gerasdf / esp8266 Goto Github PK

Sometimes devices become unresponsive, find a way to automatically recover from that.

ESP8266 Watchdogs in Arduino

Reverse Engineering EPS watchdogs

only authenticated users should be able to do it.
also enable WiFi OTA if available in the libraries

If on first startup (or missconfiguration), the user opens a specific AP (ToI:Love) the device connects to it, and uses it for accessing internet.

This could be used for automatic configuration, see #29

Think about it:

Today each call to getUpdates() only retrieves 1 update (HANDLE_MESSAGES = 1).
This makes the bot have a longer response time, since it's only doing one getUpdates() pero loop iteration.

Using getPendingUpdatesCount()(gerasdf/Universal-Arduino-Telegram-Bot#7) it's possible to bring them all looping. This is preferred to just looping, specially when using long-polls

Commands must be directed to the particular device, otherwise they won't be answered

Some possibilities to identify messages:

• Accept commands which are a reply_to_message to the last message sent by the device.
• Accept commands which are a reply_to_message to a message starting with the device's name
• Accept commands which are a callback_query.message.message_id from the last inline keyboard sent by the device (see #6 )

show all available commands and current status.
automatically call it on startup

When in a group, commands are automatically competed with @botName, as in ʼ/status@toi_botʼ. This as to be supported

Commands to reconfigure the device?
/config token blah and /config name blah

or

/token blah and /name blah ?

Or maybe via a webpage?

If the token is wrong the user may completely lose control of the bot.

Implement a confirmation mechanism over the new bot token.

• receive /settoken
• ask for confirmation over the new bot token
• if confirmation is received, keep new token and make it persistent
• if no confirmation is received, keep the old token and discard the new

dupplicate of #7 (different approach?)

If all devices use a single Telegram bot to communicate, it's not clear how the user can talk to them independently.

It may be possible to filter the incoming messages by using the name of the devices in the messages, such as changing

polarity

for

caldera_1 polarity

and then even shortening the names.

But I'm not sure it's possible, as if one device pops the message from Telegram, the other devices won't see it. And if devices just peek the messages (is it even possible?) who will finally pop it and remove it from the list?

so maybe we need one different Telegram Bot per device? or is there anything like telegram-users-for-devices?

if /settoken doesn't clean updates, and the device is the only one in the bot, the /settoken command will persist and will be again effective on next reconnection to the same token.

Specially when coming back from a timeout. It's kind of needed or it otherwise enters in a bouncing loop going form one to the other.

• token
• owner id
• name
• wifi password
• polarity

sometimes the module reboots after reconnecting to WiFi

[321] numNewMessages: 0 last: 888666378 mem: 30960 - 31 - 20312
[321] .[322] .[323] .[324] .[325] .[326] .[327] .[328] .[329] .[330] .[331] .[332] .[333] .[334] .[335] .[336] .[337] .[338] .[339] .[340] .[341] .[342] .[343] .[344] .[345] .[346] .[347] .[348] .[349] .[350] .[351] .[352] .[353] .[354] .[355] .[356] .[357] .[358] .[359] .[360] .[361] .[362] .[363] .[364] .[365] .[366] .[367] .[368] .[369] .[370] .[371] .[372] .[373] .[374] .[375] .[376] .Connected!
[381] numNewMessages: 1 last: 888666380 mem: 11016 - 30 - 7400
[381] Received "allstatus" from 25235518
[381] *ex-Caldera2*: status: *Ok* relay: *Off* polarity: *+*

Exception (29):
epc1=0x4021cfd2 epc2=0x00000000 epc3=0x00000000 excvaddr=0x00000008 depc=0x00000000

>>>stack>>>
3fffffa0:  feefeffe feefeffe feefeffe 40226c88  
<<<stack<<<
last failed alloc call: 4021E111(288)
ets Jan  8 2013,rst cause:2, boot mode:(3,7)
load 0x4010f000, len 1392, room 16
tail 0
chksum 0xd0
csum 0xd0
v3d128e5c
~ld
   I'm ex-Caldera2

It seems to be an out of memory issue

This info will answer information about the device itself, not the sensors

• IP
• WiFi network and information
• Uptime (remove from /status) ?
• Software version
• RAM usage
• Flash usage (if used)

Persist configuration

More info about OSS licenses found here

A possible solution is to maintain a list of the last 20 messages in the servers. The offset in a getUpdates call retrieve messages starting at that update_id, and clears all preceding. Incrementing updates_id late by 20 updates will keep 20 in the server.

