Other consequences of textwrap.wrap() changes in 3.15

Bug report

Bug description:

It seems like the change in #139902 had other effects than what the PR intended. If those changes are intentional, it would be good if they were documented. And if they were not intentional, then this seem like a bug. This change has since been backported (#139902, #139903), though I don't believe those backports have been released.

Given the following code:

import textwrap
textwrap.wrap(' ABCDEFG', width=1)

In 3.14.0, this returns [' ', 'A', 'B', 'C', 'D', 'E', 'F', 'G']
In 3.15.0a1, this returns ['A', 'B', 'C', 'D', 'E', 'F', 'G']

Since the PR only mentions trailing whitespace, it appears this change was unintended.

The documentation reads as follows.

drop_whitespace
(default: True) If true, whitespace at the beginning and ending of every line (after wrapping but before indenting) is dropped. Whitespace at the beginning of the paragraph, however, is not dropped if non-whitespace follows it. If whitespace being dropped takes up an entire line, the whole line is dropped.

I am not sure how to interpret this. This line would seem to indicate the space should be preserved.

Whitespace at the beginning of the paragraph, however, is not dropped if non-whitespace follows it.

CPython versions tested on:

3.15

Operating systems tested on:

Linux

Linked PRs

• gh-140639

Start by running the textwrap.wrap(' ABCDEFG', width=1) example on the reported Python versions and read the documented drop_whitespace behavior. Compare the consequences of issue #139902 with linked PR #140639; done means resolving whether the changed leading-space behavior is intentional and fixing or documenting it accordingly.