Questions about two different results with the same coding target.

[han5onwi11y]

Thank you for reading this message and thanks to the authors of this reed-solomon Python repository.

If we take (255, 223) coding as an example and assume GF(2^8), the information before coding has 223 bytes, and the reed-solomon ecc has 32 bytes. Hence, the total encoded length is 255 bytes.

First, the information before coding has a file named "input.bin", which has 233 bytes.

input.bin	column 1	column 2	column 3	column 4	column 5	column 6	column 7	column 8	column 9	column 10	column 11	column 12	column 13	column 14	column 15	column 16
row 1	0x12	0x34	0x56	0x78	0x90	0x11	0x22	0x33	0x44	0x55	0x66	0x77	0x88	0x99	0x00	0xff
row 2	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff	0xff
row 3	0x32	0x3a	0x63	0x75	0x3c	0x77	0xb3	0x83	0x86	0x86	0x5c	0x00	0x00	0x20	0x00	0x00
row 4	0x00	0x00	0x20	0x02	0x00	0x10	0x00	0x00	0x01	0x00	0x00	0x00	0xb0	0xf1	0x01	0x00
row 5	0x00	0x00	0x60	0x08	0x00	0x00	0x18	0x30	0x97	0xe5	0x00	0x00	0x53	0xe3	0x0c	0x00
row 6	0x00	0x1a	0xd8	0x28	0xcd	0xe1	0x78	0x10	0x96	0xe5	0x06	0x00	0xa0	0xe1	0x31	0xff
row 7	0x2f	0xe1	0x00	0x20	0x50	0xe2	0xe5	0xff	0xff	0xca	0x05	0x00	0x00	0x0a	0xcc	0x01
row 8	0x9f	0xe5	0x0a	0x10	0xa0	0xe1	0x00	0x00	0x8f	0xe0	0xdc	0xa0	0x90	0xe5	0x01	0x10
row 9	0x41	0xe2	0x94	0x90	0x90	0xe5	0x01	0x20	0x02	0xe0	0x00	0x30	0xa0	0xe3	0xc8	0xd0
row10	0x4d	0xe2	0x03	0x00	0x92	0xe1	0x03	0x00	0x00	0x0a	0x50	0x03	0x9f	0xe5	0x00	0x00
row11	0x8f	0xe0	0xa0	0x90	0xe5	0x01	0x10	0x41	0xe2	0x94	0x90	0x90	0xe5	0x01	0x20	0x02
row12	0xe0	0x00	0x30	0xa0	0xe3	0xc8	0xd0	0x4d	0xe2	0x03	0x00	0x92	0xe1	0x03	0x08	0x10
row13	0x9d	0xe5	0x00	0x00	0x51	0xe3	0x04	0x00	0x00	0x0a	0x08	0x20	0x9d	0xe5	0x04	0x00
row14	0xa0	0x08	0x10	0x9d	0xe5	0x00	0x00	0x51	0xe3	0x04	0x00	0x00	0x0a	0x08	0x20	-

Second, I use the MATLAB to generate the RS code, and the result of 32-byte RS code is below. Please note that this 32-byte RS code generated by MATLAB is my desired result. However, the Python script provided by his Github repository has different results (see the third part).

m = 8;
n = 255;
k = 223;
prim_poly = 285;
genpoly = rsgenpoly(n,k,prim_poly,1);
fp = fopen('input.bin','r','b') ;
data = fread(fp, 'ubit8')';
info_bytes = gf(data ,m);
encode_bytes = rsenc(info_bytes,n,k,genpoly) ;
encode_bytes.x %print this, then see the 255 bytes with rs code

rs code	column 1	column 2	column 3	column 4	column 5	column 6	column 7	column 8	column 9	column 10	column 11	column 12	column 13	column 14	column 15	column 16
row 1	0x28	0x49	0x62	0x92	0x1d	0x89	0x5c	0x81	0x02	0xde	0x7b	0x1f	0xb3	0xc1	0xa8	0x48
row 2	0xd4	0xe9	0xad	0x47	0xf4	0x70	0x58	0x8b	0x97	0x1d	0x8a	0x83	0x42	0x32	0x29	0x7d

Third, I use the Python script provided by his Github repository.

from reedsolo import RSCodec
import reedsolo as rs
inputfile = open('input.bin', "rb")
n = 255
rsc = RSCodec(32)
rs.init_tables(0x11d)
ecc_encode = rsc.encode(inputfile.read())

If you print ecc_encode out, you can see the 32-byte RS code below.

rs code	column 1	column 2	column 3	column 4	column 5	column 6	column 7	column 8	column 9	column 10	column 11	column 12	column 13	column 14	column 15	column 16
row 1	0xfd	0xe2	0xaa	0x47	0x93	0xad	0x57	0x5b	0xca	0xd8	0x17	0x19	0xe2	0x38	0x52	0x7d
row 2	0x74	0xa3	0x98	0xf9	0x23	0xc3	0xa3	0xba	0x3d	0x80	0x80	0x02	0x61	0xf5	0x91	0x29

The result generated by Python script is not corrected (at least, it is not my desired).

Could you help me correct the Python script? Thank you.

[lrq3000]

Hello,

Normally our library is mathematically standardized, I developed two different codebases (see also unireedsolomon) from two different approaches to ensure that it is universal. So either MATLAB's codec is doing some assumptions that you did not input into our codec (such as a non-zero fcr value), or there is something different with the data loading, so that the data you feed into RSCodec is actually different from the one you feed into the MATLAB's RS codec.

In particular, I am referring to this MATLAB line:

data = fread(fp, 'ubit8')';

There is no equivalent in your Python code.

However, since you do not provide here the original input data, I cannot test further. It is not sufficient you provide the output of both codecs encodings, we would need to see the original file, so that we can try to reproduce on our own. You provide a dump above but it would help if you could provide a code to generate it either in MATLAB or Python, or just attach it here or via a 3rd party website, so that we are exactly on the same page, as otherwise reproducing the dump is I think hazardous.