Berry Scripting Language ~

Useful resources:

• First time user of Berry: Berry Introduction (in 20 minutes of less)
• Language fast reference PDF (7 pages) Berry Short manual
• Full language documentation The Berry Script Language Reference Manual
• Tasmota extension of Berry, see below
• Full examples in the Berry Cookbook

Introduction to Berry~

Berry is the next generation scripting for Tasmota. It is based on the open-source Berry project, delivering an ultra-lightweight dynamically typed scripting language designed for lower-performance embedded devices.

Github Manual

Berry Scripting allows simple and also advanced extensions of Tasmota, for example:

• simple scripting
• advanced rules, beyond what is possible with native rules
• advanced automations

Berry Scripting takes it one step further and allows to build dynamic extensions to Tasmota, that would previously require native code:

• build light animations
• build I²C drivers
• build complete Tasmota drivers
• integrate native libraries like lvgl see LVGL

About the Berry language~

Berry has the following advantages:

• Lightweight: A well-optimized interpreter with very little resources. Ideal for use in microprocessors.
• Fast: optimized one-pass bytecode compiler and register-based virtual machine.
• Powerful: supports imperative programming, object-oriented programming, functional programming.
• Flexible: Berry is a dynamic type script, and it's intended for embedding in applications. It can provide good dynamic scalability for the host system.
• Simple: simple and natural MicroPython-eque syntax, supports garbage collection and easy to use FFI (foreign function interface).
• RAM saving: With compile-time object construction, most of the constant objects are stored in read-only code data segments, so the RAM usage of the interpreter is very low when it starts.

Tasmota Port~

Berry Scripting in only supported on Tasmota32 for ESP32. The RAM usage starts at ~10kb and will be later optimized. Berry uses PSRAM on ESP32 if available (PSRAM is external RAM attached to ESP32 via SPI, it is slower but larger than internal RAM.

Quick Start~

Click on Configuration then Berry Scripting Console and enjoy the colorful Berry console, also called REPL (Read-Eval-Print-Loop).

The console is not designed for big coding tasks but it's recommended to use a code editor when dealing with many, many lines of code. An extension for Visual Studio Code exists to make writing Berry scripts even easier with colored syntax. Download the entire folder and copy to VSCode extensions folder.

REPL Console~

Try typing simple commands in the REPL. Since the input can be multi-lines, press Enter twice or click "Run" button to run the code. Use Up and Down to navigate through history of previous commands.

> 1+1
2

> 2.0/3
0.666667

> print('Hello Tasmota!')
Hello Tasmota!

Note: Berry's native print() command displays text in the Berry Console and in the Tasmota logs. To log with finer control, you can also use the log() function which will not display in the Berry Console.

> print('Hello Tasmota!')
  log('Hello again')
Hello Tasmota!

Meanwhile the Tasmota log shows:

> tasmota.cmd("Dimmer 60")
{'POWER': 'ON', 'Dimmer': 60, 'Color': '996245', 'HSBColor': '21,55,60', 'Channel': [60, 38, 27]}
The light is bright

Save your Scripts~

Berry can autostart your scripts. See a short desciption in the Section about the Filesystem: https://tasmota.github.io/docs/UFS/#autoexecbe Your can use the Filemanager to edit or save files with your berry scripts.

Iterate without rebooting~

Since v13.0.0.1 you can restart the entire Berry VM with a click in the Berry console. This feature requires to compile with #define USE_BERRY_DEBUG which is anyways highly recommended when coding in Berry. Be aware that restarting the Berry VM loses all context, and may generate negative side effects that we haven't yet identified. When restarting the VM, autoexec.be is ran again.

Instead of using the Web UI, you can also use the BrRestart command which does not require #define USE_BERRY_DEBUG.

Lights and Relays~

Berry provides complete support for Relays and Lights.

You can control individual Relays or lights with tasmota.get_power() and tasmota.set_power().

tasmota.get_power() returns an array of booleans representing the state of each relays and light (light comes last).

tasmota.set_power(relay, onoff) changes the state of a single relay/light.

> tasmota.get_power()
[false, true, false]

> tasmota.set_power(0, true)
true

> tasmota.get_power()
[true, true, false]

For light control, light.get() and light.set accept a structured object containing the following arguments:

When setting attributes, they are evaluated in the following order, the latter overriding the previous: power, ct, hue, sat, rgb, channels, bri.

  # set to yellow, 25% brightness
> light.set({"power": true, "hue":60, "bri":64, "sat":255})
{'bri': 64, 'hue': 60, 'power': true, 'sat': 255, 'rgb': '404000', 'channels': [64, 64, 0]}

  # set to RGB 000080 (blue 50%)
> light.set({"rgb": "000080"})
{'bri': 128, 'hue': 240, 'power': true, 'sat': 255, 'rgb': '000080', 'channels': [0, 0, 128]}

  # set bri to zero, also powers off
> light.set({"bri": 0})
{'bri': 0, 'hue': 240, 'power': false, 'sat': 255, 'rgb': '000000', 'channels': [0, 0, 0]}

  # changing bri doesn't automatically power
> light.set({"bri": 32, "power":true})
{'bri': 32, 'hue': 240, 'power': true, 'sat': 255, 'rgb': '000020', 'channels': [0, 0, 32]}

  # set channels as numbers (purple 12%)
> light.set({"channels": [32,0,32]})
{'bri': 32, 'hue': 300, 'power': true, 'sat': 255, 'rgb': '200020', 'channels': [32, 0, 32]}

Rules~

The rule function have the general form below where parameters are optional:

def function_name(value, trigger, msg)
end

> def dimmer_over_50()
    print("The light is bright")
  end
  tasmota.add_rule("Dimmer>50", dimmer_over_50)

> tasmota.cmd("Dimmer 30")
{'POWER': 'ON', 'Dimmer': 30, 'Color': '4D3223', 'HSBColor': '21,55,30', 'Channel': [30, 20, 14]}

> tasmota.cmd("Dimmer 60")
{'POWER': 'ON', 'Dimmer': 60, 'Color': '996245', 'HSBColor': '21,55,60', 'Channel': [60, 38, 27]}
The light is bright

The same function can be used with multiple triggers.

If the function to process an ADC input should be triggered both by the tele/SENSOR message and the result of a Status 10 command:

tasmota.add_rule("ANALOG#A1", rule_adc_1)
tasmota.add_rule("StatusSNS#ANALOG#A1", rule_adc_1)

Or if the same function is used to process similar triggers:

import string

def rule_adc(value, trigger)
  var i=string.find(trigger,"#A")
  var tr=string.split(trigger,i+2)
  var adc=number(tr[1])
  print("value of adc",adc," is ",value)
end

tasmota.add_rule("ANALOG#A1",rule_adc)
tasmota.add_rule("ANALOG#A2",rule_adc)

Another way to address the same using anonymous functions created dynamically

def rule_adc(adc, value)
  print("value of adc",adc," is ",value)
end
tasmota.add_rule("ANALOG#A1", def (value) rule_adc(1,value) end )
tasmota.add_rule("ANALOG#A2", def (value) rule_adc(2,value) end )

Multiple triggers AND logic~

It is possible to combine multiple triggers in a AND logic as an array:

tasmota.add_rule(["ANALOG#A1>300","ANALOG#A1<500"], def (values) rule_adc_in_range(1,values) end )

Triggers can be of different types too:

tasmota.add_rule(["ANALOG#A1>300","BME280#Temperature>28.0"], def (values) rule_adc_and_temp(1,values) end )

In that case, the value and trigger arguments passed to the rule function are also lists:

def function_name(values:list_of_string, triggers:list_of_string, msg)
end

Teleperiod rules~

Teleperiod rules are supported with a different syntax from Tasmota rules. Instead of using Tele- prefix, you must use Tele#. For example Tele#ANALOG#Temperature1 instead of Tele-ANALOG#Temperature1

Rules operators~

Timers~

Berry code, when it is running, blocks the rest of Tasmota. This means that you should not block for too long, or you may encounter problems. As a rule of thumb, try to never block more than 50ms. If you need to wait longer before the next action, use timers. As you will see, timers are very easy to create thanks to Berry's functional nature.

