations_app()
# Only ext_database and ext_migrate are initialized
# No Redis, Celery, or other heavy services required

Debug timing output (when DEBUG=true):

INFO: Loaded ext_timezone (0.12 ms)
INFO: Loaded ext_logging (1.45 ms)
INFO: Loaded ext_database (23.67 ms)
INFO: Loaded ext_redis (5.34 ms)
INFO: Loaded ext_celery (45.89 ms)
...
INFO: Finished create_app (312.45 ms)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment