Support Anova Nano

Question

Support Anova Nano

c3V6a2Vy opened this issue 4 years ago · comments

Borrowed a Anova NANO and found out the library is not working. A quick sniffing reveals change in device_notification_uuid and device_primary_uuid

c3V6a2Vy · Answer 1 · Wed Jun 03 2020 15:42:30 GMT+0800 (China Standard Time)

using the standard discovery can find the device, without needing to press any button. And quite opposite, if you long press the "target temp" button til it blinks the BLE scan won't be able to find the device at all.

from pyanova import pyanova
pa = pyanova.PyAnova(auto_connect=False)
devices = pa.discover(list_all=True, timeout=6)

yields: {'name': None, 'address': u'58:93:D8:9C:51:4A'}
However, connecting with the address directly failed at:

>>> pa.connect_device( {'name': None, 'address': u'58:93:D8:9C:51:4A'})
2020-06-03 00:41:54,678 - INFO - Starting PyAnova BLE adapter
2020-06-03 00:41:55,316 - INFO - Connecting to Anova device: {'name': None, 'address': u'58:93:D8:9C:51:4A'}
2020-06-03 00:41:55,897 - INFO - Connected to: {'name': None, 'address': u'58:93:D8:9C:51:4A'}
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/dist-packages/pyanova-0.1.3-py2.7.egg/pyanova/pyanova.py", line 209, in connect_device
    self._dev.subscribe(notification_uuid, callback=PyAnova.indication_callback, indication=True)
  File "/usr/local/lib/python2.7/dist-packages/pygatt/device.py", line 144, in subscribe
    self._notification_handles(uuid)
  File "/usr/local/lib/python2.7/dist-packages/pygatt/device.py", line 120, in _notification_handles
    value_handle = self.get_handle(uuid)
  File "/usr/local/lib/python2.7/dist-packages/pygatt/device.py", line 208, in get_handle
    raise exceptions.BLEError(message)
pygatt.exceptions.BLEError: No characteristic found matching 0000ffe1-0000-1000-8000-00805f9b34fb

c3V6a2Vy · Answer 2 · Wed Jun 03 2020 17:17:32 GMT+0800 (China Standard Time)

Through pygatt we found the following characteristics:

import pygatt
adapter = pygatt.GATTToolBackend()
adapter.start()
dev = adapter.connect(u'58:93:D8:9C:51:4A')
char_list = dev.discover_characteristics()
for c,v in char_list.items(): print(v.__dict__)

yield

...
{'descriptors': {}, 'handle': 14, 'uuid': '0e140002-0af1-4582-a242-773e63054c68'}
{'descriptors': {}, 'handle': 18, 'uuid': '0e140003-0af1-4582-a242-773e63054c68'}
{'descriptors': {}, 'handle': 5, 'uuid': '00002a01-0000-1000-8000-00805f9b34fb'}
{'descriptors': {}, 'handle': 11, 'uuid': '0e140001-0af1-4582-a242-773e63054c68'}
{'descriptors': {}, 'handle': 7, 'uuid': '00002a04-0000-1000-8000-00805f9b34fb'}
{'descriptors': {}, 'handle': 3, 'uuid': '00002a00-0000-1000-8000-00805f9b34fb'}

which pretty much aligns with the one we discovered in Windows.

0e140001-0af1-4582-a242-773e63054c68 seemed to be the primary characteristics since it's user description was "tx-char" with handle 0xA . 0e140003-0af1-4582-a242-773e63054c68 was marked as "rx-char" so we tried this and the asyn-char as well for receiving. None of the combination worked.

c3V6a2Vy · Answer 3 · Wed Jun 03 2020 17:19:53 GMT+0800 (China Standard Time)

Will need to get my old Anova back to do more analysis on what that was like. And though unlikely, it could also be that the commands had changed.

garret · Answer 4 · Tue Jul 14 2020 21:23:00 GMT+0800 (China Standard Time)

Hey @c3V6a2Vy did you have any luck in sorting out how to control the Nano? I have one and please let me know if you need some help for testing stuff.

c3V6a2Vy · Answer 5 · Sun Sep 06 2020 07:58:09 GMT+0800 (China Standard Time)

Writing to handle 18 in seemed to 0e140002-0af1-4582-a242-773e63054c68 be not working...

handle = 14
dev.char_write_handle(handle , bytearray("%s\r"%('status'), 'utf8'))

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/device.py", line 17, in wrapper
    return func(self, *args, **kwargs)
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/device.py", line 48, in char_write_handle
    self._backend.char_write_handle(self, handle, *args, **kwargs)
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/gatttool.py", line 50, in wrapper
    return func(self, *args, **kwargs)
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/gatttool.py", line 574, in char_write_handle
    self.sendline(cmd)
  File "/usr/lib/python3.4/contextlib.py", line 66, in __exit__
    next(self.gen)
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/gatttool.py", line 191, in event
    self.wait(event, timeout)
  File "/usr/local/lib/python3.4/dist-packages/pygatt-4.0.5-py3.4.egg/pygatt/backends/gatttool/gatttool.py", line 157, in wait
    raise NotificationTimeout()
pygatt.exceptions.NotificationTimeout: None

