Raw message reading terminated prematurely when 0x34 (end byte) is part of data about esphome_samsung_ac HOT 6 CLOSED

Thanks for reporting this! And thank you that you attached example data. That is a good start to find a better way.

The current solution is a pretty basic approach.

A far better approach would it be to fill the buffer and check if we found a valid packed until we found one or the buffer has reached its max size.

from esphome_samsung_ac.

Tried this in 'my' fork in the second part of void Samsung_AC::loop(), and it seems to work fine (declared bytes_ and size_ as uint16_t in samsung_ac.h):
last_transmission_ = now;
while (available())
{
uint8_t c;
read_byte(&c);
if (c == 0x32 && !receiving_) // start-byte found
{
receiving_ = true;
bytes_ = 0;
size_ = 0;
data_.clear();
}
if (receiving_)
{
data_.push_back(c);
bytes_++;
switch (bytes_)
{
case 1: // start byte found
break;
case 2: // first part of size found
{
size_ = c;
break;
}
case 3: // second part of size found
{
size_ = (int)size_ << 8 | c;
ESP_LOGV(TAG, "Message size in packet: %d", size_);
break;
}
default: // subsequent bytes
{
if (bytes_ >= (size_+2)) // end byte found
{
receiving_ = false;
process_message(data_, this);
}
break;
}
}
}
}

from esphome_samsung_ac.

The problem with that solution is that it only works for the NASA protocol. I will try to find a more general solution next week.

from esphome_samsung_ac.

I now have commited a solution to that. May you have a look?

from esphome_samsung_ac.

Hi, checked for NASA protocol on my SAMSUNG heatpump and all seems to work fine, data is read correctly. So regression test wise: looks good. However, I don't think that my current data stream includes the 0x34 end byte as part of the data, so we have no 100% proof that the original issue is definitively solved. And I do not know how to manipulate the data stream to create a test case. So what shall we do with this issue, close it for now?

from esphome_samsung_ac.

👍🏻

from esphome_samsung_ac.