Private messages should work with names (regular send-message plugin)

[janagoetze]

Sending private messages with the @ syntax (@<user> <message>) should trigger on names. Currently triggers on User IDs (that are not visible in the browser window)

cf.

slurk/slurk/views/static/plugins/send-message.js

Lines 17 to 18 in df284b5

	} else if (text.match("^@")) {
	let user = text.substr(1, text.indexOf(" ") - 1);

slurk/slurk/views/static/plugins/send-message.js

Line 34 in df284b5

submit_private_text(user, message, () => {

slurk/slurk/views/static/js/plugins.js

Lines 132 to 133 in df284b5

	function submit_private_text(receiver, text, resolve, reject) {
	socket.emit('text', { room: self_room, receiver_id: receiver, message: text }, (success, error) => {

[wencke-lm]

I noted the same a while ago. Tim said, that this part of the code was originally only meant for easier debugging.

The most logical thing would probably be to translate the name to id, directly where it is received in send-message.js. The variable user map, is instantiated when displaying the current users of a room.

+    let user_name = text.substr(1, text.indexOf(" ") - 1);
+    let user = "";
+    for (let u in user_map) {
+        if (user_map[u] === user_name) {
+            user = u;
+       }
+    }

Disadvantages when using this variable would be, that it only contains the users inside a room, so when someone tries to address someone in a different room, normally they received an error User <id> is not in this room, but in this way, they would receive the error User "" does not exist. And I think some time in the future one might allow writing messages to users in other rooms, then there would be two places in the code that need updating.

The resolution of the passed receiver happens here:

slurk/slurk/views/chat/events.py

Line 169 in df284b5

receiver = db.query(User).get(receiver_id)

One could try to retrieve by name there instead, but that would affect all private messages, also those sent by bots.

receiver = db.query(User).filter_by(name=receiver_id).one_or_none()

One could also check for both.

  receiver = db.query(User).filter(
      or_(
          User.name == receiver_id,
          User.id == receiver_id
      )
  ).one_or_none()

At the moment names are not restricted to be unique. One would definitely have to change this. Also for the last approach, names should not take the form of ids, so one would have to add the restriction, that names can't only be numbers.

slurk/slurk/models/user.py

Line 13 in df284b5

name = Column(String, nullable=False)