Actually, subscription attempts to 0e140002-0af1-4582-a242-773e63054c68 would failed as well with indication turned on:

dev.subscribe("0e140002-0af1-4582-a242-773e63054c68", indication=True)

c3V6a2Vy · Answer 6 · Sun Sep 06 2020 08:03:09 GMT+0800 (China Standard Time)

Hey @c3V6a2Vy did you have any luck in sorting out how to control the Nano? I have one and please let me know if you need some help for testing stuff.

thanks mate, sorry for the late reply, didn't see your message nor have time to make more progress. I am going through the some details... stay tuned.

c3V6a2Vy · Answer 7 · Sun Sep 06 2020 13:29:36 GMT+0800 (China Standard Time)

ok you can write to handle 11 (character: 0e140001-0af1-4582-a242-773e63054c68) but nothing happened.

 dev.char_write_handle(11, bytearray("%s\r"%('set unit c'), 'utf8'))

garret · Answer 8 · Mon Sep 07 2020 16:57:13 GMT+0800 (China Standard Time)

@c3V6a2Vy sorry I am not expert at all in the issue as I am not a programmer. Are you perhaps making any progress? I have a nano so if you need a double check I can help but I would need a detailed guide... :)

c3V6a2Vy · Answer 9 · Mon Sep 07 2020 17:09:26 GMT+0800 (China Standard Time)

From some secret details I managed to trace down the sendTemperature command and found it's actually sending Protobuf msg bytes now instead of the string protocol used in the first generation.

Good news: we (sort of) know what comprises a valid command.
Not so good news: we have to find out its proto definition and have to port it to this library. I cannot find the original "proto" definitions in the package (which makes sense), so need to find a way to convert the generated codes back to proto definitions. I think it's doable just takes time..

garret · Answer 10 · Wed Nov 11 2020 01:39:39 GMT+0800 (China Standard Time)

I guess unfortunately there are still no updates on the nano support? :(

Myles Keller · Answer 11 · Fri Mar 05 2021 06:59:28 GMT+0800 (China Standard Time)

From some secret details I managed to trace down the sendTemperature command and found it's actually sending Protobuf msg bytes now instead of the string protocol used in the first generation.

Good news: we (sort of) know what comprises a valid command.
Not so good news: we have to find out its proto definition and have to port it to this library. I cannot find the original "proto" definitions in the package (which makes sense), so need to find a way to convert the generated codes back to proto definitions. I think it's doable just takes time..

could you describe in a little more detail how you're going about converting the definitions?
i've been following this issue since realizing that my Nano wouldn't work with existing packages, and have also decompiled the anova app and started poking around.
depending on the process you're using to convert the proto definitions, i'm hoping that maybe Xposed Framework could access the information you need directly since it hooks into an app and can modify its flow at runtime.

c3V6a2Vy · Answer 12 · Wed Mar 10 2021 06:14:30 GMT+0800 (China Standard Time)

@myleskeller @garret
sorry for the lack of updates.. was busy with other projects.

i was combing through the decompiled codes from the app code, confirmed with the uuid and handle, and i was able to connect to the device via BT.

the challenge was the content being different from the the 1st gen device (which is basically byte array of string commands), it's sending protobuf encoded messages to the device. Since this lib is in Python but the app was in Java, I didn't really know how to invoke the compiled protobuf encoding code in the Java (via Xposed framework maybe, looking into that).

So, the more plausible way (at the time from my last update) was to decode (reconstruct) the original protobuf definition and compile it again into Python, which it seemed harder than I thought in my later experiment. Any suggestion here?

c3V6a2Vy · Answer 13 · Mon Mar 15 2021 09:57:56 GMT+0800 (China Standard Time)

will give this guy a try: https://github.com/marin-m/pbtk

Sean Jackson · Answer 14 · Sat Mar 20 2021 00:09:30 GMT+0800 (China Standard Time)

I've recently acquired an Anova Pro, and have followed the steps in this issue, as it appears to be the exact same problem. Did you manage to get the binary/hex of the protobuf messages being sent? Presumably if the above doesn't work, we could use something like Online Protobuf Decoder and determine the field names via process of elimination / looking at the decompiled app code.

Daniel Engelke · Answer 15 · Mon Jan 03 2022 05:10:15 GMT+0800 (China Standard Time)

I've done some reverse engineering of the decompiled app and started the project https://github.com/dengelke/node-sous-vide which is still a WIP but is working with the Nano. Although not tested with the Pro there it shouldn't require much or any work to get running.

c3V6a2Vy · Answer 16 · Mon Jan 03 2022 14:55:20 GMT+0800 (China Standard Time)

I've done some reverse engineering of the decompiled app and started the project https://github.com/dengelke/node-sous-vide which is still a WIP but is working with the Nano. Although not tested with the Pro there it shouldn't require much or any work to get running.

