如何将十六进制字符串转换为Float（小尾数法）

Question

I just learned Python (3.x) and I am stuck with HEX String conversion to Float. I have this HEX String values: '0x22354942F31AFA42CE6A494311518A43082CAF437C6BD4C35F78FA433BF10F442A5222448D3D3544200749C438295C4468AF6E4406B4804450518A4423B0934450E99CC4' And I want to turn it into float.

I have tried to use this code:

bs=bytes.fromhex(row[2:])
fmt = '<' + ('H' * (len(bs) // 2))
res=struct.unpack(fmt, bs)

and it gives me the result of 13602.0,16969.0,6899.0,17146.0,27342.0,17225.0,20753.0,17290.0,11272.0,17327.0,27516.0,50132.0,30815.0,17402.0,61755.0,17423.0,21034.0,17442.0,15757.0,17461.0,1824.0,50249.0,10552.0,17500.0,44904.0,17518.0,46086.0,17536.0,20816.0,17546.0,45091.0,17555.0,59728.0,50332.0

After checking it, I found out that the code that what I currently have is float in base 16, while I need it in base 32 (or maybe not because I am not sure what base/format), with expected float results as 50.3018875, 125.052635,201.4172,276.633331,350.344,424.839722,500.9404,575.7692,649.2838,724.961731,804.1113,880.644043,954.7407,1029.62573,106.541,1181.50427,1255.291 the values which I got from this Calculator Converter . What should I change in the coding to get the expected results? Thank you.

Answer 1

Let's break things down here, because you seem to be confused a bit with all of the juggling of representations. You have some hexadecimal string (that's base 16 encoding) of some binary data. That's your 0x22354942F31AFA42CE6A494311... . You correctly identified that you can convert this from its encoded form to python bytes with bytes.fromhex :

hex_encoded = '0x22354942F31AFA42CE6A494311518A43082CAF437C6BD4C35F78FA433BF10F442A5222448D3D3544200749C438295C4468AF6E4406B4804450518A4423B0934450E99CC4'
binary_data = bytes.fromhex(hex_encoded[2:])  # we do 2: to remove the leading '0x'

At this point, unless we know how binary_data was constructed we can't do anything. But we can take some guesses. You know the first few numbers are floating points: 50.3018875, 125.052635, 201.4172, ... . Typically floats are encoded using the IEEE 754 standard . This provides 3 different encodings of a floating point number: binary16 (16 bits), float (32 bits), and double (64 bits). You can see these in the struct documentation , they are format codes 'e' , 'f' , and 'd' , respectively. We can try each to see which of (if any) your binary data is encoded as. By trial and error, we discover your data was encoded as 32-bit floats, so you can decode them with:

FLOAT = 'f'
fmt = '<' + FLOAT * (len(binary_data) // struct.calcsize(FLOAT))
numbers = struct.unpack(fmt, binary_data)
print(numbers)

Why did what you tried not work? Well you used the format code 'H' which is for an unsigned short. This is an integer , which is why you were getting back numbers with no fractional part!