All times are in milliseconds. You can know the current running time in milliseconds since the last boot:

> tasmota.millis()
9977038

> def t() print("Booh!") end

> tasmota.set_timer(5000, t)
[5 seconds later]
Booh!

A word on functions and closure~

Berry is a functional language, and includes the very powerful concept of a closure. In a nutshell, it means that when you create a function, it can capture the values of variables when the function was created. This roughly means that it does what intuitively you would expect it to do.

When using Rules or Timers, you always pass Berry functions.

cron recurrent calls~

You can choose to run some function/closure at regular intervals specified as cron style format with the first field representing seconds.

> def f() print("Hi") end
> tasmota.add_cron("*/15 * * * * *", f, "every_15_s")
Hi
Hi      # added every 15 seconds
> tasmota.remove_cron("every_15_s")     # cron stops

Like timers, you need to create a closure if you want to register a method of an instance. Example:

class A
    var name
    def init(name)
        self.name = name
    end
    def p()
        print("Hi,", self.name)
    end
end

> bob = A("bob")
> bob.p()
Hi, bob
> tasmota.add_cron("*/15 * * * * *", /-> bob.p(), "hi_bob")
Hi, bob
Hi, bob
Hi, bob
> tasmota.remove_cron("hi_bob")     # cron stops

You can get the timestamp for the next event by using tasmota.next_cron(id) which returns an epoch in seconds.

Loading Filesystem~

Berry files can exist in 2 forms, either a source file (extension .be) or a pre-compiled bytecode (extension .bec). Pre-compiled are usually smaller and load slightly faster (although compilation is fast enough in most use cases). It's usually more flexible and simpler to use source code (.be).

You can upload Berry code in the filesystem using the Consoles - Manage File system menu and load them at runtime. Make careful to use *.be extension for those files.

To load a Berry file, use the load(filename) function where filename is the name of the file with .be or .bec extension; if the file has no extension '.be' is automatically appended.

The loading behavior is as follows:

• load("hello") and load("hello.be") loads hello.be and tries hello.bec if the first does not exist. If both hello.be and hello.bec exist, hello.bec is deleted to avoid confusion between versions.
• load("hello.bec") loads only hello.bec and fails if only hello.be is present

To compile to .bec use tasmota.compile("hello.be"). If all is good, it returns true and creates hello.bec. But beware that if you use load() the .bec file is deleted.

Note: tasmota.compile() is different than native Berry compile()

Creating a Tasmota Driver~

You can easily create a complete Tasmota driver with Berry.

A Driver responds to messages from Tasmota. For each message type, the method with the same name is called. Actually you can register any class as a driver, it does not need to inherit from Driver; the call mechanism is based on names of methods that must match the name of the event to be called.

Driver methods are called with the following parameters: f(cmd, idx, payload, raw). cmd is a string, idx an integer, payload a Berry object representation of the JSON in payload (if any) or nil, raw is a string. These parameters are meaningful to a small subset of events:

• every_second(): called every second
• every_50ms(): called every 50ms (i.e. 20 times per second)
• every_100ms(): called every 100ms (i.e. 10 times per second)
• every_250ms(): called every 250ms (i.e. 4 times per second)
• web_sensor(): display sensor information on the Web UI
• json_append(): display sensor information in JSON format for TelePeriod reporting
• web_add_button(): (deprecated) synonym of web_add_console_button()
• web_add_main_button(), web_add_management_button(), web_add_console_button(), web_add_config_button(): add a button to Tasmotas Web UI on a specific page
• web_add_handler(): called when Tasmota web server started, and the right time to call webserver.on() to add handlers
• button_pressed(): called when a button is pressed
• save_before_restart(): called just before a restart
• mqtt_data(topic, idx, data, databytes): called for MQTT payloads matching mqtt.subscribe. idx is zero, and data is normally unparsed JSON.
• set_power_handler(cmd, idx): called whenever a Power command is made. idx is a combined index value, with one bit per relay or light currently on. cmd can be ignored.
• any_key(cmd, idx): called when an interaction with Button or Switch occurs. idx is encoded as follows: device_save << 24 | key << 16 | state << 8 | device
• display(): called by display driver with the following subtypes: init_driver, model, dim, power.

Then register the driver with tasmota.add_driver(<driver>).

There are basically two ways to respond to an event:

Example

Define a class and implement methods with the same name as the events you want to respond to.

class MyDriver
  def every_second()
    # do something
  end
end

d1 = MyDriver()

tasmota.add_driver(d1)

Fast Loop~

Beyond the events above, a specific mechanism is available for near-real-time events or fast loops (200 times per second, or 5ms).

Special attention is made so that there is no or very little impact on performance. Until a first callback is registered, performance is not impacted and Berry is not called. This protects any current use from any performance impact.

Once a callback is registered, it is called separately from Berry drivers to ensure minimal overhead.

tasmota.add_fast_loop(cl:function) -> nil registers a callback to be called in fast loop mode.

The callback is called without any parameter and does not need to return anything. The callback is called at each iteration of Tasmota event loop. The frequency is set to 200Hz or 5ms.

Note: since v13.1.0.2, the frequency of fast_loop does not depend anymore on the value of the Sleep <x> command.

tasmota.remove_fast_loop(cl:function) -> nil removes a previously registered function or closure. You need to pass the exact same closure reference.

Warning, if you need to register a method from an instance, you need a closure:

class my_driver
  def every_100ms()
    # called every 100ms via normal way
  end

  def fast_loop()
    # called at each iteration, and needs to be registered separately and explicitly
  end

  def init()
    # register fast_loop method
    tasmota.add_fast_loop(/-> self.fast_loop())
    # variant:
    # tasmota.add_fast_loop(def () self.fast_loop() end)
  end
end

tasmota.add_driver(my_driver())                     # register driver
tasmota.add_fast_loop(/-> my_driver.fast_loop())    # register a closure to capture the instance of the class as well as the method

Tasmota Only Extensions~

log(msg:string [, level:int = 3]) -> string~

Logs a message to the Tasmota console. Optional second argument is log_level (0..4), default is 2 (matching build time LOG_LEVEL_INFO).

> log("A")
A

load(filename:string) -> bool~

Loads a Berry script from the filesystem, and returns true if loaded successfully, false if file not found, or raises an exception in runtime. Filename does not need to start with /, but needs to end with .be (Berry source code) or .bec (precompiled bytecode). If the .be extension is missing, it is automatically added.

The behavior for .bec files changed in v13.4: - when loading <file>.be, the .be file is loaded in priority and any .bec file with same prefix is removed (to avoid inconsistencies). If no .be file is present, it tried to load .bec file. - when loading <file>.bec, only .bec files are loaded and .be is ignored. - to create .bec files, you need to use tasmota.compile (see below)

tasmota.compile(filename:string) -> bool~

Loads a .be file, compiles it and saves a .bec file containing the compiled bytecode.

