Bug, Unhandled case when a section is empty

[jspaezp]

Issue

# >>> Call Args:  ['tests/data/format/numpydoc/numpydoc_parameter_spacing.py', '--style', 'numpydoc', '--style', 'pep257']

This call fails with an index error because section[0] has length 0, therefore section[0][0] returns an index error.

Fix

Modifying:

pydocstringformatter/pydocstringformatter/_formatting/base.py

Lines 275 to 282 in c749fc6

	# Check that indent on first line of section didn't get weird
	first_section = True
	for section in new_sections.values():
	if first_section:
	section[0] = section[0].lstrip()
	first_section = False
	elif not section[0][0].isspace():
	section[0] = f"{' ' * indent_length:s}{section[0]:s}"

To:

 # Check that indent on first line of section didn't get weird 
 first_section = True 
 for section in new_sections.values(): 
     if first_section: 
         section[0] = section[0].lstrip() 
         first_section = False 
     elif len(section[0]) > 0 and not section[0][0].isspace(): # <<<<<< this is the fix
         section[0] = f"{' ' * indent_length:s}{section[0]:s}" 

Related:

Breaking call:

# >>> Call Args:  ['tests/data/format/summary_splitter/regression_151.py', '--style', 'numpydoc', '--linewrap-full-docstring']

Fix:

pydocstringformatter/pydocstringformatter/_formatting/formatters_default.py

Lines 91 to 101 in c749fc6

	summary_lines = summary.splitlines()
	new_summary = "\n".join(
	textwrap.wrap(
	summary_lines[0],
	width=line_length,
	initial_indent=" " * (indent_length + quotes_length),
	subsequent_indent=" " * indent_length,
	replace_whitespace=True,
	)
	)[indent_length + quotes_length :]

        summary_lines = summary.splitlines()

        # Here is the fix, checks if the summary line is empty before going though
        if len(summary_lines) == 0:
            summary_lines = [""]

        new_summary = "\n".join(
            textwrap.wrap(
                summary_lines[0],
                width=line_length,
                initial_indent=" " * (indent_length + quotes_length),
                subsequent_indent=" " * indent_length,
                replace_whitespace=True,
            )
        )[indent_length + quotes_length :]

[DanielNoord]

Sounds good! Could you open a PR with the proposed fix and a test?

[DanielNoord]

Reopening as the test in #162 shows this can still reproduce in some edge cases.

I can't immediately think of a fix for #162 as it also needs to raise some sort of user warning like we saw in #278