Added tutorial on ThreadedSubscriptionProducer

This makes it easier to see how to use sensors and other external things with RTCBot.

Also fixed a couple little bugs in closing data channels that are being used to send data.

Author: dkumor

README.md

@@ -74,7 +74,7 @@ async def index(request):
     with open("index.html", "r") as f:
         return web.Response(content_type="text/html", text=f.read())
 
-async def cleanup(app):
+async def cleanup(app=None):
     await conn.close()
     camera.close()
 

docs/installing.rst

@@ -22,16 +22,16 @@ Then, you can complete the installation with pip::
 
     sudo pip3 install rtcbot
 
+.. warning::
+    You might need to reboot your Pi for RTCBot to work! If rtcbot freezes on import, it means that you need to start PulseAudio.
+
 .. note::
     It is recommended that you use the Pi 4 with RTCBot. While it was tested to work down to the Raspberry Pi 3B, it was observed to have
     extra latency, since the CPU had difficulty keeping up with encoding the video stream while processing controller input.
     This is because RTCBot currently cannot take advantage of the Pi's hardware acceleration, 
     meaning that all video encoding is done in software.
 
 .. note::
-    You might need to reboot your Pi for RTCBot to work!
-
-.. warning::
     These instructions were made with reference to Raspbian Buster.
     While the library *does* work on Raspbian Stretch,
     you'll need to install aiohttp through pip, and avoid installing opencv.
@@ -50,8 +50,8 @@ Then, you can complete the installation with pip::
 
     sudo pip3 install rtcbot
 
-.. note::
-    You might need to reboot, or manually start pulseaudio if it was not previously installed.
+.. warning::
+    You might need to reboot, or manually start PulseAudio if it was not previously installed. If RTCBot freezes on import, it means that PulseAudio is not running.
 
 Mac
 +++++++++++

docs/javascript.rst

@@ -69,7 +69,7 @@ Next, to establish the connection with Python, you include the Python counterpar
         return web.json_response(response)
 
 
-    async def cleanup(app):
+    async def cleanup(app=None):
         if conn is not None:
             await conn.close()
 

examples/index.rst

@@ -19,3 +19,4 @@ The full code for each of the tutorials can be seen in the `examples directory <
     remotecontrol/README.md
     mobile/README.md
     offloading/README.md
+    threads/README.md

examples/mobile/README.md

@@ -125,7 +125,7 @@ async def index(request):
     </html>
     """)
 
-async def cleanup(app):
+async def cleanup(app=None):
     global ws
     if ws is not None:
         c = ws.close()
@@ -171,10 +171,9 @@ finally:
 
 With these two pieces of code, you first start the server, then start the robot, and finally open `http://localhost:8080` in the browser to view a video stream coming directly from the robot, even if the robot has an unknown IP.
 
-
 ## rtcbot.dev
 
-The above example requires you to have your own internet-accessible server at a known IP address to set up the connection, if your remote code is not on your local network. The server's only real purpose is to help *establish* a connection - once the connection is established, it does not do anything.
+The above example requires you to have your own internet-accessible server at a known IP address to set up the connection, if your remote code is not on your local network. The server's only real purpose is to help _establish_ a connection - once the connection is established, it does not do anything.
 
 For this reason, I am hosting a free testing server online at `https://rtcbot.dev` that performs the equivalent of the following operation from the above server code:
 
@@ -226,11 +225,12 @@ and the local browser's connection code becomes:
 
 ```js
 let response = await fetch("https://rtcbot.dev/myRandomSequence11", {
-    method: "POST",
-    cache: "no-cache",
-    body: JSON.stringify(offer)
+  method: "POST",
+  cache: "no-cache",
+  body: JSON.stringify(offer),
 });
 ```
+
 With `rtcbot.dev`, you no longer need your local server code to run websockets or a connection service. Its only purpose is to give the browser the html and javascript necessary to establish a connection. We will get rid of the browser entirely in the next tutorial.
 
 ## If it doesn't work over 4G
@@ -255,6 +255,7 @@ There are two options through which to setup a TURN server: [coTURN](https://git
 The Pion server is easy to set up on Windows,Mac and Linux - all you need to do is [download the executable](https://github.com/pion/turn/releases/tag/1.0.3), and run it from the command line as shown.
 
