Convert endianness #2506
Convert endianness #2506
Conversation
There was a problem hiding this comment.
Please read the standard:
'''
4.2 Data Encoding
MODBUS uses a ‘big-Endian’ representation for addresses and data items. This means
that when a numerical quantity larger than a single byte is transmitted, the most significant
byte is sent first. So for example
Register size value
16 - bits 0x1234 the first byte sent is 0x12 then 0x34
'''
Your pull request would make sense, for int32 etc. that is changing the order of the registers (words) according to "little" or "big".
pymodbus/client/mixin.py
Outdated
| ) -> int | float | str | list[bool]: | ||
| """Convert registers to int/float/str. | ||
|
|
||
| :param registers: list of registers received from e.g. read_holding_registers() | ||
| :param data_type: data type to convert to | ||
| :param endian: endianness of the byte order |
There was a problem hiding this comment.
This is wrong, byte order is always big.
What is not defined by the standard is the word order.
pymodbus/client/mixin.py
Outdated
| :returns: int, float, str or list[bool] depending on "data_type" | ||
| :raises ModbusException: when size of registers is not 1, 2 or 4 | ||
| """ | ||
| byte_list = bytearray() | ||
| for x in registers: | ||
| byte_list.extend(int.to_bytes(x, 2, "big")) | ||
| byte_list.extend(int.to_bytes(x, 2, endian)) |
There was a problem hiding this comment.
this is wrong, a register is always "big".
pymodbus/client/mixin.py
Outdated
| @@ -725,12 +726,13 @@ def convert_from_registers( | |||
|
|
|||
| @classmethod | |||
| def convert_to_registers( | |||
| cls, value: int | float | str | list[bool], data_type: DATATYPE | |||
| cls, value: int | float | str | list[bool], data_type: DATATYPE, endian: Literal["big", "little"] = "big" | |||
pymodbus/client/mixin.py
Outdated
| ) -> list[int]: | ||
| """Convert int/float/str to registers (16/32/64 bit). | ||
|
|
||
| :param value: value to be converted | ||
| :param data_type: data type to be encoded as registers | ||
| :param endian: endianness to encode bytes to |
pymodbus/client/mixin.py
Outdated
| @@ -749,7 +751,7 @@ def convert_to_registers( | |||
| else: | |||
| byte_list = struct.pack(f">{data_type.value[0]}", value) | |||
| regs = [ | |||
| int.from_bytes(byte_list[x : x + 2], "big") | |||
| int.from_bytes(byte_list[x : x + 2], endian) | |||
|
You are right about the modbus protocol stating each register should be big-endian, but we have encountered situations where it was not. I like the new implementation convert_from_registers/convert_to_registers but we can only use it if we can also use little endian byte_order. Proposing a new solution and like to hear your opinion. |
|
Pymodbus implements the modbus standard NOT all the non-standard implementations! Apart from that it is hard to believe any devices would change the byte order, because it would pretty much break compatibility with every professional modbus software. If you want a different byte order then please convert the bytes in the app. Again the word order is different, because that is only a de facto standard so that could/should be open in pymodbus. |
pymodbus/client/mixin.py
Outdated
| ) -> int | float | str | list[bool]: | ||
| """Convert registers to int/float/str. | ||
|
|
||
| :param registers: list of registers received from e.g. read_holding_registers() | ||
| :param data_type: data type to convert to | ||
| :param byte_order: Literal[big, small] order of to bytes in 16 bit register encoding (ALMOST always big) |
There was a problem hiding this comment.
Byte order is fixed by the standard.
pymodbus/client/mixin.py
Outdated
| byte_list.extend(int.to_bytes(x, 2, "big")) | ||
| byte_list.extend(int.to_bytes(x, 2, byte_order)) | ||
| if word_order == "little": | ||
| byte_list = byte_list[::-1] |
There was a problem hiding this comment.
This does not convert the order of all registers.
No description provided.