D414 false positive on function with Numpy convention

[1oglop1]

Hi I have this function and pydocstyle recognises argument attributes in docstring as a section but Attributes section belongs to a class This is related to #318 numpy style reference: https://numpydoc.readthedocs.io/en/latest/format.html#docstring-standard

def parse_attributes(attributes):
    """
    Parse resource attributes.

    Parameters
    ----------
    attributes
        Resource attributes.

    """
    print(attributes)

[wulmer]

Same with parameters which results in D405,406,407,410,411,414, all of them twice except 406:

"""Demo module for parameters bug."""

def foo(parameters):
    """Do something.
    
    Parameters
    ----------
    parameters:
       a list of parameters
    """

Linting above file gives following output (with pydocstyle 4.0.1):

foo.py:4 in public function `foo`:
        D410: Missing blank line after section ('Parameters')
foo.py:4 in public function `foo`:
        D414: Section has no content ('Parameters')
foo.py:4 in public function `foo`:
        D405: Section name should be properly capitalized ('Parameters', not 'parameters')
foo.py:4 in public function `foo`:
        D411: Missing blank line before section ('Parameters')
foo.py:4 in public function `foo`:
        D407: Missing dashed underline after section ('Parameters')
foo.py:4 in public function `foo`:
        D406: Section name should end with a newline ('Parameters', not 'parameters:')
foo.py:4 in public function `foo`:
        D410: Missing blank line after section ('Parameters')
foo.py:4 in public function `foo`:
        D414: Section has no content ('Parameters')
foo.py:4 in public function `foo`:
        D405: Section name should be properly capitalized ('Parameters', not 'parameters')
foo.py:4 in public function `foo`:
        D411: Missing blank line before section ('Parameters')
foo.py:4 in public function `foo`:
        D407: Missing dashed underline after section ('Parameters')```

Would it be a solution to introduce a state machine which marks already passed sections as such so that they will not be mixed up with any following attribute/parameter definitions?

In the case above, the parser could mark that it has already seen a section `Parameters` and then not check for a `Parameters` section again. In case someone really has two parameter sections, this will of course pass unnoticed, but all those false positives right now are IMHO the larger annoyance.

[rmorshea]

Oddly enough, if you omit the . at the end of the parameter description for params like attributes: these errors go away for some reason.