Note: tasmota.compile is different from Berry native compile function.

save(filename:string, f:closure) -> nil~

Internally used function to save bytecode. It's a wrapper to the Berry's internal API be_savecode(). There is no check made on the filename.

There is generally no need to use this function, it is used internally by load().

tasmota object~

A root level object called tasmota is created and contains numerous functions to interact with Tasmota.

Functions used to retrieve Tasmota configuration~

Functions for time, timers or cron~

Functions to create custom Tasmota command~

Functions to add custom responses to JSON and Web UI to sensors~

See examples in the Berry-Cookbook

Functions to manage Relays and Lights~

Low-level access to Tasmota globals and settings.~

Use with care and only if you know what you are doing.

The construct is to use tasmota.global or tasmota.settings to read or write attributes.

mqtt module~

Use with import mqtt.

Since v11.1.0.1, there is an easier way than registering a driver, and listening to mqtt_data event. You can now just attach a function or closure to a MQTT topic, and it does the magic for you.

The function you attach to a topic pattern received only the matching MQTT messages, not all messages unlike mqtt_data() would.

The function takes the same parameters as mqtt_data():

• topic: full topic received from the broker
• idx: not used
• payload_s: payload as string, usually converted to JSON with import json json.load(payload_s)
• payload_b: payload as a binary payload, bytes() array

the function should return true if the event was parsed or if the event should not trigger a Tasmota command. If you return nil or nothing, it is considered as true which is the usual behavior you want (i.e. not trigger a Tasmota command from random MQTT messages).

light object~

Module light is automatically imported via a hidden import light command.

gpio module~

This module allows to retrieve the GPIO configuration set in the templates. You need to distinguish between logical gpio (like PWM, or I2C) and physical gpio which represent the GPIO number of the physical pin. gpio.pin() transforms a logical GPIO to a physical GPIO, or -1 if the logical GPIO is not set.

Currently there is limited support for GPIO: you can only read/write in digital mode and set the GPIO mode.

Any internal error or using unsupported GPIO yields a Berry exception.

An H-bridge is an electronic circuit that switches the polarity of a voltage applied to a load. These circuits are often used in robotics and other applications to allow DC motors to run forwards or backwards.

See the Berry cookbook for H-bridge control

DAC GPIOs~

DAC is limited to specific GPIOs:

• ESP32: only GPIO 25-26
• ESP32-S2: only GPIO 17-18
• ESP32-C3: not supported

> gpio.pin_mode(25, gpio.DAC)   # sets GPIO25 to a DAC pin
> gpio.dac_voltage(25, 1250)    # set voltage to 1250mV
1255

I2S~

DAC can also be used via Esp8266Audio through the ESP32 I2S -> DAC bridge.

class MP3_Player : Driver
  var audio_output, audio_mp3, fast_loop_closure
  def init()
    self.audio_output = AudioOutputI2S()
    self.audio_mp3 = AudioGeneratorMP3()
    self.fast_loop_closure = def () self.fast_loop() end
    tasmota.add_fast_loop(self.fast_loop_closure)
  end

  def play(mp3_fname)
    if self.audio_mp3.isrunning()
      self.audio_mp3.stop()
    end
    var audio_file = AudioFileSourceFS(mp3_fname)
    self.audio_mp3.begin(audio_file, self.audio_output)
    self.audio_mp3.loop()    #- start playing now -#
  end

  def fast_loop()
    if self.audio_mp3.isrunning()
      if !self.audio_mp3.loop()
        self.audio_mp3.stop()
        tasmota.remove_fast_loop(self.fast_loop_closure)
      end
    end
  end
end

mp3_player = MP3_Player()
mp3_player.play("/pno-cs.mp3")

energy module~

The energy module provides ways to read current energy counters and values (if you're creating your own automation) or updating the energy counters (if you're writing a driver).

It relies on a new Berry feature that provides a direct mapping between the internal C structure called struct Energy and the energy module in Berry.

For example, if you want to read or update an energy value:

> energy.active_power
0
> energy.active_power = 460
> energy.active_power
460

# internally it updates the C value `Energy.active_power[0]` (float)

You don't need to do import energy since Tasmota does it for you at boot.

The special energy.read() function dumps all current values to a single map. Be aware that the object is very long. Prefer accessing individual attributes instead.

List of energy attributes that you can read or write:

Energy driver in Berry~

Since v14.2.0, it is possible to implement an Energy driver in pure Berry. The Berry driver is enabled when an OPTION_A 9 GPIO is configured:

• by default, the energy driver has zero consumption.
• the berry code can is energy.driver_enabled() to check if the virtual Berry Energy driver is active (i.e. OPTION_A 9 is configured)
• the following values need to be configured: energy.phase_count (default 1), energy.voltage, energy.current, energy.power_factor (typically 1.0 or less), energy.frequency (default nan)
• the most important value is energy.active_power (in Watt) which is added to the daily power consumption

Example test code in autoexec.be:

if energy.driver_enabled()
  energy.phase_count = 1
  energy.voltage = 240
  energy.power_factor = 1.0
  energy.current = 1.5
  energy.frequency = 50
  energy.active_power = 360
end

wire object for I²C~

Berry Scripting provides 2 objects: wire1 and wire2 to communicate with both I²C buses.

Use wire1.scan() and wire2.scan() to scan both buses:

> wire1.scan()
[]

> wire2.scan()
[140]

You generally use tasmota.wire_scan() to find a device and the corresponding I²C bus.

> mpuwire = tasmota.wire_scan(0x68, 58)
> mpuwire
<instance: Wire()>

Low-level commands if you need finer control:

path module~

A simplified version of os.path module of standard Berry which is disabled in Tasmota because we don't have a full OS.

The default file-system is the ESP32 internal flash. If you have a SD card mounted, it is mapped to the /sd/ subdirectory.

Example:

import path
print(path.listdir("/sd/"))
# outputs a list of filenames at the root dir of the SD card

persist module~

Easy way to persist simple values in Berry and read/write any attribute. Values are written in JSON format in _persist.json file. Be aware that persist cannot detect any change in sub-objects like lists or maps; in such case you can call persist.dirty() to indicate that data needs to be saved.

> import persist    
> persist.a = 1    
> persist.b = "foobar"    
> print(persist)    
<instance: Persist({'a': 1, 'b': 'foobar'})>    
> persist.save()   # save to _persist.json

introspect module~

Allows to do introspection on instances and modules, to programmatically list attributes, set and get them.

> class A var a,b def f() return 1 end end
> ins=A()
> ins.a = "foo"
> import introspect

> introspect.members(ins)
['b', 'a', 'f']

> introspect.get(ins, "a")
foo

> introspect.set(ins, "a", "bar")
bar

> ins.a
bar

webclient class~

Class webclient provides an implementation of an HTTP/HTTPS web client and make requests on the LAN or over the Internet.

