Pseudo code to find Bip39 checksum?

Question

I'm trying to find the checksum for a 12-word BIP39.

I can find the first eleven words just fine, but my 12th word is always off by a few positions.

For example, take the 128-bit entropy e84eaa26330ccbb2f866a1dc155e94ba.

In binary that is 11101000010011101010101000100110001100110000110011001011101100101111100001100110101000011101110000010101010111101001010010111010.

The first eleven 11-bit words are:

11101000010 (bin) => 1858 (decimal) => tribe
01110101010 (bin) => 938  (decimal) => inspire
10001001100 (bin) => 1100 (decimal) => maze
01100110000 (bin) => 816  (decimal) => gravity
11001100101 (bin) => 1637 (decimal) => smile
11011001011 (bin) => 1739 (decimal) => sunny
11100001100 (bin) => 1804 (decimal) => thunder
11010100001 (bin) => 1697 (decimal) => stage
11011100000 (bin) => 1760 (decimal) => swing
10101010111 (bin) => 1367 (decimal) => priority
10100101001 (bin) => 1321 (decimal) => pioneer

There are still 7 unused bits, the last ones: 0111010.

The SHA-256 hash of the hex string is 5355d54a3d673c4b1ac20b839ead09af3c6fea6dc24199b477f6ff64e7a68262. The first 4 bits of that are represented by the first hex character, 5, which in binary is 0101.

I append those 4 bytes to my binary string: 111010000100111010101010001001100011001100001100110010111011001011111000011001101010000111011100000101010101111010010100101110100101

Now I have the last 11-bit word:

01110100101 (bin) => 933 (decimal) => input

But when I try to validate that against Ian Coleman's Bip39 I see the checksum should be 0011 (not 0101) and that would make the last word be

01110100011 (bin) => 931 (decimal) => inner

What am I doing wrong?

Answer 1

The SHA-256 hash of the hex string is 5355d54a3d673c4b1ac20b839ead09af3c6fea6dc24199b477f6ff64e7a68262. The first 4 bits of that are represented by the first hex character, 5, which in binary is 0101.

You need to take the shasum of the raw binary, not the hex string:

$ echo "11101000010011101010101000100110001100110000110011001011101100101111100001100110101000011101110000010101010111101001010010111010" | shasum -a 256 -b

Gives output

3043b02ceb37bb00ab48c91f76bb23d487e1ec822c90e1575967e8838a655f7c *-

When converting this hexstring to binary we get:

0011000001000011101100000010110011101011001101111011101100000000101010110100100011001001000111110111011010111011001000111101010010000111111000011110110010000010001011001001000011100001010101110101100101100111111010001000001110001010011001010101111101111100

The first four bits of this string are 0011, which is consistent with the output given by the Ian Coleman tool.

Answer 2

when I want "$ echo "11101000010011101010101000100110001100110000110011001011101100101111100001100110101000011101110000010101010111101001010010111010" | shasum -a 256 -b" convert python language to "sha256_hash = hashlib.sha256(data.encode()).hexdigest()", it return "7529ea5a2f86b91b4b72e41aa36c190e8173e2fef6bbda828f74e55022eefec8", wrong result, BUT I don't know which error came from?