112

We wrote a small wrapper to a twitter app and published this information to http://pypi.python.org. But setup.py just contained a single field for specifying email / name of the author. How do I specify multiple contributors / email list, to the following fields since we would like this package to be listed under our names, much similar to how it shows up in http://rubygems.org.

author='foo',
author_email='foo.bar@gmail.com',
Neuron
  • 5,141
  • 5
  • 38
  • 59
priya
  • 24,861
  • 26
  • 62
  • 81
  • 1
    Do they accept comma or semi-colon separated emails? – heltonbiker Apr 03 '12 at 19:16
  • See [cpython issue 51241](https://github.com/python/cpython/issues/51241) for a discussion on multiple author names and depreciation of the maintainer argument. The issue has been further discussed in the [pep-0345](https://peps.python.org/pep-0345/#author-optional) under the author section. – Paul Rougieux Dec 07 '22 at 15:33

3 Answers3

112

As far as I know, setuptools doesn't support using a list of strings in order to specify multiple authors. Your best bet is to list the authors in a single string:

author='Foo Bar, Spam Eggs',
author_email='foobar@baz.com, spameggs@joe.org',

I'm not sure if PyPI validates the author_email field, so you may run into trouble with that one. In any case, I would recommend you limit these to a single author and mention all contributors in the documentation or description.

Some sources:

This has been registered as a bug, actually, but it seems like support for multiple authors was not implemented. Here is an alternative solution. Here is an idea for how to provide a contact email for a project with multiple authors.

Neuron
  • 5,141
  • 5
  • 38
  • 59
Brian Gesiak
  • 6,648
  • 4
  • 35
  • 50
  • 1
    +1 for the edit and the related links... The discussion in the bug is really interesting, but it's sad that no consensus seem to have been found and that PEP 345 still does not speak of best practices and multiple authors – Stefano Jan 29 '15 at 10:23
  • An idea I got from the Python Bug Tracker link was to dedicate a CONTRIBUTORS or ACK file separately from the PyPI page so users have a single point of contact/complaint/issue request. – Saif Ul Islam Aug 19 '21 at 08:40
3

I'm sort of just piggybacking off of @modocache's answer, in case you want some specifics.

Throughout this answer, I'll be refering to a python3.6 version of the FOO-PYTHON-ENV\Lib\distutils\dist.py file

To reiterate, you cannot use a list in the author field. Here's why:

Spoiler: Two methods belonging to the DistributionMetadata class are the reason --

def _read_field(name):
    value = msg[name]
    if value == 'UNKNOWN':
        return None
    return value

def _read_list(name):
    values = msg.get_all(name, None)
    if values == []:
        return None
    return values

Here's where you'll hit an error if you try to stick a list in the author field:

class DistributionMetadata:

#*...(R E D A C T E D)...*#

    def read_pkg_file(self, file):
        """Reads the metadata values from a file object."""
    #*...(R E D A C T E D)...*#
        # ####################################
        # Note the usage of _read_field() here
        # ####################################
        self.name = _read_field('name')
        self.version = _read_field('version')
        self.description = _read_field('summary')
        # we are filling author only.
        self.author = _read_field('author')
        self.maintainer = None
        self.author_email = _read_field('author-email')
        self.maintainer_email = None
        self.url = _read_field('home-page')
        self.license = _read_field('license')
    #*...(R E D A C T E D)...*#
        # ###################################
        # Note the usage of _read_list() here
        # ###################################
        self.platforms = _read_list('platform')
        self.classifiers = _read_list('classifier')
    #*...(R E D A C T E D)...*#

& Here's the whole thing:

class DistributionMetadata:
        """Dummy class to hold the distribution meta-data: name, version,
        author, and so forth.
        """

        _METHOD_BASENAMES = ("name", "version", "author", "author_email",
                     "maintainer", "maintainer_email", "url",
                     "license", "description", "long_description",
                     "keywords", "platforms", "fullname", "contact",
                     "contact_email", "classifiers", "download_url",
                     # PEP 314
                     "provides", "requires", "obsoletes",
                     )

    def __init__(self, path=None):
        if path is not None:
            self.read_pkg_file(open(path))
        else:
            self.name = None
            self.version = None
            self.author = None
            self.author_email = None
            self.maintainer = None
            self.maintainer_email = None
            self.url = None
            self.license = None
            self.description = None
            self.long_description = None
            self.keywords = None
            self.platforms = None
            self.classifiers = None
            self.download_url = None
            # PEP 314
            self.provides = None
            self.requires = None
            self.obsoletes = None

    def read_pkg_file(self, file):
        """Reads the metadata values from a file object."""
        msg = message_from_file(file)

        def _read_field(name):
            value = msg[name]
            if value == 'UNKNOWN':
                return None
            return value

        def _read_list(name):
            values = msg.get_all(name, None)
            if values == []:
                return None
            return values

        metadata_version = msg['metadata-version']
        self.name = _read_field('name')
        self.version = _read_field('version')
        self.description = _read_field('summary')
        # we are filling author only.
        self.author = _read_field('author')
        self.maintainer = None
        self.author_email = _read_field('author-email')
        self.maintainer_email = None
        self.url = _read_field('home-page')
        self.license = _read_field('license')

        if 'download-url' in msg:
            self.download_url = _read_field('download-url')
        else:
            self.download_url = None

        self.long_description = _read_field('description')
        self.description = _read_field('summary')

        if 'keywords' in msg:
            self.keywords = _read_field('keywords').split(',')

        self.platforms = _read_list('platform')
        self.classifiers = _read_list('classifier')

        # PEP 314 - these fields only exist in 1.1
        if metadata_version == '1.1':
            self.requires = _read_list('requires')
            self.provides = _read_list('provides')
            self.obsoletes = _read_list('obsoletes')
        else:
            self.requires = None
            self.provides = None
            self.obsoletes = None
Rob Truxal
  • 5,856
  • 4
  • 22
  • 39
1

Consider using flit to build the package, as this build system supports multiple authors and maintainers. Store this metadata in pyproject.toml as follows:

[build-system]
requires = ["flit_core >=3.2,<4"]
build-backend = "flit_core.buildapi"

[project]
...
authors = [
    {name = "First1 Last1", email = "name1@foo.bar"},
    {name = "First2 Last2", email = "name2@foo.bar"},
]
maintainers = [
    {name = "First1 Last1", email = "name1@foo.bar"},
    {name = "First2 Last2", email = "name2@foo.bar"},
]
Mike T
  • 41,085
  • 18
  • 152
  • 203