def all_subjects(self):
q = []
for rule in self.subjects_acceptable:
if rule[1]:
q.append(Q('parents', 'eq', Subject.load(rule[0][-1])))
if len(rule[0]) == 1:
potential_parents = Subject.find(Q('parents', 'eq', Subject.load(rule[0][-1])))
for parent in potential_parents:
q.append(Q('parents', 'eq', parent))
for sub in rule[0]:
q.append(Q('_id', 'eq', sub))
return Subject.find(reduce(lambda x, y: x | y, q)) if len(q) > 1 else (Subject.find(q[0]) if len(q) else Subject.find())
def main():
dry_run = '--dry' in sys.argv
if not dry_run:
script_utils.add_file_logger(logger, __file__)
set_up_storage([Subject], storage.MongoStorage)
with TokuTransaction():
update_taxonomies('bepress_taxonomy.json')
# Now that all subjects have been added to the db, compute and set
# the 'children' field for every subject
logger.info('Setting "children" field for each Subject')
for subject in Subject.find():
subject.children = Subject.find(Q('parents', 'eq', subject))
subject.save()
if dry_run:
raise RuntimeError('Dry run, transaction rolled back')
def _create(cls, target_class, text=None, parents=[], *args, **kwargs):
try:
subject = Subject.find_one(Q('text', 'eq', text))
except NoResultsFound:
subject = target_class(*args, **kwargs)
subject.text = text
subject.parents = parents
subject.save()
return subject
def _create(cls, target_class, text=None, parents=[], *args, **kwargs):
try:
subject = Subject.find_one(Q('text', 'eq', text))
except NoResultsFound:
subject = target_class(*args, **kwargs)
subject.text = text
subject.parents = parents
subject.save()
return subject
def get_subjects(self):
ret = []
for subj_list in self.subjects:
subj_hierarchy = []
for subj_id in subj_list:
subj = Subject.load(subj_id)
if subj:
subj_hierarchy += ({'id': subj_id, 'text': subj.text}, )
if subj_hierarchy:
ret.append(subj_hierarchy)
return ret
def setUp(self):
super(TestTaxonomy, self).setUp()
# Subject 1 has 3 children
self.subject1 = SubjectFactory()
self.subject1_child1 = SubjectFactory(parent=self.subject1)
self.subject1_child2 = SubjectFactory(parent=self.subject1)
self.subjects = Subject.find()
self.url = '/{}taxonomies/'.format(API_BASE)
self.res = self.app.get(self.url)
self.data = self.res.json['data']
def test_taxonomy_filter_top_level(self):
top_level_subjects = Subject.find(Q('parents', 'eq', []))
top_level_url = self.url + '?filter[parents]=null'
res = self.app.get(top_level_url)
assert_equal(res.status_code, 200)
data = res.json['data']
assert_equal(len(top_level_subjects), len(data))
assert len(top_level_subjects) > 0
for subject in data:
assert_equal(subject['attributes']['parents'], [])
def test_taxonomy_filter_top_level(self):
top_level_subjects = Subject.find(
Q('parents', 'eq', [])
)
top_level_url = self.url + '?filter[parents]=null'
res = self.app.get(top_level_url)
assert_equal(res.status_code, 200)
data = res.json['data']
assert_equal(len(top_level_subjects), len(data))
assert len(top_level_subjects) > 0
for subject in data:
assert_equal(subject['attributes']['parents'], [])
def setUp(self):
super(TestTaxonomy, self).setUp()
# Subject 1 has 3 children
self.subject1 = SubjectFactory()
self.subject1_child1 = SubjectFactory(parents=[self.subject1])
self.subject1_child2 = SubjectFactory(parents=[self.subject1])
# Subject 2 has a child whose parent is both subject 1 and subject 2
self.subject2 = SubjectFactory()
self.subject2_child1_subject1_child3 = SubjectFactory(parents=[self.subject1, self.subject2])
self.subjects = Subject.find()
self.url = '/{}taxonomies/'.format(API_BASE)
self.res = self.app.get(self.url)
self.data = self.res.json['data']
def test_taxonomy_filter_by_parent(self):
children_subjects = Subject.find(
Q('parents', 'eq', self.subject1._id)
)
children_url = self.url + '?filter[parents]={}'.format(self.subject1._id)
res = self.app.get(children_url)
assert_equal(res.status_code, 200)
data = res.json['data']
assert_equal(len(children_subjects), len(data))
for subject in data:
parents_ids = []
for parent in subject['attributes']['parents']:
parents_ids.append(parent['id'])
assert_in(self.subject1._id, parents_ids)
def update_taxonomies(filename):
# Flat taxonomy is stored locally, read in here
with open(
os.path.join(
settings.APP_PATH,
'website', 'static', filename
)
) as fp:
taxonomy = json.load(fp)
for subject_path in taxonomy.get('data'):
subjects = subject_path.split('_')
text = subjects[-1]
# Search for parent subject, get id if it exists
parent = None
if len(subjects) > 1:
try:
parent = Subject.find_one(Q('text', 'eq', subjects[-2]))
except NoResultsFound:
pass
try:
subject = Subject.find_one(Q('text', 'eq', text))
logger.info('Found existing Subject "{}":{}{}'.format(
subject.text,
subject._id,
u' with parent {}:{}'.format(parent.text, parent._id) if parent else ''
))
except (NoResultsFound):
# If subject does not yet exist, create it
subject = Subject(
text=text,
parents=[parent] if parent else [],
)
logger.info(u'Creating Subject "{}":{}{}'.format(
subject.text,
subject._id,
u' with parent {}:{}'.format(parent.text, parent._id) if parent else ''
))
else:
# If subject does exist, append parent_id if not already added
if parent and parent not in subject.parents:
subject.parents.append(parent)
logger.info(u'Adding parent "{}":{} to Subject "{}":{}'.format(
parent.text, parent._id,
subject.text, subject._id
))
subject.save()
def all_subjects(self):
q = []
for rule in self.subjects_acceptable:
if rule[1]:
q.append(Q('parents', 'eq', Subject.load(rule[0][-1])))
if len(rule[0]) == 1:
potential_parents = Subject.find(
Q('parents', 'eq', Subject.load(rule[0][-1])))
for parent in potential_parents:
q.append(Q('parents', 'eq', parent))
for sub in rule[0]:
q.append(Q('_id', 'eq', sub))
return Subject.find(reduce(lambda x, y: x | y, q)) if len(q) > 1 else (
Subject.find(q[0]) if len(q) else Subject.find())
def top_level_subjects(self):
if len(self.subjects_acceptable) == 0:
return Subject.find(Q('parents', 'eq', []))
tops = set([sub[0][0] for sub in self.subjects_acceptable])
return [Subject.load(sub) for sub in tops]
def get_queryset(self):
return Subject.find(self.get_query_from_request())
def get_queryset(self):
return Subject.find(self.get_query_from_request())
def tearDown(self):
super(TestTaxonomy, self).tearDown()
Subject.remove()
def top_level_subjects(self):
if len(self.subjects_acceptable) == 0:
return Subject.find(Q('parents', 'eq', []))
tops = set([sub[0][0] for sub in self.subjects_acceptable])
return [Subject.load(sub) for sub in tops]
