def find_child_and_grandchild(grandpa, childIndex=0):
parent = Subject.find(Q('parents', 'eq', grandpa))[childIndex]
try:
child = Subject.find(Q('parents', 'eq', parent))[0]
except IndexError:
return find_child_and_grandchild(grandpa, childIndex=childIndex + 1)
return parent, child
def find_child_and_grandchild(grandpa, childIndex=0):
parent = Subject.find(Q('parents', 'eq', grandpa))[childIndex]
try:
child = Subject.find(Q('parents', 'eq', parent))[0]
except IndexError:
return find_child_and_grandchild(grandpa, childIndex=childIndex+1)
return parent, child
def create_subject_rules():
top_levels = Subject.find(Q('parents', 'eq', []))
subA = top_levels[0]
subB = top_levels[1]
subC = top_levels[2]
children_of_A = Subject.find(Q('parents', 'eq', subA))
subD = children_of_A[0]
subE = children_of_A[1]
subF, subG = find_child_and_grandchild(subB)
rules = [([subA._id, subD._id], False), ([subA._id, subE._id], True),
([subB._id, subF._id, subG._id], True), ([subC._id], True)]
return rules
def to_representation(self, subject):
if not isinstance(subject, Subject):
subject = Subject.load(subject)
if subject is not None:
return {'id': subject._id,
'text': subject.text}
return None
def get_queryset(self):
parent = self.request.query_params.get(
'filter[parents]', None) or self.request.query_params.get(
'filter[parent]', None)
provider = get_object_or_error(PreprintProvider,
self.kwargs['provider_id'],
display_name='PreprintProvider')
if parent:
if parent == 'null':
return provider.top_level_subjects
if provider.subjects.exists():
return provider.subjects.filter(parent___id=parent)
else:
# TODO: Delet this when all PreprintProviders have a mapping
# Calculate this here to only have to do it once.
allowed_parents = [
id_ for sublist in provider.subjects_acceptable
for id_ in sublist[0]
]
allows_children = [
subs[0][-1] for subs in provider.subjects_acceptable
if subs[1]
]
return [
sub
for sub in Subject.find(MQ('parent___id', 'eq', parent))
if provider.subjects_acceptable == []
or self.is_valid_subject(allows_children=allows_children,
allowed_parents=allowed_parents,
sub=sub)
]
return provider.all_subjects
def create_subject_rules():
top_levels = Subject.find(Q('parents', 'eq', []))
subA = top_levels[0]
subB = top_levels[1]
subC = top_levels[2]
children_of_A = Subject.find(Q('parents', 'eq', subA))
subD = children_of_A[0]
subE = children_of_A[1]
subF, subG = find_child_and_grandchild(subB)
rules = [
([subA._id, subD._id], False),
([subA._id, subE._id], True),
([subB._id, subF._id, subG._id], True),
([subC._id], True)
]
return rules
def get_queryset(self):
parent = self.request.query_params.get('filter[parents]', None)
provider = PreprintProvider.load(self.kwargs['provider_id'])
if parent:
if parent == 'null':
return provider.top_level_subjects
# Calculate this here to only have to do it once.
allowed_parents = [id_ for sublist in provider.subjects_acceptable for id_ in sublist[0]]
allows_children = [subs[0][-1] for subs in provider.subjects_acceptable if subs[1]]
return [sub for sub in Subject.find(Q('parents', 'eq', parent)) if provider.subjects_acceptable == [] or self.is_valid_subject(allows_children=allows_children, allowed_parents=allowed_parents, sub=sub)]
return provider.all_subjects
def get_subject_id(name):
if not name in SUBJECTS_CACHE:
subject = None
try:
subject = Subject.find_one(Q('text', 'eq', name))
except NoResultsFound:
raise Exception('Subject: "{}" not found'.format(name))
else:
SUBJECTS_CACHE[name] = subject._id
return SUBJECTS_CACHE[name]
def get_subject_id(name):
if not name in SUBJECTS_CACHE:
subject = None
try:
subject = Subject.find_one(Q('text', 'eq', name))
except NoResultsFound:
raise Exception('Subject: "{}" not found'.format(name))
else:
SUBJECTS_CACHE[name] = subject._id
return SUBJECTS_CACHE[name]
def migrate_preprint(preprint):
logger.info('Preparint to migrate {}'.format(preprint['_id']))
new_hiers = []
for hier in preprint['subjects']:
leaf = get_leaf(hier)
new_leaf_name = PLOS_TO_BP_MAP[database['plos_subject'].find_one(
{'_id': leaf})['text']]
if new_leaf_name != '<drop>':
new_hiers.append(
Subject.find_one(Q('text', 'eq', new_leaf_name)).hierarchy)
logger.info('Setting subjects on {} to {}'.format(preprint['_id'],
new_hiers))
database['preprintservice'].find_and_modify(
{'_id': preprint['_id']}, {'$set': {
'subjects': new_hiers
}})
def migrate_preprint(preprint):
logger.info('Preparint to migrate {}'.format(preprint['_id']))
new_hiers = []
for hier in preprint['subjects']:
leaf = get_leaf(hier)
new_leaf_name = PLOS_TO_BP_MAP[database['plos_subject'].find_one({'_id': leaf})['text']]
if new_leaf_name != '<drop>':
new_hiers.append(
Subject.find_one(Q('text', 'eq', new_leaf_name)).hierarchy
)
logger.info('Setting subjects on {} to {}'.format(preprint['_id'], new_hiers))
database['preprintservice'].find_and_modify(
{'_id': preprint['_id']},
{'$set':{
'subjects': new_hiers
}}
)
def load_bepress(f_path):
assert Subject.find().count() == 0
logger.info('Loading BePress...')
with open(f_path) as fp:
bpress = json.load(fp)
validate_map_bepress_correctness(set(bpress.keys()))
logger.info('Populating Subjects...')
for text in bpress.keys():
Subject(text=text).save()
assert Subject.find().count() == len(bpress.keys())
logger.info('Setting parents...')
for s in Subject.find():
if bpress[s.text]['lineage']:
s.parents = [
Subject.find_one(
Q('text', 'eq', bpress[s.text]['lineage'][-1]))
]
s.save()
logger.info('Setting children...')
for s in Subject.find():
s.children = Subject.find(Q('parents', 'eq', s))
s.save()
logger.info('Successfully imported BePress taxonomy.')
def load_bepress(f_path):
assert Subject.find().count() == 0
logger.info('Loading BePress...')
with open(f_path) as fp:
bpress = json.load(fp)
validate_map_bepress_correctness(set(bpress.keys()))
logger.info('Populating Subjects...')
for text in bpress.keys():
Subject(text=text).save()
assert Subject.find().count() == len(bpress.keys())
logger.info('Setting parents...')
for s in Subject.find():
if bpress[s.text]['lineage']:
s.parents = [Subject.find_one(Q('text', 'eq', bpress[s.text]['lineage'][-1]))]
s.save()
logger.info('Setting children...')
for s in Subject.find():
s.children = Subject.find(Q('parents', 'eq', s))
s.save()
logger.info('Successfully imported BePress taxonomy.')
def to_representation(self, subject):
if not isinstance(subject, Subject):
subject = Subject.load(subject)
if subject is not None:
return {'id': subject._id, 'text': subject.text}
return None