 **Linux/Mac**:
+
 ```bash
 chmod +x ./simple-turn-linux-amd64 # allow executing the downloaded file
 export USERS='myusername=mypassword'
@@ -264,6 +265,7 @@ export UDP_PORT=3478
 ```
 
 **Windows**: You can run the following from powershell:
+
 ```powershell
 $env:USERS = "myusername=mypassword"
 $env:REALM = "my.server.ip"
@@ -272,6 +274,7 @@ $env:UDP_PORT = 3478
 ```
 
 With the Pion server running, you will need to let both Python and Javascript know about it when creating your `RTCConnection`:
+
 ```python
 from aiortc import RTCConfiguration, RTCIceServer
 
@@ -286,7 +289,7 @@ myConnection = RTCConnection(rtcConfiguration=RTCConfiguration([
 var conn = new rtcbot.RTCConnection(true, {
                 iceServers:[
                     { urls: ["stun:stun.l.google.com:19302"] },
-                    { urls: "turn:my.server.ip:3478?transport=udp", 
+                    { urls: "turn:my.server.ip:3478?transport=udp",
                         username: "myusername", credential: "mypassword", },
                 ]);
 ```
@@ -296,6 +299,7 @@ var conn = new rtcbot.RTCConnection(true, {
 Setting up a coTURN server takes a bit more work and is only supported on Linux and Mac. The following steps will assume a Linux system running Ubuntu.
 
 Install coTURN and stop the coTURN service to modify config files with
+
 ```bash
 sudo apt install coturn
 sudo systemctl stop coturn
@@ -304,6 +308,7 @@ sudo systemctl stop coturn
 Edit the file `/etc/default/coturn` by uncommenting the line `TURNSERVER_ENABLED=1`. This will allow coTURN to start in daemon mode on boot.
 
 Edit another file `/etc/turnserver.conf` and add the following lines. Be sure to put your system's public facing IP address in place of `<PUBLIC_NETWORK_IP>`, your domain name in place of `<DOMAIN>`, and your own credentials in place of `<USERNAME>` and `<PASSWORD>`.
+
 ```
 listening-port=3478
 tls-listening-port=5349
@@ -323,13 +328,15 @@ lt-cred-mech
 ```
 
 Restart the coTURN service, check that it's running, and reboot.
+
 ```bash
 sudo systemctl start coturn
 sudo systemctl status coturn
 sudo reboot
 ```
 
 With the coTURN server running, you will need to let both Python and Javascript know about it when creating your `RTCConnection`:
+
 ```python
 from aiortc import RTCConfiguration, RTCIceServer
 
@@ -344,7 +351,7 @@ myConnection = RTCConnection(rtcConfiguration=RTCConfiguration([
 var conn = new rtcbot.RTCConnection(true, {
                 iceServers:[
                     { urls: ["stun:stun.l.google.com:19302"] },
-                    { urls: "turn:<PUBLIC_NETWORK_IP:3478?transport=udp", 
+                    { urls: "turn:<PUBLIC_NETWORK_IP:3478?transport=udp",
                         username: "myusername", credential: "mypassword", },
                 ]);
 ```

examples/mobile/server.py

@@ -84,7 +84,7 @@ async def index(request):
     )
 
 
-async def cleanup(app):
+async def cleanup(app=None):
     global ws
     if ws is not None:
         c = ws.close()

examples/offloading/README.md

@@ -5,9 +5,7 @@ and an average desktop.
 
 The ideal situation would be if you could strap an entire desktop to your robot. With RTCBot, we can do the next best thing: we can stream the robot's inputs to a desktop, which can then perform computation, and send back commands.
 
-In this tutorial, we will go back to a single file for both server and robot for simplicitly. The tutorial is split into two parts. In the first part, we set up a connection to the robot from Python, allowing you to control the robot with an xbox controller without a browser.
-
-The second part of the tutorial adds a neural network into the mix, which tries to imitate the xbox controller's controls.
+In this tutorial, we will go back to a single file for both server and robot for simplicitly. We set up a connection to the robot from Python, allowing you to control the robot with an xbox controller without a browser.
 
 ```eval_rst
 .. note::
@@ -130,4 +128,3 @@ Control message: {'timestamp': 1553379217.004892, 'code': 'ABS_X', 'state': -125
 .. warning::
    The output for the `Gamepad` object is currently different in Javascript and in Python. Make sure you don't mix them up!
 ```
-

examples/offloading/robot.py

@@ -23,7 +23,7 @@ async def connect(request):
     return web.json_response(serverResponse)
 
 
-async def cleanup(app):
+async def cleanup(app=None):
     await conn.close()
     cam.close()
 

examples/remotecontrol/README.md

@@ -68,7 +68,7 @@ async def index(request):
     </html>
     """)
 
-async def cleanup(app):
+async def cleanup(app=None):
     await conn.close()
 
 app = web.Application()
@@ -152,7 +152,7 @@ We now add keyboard support. This is done with the `rtcbot.Keyboard` javascript
      )
 
 
- async def cleanup(app):
+ async def cleanup(app=None):
      await conn.close()
 
  app = web.Application()
@@ -349,7 +349,7 @@ async def index(request):
     </html>
     """)
 
-async def cleanup(app):
+async def cleanup(app=None):
     await conn.close()
 
 app = web.Application()

examples/remotecontrol/gamepad.py

@@ -72,7 +72,7 @@ async def index(request):
     )
 
 
-async def cleanup(app):
+async def cleanup(app=None):
     await conn.close()