Thanks @dengelke ! Are 544.js and 563.js decompiled codes from the Android app and then later translated into JS? I had the similar code from Java but was struggling to reverse engineer that into Python. Your decompiled JS code will help a lot, I will try to translate the JS into Py with something like https://github.com/PiotrDabkowski/Js2Py

Daniel Engelke · Answer 17 · Mon Jan 03 2022 17:19:04 GMT+0800 (China Standard Time)

Will write out a guide on how I decompiled it - the android app itself is written in JS using React Native so these are just modules from the app which were modified to work in a Node.js environment instead of React Native.

Currently refactoring the code to get rid of 544.js and 563.js using https://www.npmjs.com/package/protobufjs which will give the original protobuf definitions and a lot less code to worry about.

Not sure about converting the code from Js to Py, would suggest you play around with it and if you have a Nano or Pro try and use it. Big changes you need to look at for:

There are now two characteristics you need to use, one to write commands to, one to read responses from look at src/sendDeviceCommand.ts to see how this works compared to the original Anova which only had one bluetooth channel
If you have a look at src/Command.ts one of the device commands to see the current temperature is [1, 2, 4, 0] which just needs to be converted to a buffer and sent to the write characteristic by Buffer.from(commandArray) you will then get a response on the read characteristic, which needs to be converted to an array then decoded using the protobuf library

Good luck and let me know if you have any questions

c3V6a2Vy · Answer 18 · Tue Jan 04 2022 03:34:04 GMT+0800 (China Standard Time)

Currently refactoring the code to get rid of 544.js and 563.js using https://www.npmjs.com/package/protobufjs which will give the original protobuf definitions and a lot less code to worry about.

https://github.com/dengelke/node-sous-vide/blob/master/src/proto/messages.proto

With this I can compile the Python codes directly :)

Daniel Engelke · Answer 19 · Tue Jan 04 2022 04:15:36 GMT+0800 (China Standard Time)

@c3V6a2Vy glad the proto definitions could be useful, these aren't the complete definitions - just the ones we required to get this package to work nicely.

Now having a better look at the decompiled code I have, seems the Pro although having a similar Bluetooth configuration has a completely different set of commands so although this package will let you connect to it, the commands won't work :/

Daniel Engelke · Answer 20 · Tue Jan 04 2022 04:35:03 GMT+0800 (China Standard Time)

For detailed decompilation instructions for the Android APK please give https://github.com/dengelke/node-sous-vide/blob/master/Decompilation.md a read

Lezgin Bakircioglu · Answer 21 · Fri May 26 2023 04:37:52 GMT+0800 (China Standard Time)

I would love to get my anova nano to work, any progress on this ? :-) Big thanks for the contribution so far to get anova working!

Mitja Mueller-Jend · Answer 22 · Fri Jan 19 2024 13:55:47 GMT+0800 (China Standard Time)

I think I managed to translate @dengelke's node-sous-vide to python: https://github.com/filmkorn/pyanova-nano

I've never touched any BLE client code nor asyncio code, so I'd love if someone could give this a review?

Sadly I could not get pyanova to run on Windows, so I basically started from scratch using bleak instead of pygatt/bluez. That doesn't mean the logic can't be integrated here.

Daniel Engelke · Answer 23 · Sat Jan 20 2024 10:20:33 GMT+0800 (China Standard Time)

I think I managed to translate @dengelke's node-sous-vide to python: https://github.com/filmkorn/pyanova-nano

I've never touched any BLE client code nor asyncio code, so I'd love if someone could give this a review?

Sadly I could not get pyanova to run on Windows, so I basically started from scratch using bleak instead of pygatt/bluez. That doesn't mean the logic can't be integrated here.

Nice work @filmkorn!

c3V6a2Vy · Answer 24 · Mon Jan 22 2024 14:30:43 GMT+0800 (China Standard Time)

nice one! @filmkorn . taking a look. i borrowed friend's nano yesterday hopefully i can run some tests on it too.

works fine for me on windows 11 build 22621, Python 3.12.1

asyncio.run(print_device_sensors())
SensorValues(water_temp=23.2, water_temp_units='C', heater_temp=22.0, heater_temp_units='C', triac_temp=23.0, triac_temp_units='C', internal_temp=29.0, internal_temp_units='C', water_low=False, water_leak=False, motor_speed=0)

bleak seemed to be much more cross-platform friendly and easy to operate than bluez, the pygatt/bluez stuffs was used mainly because it's the only thing I can get it to compile on the SoC that I used to use, now that I lost that SoC too so I think it's time to move on to something nicer.