Documentation: more info on usage of the code

[chinezbrun]

is a good approach and i will like to use it in my automation but i need more info on the usage of the code.

i.e in example provided the is a syntax like:

register = HoldingRegisters.BMSBatteryCapacity
BMSBatteryCapacity is an example but where we can find the rest?

similar for WritableRegisters :)

[githubDante]

All registers are accessible as class properties and are listed even in the REPL on Tab+Tab

>>> from deye_controller.modbus.protocol import HoldingRegisters, WritableRegisters
>>> HoldingRegisters.BMS<Tab><Tab>
HoldingRegisters.BMSBatteryAlarm           HoldingRegisters.BMSBatterySOH             HoldingRegisters.BMSChargedVoltage         HoldingRegisters.BMSMaxChargingCurrent
HoldingRegisters.BMSBatteryCapacity        HoldingRegisters.BMSBatterySymbol          HoldingRegisters.BMSChargingCurrentLimit   HoldingRegisters.BMSMaxDischargeCurrent
HoldingRegisters.BMSBatteryCurrent         HoldingRegisters.BMSBatteryTemp            HoldingRegisters.BMSDischargeCurrentLimit  HoldingRegisters.BMSType
HoldingRegisters.BMSBatteryFaultLocation   HoldingRegisters.BMSBatteryVoltage         HoldingRegisters.BMSDischargedVoltage      

On read the values received from the inverter must be set as a value to the respective register e.g.

>>> HoldingRegisters.BMSBatteryTemp.value = 1410

You can use the group_registers() / map_response() helpers from utils to do that. The actual (human readable) values are returned from a call to the format() method e.g.

>>> HoldingRegisters.BMSBatteryTemp.format()
41.0

For writing the values are set via the set() method e.g.

>>> WritableRegisters.SellModeVolts1.set(51)

and then the actual modbus values which must be sent to the inverter are exposed as a property

>>> WritableRegisters.SellModeVolts1.to_modbus
5100

[chinezbrun]

thanks for your answer.
indeed if you ar digging you can find them. the point was how to find info faster. but anyhow i get the point,
i understand that you are upload the parameters on PostgreSQL. How? I have MariaDB and
and i'm looking to have them there.

[githubDante]

How ... A call to a DB stored procedure (via asyncpg) with the collected data. The data is stored in the following format:

<date> | <inverter serial: numeric> | <metric_1: double precision[]> | <metric_2: double precision[]> | ...

Each index in the metric arrays corresponds to an hour/minute integer and is defined by:

    current_t := now();
    h_idx := extract(hour from current_t);
    m_idx := extract(minute from current_t);
    
    _idx_ := h_idx * 12 + m_idx / 5;

where the variables are declared as:

    declare h_idx int;
    declare m_idx int;
    declare _idx_ int;
    declare current_t timestamp without time zone;

This gives you 288 records/24h for each metric (column) and everything is stored in a single row, so you have really quick access to the data for the respective day.
MariaDB doesn't seem like the right tool for the task as it does not support arrays as far as I know.

[chinezbrun]

thanks! Could your script provide output in JSON format?

[githubDante]

Not currently, but could be added easily (assuming we are talking about deye-read). Any suggestions about the JSON structure ?

[chinezbrun]

i will think about :) thanks you for openness !

[githubDante]

JSON output is supported in version 0.1.3. The output is available in the examples dir

[chinezbrun]

wow. super fast. looks good.
sorry that i did not have the chance to provide you a suggestion for JSON structure...
but instead of:

 {
  "logger": 2719999999,
  "serial": "2299999999",
  "data": [
  {
      "device_type": {
        "addr": 0,
        "value": "Hybrid3Phase",
        "unit": ""
      }
  }
  ,
    {
      "modbus_address": {
        "addr": 1,
        "value": 1,
        "unit": ""
      }
    }
  
  ]
}

why not:

{
  "logger": 2719999999,
  "serial": "2299999999",
  "data": [{
      "device_type": {
        "addr": 0,
        "value": "Hybrid3Phase",
        "unit": ""
      }
  ,
      "modbus_address": {
        "addr": 1,
        "value": 1,
        "unit": ""
      }
  }
]
}

and i believe will simplify the future reading of JSON data

[githubDante]

Current implementation allows you to read the output like this:

for item in output['data']:
    for k,v in item.items():
        if v['addr'] in [145, 214]:
            do_something_with(v['value'])

or if filtering by description is required:

for item in output['data']:
    for k,v in item.items():
        if k in ['solar_sell', 'bms_battery_SOC']:
            do_something_with(v['value'])
     

I don't see a problem with that. A loop over the ['data'] key would be needed anyway.

[chinezbrun]

clear. which is the file path? can be changed?

[githubDante]

Currently the output is written directly to stdout. On Linux it's easy to write it wherever you want by using > <path>/<filename> but probably we are talking for universal method for all platforms.

Expect an --out option in 0.1.4

[chinezbrun]

cool.good job!

[githubDante]

Do you know how to update from git in order to test this d91f755 :)

Just in case: pip install --upgrade git+https://github.com/githubDante/deye-controller.git

[chinezbrun]

done. what's the sintax?

[githubDante]

Sorry. Apparently update directly from git does not update the console scripts. Version 0.1.4 is already on pypi.org

pip install --upgrade deye-controller

and should be OK. The syntax is

deye-read --json --out <filename> <ip_addr> <logger_sn>