In this way all devices/instances will always see the last 20 (configurage), and the user can talk to any of them... I think.

Devices need to know if a message is for them or not, see issue #6 for an idea.

Additionally, a device could clean all consecutive message for it starting on the first of the 20. This minimize the number of message in the server. Also, if too many messages for one device are accumulated, other devices can detect a device is failing (or at least have a hint).

Unify how messages are sent (refactor). Start all messages with device name, such as

Termometer_1: Temperature is 25°C

See #10

Make it possible to configure light sensor polarity

Telegram command

Report current polarity on status and startup

When WiFi is lost, any alarm is lost.

Count and save them (save also time?) and report when WiFi comes back.
Also report that WiFi was gone (for how long?)

Decide what's better. To support a single user, who needs to be registered, or to simultaneously support multiple users, if at least to command the devices.

One user:

• All messages are directed to s/he
• Message only coming from s/he are accepted

Many users:

• Who gets alerts?
• Messages are answered to whoever sent them
• Can anybody take over a device?

Answer something so the pending icon goes away.

See if better answers are switable
Maybe don't answer messages at all

decide whether to use it or not,

see https://core.telegram.org/bots/api#getting-updates, timeout parameter is suggested to be greater than 0, and the default for TelegramBot is 0. It has to be changed using:

bot.longPoll = 1

longPoll is in seconds

See https://core.telegram.org/bots/api#setmycommands
When commands are used, the UI for the bot is a bit better, commands are auto completed, help and settings are shown on bot's description and welcome messages, etc.

Though this is not compatible with issue #7, it may still be usefull for global configuration things

When the user first plugs in a device (or on missconfiguration or factory reset), there should be an easy setup procedure.

Internet access can be assumed (see #28 and as now implemented using AutoConnect portal)

It should change the name of the device as it reports it in the messages.
It must be persistent across boots

This command sholuld let the user change the bot token. Is used to install new devices.
It has to be persistent. Mabe it can only come fom a hardcoded user id?

maybe use chipID?

ToI %08x?

When a new user requests ownership, send the old user a message to validate the change.
Also, have a owner (instead of default_user)

If compiled out it should save memory space, and should unify logging.

Maybe using a helper function

To save memory.

Try first to connect using regular wifi and saved credentials.
The, allocate, use and destroy portal on success

on
off
toggle ?

If only one of the bot (tokens) is invited to a group chat, it won't be possible to send messages to the chat for the other token, so the process is kind of broken.

Related to Issue #22

See AutoConnect documentation on how to do it

Once /settoken is received, show the keyobard and then do not accept any other command other than /confirmtoken until it's confirmed, or timeout and go back to previous token and continue accepting commands as usual

Originally posted by @gerasdf in #22 (comment)

help message (automatically invoked on startup) should answer with an Keyboard (https://core.telegram.org/bots#keyboards) with all available commands. The user then clicks on the keyboard to talk to a specific instance. The help message message_id serves to identify to what instance the user is speaking.

https://api.telegram.org/bot648272766:AAEkW5FaFMeHqWwuNBsZJckFEOdhlSVisEc/sendMessage?chat_id=25235518&text=hola&&reply_markup={%22inline_keyboard%22:[[{%22text%22:%22hola%22,%22callback_data%22:%22hola%22},{%22text%22:%22chau%22,%22callback_data%22:%22chau%22}]]}

https://api.telegram.org/bot648272766:AAEkW5FaFMeHqWwuNBsZJckFEOdhlSVisEc/getUpdates?offset=0