napkin 0.5.2

Python DSL for writing PlantUML sequence diagram

Napkin

Napkin is a tool to "write" sequence diagrams effectively as Python code.

Motivation

The sequence diagrams are useful tool to capture the behavioural aspect of the design. PlantUML is a great tool to draw nice sequence diagrams with simple human readable plain text.

However, the syntax of PlantUML is hard to use when there are nested calls, where lifeline with multiple activation/deactivations are involved. Unfortunately, this situation is quite common in sequence diagram for S/W.

For example, consider the following common sequence diagram, which is from Figure 4.2, UML Distilled 3E:

The PlainUML script for the diagram will be as follows:

@startuml
participant User
participant Order
participant OrderLine
participant Product
participant Customer

User -> Order : calculatePrice()
activate Order
Order -> OrderLine : calculatePrice()
activate OrderLine
OrderLine -> Product : getPrice(quantity:number)
OrderLine -> Customer : getDiscountedValue(Order)
activate Customer
Customer -> Order : getBaseValue()
activate Order
Customer <-- Order: value
deactivate Order
OrderLine <-- Customer: discountedValue
deactivate Customer
deactivate OrderLine
deactivate Order
@enduml

It is quite hard to follow especially as there are multiple level of nested actviation/deactivation.

What if we express the same thing as the following Python code ?

@napkin.seq_diagram()
def distributed_control(c):
    user = c.object('User')
    order = c.object('Order')
    orderLine = c.object('OrderLine')
    product = c.object('Product')
    customer = c.object('Customer')

    with user:
        with order.calculatePrice():
            with orderLine.calculatePrice():
                product.getPrice('quantity:number')
                with customer.getDiscountedValue(order):
                    order.getBaseValue().ret('value')
                    c.ret('discountedValue')

distributed_control is normal function accepting a context object, c to access APIs. The function defines objects and the control starts with user object, which then calls orderLine.calculatePrice(). Basically, the sequence diagram is expressed as "almost" normal python code.

There are several advantages in using Python instead of using other special syntax language:

• Easy to write/maintain scripts for the correct diagrams
• Many common mistakes are detected as normal Python error. For example, method call to an undefined object will be just normal Python error.(This can be even checked by IDE without running scripts).
• Any Python editor can become sequence diagram editor
• There can be many interesting usages by taking advantage of Python as general language. For example, we can build a library for patterns.

Installation

Install and update using pip

pip install -U napkin

Hello world

Write a simple script called hello.py as follows:

import napkin

@napkin.seq_diagram()
def hello_world(c):
    user = c.object('user')
    world = c.object('world')
    with user:
        world.hello()

Then, the following command will generate hello_world.uml:

$ napkin hello.py

More examples

See example