Features:

• Support HTTP and HTTPS requests to IPv4 addresses and domain names, to arbitrary ports, via a full URL.
• Support for HTTPS and TLS via BearSSL (which is much lighter than default mbedTLS)
• HTTPS (TLS) only supports cipher ECDHE_RSA_WITH_AES_128_GCM_SHA256 which is both secure and widely supported
• Support for URL redirections
• Ability to set custom User-Agent
• Ability to set custom headers
• Ability to set Authentication header
• Support for Chunked encoding response (so works well with Tasmota devices)
• Support for GET, POST, PUT, PATCH, DELETE methods

The current implementation is based on a fork of Arduino's HttpClient customized to use BearSSL

Current limitations (if you need extra features please open a feature request on GitHub):

• Payload sent to server (POST) can include either text or binary
• Only supports text responses (html, json...) but not binary content yet (no NULL char allowed). However you can download binary content to the file-system with write_file
• Maximum response size is 32KB, requests are dropped if larger
• HTTPS (TLS) is in 'insecure' mode and does not check the server's certificate; it is subject to Man-in-the-Middle attack
• No support for compressed response - this should not be a problem since the client does not advertize support for compressed responses

> cl = webclient()
> cl.begin("http://ota.tasmota.com/tasmota32/release/")
<instance: webclient()>

> r = cl.GET()
> print(r)
200

> s = cl.get_string()
> print(s)
<pre>
<b></b>Alternative firmware for ESP32 based devices with web UI,
[.../...]

> cl = webclient()
> cl.begin("https://raw.githubusercontent.com/tasmota/autoconf/main/esp32/M5Stack_Fire_autoconf.zip")
<instance: webclient()>

> r = cl.GET()
> print(r)
200

> cl.write_file("M5Stack_Fire_autoconf.zip")
950

Managing redirects~

HTTP redirects (301/302) are not followed by default. You can use wc.set_follow_redirects(true) to have redirects automatically followed for HEAD and GET. There is a default limit of 10 successive redirects, this prevents from infinite loops.

For the examples, we use http://ota.tasmota.com/tasmota32 which is redirected to http://ota.tasmota.com/tasmota32/

cl = webclient()
cl.set_follow_redirects(true)
cl.begin("http://ota.tasmota.com/tasmota32")
r = cl.GET()
print(r)
s = cl.get_string()
print(s)

Alternatively, you can manage yourself redirects and retrieve the Location header

cl = webclient()
cl.set_follow_redirects(false)
cl.collect_headers("Location")
cl.begin("http://ota.tasmota.com/tasmota32")
r = cl.GET()
print(r)
if r == 301 || r == 302
  print("Location:", cl.get_header("Location"))
elif r == 200
  s = cl.get_string()
  print(s)
end
cl.close()

Main functions:

Request customization:

Static utility methods:

webserver module~

Module webserver provides functions to enrich Tasmota's Web UI. It is tightly linked to Tasmota page layout.

Functions used to add UI elements like buttons to Tasmota pages, and analyze the current request. See above Driver to add buttons to Tasmota UI.

Low-level functions if you want to display custom pages and content:

Module webserver also defines the following constants:

• Tasmota's web server states: webserver.HTTP_OFF, webserver.HTTP_USER, webserver.HTTP_ADMIN, webserver.HTTP_MANAGER, webserver.HTTP_MANAGER_RESET_ONLY
• Tasmota's pages: webserver.BUTTON_CONFIGURATION, webserver.BUTTON_INFORMATION, webserver.BUTTON_MAIN, webserver.BUTTON_MANAGEMENT, webserver.BUTTON_MODULE
• Methods received by handler: webserver.HTTP_ANY, webserver.HTTP_GET, webserver.HTTP_OPTIONS, webserver.HTTP_POST

See the Berry Cookbook for examples.

tcpclient class~

Simple tcp client supporting string and binary transfers:

• create an instance of the client with var tcp = tcpclient()
• connect to the server tcp.connect(address:string, port:int [, timeout_ms:int]) -> bool Address can be numerical IPv4 or domain name. Returns true if the connection succeeded. Optional timeout in milliseconds. The default timeout is USE_BERRY_WEBCLIENT_TIMEOUT (2 seconds).
• check if the socket is connected with tcp.connected()
• send content with tcp.write(content:string or bytes) -> int. Accepts either a string or a bytes buffer, returns the number of bytes sent. It's your responsibility to resend the missing bytes
• check if bytes are available for reading tcp.available() -> int. Returns 0 if nothing was received. This is the call you should make in loops for polling.
• read incoming content as string tcp.read() -> string or as bytes tcp.readbytes() -> bytes. It is best to call tcp.available() first to avoid creating empty response objects when not needed
• close the socket with tcp.close()

Full example:

tcp = tcpclient()
tcp.connect("192.168.2.204", 80)
print("connected:", tcp.connected())
s= "GET / HTTP/1.0\r\n\r\n"
tcp.write(s)
print("available1:", tcp.available())
tasmota.delay(100)
print("available1:", tcp.available())
r = tcp.read()
tcp.close()
print(r)

tcpclientasync class~

Variant of tcpclient using only non-blocking calls in full asynchronous mode. This allows to have multiple concurrent connections with fine-grained control over timeouts and no blocking of Tasmota. This is especially useful for Matter Border Router for ESP8266 Tasmota based devices via HTTP.

All calls return immediately, so you need to poll the API periodically to send/receive data, and manage timeouts yourself.

Typical equence:

• create an instance of the client with var tcp = tcpclientasync()
• connect to the server tcp.connect(address:string, port:int) -> bool. Address should be numerical IPv4 or IPv6 if you want the call to return immediately (i.e. do DNS resolution ahead of time), otherwise a DNS resolution might take some time and fail. If DNS failed, this call returns false.
• regularly call connected() waiting for true to detect when the connection is established. While connected() returns nil then connection is in-progress. If connected() changes to false then the connection was refused by the host.
• if the connection is not established after a definite amount of time, you should declare 'timeout' and call close()
• to send data: first call listening() to ensure that the socket is ready to send data. Note: the socket is always listening when the connection was just established. Then call write() to send you data (string or bytes), this call returns the actual amount of data sent; if it is lower than your content, you need to handle yourself re-sending the remaining data. Note: ensuring that you send less than the MTU should keep you from happening (~1430 bytes max).
• to receive data: first call available() to check if some data is ready to be received. Then call read() or readbytes() to get the buffer as string or bytes. You can limit the amount of data received, but in such case, the extra data is discarded and lost.
• regularly call connected() to check if the connection is still up
• finally call close() to close the connection on your side and free resources. It is implicitly called if the connection was closed from the peer.

Full example:

def try_connect(addr, port)
  import string
  var tcp = tcpclientasync()
  var now = tasmota.millis()
  var r = tcp.connect(addr, port)
  print(string.format("Time=%5i state=%s", tasmota.millis()-now, str(tcp.connected())))
  print(tcp.info())
  tasmota.delay(50)
  print(string.format("Time=%5i state=%s", tasmota.millis()-now, str(tcp.connected())))
  print(tcp.info())
  tasmota.delay(150)
  print(string.format("Time=%5i state=%s", tasmota.millis()-now, str(tcp.connected())))
  print(tcp.info())
  tasmota.delay(500)
  print(string.format("Time=%5i state=%s", tasmota.millis()-now, str(tcp.connected())))
  print(tcp.info())
  return tcp
end
tcp = try_connect("192.168.1.19", 80)

tcpserver class~

Simple tcp server (socket) listening for incoming connection on any port.

• create an instance of the tcpserver on a specific port with s = tcpserver(8888)
• periodically call s.hasclient() to know if a new client has connected
• if the previous returned true, call var c = s.accept() or var c = s.acceptasync() to accept the connection. It returns an instance of tcpclient or tcpclientasync; it responds to the same APIs as outgoing TCP connection and allows text and binary transfers.
• you can call c.close() to close the connection, or call c.connected() to know if it's still connected (i.e. the client hasn't closed the connection on their side)
• close the server with s.close(). This will prevent the server from receinving any new connection, but existing connections are kept alive.

