def create_from_subjects_acceptable(custom_provider, add_missing=False, missing=None):
tries = 0
subjects_to_copy = list(rules_to_subjects(custom_provider.subjects_acceptable))
if missing and add_missing:
subjects_to_copy = subjects_to_copy + missing
while len(subjects_to_copy):
previous_len = len(subjects_to_copy)
tries += 1
if tries == 10:
raise RuntimeError('Unable to map subjects acceptable with 10 iterations -- subjects remaining: {}'.format(subjects_to_copy))
for subj in list(subjects_to_copy):
if map_custom_subject(custom_provider, subj.text, subj.parent.text if subj.parent else None, subj.text):
subjects_to_copy.remove(subj)
elif add_missing and subj.parent and subj.parent not in subjects_to_copy:
# Dirty
subjects_to_copy.append(subj.parent)
previous_len += 1
else:
logger.warn('Failed. Retrying next iteration')
new_len = len(subjects_to_copy)
if new_len == previous_len:
raise RuntimeError('Unable to map any custom subjects on iteration -- subjects remaining: {}'.format(subjects_to_copy))
def test_rules_to_subjects(self):
rules = [[[self.parent_one._id, self.child_one_1._id], False]]
subject_queryset_ideal = Subject.objects.filter(
Q(id=self.parent_one.id) | Q(id=self.child_one_1.id))
returned_subjects = rules_to_subjects(rules)
self.assertItemsEqual(subject_queryset_ideal, returned_subjects)
def create_from_subjects_acceptable(custom_provider, add_missing=False, missing=None):
tries = 0
subjects_to_copy = list(rules_to_subjects(custom_provider.subjects_acceptable))
if missing and add_missing:
subjects_to_copy = subjects_to_copy + missing
while len(subjects_to_copy):
previous_len = len(subjects_to_copy)
tries += 1
if tries == 10:
raise RuntimeError('Unable to map subjects acceptable with 10 iterations -- subjects remaining: {}'.format(subjects_to_copy))
for subj in list(subjects_to_copy):
if map_custom_subject(custom_provider, subj.text, subj.parent.text if subj.parent else None, subj.text):
subjects_to_copy.remove(subj)
elif add_missing and subj.parent and subj.parent not in subjects_to_copy:
# Dirty
subjects_to_copy.append(subj.parent)
previous_len += 1
else:
logger.warn('Failed. Retrying next iteration')
new_len = len(subjects_to_copy)
if new_len == previous_len:
raise RuntimeError('Unable to map any custom subjects on iteration -- subjects remaining: {}'.format(subjects_to_copy))
def test_rules_to_subjects(self):
rules = [
[[self.parent_one._id, self.child_one_1._id], False]
]
subject_queryset_ideal = Subject.objects.filter(Q(id=self.parent_one.id) | Q(id=self.child_one_1.id))
returned_subjects = rules_to_subjects(rules)
self.assertItemsEqual(subject_queryset_ideal, returned_subjects)
def validate_input(custom_provider, data, provider_type='osf.preprintprovider', copy=False, add_missing=False):
# This function may be run outside of this command (e.g. in the admin app) so we
# need to make sure that BEPRESS_PROVIDER is set
global BEPRESS_PROVIDER
BEPRESS_PROVIDER = AbstractProvider.objects.filter(_id='osf', type='osf.preprintprovider').first()
logger.info('Validating data')
includes = data.get('include', [])
excludes = data.get('exclude', [])
customs = data.get('custom', {})
merges = data.get('merge', {})
if copy:
included_subjects = rules_to_subjects(custom_provider.subjects_acceptable)
else:
assert not set(includes) & set(excludes), 'There must be no overlap between includes and excludes'
for text in includes:
assert Subject.objects.filter(provider=BEPRESS_PROVIDER, text=text).exists(), 'Unable to find included subject with text {}'.format(text)
included_subjects = Subject.objects.filter(provider=BEPRESS_PROVIDER, text__in=includes).include_children()
logger.info('Successfully validated `include`')
for text in excludes:
try:
Subject.objects.get(provider=BEPRESS_PROVIDER, text=text)
except Subject.DoesNotExist:
raise RuntimeError('Unable to find excluded subject with text {}'.format(text))
assert included_subjects.filter(text=text).exists(), 'Excluded subject with text {} was not included'.format(text)
included_subjects = included_subjects.exclude(text__in=excludes)
logger.info('Successfully validated `exclude`')
for cust_name, map_dict in customs.items():
assert not included_subjects.filter(text=cust_name).exists(), 'Custom text {} already exists in mapped set'.format(cust_name)
assert Subject.objects.filter(provider=BEPRESS_PROVIDER, text=map_dict.get('bepress')).exists(), 'Unable to find specified BePress subject with text {}'.format(map_dict.get('bepress'))
if map_dict.get('parent'): # Null parent possible
assert map_dict['parent'] in set(customs.keys()) | set(included_subjects.values_list('text', flat=True)), 'Unable to find specified parent with text {} in mapped set'.format(map_dict['parent'])
# TODO: hierarchy length validation? Probably more trouble than worth here, done on .save
logger.info('Successfully validated `custom`')
included_subjects = included_subjects | Subject.objects.filter(text__in=[map_dict['bepress'] for map_dict in customs.values()])
for merged_from, merged_into in merges.items():
assert not included_subjects.filter(text=merged_from).exists(), 'Cannot merge subject "{}" that will be included'.format(merged_from)
assert merged_into in set(included_subjects.values_list('text', flat=True)) | set(customs.keys()), 'Unable to determine merge target for "{}"'.format(merged_into)
included_subjects = included_subjects | Subject.objects.filter(text__in=merges.keys())
missing_subjects = Subject.objects.filter(id__in=set([hier[-1].id for ps in Preprint.objects.filter(provider=custom_provider) for hier in ps.subject_hierarchy])).exclude(id__in=included_subjects.values_list('id', flat=True))
if not add_missing:
assert not missing_subjects.exists(), 'Incomplete mapping -- following subjects in use but not included:\n{}'.format(list(missing_subjects.values_list('text', flat=True)))
if isinstance(custom_provider, PreprintProvider):
assert custom_provider.share_title not in [None, '', 'bepress'], 'share title not set; please set the share title on this provider before creating a custom taxonomy.'
