Welcome to nautilus’s documentation!¶
Contents:
Getting Started¶
Services¶
Services are the building block of nautilus clouds. Very simply, a nautilus service is a standalone process that responds to actions sent over the global event queue, maintains and mutates some internal state according to the action, and provides a summary of that internal state via a GraphQL API.
Nautilus provides extendible implementations of common services as well as a base Service class which all act as good starting points for your own services:
Action Handlers¶
Action handlers describe how your service mutates its internal state in
response to the arrival of an action from the queue. They are defined as
a function of two arguments: type and payload. Type is a string that
classifies the event and payload is a dictionary representing the
data associated with the event. For example,
def action_handler(action_type, payload):
# if the payload represents a new recipe to add to the list
if action_type == 'create_recipe':
# create a new instance of the recipe model
recipe = Recipe(**payload)
# save the new model
recipe.save()
# otherwise if the payload is the id of a recipe to be deleted
elif action_type == 'delete_recipe':
# find the matching recipe
recipe = Recipe.query.first(Recipe.id == payload)
# remove the recipe from the database
recipe.remove()
Combining Action Handlers¶
As your services get more complex, you’ll want to split your action handler into
separate functions which each get called. Nautilus provides a function called
combineActionHandlers which serves just this case:
from nautilus.network import combineActionHandlers
def action_handler1(type, payload):
print("first handler fired!")
def action_handler2(type, payload):
print("second handler fired!")
combined_handler = combineActionHandlers(
action_handler1,
action_handler2
)
Service API¶
For the most part, the generation of a Graphql schema takes care of itself thanks to the excellent Graphene community. To reduce the overall boilerplate, Nautilus provides a schema factory that generates a schema to match a model:
Summarizing an External Service in A Schema¶
Although a service should never rely on information that it does not maintain, there are very rare cases (like the api gateway) where it is necessary to show another service’s data in a schema. In this case, nautilus provides a special base class for the object type that represents remote data.
Designating a GraphQL Equivalent of an SQLAlchemy Type¶
Due to the early age of the Graphene/GraphQL community, not all SQLAlchemy types have conversions already specified. If you encounter such a type, consider posting a PR. If that’s not your style, you can always register it yourself:
from nautilus.models import BaseModel
from nautilus.api import convert_sqlalchemy_type
from sqlalchemy import Column
from sqlalchemy.dialects.postgresql import UUID
from graphene.core.types.scalars import String
class Product(BaseModel):
id = Column(UUID, primary_key = True)
@convert_sqlalchemy_type.register(UUID)
def convert_column_to_string(type, column):
return String(description = column.doc)