Full example:

> s = tcpserver(8888)    # listen on port 8888
> s.hasclient()
false

# in parallel connect on this port with `nc <ip_address> 8888`

> s.hasclient()
true               # we have an incoming connection
> c = s.accept()
> c
<instance: tcpclient()>

# send 'foobar' from the client
> c.read()
foobar

# send 'foobar2' again from the client
> c.readbytes()
bytes('666F6F626172320A')

> c.close()
# this closes the connection

udp class~

Class udp provides ability to send and received UDP packets, including multicast addresses.

You need to create an object of class udp. Such object can send packets and listen to local ports. If you don't specify a local port, the client will take a random source port. Otherwise the local port is used as source port.

When creating a local port, you need to use udp->begin(<ip>, <port)>. If <ip> is empty string "" then the port is open on all interfaces (wifi and ethernet).

Sending udp packets~

> u = udp()
> u.begin("", 2000)    # listen on all interfaces, port 2000
true
> u.send("192.168.1.10", 2000, bytes("414243"))   # send 'ABC' to 192.168.1.10:2000, source port is 2000
true

Receive udp packets~

You need to do polling on udp->read(). If no packet was received, the call immediately returns nil.

> u = udp()
> u.begin("", 2000)    # listen on all interfaces, port 2000
true
> u.read()     # if no packet received, returns `nil`
>

> u.read()     # if no packet received, returns `nil`
bytes("414243")    # received packet as `bytes()`

Simple UDP server printing received packets~

class udp_listener
  var u
  def init(ip, port)
    self.u = udp()
    print(self.u.begin_multicast(ip, port))
    tasmota.add_driver(self)
  end
  def every_50ms()
    import string
    var packet = self.u.read()
    while packet != nil
      tasmota.log(string.format(">>> Received packet ([%s]:%i): %s", self.u.remote_ip, self.u.remote_port, packet.tohex()), 2)
      packet = self.u.read()
    end
  end
end

# listen on port 2000 for all interfaces
# udp_listener("", 2000)

Send and receive multicast~

IPv4 example, using the udp_listener listener above.

On receiver side:

udp_listener("224.3.0.1", 2000)

On sender side:

u = udp()
u.begin_multicast("224.3.0.1", 2000)
u.send_multicast(bytes().fromstring("hello"))

# alternatively
u = udp()
u.begin("", 0)      # send on all interfaces, choose random port number
u.send("224.3.0.1", 2000, bytes().fromstring("world"))

The receiver will show:

>>> Received packet ([192.168.x.x]:2000): 68656C6C6F
>>> Received packet ([192.168.x.x]:64882): 776F726C64

This works the same with IPv6 using an address like "FF35:0040:FD00::AABB"

mdns module~

Module import mdns support for mdns (Multicast DNS, aka Bonjour protocol) announces. This is needed for Matter Wifi support.

This feature requires #define USE_DISCOVERY compile option (not included in standard builds).

Example (announce of a Matter Wifi device):

import mdns
mdns.start()
mdns.add_service("_matterc","_udp", 5540, {"VP":"65521+32768", "SII":5000, "SAI":300, "T":1, "D":3840, "CM":1, "PH":33, "PI":""})

Addressable leds (WS2812, SK6812)~

There is native support for addressable leds via NeoPixelBus, with support for animations. Currently supported: WS2812, SK6812.

Details are in Berry leds

serial class~

The serial class provides a low-level interface to hardware UART. The serial GPIOs don't need to be configured in the template.

# gpio_rx:4 gpio_tx:5
ser = serial(4, 5, 9600, serial.SERIAL_7E1)

ser.write(bytes(203132))   # send binary 203132
ser.write(bytes().fromstring("Hello))   # send string "Hello"

msg = ser.read()   # read bytes from serial as bytes
print(msg.asstring())   # print the message as string

Supported serial message formats: SERIAL_5N1, SERIAL_6N1, SERIAL_7N1, SERIAL_8N1, SERIAL_5N2, SERIAL_6N2, SERIAL_7N2, SERIAL_8N2, SERIAL_5E1, SERIAL_6E1, SERIAL_7E1, SERIAL_8E1, SERIAL_5E2, SERIAL_6E2, SERIAL_7E2, SERIAL_8E2, SERIAL_5O1, SERIAL_6O1, SERIAL_7O1, SERIAL_8O1, SERIAL_5O2, SERIAL_6O2, SERIAL_7O2, SERIAL_8O2

display module~

The display module provides a simple API to initialize the Universal Display Driver with data provided as a string. It is used by autoconf mechanism.

uuid module~

The uuid module allows to generate uuid4 random ids.

> import uuid
> uuid.uuid4()
1a8b7f78-59d8-4868-96a7-b7ff3477d43f

crc module~

The crc module allows to compute crc32/16/8 from bytes() arrays.

> import crc
> crc.crc32(0xFFFFFFFF, bytes("AABBCC"))
-1091314015
> crc.crc16(0xFFFF, bytes("AABBCC"))
20980
> crc.crc8(0xFF, bytes("AABBCC"))
139

tasmota_log_reader class~

The tasmota_log_reader class allows you to read and potentially parse the Tasmota logs. It keeps track of what logs were already read in the past and feeds you with new log lines if some are available. It is for example used by the LVGL tasmota_log widget to display logs on a display.

Note: calling tasmota_log_reader can be expensive in string allocations, and adds pressure on the garbage collector. Use wisely.

Example:

var lr = tasmota_log_reader()

# do this regularly
var ret = lr.get_log(2)    # read at log level 2
if ret != nil
  var lines = r.split('\n')  # extract as a list of lines
  # do whatever you need
end

ULP module~

The ULP module exposes the third computing unit of the ESP32, which is a simple finite state machine (FSM) that is designed to perform measurements using the ADC, temperature sensor and even external I2C sensors. This small ultra low power coprocessor can run in parallel to the main cores and in deep sleep mode, where it is capable to wake up the system, i.e. in reaction to sensor measurements. The binding to Berry consists of some lightweight wrapper functions and the communication with the main cores works by accessing the RTC_SLOW_MEM from both sides, which is the same way as in any other ESP32 ULP project.

# simple LED blink example
import ULP
ULP.wake_period(0,500000) # off time
ULP.wake_period(1,200000) # on time 
c = bytes("756c70000c006c00000000001000008000000000000000000000000010008072010000d0e5af2c72340040802705cc190005681d10008072e1af8c720100006821008072040000d0120080720800207004000068010005825c0000800405681d00000092680000800505681d0100009268000080000000b0")
ULP.load(c)
ULP.run()

More infos (including suggestions for a toolchain) on the ULP page.

re regex module~

Use with import re.

There are two ways to use regex, first is to call directly the module which triggers a compilation of the regex at each call. The second one is to pre-compile the regex once into an object which is much more efficient if you need to use the regex multiple times. Any error in the compilation of the regex pattern yields an exception.

> import re

# first series are all-in-one, patterns are compiled on the fly

> re.search("a.*?b(z+)", "zaaaabbbccbbzzzee")
['aaaabbbccbbzzz', 'zzz']
> re.match("a.*?b(z+)", "aaaabbbccbbzzzee")
['aaaabbbccbbzzz', 'zzz']
> re.split('/', "foo/bar//baz")
['foo', 'bar', '', 'baz']
> re.searchall('<([a-zA-Z]+)>', '<abc> yeah <xyz>')
[['<abc>', 'abc'], ['<xyz>', 'xyz']]

# below are pre-compiled patterns, which is much faster if you use the
# pattern multiple times

> rr = re.compile('<([a-zA-Z]+)>')
> rr.searchall('<abc> yeah <xyz>')
[['<abc>', 'abc'], ['<xyz>', 'xyz']]

> rr = re.compile("/")
> rr
<instance: re_pattern()>

> rr.split("foo/bar//baz")
['foo', 'bar', '', 'baz']
> rr.split("/b")
['', 'b']

Note: for match and search, the first element in the list contains the global match of the pattern. Additional elements correspond to the sub-groups (in parenthesis).

