def show(id):
try:
paper = Paper.objects(id=id).get()
abstract = paper.abstract
tokens = ''.join(c for c in abstract if c.isalnum() or c.isspace()).split()
tokens.extend([p.plural(token) for token in tokens])
two_word = [ " ".join(tokens[i:i+2]) for i in range(len(tokens)-1)]
two_word_plural = [ p.plural(w) for w in two_word ]
tokens.extend(two_word)
tokens.extend(two_word_plural)
tokens.extend([t[0].upper()+t[1:] if t[0].islower() else t[0].lower()+t[1:] for t in tokens])
terms = [term.serialize() for term in Term.objects(name__in=tokens)]
abstract = [abstract]
for term in terms:
# print('term {}; len {}'.format(term['name'], len(abstract)))
name = term['name']
for i in range(len(abstract)):
part = abstract[i]
if isinstance(part, str):
re1 = re.compile(name, re.I)
# print("search {} in {}".format(name, part))
m = re.search(re1, part)
if m:
term_here = term.copy()
term_here['here'] = m.group()
part = re.split(re1, part)
new_part = [term_here] * (len(part) * 2 - 1)
new_part[0::2] = part
part = new_part
abstract = abstract[:i] + part + abstract[i+1:]
continue
plural = p.plural(name)
re2 = re.compile(plural, re.I)
# print("search {} in {}".format(name, part))
m = re.search(re2, part)
if m:
term_here = term.copy()
term_here['here'] = m.group()
part = re.split(re2, part)
new_part = [term_here] * (len(part) * 2 - 1)
new_part[0::2] = part
part = new_part
abstract = abstract[:i] + part + abstract[i+1:]
return jsonify(response=abstract)
except Exception as e:
logging.warning(e)
return jsonify(response=list(), error=True)
def recommend_from_user(user):
bag = defaultdict(int)
for subscription in user.subscriptions:
all_tokens = []
papers = subscription.papers
num_paper = len(papers)
# voc = Vocabulary()
for paper in papers:
try:
abstract = paper.abstract
words = list(set([x.strip().strip(",.").lower() for x in abstract.split()]))
for word in words:
if word not in common_words:
bag[word] += 1
except Exception as e:
logging.warning(e)
for paper in papers:
# Map to terms
abstract = paper.abstract
tokens = ''.join(c for c in abstract if c.isalnum() or c.isspace()).split()
tokens.extend([p.plural(token) for token in tokens])
# two_word = [ " ".join(tokens[i:i+2]) for i in range(len(tokens)-1)]
# two_word_plural = [ p.plural(w) for w in two_word ]
# tokens.extend(two_word)
# tokens.extend(two_word_plural)
tokens.extend([t[0].upper()+t[1:] if t[0].islower() else t[0].lower()+t[1:] for t in tokens])
all_tokens.extend(tokens)
# print(Term.objects(name__in=all_tokens).count())
all_tokens = set(all_tokens)
try:
all_tokens.remove('toes')
all_tokens.remove('Toes')
except:
pass
terms = [term.name for term in Term.objects(name__in=all_tokens)]
for term in terms:
bag[term] += num_paper
for history in SearchHistory.objects(user=user.id):
keyword = history.item.keyword
bag[keyword] += 100
return [(x, bag.__getitem__(x)) for x in bag]
# return [(x, bag.__getitem__(x)) for x in sorted(bag, key=bag.__getitem__, reverse=True)]
def parse():
print("Start to import GO terms")
count = 0
name, definition, oid, namespace, tree_number_list, synonyms = None, None, None, None, [], []
with open("resources/go.obo", 'r') as f:
for line in f:
if line == "[Term]\n" or line == "\n":
if oid and oid.find("GO") == 0:
term = Term(name=name, definition=definition, oid=oid, namespace=namespace, tree_number_list=tree_number_list, synonyms= synonyms, source="GO")
term.save()
count += 1
if count % 100 == 0:
print(".", end='',flush=True)
name, definition, oid, namespace, tree_number_list, synonyms = None, None, None, None, [], []
else:
if line.find('name:') == 0:
name = line.strip()[6:]
if line.find('id:') == 0:
oid = line.strip()[4:]
elif line.find('def:') == 0:
definition = line.split('"')[1]
elif line.find('namespace:') == 0:
namespace = line.strip()[11:]
elif line.find('is_a:') == 0:
tree_number_list.append(line.strip()[6:16])
elif line.find('synonym:') == 0:
synonyms.append(line.split('"')[1])
print("\nFinished importing GO terms")
print("Start to fetch ancestor objects")
count = 0
for term in Term.objects(source="GO"):
tree_number_list = term.tree_number_list
for number in tree_number_list:
try:
object = Term.objects(oid=number).get()
term.update(push__ancestors=object)
except Exception as e:
logging.warning(e)
logging.warning(number)
count += 1
if count % 100 == 0:
print(".", end='',flush=True)