Why is passing bytes to class str constructor special?

Question

Offical Python3 docs say this about passing bytes to the single argument constructor for class str:

Passing a bytes object to str() without the encoding or errors arguments falls under the first case of returning the informal string representation (see also the -b command-line option to Python).

Ref: https://docs.python.org/3/library/stdtypes.html#str

informal string representation -> Huh?

Using the Python console (REPL), and I see the following weirdness:

>>> ''
''
>>> b''
b''
>>> str()
''
>>> str('')
''
>>> str(b'')
"b''"  # What the heck is this?
>>> str(b'abc')
"b'abc'"
>>> "x" + str(b'')
"xb''"  # Woah.

(The question title can be improved -- I'm struggling to find a better one. Please help to clarify.)

Answer 1

The concept behind str seems to be that it returns a "nicely printable" string, usually in a human understandable form. The documentation actually uses the phrase "nicely printable":

If neither encoding nor errors is given, str(object) returns object.__str__(), which is the “informal” or nicely printable string representation of object. For string objects, this is the string itself. If object does not have a __str__() method, then str() falls back to returning repr(object).

With that in mind, note that str of a tuple or list produces string versions such as:

>>> str( (1, 2) )
'(1, 2)'
>>> str( [1, 3, 5] )
'[1, 3, 5]'

Python considers the above to be the "nicely printable" form for these objects. With that as background, the following seems a bit more reasonable:

>>> str(b'abc')
"b'abc'"

With no encoding provided, the bytes b'abc' are just bytes, not characters. Thus, str falls back to the "nicely printable" form and the six character string b'abc' is nicely printable.