The regex engine is based on re1.5 also used in Micropython.

crypto module~

Module import crypto support for common cryptographic algorithms.

Currently supported algorithms:

• AES CTR 256 bits - requires #define USE_BERRY_CRYPTO_AES_CTR
• AES GCM 256 bits
• AES CCM 128 or 256 bits
• AES CBC 128 bits
• Elliptic Curve C25519 - requires #define USE_BERRY_CRYPTO_EC_C25519
• Elliptic Curve P256 (secp256r1) - requires #define USE_BERRY_CRYPTO_EC_P256
• HKDF key derivation with HMAC SHA256 - requires #define USE_BERRY_CRYPTO_HKDF_SHA256
• HMAC SHA256
• MD5
• PKKDF2 with HMAC SHA256 key derivation - requires #define USE_BERRY_CRYPTO_PBKDF2_HMAC_SHA256
• SHA256
• JWT RS256 (RSASSA-PKCS1-v1_5 with SHA256) - requires #define USE_BERRY_CRYPTO_RSA

crypto.AES_CTR class~

Encrypt and decrypt, using AES CTR (Counter mode) with 256 bits keys.

Test vectors from https://datatracker.ietf.org/doc/html/rfc4231

# Test case from https://www.ietf.org/rfc/rfc3686.txt
import crypto
key = bytes("F6D66D6BD52D59BB0796365879EFF886C66DD51A5B6A99744B50590C87A23884")
iv = bytes("00FAAC24C1585EF15A43D875")
cc = 0x000001
aes = crypto.AES_CTR(key)
plain = bytes("000102030405060708090A0B0C0D0E0F101112131415161718191A1B1C1D1E1F")
cipher = aes.encrypt(plain, iv, cc)
assert(cipher == bytes("F05E231B3894612C49EE000B804EB2A9B8306B508F839D6A5530831D9344AF1C"))
plain2 = aes.decrypt(cipher, iv, cc)
assert(plain == plain2)

crypto.AES_GCM class~

Encrypt, decrypt and verify, using AES GCM (Gallois Counter Mode) with 256 bits keys.

Example taken from https://wizardforcel.gitbooks.io/practical-cryptography-for-developers-book/content/symmetric-key-ciphers/aes-encrypt-decrypt-examples.html

import crypto

key = bytes('233f8ce4ac6aa125927ccd98af5750d08c9c61d98a3f5d43cbf096b4caaebe80')
ciphertext = bytes('1334cd5d487f7f47924187c94424a2079656838e063e5521e7779e441aa513de268550a89917fbfb0492fc')
iv = bytes('2f3849399c60cb04b923bd33265b81c7')
authTag = bytes('af453a410d142bc6f926c0f3bc776390')

# decrypt ciphertext with key and iv
aes = crypto.AES_GCM(key, iv)
plaintext = aes.decrypt(ciphertext)
print(plaintext.asstring())
# 'Message for AES-256-GCM + Scrypt encryption'

tag = aes.tag()
print(tag == authTag)
# true

crypto.AES_CCM class~

Encrypt and decrypt, using AES CCM with 256 bits keys.

Example from Matter:

# raw_in is the received frame
raw_in = bytes("00A0DE009A5E3D0F3E85246C0EB1AA630A99042B82EC903483E26A4148C8AC909B12EF8CDB6B144493ABD6278EDBA8859C9B2C")

payload_idx = 8     # unencrypted header is 8 bytes
tag_len = 16        # MIC is 16 bytes

p = raw[payload_idx .. -tag_len - 1]   # payload
mic = raw[-tag_len .. ]                # MIC
a = raw[0 .. payload_idx - 1]          # AAD

i2r = bytes("92027B9F0DBC82491D4C3B3AFA5F2DEB")   # key
# p   = bytes("3E85246C0EB1AA630A99042B82EC903483E26A4148C8AC909B12EF")
# a     = bytes("00A0DE009A5E3D0F")
n   = bytes("009A5E3D0F0000000000000000")         # nonce / IV
# mic = bytes("8CDB6B144493ABD6278EDBA8859C9B2C")

# expected cleartext
clr = bytes("05024FF601001536001724020024031D2404031818290324FF0118")

# method 1 - with distinct calls
import crypto
aes = crypto.AES_CCM(i2r, n, a, size(p), 16)
cleartext = aes.decrypt(p)
tag = aes.tag()

assert(cleartext == clr)
assert(tag == mic)

# method 2 - single call
raw = raw_in.copy()      # copy first if we want to keep the encrypted version
var ret = crypto.AES_CCM.decrypt1(i2r, n, 0, size(n), raw, 0, payload_idx, raw, payload_idx, size(raw) - payload_idx - tag_len, raw, size(raw) - tag_len, tag_len)

assert(ret)
assert(raw[payload_idx .. -tag_len - 1] == clr)

crypto.AES_CBC class~

Encrypt and decrypt, using AES CBC with 128 bits keys.

Example:

var b = bytes().fromstring("hello world_____") # 16-byte aligned
var key = bytes().fromstring("1122334455667788") # 16 bytes
var iv = bytes().fromstring("8877665544332211") # 16 bytes

print("data:",b.asstring()) # "hello world_____"
import crypto
aes = crypto.AES_CBC()
aes.encrypt1(key, iv, b)
print("cipher:",b)
iv = bytes().fromstring("8877665544332211")
aes.decrypt1(key, iv, b)
print("decrypted data:",b.asstring()) # "hello world_____"

crypto.EC_C25519 class~

Provides Elliptic Curve C25519 Diffie-Hellman key agreement. Requires #define USE_BERRY_CRYPTO_EC_C25519

Example from test vectors https://www.rfc-editor.org/rfc/rfc7748:

import crypto

# alice side
alice_priv_key = bytes("77076d0a7318a57d3c16c17251b26645df4c2f87ebc0992ab177fba51db92c2a")
alice_pub_key = bytes("8520f0098930a754748b7ddcb43ef75a0dbf3a0d26381af4eba4a98eaa9b4e6a")
assert(crypto.EC_C25519().public_key(alice_priv_key) == alice_pub_key)

# bob side
bob_priv_key = bytes("5dab087e624a8a4b79e17f8b83800ee66f3bb1292618b6fd1c2f8b27ff88e0eb")
bob_pub_key = bytes("de9edb7d7b7dc1b4d35b61c2ece435373f8343c85b78674dadfc7e146f882b4f")
assert(crypto.EC_C25519().public_key(bob_priv_key) == bob_pub_key)

# shared key computed by alice
ref_shared_key = bytes("4a5d9d5ba4ce2de1728e3bf480350f25e07e21c947d19e3376f09b3c1e161742")
alice_shared_key = crypto.EC_C25519().shared_key(alice_priv_key, bob_pub_key)
bob_shared_key = crypto.EC_C25519().shared_key(bob_priv_key, alice_pub_key)
assert(alice_shared_key == ref_shared_key)
assert(bob_shared_key == ref_shared_key)

crypto.EC_P256 class~

Provides Elliptic Curve Prime256 (secp256r1) Diffie-Hellman key agreement and various functions on P256 curve. Requires #define USE_BERRY_CRYPTO_EC_P256

