Skip to content

dhmit/chat-lab

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Graffiti Chat Lab

This lab provides a basic working chat application that supports direct messaging and channel-based discussion. You can try out the app here.

The app is built using the Graffiti infrastructure which enables the sending and receiving of messages in realtime. You can find reference material on Graffiti's Vue interface here, which is also outlined below.

We encourage you to read through the code for this repo — there are a lot of comments to explain how everything is working.

Graffiti Overview

Graffiti functions can be referenced with this.$gf in your javascript code and with just $gf in your HTML. The first thing a Graffiti app should do is provide a place for the user to log in using the $gf.toggleLogIn function (see the log in button at the top of the index.html file).

Once the user is logged in, the $gf.me variable will be equal to the user's unique ID - in Graffiti this is called an "actor" to align with the ActivityPub Specification. You will also be able to access functions that create and destroy data, described later in this document: $gf.post and $gf.remove.

Graffiti apps are designed to be both interoperable and extensible. That means that you should be able to send messages from one chat app and receive them in another while also having the freedom to add features to your chat app that might extend or change the data schema that it uses.

To meet both of those design goals, Graffiti apps uses the following work flow:

  • Find a large collection of potentially relevant data.
  • Filter the large data collection for data that your app can use.
  • Process the filtered data to display in your interface.

In this chat app, this work flow is enacted to:

  • Fetch all data in your "inbox" or a particular channel, filter that data for message data, and display those messages.
  • Fetch all data related to a user's identity, filter that data for the user's most recent profile data, and display name of that profile.

We're hoping that you can use your existing knowledge of Vue to take care of the "displaying" step — reading through the template source code should give you a good idea of how it's done. So this documentation is going to go in depth on the fetching and filtering steps. Finally, we'll go over the (simpler!) job of creating, modifying, and destroying data, as well as making private messages.

Fetching Data

Data in Graffiti is organized by "context". A context is essentially just a string (technically a "URI"). When you create data, you must assign that data to one or more contexts. Then you, or others, can query for all data that has been assigned to a particular context.

The results of the context query are streamed in realtime, which is essential for a chat application. You can stream these results manually using the asynchronous generator $gf.objects, however Graffiti provides tooling to get a reactive Vue array which is what we use in the template app.

In the setup() function of the template app, we use useObjects to declare a reactive array called messagesRaw. If we're using channel-based chat, the context we pass to useObjects is simply the name of the channel. If we're using private messagine, the context we pass to useObjects is $gf.me — this context is a good place to store any user-specific data and in this case the context is acting simultaneously as an inbox and outbox of private messages.

Filtering Data

Each unit of Graffiti data is a special JSON object that has the following default properties:

  • published: This is the timestamp that the data was published.
  • updated: This is the timestamp that the data was last updated.
  • id: This is a unique identifier of the data object itself.
  • actor: This is a unique identifier corresponding to the creator of the data. Objects you create will have the same actor value as $gf.me.
  • context: This is an array of all the contexts that the data has been assigned to.

The other object properties are entirely up to the application developer - you! There are many ways you could try to represent a message in a JSON object, but to be interoperable it's best to try and use a standard schema if one exists.

Once you've defined a schema for the data your app uses, you can use Array.filter() or whatever javascript you would like to filter the data that you've fetched. In the template app, we apply filters as Vue computed properties which reactively re-filter whenever the source data changes.

Messages

Messages in the template app follow the ActivityPub Note schema, which is intended for short pieces of text like messages and tweets. All this means is that our messages have a property type that is equal to 'Note' and a property content that stores message text.

Profile

To assign names to users in the template app, we use an ActivityPub Profile. This object has a property type that is equal to 'Profile' and we use the property name to store the user's name. Some apps might use other fields like icon to store a profile image, or summary to share a quick bio about themselves. That's OK! We don't need to filter out objects that have extra fields.

Other Message Types

There are many standard objects, activities, and properties that are already laid out in the ActivityVocabulary spec. If you're adding new functionality to the chat app, we recommend checking if there's something that already fits your use case so that your app will naturally interoperate with other apps that added a similar feature. For example, you could:

Creating and Modifying Data

To add data to the Graffiti server, all you need to do is call $gf.post on the base JSON object want to add. The actor, id, published, and updated fields are added automatically, but you need to manually specify the context(s) that you would like the data to appear in. See the sendMessage method in the source code.

To edit data you can simply modify the object and it will sync with the server. See the saveEditMessage method in the source code.

To delete a message you need to call $gf.remove on it. Be careful this can't be undone! Make sure to protect your users from accidental deletion. See the removeMessage method in the source code.

Note that you can only modify and delete your own data. To "Like" an object you could create a Like activities that references the id of the object you are liking. Similarly, to delete someone else's message you could create a Flag activies that references the id of the object you are trying to delete. Your interface can then react appropriately, either by displaying a like, or removing the message from your display.

Private Messaging

To create a private message, you need to add a bto or bcc property to the object. These properties must have values that are arrays of Graffiti Actor URIs, corresponding to the user's that you want to be able to view the private message. See the sendMessage method in the source code.

If either bto or bcc is not included, the object is public, if one or both fields are empty arrays, the object can only be seen by it's creator. You can use objects with empty arrays, bto: [], to store interoperable application state. For example, to mark which messages the user has read or to keep track of the last chat they had open.

There is no functional difference between the bto and bcc, they simply mirror the semantic meaning of to and cc fields in emails.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • JavaScript 67.4%
  • HTML 32.6%