providers: provide DataCite-like DOI locally

- Generate a random, configurable length,
  base32, URI-friendly, hyphen-separated,
  optionally checksummed DOI

invenio_pidstore/__init__.py

@@ -210,20 +210,29 @@
 
 Providers
 ---------
-Providers adds extra functionality persistent identifiers. Use cases for this
-includes automatically creating the persistent identifier or retrieving the
-persistent identifier from an external service.
+Providers wrap the creation of persistent identifiers with extra functionality.
+Use cases for this include automatically creating the persistent identifier or
+retrieving the persistent identifier from an external service.
+
+PIDStore comes by default with two providers:
+:py:class:`invenio_pidstore.providers.recordid.RecordIdProvider` which
+creates Invenio legacy integer record identifiers and
+:py:class:`invenio_pidstore.providers.datacite.DataCiteProvider` which
+creates checksummed random alphanumeric 10-character strings:
 
-PIDStore comes by default with a
-:py:class:`invenio_pidstore.providers.recordid.RecordIdProvider` which will
-create Invenio legacy integer record identifiers:
 
 >>> from invenio_pidstore.providers.recordid import RecordIdProvider
 >>> provider = RecordIdProvider.create()
 >>> provider.pid.pid_type
 'recid'
 >>> provider.pid.pid_value
 '1'
+>>> from invenio_pidstore.providers.datacite import DataCiteProvider
+>>> provider = DataCiteProvider.create()
+>>> provider.pid.pid_type
+'doi'
+>>> provider.pid.pid_value
+'3sbk2-5j060'
 