Example:

import crypto
priv = bytes("f502fb911d746b77f4438c674e1c43650b68285dfcc0583c49cd6ed88f0fbb58")
p = crypto.EC_P256()
pub = p.public_key(priv)
assert(pub == bytes("04F94C20D682DA29B7E99985D8DBA6ABEA9051D16508742899835098B1113D3D749466644C47B559DB184556C1733C33E5788AE250B8FB45F29D4CF48FF752C1ED"))

import crypto
priv = bytes("4E832960415F2B5FA2B1FDA75C1A8F3C84BAEB189EDC47211EF6D27A21FC0ED8")
p = crypto.EC_P256()
pub = p.public_key(priv)
assert(pub == bytes("042166AE4F89981472B7589B8D79B8F1244E2EEE6E0A737FFBFED2981DA3E193D6643317E054D2A924F2F56F1BF4BECA13192B27D8566AF379FBBF8615A223D899"))
print("x=",pub[1..32])
print("y=",pub[33..65])

import crypto
p = crypto.EC_P256()
priv_A = bytes("f502fb911d746b77f4438c674e1c43650b68285dfcc0583c49cd6ed88f0fbb58")
pub_A = bytes("04F94C20D682DA29B7E99985D8DBA6ABEA9051D16508742899835098B1113D3D749466644C47B559DB184556C1733C33E5788AE250B8FB45F29D4CF48FF752C1ED")
priv_B = bytes("4E832960415F2B5FA2B1FDA75C1A8F3C84BAEB189EDC47211EF6D27A21FC0ED8")
pub_B = bytes("042166AE4F89981472B7589B8D79B8F1244E2EEE6E0A737FFBFED2981DA3E193D6643317E054D2A924F2F56F1BF4BECA13192B27D8566AF379FBBF8615A223D899")

shared_1 = p.shared_key(priv_A, pub_B)
shared_2 = p.shared_key(priv_B, pub_A)
assert(shared_1 == shared_2)

crypto.HKDF_SHA256 class~

Provides HKDF using HMAC SHA256 key derivation. Turns 'ikm' (input keying material) of low entropy and creates a pseudo random key. Requires #define USE_BERRY_CRYPTO_HKDF_SHA256

Test vectors from https://www.rfc-editor.org/rfc/rfc5869

import crypto

# Test Case 1
hk = crypto.HKDF_SHA256()
ikm = bytes("0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B0B")
salt = bytes("000102030405060708090A0B0C")
info = bytes("F0F1F2F3F4F5F6F7F8F9")
k = hk.derive(ikm, salt, info, 42)
assert(k == bytes("3CB25F25FAACD57A90434F64D0362F2A2D2D0A90CF1A5A4C5DB02D56ECC4C5BF34007208D5B887185865"))

# Test Case 2
hk = crypto.HKDF_SHA256()
ikm  = bytes("000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f202122232425262728292a2b2c2d2e2f303132333435363738393a3b3c3d3e3f404142434445464748494a4b4c4d4e4f")
salt = bytes("606162636465666768696a6b6c6d6e6f707172737475767778797a7b7c7d7e7f808182838485868788898a8b8c8d8e8f909192939495969798999a9b9c9d9e9fa0a1a2a3a4a5a6a7a8a9aaabacadaeaf")
info = bytes("b0b1b2b3b4b5b6b7b8b9babbbcbdbebfc0c1c2c3c4c5c6c7c8c9cacbcccdcecfd0d1d2d3d4d5d6d7d8d9dadbdcdddedfe0e1e2e3e4e5e6e7e8e9eaebecedeeeff0f1f2f3f4f5f6f7f8f9fafbfcfdfeff")
k = hk.derive(ikm, salt, info, 82)
assert(k == bytes("b11e398dc80327a1c8e7f78c596a49344f012eda2d4efad8a050cc4c19afa97c59045a99cac7827271cb41c65e590e09da3275600c2f09b8367793a9aca3db71cc30c58179ec3e87c14c01d5c1f3434f1d87"))

# Test Case 3
hk = crypto.HKDF_SHA256()
ikm  = bytes("0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b0b")
salt = bytes()
info = bytes()
k = hk.derive(ikm, salt, info, 42)
assert(k == bytes("8da4e775a563c18f715f802a063c5a31b8a11f5c5ee1879ec3454e5f3c738d2d9d201395faa4b61a96c8"))

crypto.PBKDF2_HMAC_SHA256 class~

Provides PBKDF2 using HMAC SHA256 key derivation. Turns a password into a hash.

Test vectors from https://github.com/brycx/Test-Vector-Generation/blob/master/PBKDF2/pbkdf2-hmac-sha2-test-vectors.md

import crypto
pb = crypto.PBKDF2_HMAC_SHA256()

assert(pb.derive("password", "salt", 1, 20) == bytes('120fb6cffcf8b32c43e7225256c4f837a86548c9'))

assert(pb.derive("password", "salt", 2, 20) == bytes('ae4d0c95af6b46d32d0adff928f06dd02a303f8e'))

assert(pb.derive("password", "salt", 3, 20) == bytes('ad35240ac683febfaf3cd49d845473fbbbaa2437'))

assert(pb.derive("password", "salt", 4096, 20) == bytes('c5e478d59288c841aa530db6845c4c8d962893a0'))

assert(pb.derive("passwd", "salt", 1, 128) == bytes('55AC046E56E3089FEC1691C22544B605F94185216DDE0465E68B9D57C20DACBC49CA9CCCF179B645991664B39D77EF317C71B845B1E30BD509112041D3A19783C294E850150390E1160C34D62E9665D659AE49D314510FC98274CC79681968104B8F89237E69B2D549111868658BE62F59BD715CAC44A1147ED5317C9BAE6B2A'))

crypto.SHA256 class~

Provides SHA256 hashing function

Example test vectors from https://www.dlitz.net/crypto/shad256-test-vectors/

import crypto
h = crypto.SHA256()

# SHA256 of empty message
assert(h.out() == bytes("e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"))

# (first 16 bytes of RC4 keystream where the key = 0)
h.update(bytes("de188941a3375d3a8a061e67576e926d"))
assert(h.out() == bytes("067c531269735ca7f541fdaca8f0dc76305d3cada140f89372a410fe5eff6e4d"))

crypto.HMAC_SHA256 class~

Provides HMAC SHA256 hashing function

Test case from https://datatracker.ietf.org/doc/html/rfc4231:

import crypto
key = bytes("4a656665")
msg = bytes("7768617420646f2079612077616e7420666f72206e6f7468696e673f")
h = crypto.HMAC_SHA256(key)
h.update(msg)
hmac = h.out()
assert(hmac == bytes("5bdcc146bf60754e6a042426089575c75a003f089d2739839dec58b964ec3843"))

crypto.RSA class~

Provides RSA core features, currently only JWT RS256 signing (RSASSA-PKCS1-v1_5 with SHA256) - requires #define USE_BERRY_CRYPTO_RSA

import string
import crypto

# JWT requires base64url and not raw base64
# see https://base64.guru/standards/base64url
# input: string or bytes
def base64url(v)
  import string
  if type(v) == 'string'   v = bytes().fromstring(v) end
  var b64 = v.tob64()
  # remove trailing padding
  b64 = string.tr(b64, '=', '')
  b64 = string.tr(b64, '+', '-')
  b64 = string.tr(b64, '/', '_')
  return b64
end