logger.info('Successfully validated mapping completeness')
return list(missing_subjects) if add_missing else None
def validate_input(custom_provider, data, provider_type='osf.preprintprovider', copy=False, add_missing=False):
# This function may be run outside of this command (e.g. in the admin app) so we
# need to make sure that BEPRESS_PROVIDER is set
global BEPRESS_PROVIDER
BEPRESS_PROVIDER = AbstractProvider.objects.filter(_id='osf', type='osf.preprintprovider').first()
logger.info('Validating data')
includes = data.get('include', [])
excludes = data.get('exclude', [])
customs = data.get('custom', {})
merges = data.get('merge', {})
if copy:
included_subjects = rules_to_subjects(custom_provider.subjects_acceptable)
else:
assert not set(includes) & set(excludes), 'There must be no overlap between includes and excludes'
for text in includes:
assert Subject.objects.filter(provider=BEPRESS_PROVIDER, text=text).exists(), 'Unable to find included subject with text {}'.format(text)
included_subjects = Subject.objects.filter(provider=BEPRESS_PROVIDER, text__in=includes).include_children()
logger.info('Successfully validated `include`')
for text in excludes:
try:
Subject.objects.get(provider=BEPRESS_PROVIDER, text=text)
except Subject.DoesNotExist:
raise RuntimeError('Unable to find excluded subject with text {}'.format(text))
assert included_subjects.filter(text=text).exists(), 'Excluded subject with text {} was not included'.format(text)
included_subjects = included_subjects.exclude(text__in=excludes)
logger.info('Successfully validated `exclude`')
for cust_name, map_dict in customs.iteritems():
assert not included_subjects.filter(text=cust_name).exists(), 'Custom text {} already exists in mapped set'.format(cust_name)
assert Subject.objects.filter(provider=BEPRESS_PROVIDER, text=map_dict.get('bepress')).exists(), 'Unable to find specified BePress subject with text {}'.format(map_dict.get('bepress'))
if map_dict.get('parent'): # Null parent possible
assert map_dict['parent'] in set(customs.keys()) | set(included_subjects.values_list('text', flat=True)), 'Unable to find specified parent with text {} in mapped set'.format(map_dict['parent'])
# TODO: hierarchy length validation? Probably more trouble than worth here, done on .save
logger.info('Successfully validated `custom`')
included_subjects = included_subjects | Subject.objects.filter(text__in=[map_dict['bepress'] for map_dict in customs.values()])
for merged_from, merged_into in merges.iteritems():
assert not included_subjects.filter(text=merged_from).exists(), 'Cannot merge subject "{}" that will be included'.format(merged_from)
assert merged_into in set(included_subjects.values_list('text', flat=True)) | set(customs.keys()), 'Unable to determine merge target for "{}"'.format(merged_into)
included_subjects = included_subjects | Subject.objects.filter(text__in=merges.keys())
missing_subjects = Subject.objects.filter(id__in=set([hier[-1].id for ps in PreprintService.objects.filter(provider=custom_provider) for hier in ps.subject_hierarchy])).exclude(id__in=included_subjects.values_list('id', flat=True))
if not add_missing:
assert not missing_subjects.exists(), 'Incomplete mapping -- following subjects in use but not included:\n{}'.format(list(missing_subjects.values_list('text', flat=True)))
if isinstance(custom_provider, PreprintProvider):
assert custom_provider.share_title not in [None, '', 'bepress'], 'share title not set; please set the share title on this provider before creating a custom taxonomy.'
logger.info('Successfully validated mapping completeness')
return list(missing_subjects) if add_missing else None
def get(self, request, *args, **kwargs):
rules = json.loads(request.GET['rules'])
all_subjects = rules_to_subjects(rules)
return JsonResponse({'subjects': [sub.id for sub in all_subjects]})
