The ValueError for converting an all-whitespace string to a float misleadingly displays it as the empty string

[Zeturic]

Bug report

Actual behavior:

>>> float('\n')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: could not convert string to float: ''

Expected behavior:

>>> float('\n')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: could not convert string to float: '\n'

As long as the string isn't completely whitespace, it does display properly. For example:

>>> float('\nn')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: could not convert string to float: '\nn'

In addition to newlines, the same thing can be observed with tabs and carriage returns, or any combination of the three. I didn't test any additional whitespace characters, though they would likely behave the same way.

Your environment

• CPython versions tested on: 3.10.5
• Operating system and architecture: Windows 10 (64-bit)

[mdickinson]

Thanks for the report. This is caused by an unfortunate interaction between float_from_string_inner and PyOS_string_to_double. (The PyOS_string_to_double API leaves something to be desired, with its multi-channel reporting of errors, but it's part of the limited API so we can't change it.)

Specifically, after the PyOS_string_to_double call, if end == last in float_from_string_inner, then either the parse was successful, or the string being parsed was empty; conversely, if the parse was successful then end == last. So when we raise on end != last, we catch only parse failures on nonempty strings. The parse failure on an empty string just propagates the ValueError already raised in PyOS_string_to_double, which has an inferior error message.

#95665 should fix this.