# JWT header and claim
var header = '{"alg":"RS256","typ":"JWT"}'
var claim = '{"sub":"1234567890","name":"John Doe","admin":true,"iat":1516239022}'
var b64header = base64url(header)
var b64claim = base64url(claim)

assert(b64header == 'eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9')
assert(b64claim == 'eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6IkpvaG4gRG9lIiwiYWRtaW4iOnRydWUsImlhdCI6MTUxNjIzOTAyMn0')

# `body` is the payload to sign with RS256
var body = b64header + '.' + b64claim
assert(body == 'eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6IkpvaG4gRG9lIiwiYWRtaW4iOnRydWUsImlhdCI6MTUxNjIzOTAyMn0')

var private_key =
'-----BEGIN PRIVATE KEY-----\n'+
'MIIEvwIBADANBgkqhkiG9w0BAQEFAASCBKkwggSlAgEAAoIBAQC7VJTUt9Us8cKj\n'+
'MzEfYyjiWA4R4/M2bS1GB4t7NXp98C3SC6dVMvDuictGeurT8jNbvJZHtCSuYEvu\n'+
'NMoSfm76oqFvAp8Gy0iz5sxjZmSnXyCdPEovGhLa0VzMaQ8s+CLOyS56YyCFGeJZ\n'+
'qgtzJ6GR3eqoYSW9b9UMvkBpZODSctWSNGj3P7jRFDO5VoTwCQAWbFnOjDfH5Ulg\n'+
'p2PKSQnSJP3AJLQNFNe7br1XbrhV//eO+t51mIpGSDCUv3E0DDFcWDTH9cXDTTlR\n'+
'ZVEiR2BwpZOOkE/Z0/BVnhZYL71oZV34bKfWjQIt6V/isSMahdsAASACp4ZTGtwi\n'+
'VuNd9tybAgMBAAECggEBAKTmjaS6tkK8BlPXClTQ2vpz/N6uxDeS35mXpqasqskV\n'+
'laAidgg/sWqpjXDbXr93otIMLlWsM+X0CqMDgSXKejLS2jx4GDjI1ZTXg++0AMJ8\n'+
'sJ74pWzVDOfmCEQ/7wXs3+cbnXhKriO8Z036q92Qc1+N87SI38nkGa0ABH9CN83H\n'+
'mQqt4fB7UdHzuIRe/me2PGhIq5ZBzj6h3BpoPGzEP+x3l9YmK8t/1cN0pqI+dQwY\n'+
'dgfGjackLu/2qH80MCF7IyQaseZUOJyKrCLtSD/Iixv/hzDEUPfOCjFDgTpzf3cw\n'+
'ta8+oE4wHCo1iI1/4TlPkwmXx4qSXtmw4aQPz7IDQvECgYEA8KNThCO2gsC2I9PQ\n'+
'DM/8Cw0O983WCDY+oi+7JPiNAJwv5DYBqEZB1QYdj06YD16XlC/HAZMsMku1na2T\n'+
'N0driwenQQWzoev3g2S7gRDoS/FCJSI3jJ+kjgtaA7Qmzlgk1TxODN+G1H91HW7t\n'+
'0l7VnL27IWyYo2qRRK3jzxqUiPUCgYEAx0oQs2reBQGMVZnApD1jeq7n4MvNLcPv\n'+
't8b/eU9iUv6Y4Mj0Suo/AU8lYZXm8ubbqAlwz2VSVunD2tOplHyMUrtCtObAfVDU\n'+
'AhCndKaA9gApgfb3xw1IKbuQ1u4IF1FJl3VtumfQn//LiH1B3rXhcdyo3/vIttEk\n'+
'48RakUKClU8CgYEAzV7W3COOlDDcQd935DdtKBFRAPRPAlspQUnzMi5eSHMD/ISL\n'+
'DY5IiQHbIH83D4bvXq0X7qQoSBSNP7Dvv3HYuqMhf0DaegrlBuJllFVVq9qPVRnK\n'+
'xt1Il2HgxOBvbhOT+9in1BzA+YJ99UzC85O0Qz06A+CmtHEy4aZ2kj5hHjECgYEA\n'+
'mNS4+A8Fkss8Js1RieK2LniBxMgmYml3pfVLKGnzmng7H2+cwPLhPIzIuwytXywh\n'+
'2bzbsYEfYx3EoEVgMEpPhoarQnYPukrJO4gwE2o5Te6T5mJSZGlQJQj9q4ZB2Dfz\n'+
'et6INsK0oG8XVGXSpQvQh3RUYekCZQkBBFcpqWpbIEsCgYAnM3DQf3FJoSnXaMhr\n'+
'VBIovic5l0xFkEHskAjFTevO86Fsz1C2aSeRKSqGFoOQ0tmJzBEs1R6KqnHInicD\n'+
'TQrKhArgLXX4v3CddjfTRJkFWDbE/CkvKZNOrcf1nhaGCPspRJj2KUkj1Fhl9Cnc\n'+
'dn/RsYEONbwQSjIfMPkvxF+8HQ==\n'+
'-----END PRIVATE KEY-----\n'

# public_key for reference but not actually used here
var public_key =
'-----BEGIN PUBLIC KEY-----\n'+
'MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAu1SU1LfVLPHCozMxH2Mo\n'+
'4lgOEePzNm0tRgeLezV6ffAt0gunVTLw7onLRnrq0/IzW7yWR7QkrmBL7jTKEn5u\n'+
'+qKhbwKfBstIs+bMY2Zkp18gnTxKLxoS2tFczGkPLPgizskuemMghRniWaoLcyeh\n'+
'kd3qqGElvW/VDL5AaWTg0nLVkjRo9z+40RQzuVaE8AkAFmxZzow3x+VJYKdjykkJ\n'+
'0iT9wCS0DRTXu269V264Vf/3jvredZiKRkgwlL9xNAwxXFg0x/XFw005UWVRIkdg\n'+
'cKWTjpBP2dPwVZ4WWC+9aGVd+Gyn1o0CLelf4rEjGoXbAAEgAqeGUxrcIlbjXfbc\n'+
'mwIDAQAB\n'+
'-----END PUBLIC KEY-----\n'

# read private_key as DER binary
while (private_key[-1] == '\n') private_key = private_key[0..-2] end
var private_key_DER = bytes().fromb64(string.split(private_key, '\n')[1..-2].concat())

# comparison with what was expected
assert(private_key_DER.tob64() == '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')

# sign body
var body_b64 = bytes().fromstring(body)
var sign = crypto.RSA.rs256(private_key_DER, body_b64)
var b64sign = base64url(sign)

# check output
assert(b64sign == 'NHVaYe26MbtOYhSKkoKYdFVomg4i8ZJd8_-RU8VNbftc4TSMb4bXP3l3YlNWACwyXPGffz5aXHc6lty1Y2t4SWRqGteragsVdZufDn5BlnJl9pdR_kdVFUsra2rWKEofkZeIC4yWytE58sMIihvo9H1ScmmVwBcQP6XETqYd0aSHp1gOa9RdUPDvoXQ5oqygTqVtxaDr6wUFKrKItgBMzWIdNZ6y7O9E0DhEPTbE9rfBo6KTFsHAZnMg4k68CDp2woYIaXbmYTWcvbzIuHO7_37GT79XdIwkm95QJ7hYC9RiwrV7mesbY4PAahERJawntho0my942XheVLmGwLMBkQ')

# Final token:
var jwt_token = payload + '.' + b64sign