A good FOSS project deserves a good README. ;)

I would suggest:

• Getting rid of '...'
• Fixing the Markdown/layout

Message.seen_by and .received_by fields available in previous chat version are missing from current implementation.

Refs #45 .

Pay attention to this message from the old code base:

    # TODO(alexis): consider through a model to record timestamps when the message is seen / sent to implement
    # whatsapp like double checkmarks

We need to add reactions on messages like 👍, 👎, ❤️, etc.

• Create a model schema for reactions
• Add available reactions to library settings
• Add API method to add/remove reaction for the message
• Add live updates with message's reactions for websocket listeners.
• Add reactions as nested field for Message API.

On previous chat version we had much more features like:

• Mute/unmute room
• Replies
• Reactions, etc.

We reworked an architecture since then and dropped a lot of features.

The task is to research an old chat version, create full list of missing features (as separate GitHub tickets) and implement them.

As an example, I created a feature for reactions #44

We need to keep architecture clean and performant. All live chat changes for new features (reactions, replies, etc.) should be sent via websocket.

Implement a feature allowing room admins or the room creator to send invitations to users to join a private room.

Add room categories to better organize the public rooms, and allow users to filter the room list based on these categories.

Hi I am checking api methods and models on https://github.com/djangoflow/flutter-djangoflow/tree/main/packages/djangoflow_openapi

I see after the latest refactor few things went missing.

Critical ones:

• Room id should be String instead of int . Currently in every API we use int type but for some reason GET /api/v1/chat/rooms/{id}/ uses String. But all should be String type.
• No API get the users that are in a room(need it to display who sent which message, and which one user is me, previously there used to be RoomUser model that used to give information like id, name, avatar etc of the user)
• GET /api/v1/chat/rooms/{room_id}/messages/ does not support limit , offset parameter passing for pagination

Less critical ones:

• #53
• #54
• #55

Allow users to edit or delete their messages within a certain time frame after they’ve sent them. Include an indicator showing that a message has been edited.

Give room admins the ability to pin important messages to the top of the chat for all users to see.

Docstring for UserChat reads: This model stores attributes that should be visible across all rooms for a user.

The current name is suggestive to something more akin to a chat session whereas, quite literally, it's chat-related attributes of a user. Given that the model is already namespaced to df_chat, I would suggest renaming to UserAttributes (and use import as in external contexts) or UserState (as that's how it's currently used).

I think a change like this will significantly increase comprehension of the code and data structures.

Since users are associated with rooms via the RoomUser through-model, having the users field on the Room model is unnecessary and redundant.

Changes made need to take into consideration the various uses of the users field on the Room model (if any).

Instead of pip installing an individual requirements.txt file it is better to make folder of with different requirements.txt file and and call them in one requirements.txt file. So that contributors/users can run a single pip install requirements.txt file.

In the current approach, in order to create a message in a room, a user should make a HTTP POST request. And then the message is propagated to all the listeners in that room.

But, chatting is done in real-time. Users expect to send and receive messages real quick. An HTTP connection is not stateful and takes some time to connect. They disconnect immediately once the request is fulfilled.

On the other hand, a Websocket connection is stateful - the connection is live until either party terminates it.

So, we could communicate in real time by using Websockets instead of HTTP and should allow users to create messages via Websocket instead of HTTP.

Trying to connect without providing a token results in an error "ValueError: 'AnonymousUser' value must be a positive integer or a valid Hashids string."

I think, we should return a permission denied message via the websocket and gracefully disconnect the user.

• Modify the WebSocket URL pattern to handle all chat rooms with a single WebSocket URL.
• Update the WebSocket consumer to manage multiple rooms using the new URL pattern.
• Ensure that the chat functionality continues to work correctly after the refactoring.

Add the ability to search for messages within a room based on keywords, user, or date range.

Create user profiles with customizable avatars, display names, and a short bio. Users should be able to view each other’s profiles by clicking on their names in the chat interface.

I added chat models for the new schema. Please, check API section in README and start on implementing enpoints

MessageImage model available in previous chat version is missing from current implementation.

Refs #45

Either a security issue or an incorrect comment. In the former case, it's not a TODO!

Add the ability to search for messages within a room based on keywords, user, or date range.

The following fields available in previous chat version seem to be missing from current implementation.

• creator - user which created the room
• is_public - may be implemented via ChatType
• image
• description
• muted_by
• admins

Refs #45 .