 Creating your own provider
 ~~~~~~~~~~~~~~~~~~~~~~~~~~

invenio_pidstore/config.py

@@ -14,3 +14,10 @@
 This name will be used by the fetcher, to retrieve the record ID value from the
 JSON, and by the minter, to store it inside the JSON.
 """
+
+
+PIDSTORE_DATACITE_DOI_OPTIONS = {
+    'suffix_length': 10,
+    'split_every': 5,
+    'checksum': True
+}

invenio_pidstore/providers/datacite.py

@@ -1,7 +1,8 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of Invenio.
-# Copyright (C) 2015-2018 CERN.
+# Copyright (C) 2015-2019 CERN.
+# Copyright (C) 2019 Northwestern University.
 #
 # Invenio is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -10,6 +11,11 @@
 
 from __future__ import absolute_import
 
+import os
+import random
+import re
+
+from base32_lib import base32
 from datacite import DataCiteMDSClient
 from datacite.errors import DataCiteError, DataCiteGoneError, \
     DataCiteNoContentError, DataCiteNotFoundError, HttpError
@@ -31,20 +37,65 @@ class DataCiteProvider(BaseProvider):
     default_status = PIDStatus.NEW
     """Default status for newly created PIDs by this provider."""
 
+    doi_prefix_regexp = re.compile(
+        r"10\.\d+(\.\d+)*$"
+    )
+
     @classmethod
-    def create(cls, pid_value, **kwargs):
+    def valid_doi_prefix(cls, prefix):
+        """Matches if prefix is a DOI prefix.
+
+        NOTE: This was done here to prevent relying on idutils.
+        """
+        return cls.doi_prefix_regexp.match(prefix)
+
+    @classmethod
+    def generate_doi(cls, options=None):
+        """Generate DOI with random suffix."""
+        passed_options = options or {}
+        # WHY: A new dict needs to be created to prevent side-effects
+        options = dict(current_app.config.get(
+            'PIDSTORE_DATACITE_DOI_OPTIONS', {}
+        ))
+        options.update(passed_options)
+        prefix = options.get(
+            'prefix',
+            current_app.config.get('PIDSTORE_DATACITE_DOI_PREFIX')
+        )
+        suffix_length = options.get('suffix_length')
+        split_every = options.get('split_every')
+        checksum = options.get('checksum')
+
+        if not cls.valid_doi_prefix(prefix):
+            raise ValueError("Invalid DOI prefix: {}".format(prefix))
+
+        suffix = base32.generate(
+            length=suffix_length,
+            split_every=split_every,
+            checksum=checksum
+        )
+
+        return prefix + "/" + suffix
+
+    @classmethod
+    def create(cls, pid_value=None, doi_options=None, **kwargs):
         """Create a new record identifier.
 
         For more information about parameters,
         see :meth:`invenio_pidstore.providers.BaseProvider.create`.
 
         :param pid_value: Persistent identifier value.
+        :param doi_options: ``dict`` with optional keys:
+            ``"prefix", "suffix_length", "checksum", "checksum"``
         :params **kwargs: See
             :meth:`invenio_pidstore.providers.base.BaseProvider.create` extra
             parameters.
         :returns: A :class:`invenio_pidstore.providers.DataCiteProvider`
             instance.
         """
+        if pid_value is None:
+            pid_value = cls.generate_doi(doi_options)
+
         return super(DataCiteProvider, cls).create(
             pid_value=pid_value, **kwargs)
 

run-tests.sh

@@ -1,7 +1,9 @@
+#!/usr/bin/env sh
 # -*- coding: utf-8 -*-
 #
 # This file is part of Invenio.
-# Copyright (C) 2015-2018 CERN.
+# Copyright (C) 2015-2019 CERN.
+# Copyright (C) 2019 Northwestern University.
 #
 # Invenio is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.

setup.py

@@ -1,7 +1,8 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of Invenio.
-# Copyright (C) 2015-2018 CERN.
+# Copyright (C) 2015-2019 CERN.
+# Copyright (C) 2019 Northwestern University.
 #
 # Invenio is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -25,7 +26,7 @@
     'Flask-Menu>=0.5.1',
     'invenio-access>=1.0.0',
     'invenio-accounts>=1.0.0',
-    'mock>=1.3.0',
+    'mock>=3.0.0',
     'pydocstyle>=1.0.0',
     'pytest-cov>=1.8.0',
     'pytest-pep8>=1.0.6',
@@ -70,6 +71,8 @@
 install_requires = [
     'Flask-BabelEx>=0.9.3',
     'Flask>=0.11.1',
+    'six>=1.12.0',
+    'base32-lib>=1.0.0a1'
 ]
 
 packages = find_packages()

tests/test_providers.py

@@ -1,7 +1,8 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of Invenio.
-# Copyright (C) 2015-2018 CERN.
+# Copyright (C) 2015-2019 CERN.
+# Copyright (C) 2019 Northwestern University.
 #
 # Invenio is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -241,3 +242,71 @@ def test_datacite_sync(logger, app, db):
         assert provider.pid.status == PIDStatus.NEW
         assert logger.exception.call_args[0][0] == \
             "Failed to sync status from DataCite"
+
+
+def test_datacite_valid_doi_prefix():
+    assert DataCiteProvider.valid_doi_prefix('10.1234')
+    assert DataCiteProvider.valid_doi_prefix('10.12.34')
+    assert not DataCiteProvider.valid_doi_prefix('101234')
+    assert not DataCiteProvider.valid_doi_prefix('101234.')
+    assert not DataCiteProvider.valid_doi_prefix('10.1234/')
+    assert not DataCiteProvider.valid_doi_prefix('100.1234')
+    assert not DataCiteProvider.valid_doi_prefix('10.12E45')
+
+
+@patch('invenio_pidstore.providers.datacite.base32')
+def test_datacite_generate_doi_calls_base32_generate(patched_base32, app):
+    original_prefix = app.config.get('PIDSTORE_DATACITE_DOI_PREFIX')
+    app.config['PIDSTORE_DATACITE_DOI_PREFIX'] = '10.4321'
+    original_doi_options = app.config.get('PIDSTORE_DATACITE_DOI_OPTIONS')
+    app.config['PIDSTORE_DATACITE_DOI_OPTIONS'] = {
+        'suffix_length': 8,
+        'split_every': 4,
+        'checksum': False
+    }
+
+    DataCiteProvider.generate_doi()
+
+    patched_base32.generate.assert_called_with(
+        length=8,
+        split_every=4,
+        checksum=False
+    )
+
+    app.config['PIDSTORE_DATACITE_DOI_PREFIX'] = original_prefix
+    app.config['PIDSTORE_DATACITE_DOI_OPTIONS'] = original_doi_options
+
+
+def test_datacite_generate_doi_prefix(app):
+    original_doi_options = app.config.get('PIDSTORE_DATACITE_DOI_OPTIONS')
+    app.config['PIDSTORE_DATACITE_DOI_OPTIONS'] = {
+        'suffix_length': 8,
+        'split_every': 4,
+        'checksum': False
+    }
+    # explicit prefix
+    doi = DataCiteProvider.generate_doi(options={'prefix': '10.1234'})
+
+    assert doi.startswith('10.1234/')
+
+    # configuration prefix
+    original_prefix = app.config.get('PIDSTORE_DATACITE_DOI_PREFIX')
+    app.config['PIDSTORE_DATACITE_DOI_PREFIX'] = '10.4321'
+
+    doi = DataCiteProvider.generate_doi()
+
+    assert doi.startswith('10.4321/')
+
+    app.config['PIDSTORE_DATACITE_DOI_OPTIONS'] = original_doi_options
+    app.config['PIDSTORE_DATACITE_DOI_PREFIX'] = original_prefix
+
+
+def test_datacite_provider_create_calls_generate(app, db):
+    with app.app_context():
+        provider = DataCiteProvider.create(doi_options={'prefix': '10.1234'})
+        assert provider.pid.status == PIDStatus.NEW
+        assert provider.pid.pid_provider == 'datacite'
+        assert provider.pid.pid_value.startswith('10.1234')
+
+        with pytest.raises(ValueError):
+            DataCiteProvider.create(doi_options={'prefix': '10.123fd4'})