reconnection supervisor command

4 files changed, 105 insertions, 19 deletions

diff --git a/TODO.md b/TODO.md
index 48e0556..47c86d9 100644
--- a/TODO.md
+++ b/TODO.md
@@ -2,22 +2,36 @@
 
 ## next
 
-ci/cd: doc generation
-feature: error handling on stream according to rfc6120
-docs: jid
-docs: jabber
-docs: starttls
-docs: sasl
-docs: resource binding
-feature: starttls
-feature: sasl
-feature: resource binding
-
-## in progress
-
-feature: logging
+feat(luz): everything in rfc6120 and rfc6121
+  feat(luz): handle_online
+  feat(luz): handle_offline
+  feat(luz): handle_stanza
+  feat(luz): database
+  feat(luz): error handling on stream according to rfc6120
+  feat(luz): send message
+  feat(luz): receive message
+  feat(luz): retreive messages stored in database
+  feat(luz): get roster (online and offline)
+  feat(luz): set roster
+  feat(luz): reconnect supervisorcommand
+feat: thiserror everywhere
+feat(luz): proper stanza ids
+test: proper tests
+ci: doc generation
+docs(jid): jid
+feat(peanuts): derive macros for IntoElement and FromElement
+docs(jabber): connection, starttls, sasl, binding, bound_stream, etc.
+feat: proper logging for everything basically
+feat(luz): passwordprovider trait, to avoid storing password in struct
+feat(luz): auto-reconnect state stored in luz actor, for if e.g. server shut down
+refactor(luz): dealing properly with all the joinsets and joinhandles
+feat(peanuts): some kind of way to configure the reader and writer to log the raw xml written to the stream, probably by having a method that allows you to add a log writer to them. will need to investigate some kind of log namespacing.
+feat(jabber): storing resource within the bound_stream connection
 
 ## done
 
+feature: starttls
+feature: sasl
+feature: resource binding
 feature: jabber client connection
 feature: jid
diff --git a/luz/src/connection/mod.rs b/luz/src/connection/mod.rs
index f8cf18b..d05eb62 100644
--- a/luz/src/connection/mod.rs
+++ b/luz/src/connection/mod.rs
@@ -46,7 +46,21 @@ pub enum SupervisorCommand {
     Disconnect,
     // for if there was a stream error, require to reconnect
     // couldn't stream errors just cause a crash? lol
-    Reconnect,
+    Reconnect(State),
+}
+
+pub enum State {
+    Write(mpsc::Receiver<WriteMessage>),
+    Read(
+        (
+            SqlitePool,
+            mpsc::Sender<UpdateMessage>,
+            tokio::task::JoinSet<()>,
+            mpsc::Sender<SupervisorCommand>,
+            WriteHandle,
+            Arc<Mutex<HashMap<String, oneshot::Sender<Result<Stanza, Error>>>>>,
+        ),
+    ),
 }
 
 impl Supervisor {
@@ -101,10 +115,70 @@ impl Supervisor {
                             }
                             break;
                         },
-                        SupervisorCommand::Reconnect => {
+                        SupervisorCommand::Reconnect(state) => {
                             // TODO: please omfg
                             // send abort to read stream, as already done, consider
-                            todo!()
+                            let (read_state, mut write_state);
+                            match state {
+                                // TODO: proper state things for read and write thread
+                                State::Write(receiver) => {
+                                    write_state = receiver;
+                                    let (send, recv) = oneshot::channel();
+                                    let _ = self.reader_handle.send(ReadControl::Abort(send)).await;
+                                    if let Ok(state) = recv.await {
+                                        read_state = state;
+                                    } else {
+                                        break
+                                    }
+                                },
+                                State::Read(read) => {
+                                    read_state = read;
+                                    let (send, recv) = oneshot::channel();
+                                    let _ = self.writer_handle.send(WriteControl::Abort(send)).await;
+                                    // TODO: need a tokio select, in case the state arrives from somewhere else
+                                    if let Ok(state) = recv.await {
+                                        write_state = state;
+                                    } else {
+                                        break
+                                    }
+                                },
+                            }
+
+                            let mut jid = self.jid.lock().await;
+                            let mut domain = jid.domainpart.clone();
+                            let connection = jabber::connect_and_login(&mut jid, &*self.password, &mut domain).await;
+                            match connection {
+                                Ok(c) => {
+
+                                    let (read, write) = c.split();
+                                    let (send, recv) = oneshot::channel();
+                                    self.writer_crash = recv;
+                                    self.writer_handle =
+                                        WriteControlHandle::reconnect(write, send, write_state);
+                                    let (send, recv) = oneshot::channel();
+                                    self.reader_crash = recv;
+                                    let (db, update_sender, tasks, supervisor_command, write_sender, pending_iqs) = read_state;
+                                    self.reader_handle = ReadControlHandle::reconnect(
+                                        read,
+                                        send,
+                                        db,
+                                        update_sender,
+                                        supervisor_command,
+                                        write_sender,
+                                        tasks,
+                                        pending_iqs,
+                                    );
+                                },
+                                Err(e) => {
+                                    // if reconnection failure, respond to all current write messages with lost connection error. the received processes should complete themselves.
+                                    write_state.close();
+                                    while let Some(msg) = write_state.recv().await {
+                                        let _ = msg.respond_to.send(Err(Error::LostConnection));
+                                    }
+                                    let _ = self.sender.send(UpdateMessage::Error(e.into())).await;
+                                    break;
+                                },
+                            }
                         },
                     }
                 },
diff --git a/luz/src/connection/read.rs b/luz/src/connection/read.rs
index 3c61780..c2828ad 100644
--- a/luz/src/connection/read.rs
+++ b/luz/src/connection/read.rs
@@ -21,7 +21,6 @@ use super::{
 };
 
 pub struct Read {
-    // TODO: place iq hashmap here
     control_receiver: mpsc::Receiver<ReadControl>,
     stream: BoundJabberReader<Tls>,
     on_crash: oneshot::Sender<(
diff --git a/luz/src/lib.rs b/luz/src/lib.rs
index 4d36a81..4c5a841 100644
--- a/luz/src/lib.rs
+++ b/luz/src/lib.rs
@@ -138,7 +138,6 @@ impl Luz {
                                     self.jid.clone(),
                                     self.db.clone(),
                                     self.sender.clone(),
-                                    // TODO: iq hashmap
                                     self.pending_iqs.clone()
                                 )),
                                 None => self.tasks.spawn(msg.handle_offline(