dd : treat arg as bytes if it contains a 'B'

[matrixhead]

this fixes bytes.log test case of dd in gnu test coverage #5903.

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/dd/bytes is no longer failing!

[sylvestre]

Excellent and easier than it seems.
Could you please add a test to make sure we don't regress! thanks

[matrixhead]

@sylvestre I've added some tests could you please take a look at that

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/inotify-dir-recreate (fails in this run but passes in the 'main' branch)

[tertsdiepraam]

Before we merge this, I'd like to try a few inputs with the GNU version. Another hypothesis based on the given inputs is that GNU parses the B from the start of the string. So 1xBx would not be in bytes, but 1Bxx would be. I'll check this on my machine tomorrow.

[matrixhead]

@tertsdiepraam This is how i understood this and there is a huge chance that i might be wrong. Lets assume we have an argument called N, gnu uses an "operandxN" pattern to match the "N" where xN is the multiplicative suffix and N is recursively matched also the operand could have an optional B, it's kind of a "type annotation" saying that this operand represents bytes and if any one of these operands represent bytes then treat result as bytes. All of this essentially translates to if there is "B" in the argument then treat it as bytes.

for example if we give an argument "1x2Bx4"
it should take 1 as operand and x2Bx4 as suffix then 2B as operand and x4 as suffix and so on and since there is a B it should be treated as bytes.
in the case "1xBx"
it should result in an error because B is not a valid operand.
in the case "1Bxx"
it should result in an error because x is not a valid operand.

[tertsdiepraam]

I've run some tests and I agree with your interpretation. Thanks for the explanation!