[devdavidejesus]

When initial_indent or subsequent_indent is longer than width, the room left on the line is negative. _handle_long_word() can then leave an empty chunk, which never satisfies cur_len + l <= width, so _wrap_chunks() looped forever. This happens when the text starts with whitespace (long initial_indent), or with drop_whitespace=False (long subsequent_indent). Zero-length chunks are now always accepted; they add nothing to the line, and this matches what already happens when the room is exactly 0.

Tested on macOS with a debug build of main: the new test hangs without the change and passes with it, and test_textwrap passes.

See #160558.

• Issue: textwrap.wrap() can hang when initial_indent or subsequent_indent is longer than width #160558

• Issue: textwrap.wrap() and fill() loop forever when the indent is wider than width #158979

[devdavidejesus]

Closing in favor of #158988, which was opened earlier for gh-158979, the same bug as gh-160558. It fixes the cause in _handle_long_word() instead of working around the empty chunk, and it also avoids the extra indent-only last line that this PR still produces: textwrap.wrap('aaaa', 3, subsequent_indent='----', drop_whitespace=False) gives ['aaa', '----a', '----'] here and ['aaa', '----a'] there.

I checked #158988 with the cases from this PR: they all return, and test_textwrap passes. Sorry for the duplicate; I missed the earlier issue.