Skip to content
This repository has been archived by the owner on Jul 8, 2023. It is now read-only.

Latest commit

 

History

History
121 lines (98 loc) · 3.43 KB

integration-guide.adoc

File metadata and controls

121 lines (98 loc) · 3.43 KB

Integration Guide

Audience

This document is written for those want to integrate ZenUML to their applications. If you are an end user, please read tutorials; if your would like to contribute, please read contributor’s guide.

Integration

An important goal of this project is to be easily integrated into other projects. There are mainly two ways to do this.

As a library

ZenUML Core exposes a simple API for integration:

interface IZenUml {
  get code(): string | undefined;
  get theme(): string | undefined;
  // Resolve after rendering is finished.
  render: (code: string | undefined, theme: string | undefined) => Promise<IZenUml>
}

// client code
import ZenUml from '@ZenUML/core'
const zenUml = new ZenUml(el)
await zenUml.render('A.method', 'theme-blue')

When rendering is finished

When rendering is finished, the render method will return a promise that resolves to the instance of IZenUml. The html properties of the instance will be updated to the rendered HTML.

Note

Vue’s reactive system works against the data, props and computed properties of a component. This means a parent component is not necessarily be notified when a child component is mounted or updated. So this means we can not use the mounted or updated hook on LifelineLayer to emit rendered event.

It is tricky to know when the rendering is finished. The Lifeline components used $nextTick in their mounted and updated hooks to set the top when a creation message is sent to it. We have to wait until all those $nextTick calls are finished.

Note

It seems that (to be confirmed) the $nextTick calls are queued and executed in ONE tick. This makes it easier to control the timing when to resolve the render promise. We only need to wait ONE $nextTick call to be finished.

The diagram is rendered in the following steps:

  1. The lifeline layer;

  2. The message layer;

  3. Update lifeline layer with setTimeout for creation messages.

As an iframe

You can embed the ZenUML core renderer as an application within another app, where you store the diagram data in the host app. It takes around 5 minutes to get a basic example running.

Quick test

To test this out open https://embed.zenuml.com/embed.html. In the developer console, type in the following command.

window.postMessage( {action: 'eval', args: { code: 'ZenUML.Hello' }})

The protocol

The protocol is a simple JSON object with the following fields.

{
  "action": "eval",
  "args": {
    "code": "ZenUML.Hello",
    "style": "#diagram { background-color: red; }",
    "theme": "blue",
    "css": "https://github.com/abruzzi/zenuml-css-overrides/blob/master/zenuml-override.css"
  }
}

Example

<iframe src="https://embed.zenuml.com/embed.html" id="zenuml-iframe"
        style="width: 100%; height: 100%; border: none;">

</iframe>
<script>
  const iframe = document.getElementById('zenuml-iframe');
  const message = {
    action: 'eval',
    args: {
      code: 'ZenUML.Hello',
      style: '#diagram { background-color: red; }',
      theme: 'blue',
      css: ''
    }
  };
  setTimeout(() => {
    iframe.contentWindow.postMessage(message, '*');
  }, 1000);
</script>
document.getElementsByTagName('iframe')[0] // get iframe
    .contentWindow  // get target window
    .postMessage({action: "eval", args: { code: 'A.m' }})